1

Explaining Socio-economic Causes of Urban Unemployment

and Policy Responses in Ethiopia By Tesfaye Chofana and Tegegn Gebeyaw tesfayechofana@yahoo.com and tegegnw@gmail.com

2013 Addis Ababa

2

Acknowledgment

We are very grateful to the Organization for Social Science Research in Eastern and Southern

Africa (OSSREA) for funding the research project and providing training to facilitate the task

and supervising. We would like to express our appreciation to the Central Statistical Agency

(CSA) for providing the secondary data required for the research. We would also extend our

thanks to the respective woreda and kebele offices of Addis Ababa, Bahir Dar and Hawassa

cities for significant supports they provided during primary data collection.

3

Abstract The study explores the socioeconomic causes of urban unemployment and effects of policy

interventions. It made use of primary cross-sectional data collected from three major cities and

secondary data primarily from the CSA of Ethiopia. Mainly a quantitative approach is

followed using both descriptive and inferential methods of analysis. Despite the sound

economic growth and the deliberate effort of the government to address the problem, the urban

labor market is characterized by high and persistent unemployment. Although the rate declined

from 26 percent in 2003 to 18 percent in 2011, it is still a cause for concern. The downward

inflexible unemployment rate may signify that the rapidly growing economy for almost a

decade does not result in equivalent employment opportunity. Rapidly growing urban

population and lack of vibrant non-agricultural sector are among the contributing factors of

urban unemployment while the effect of FDI inflow on unemployment is mixed. Furthermore,

the skill-mismatch and the tendency of queuing for public or formal private sector jobs are

found to be possible causes of unemployment.

The likelihood of unemployment is associated with demographic, location and education

variables. A desirable employment effect of education at individual level is found to be more

pronounced at tertiary level of education. Relative to lower primary education, all other

categories of educational qualifications below tertiary level are associated with higher rate of

unemployment. Training has a relatively desirable effect on the labor market outcomes of some

groups of the labor force; however, it makes no difference in reducing gender and age

disparity of unemployment and in encouraging self-employment. Above all, what seems

paradoxical and that requires immediate measure is TVET is likely to increase unemployment

and to decrease self-employment after eight years of implementation practices. TVET program

is also criticized for being less relevant, less responsive, non participatory, less efficient and

effective, and is less flexible. On the other hand, the employment effect of grade ten graduates

is consistently improving. The employment effect of MSEs is found to be insignificant and only

one third of them registered positive employment growth since startup. Moreover, employment

growth effects of human capital endowments of new firms, social capitals and access to credit

is nil.

4

Indeed, as the recent years experience of the country witnesses, despite the ongoing education

policy reform and MSE development and promotion efforts of the government, further

considerations are critical to achieve the desired results from policy interventions. It is

therefore important to evaluate the existing system of education and training and taking timely

measure to improve its relevance and quality. Particularly, the unsatisfactory performance of

the TVET program reminds the need to reconsider the limitations and take timely measure so

as to link the program with the labor market demand. Another important policy implication of

the finding is the need to provide support to MSEs in terms of market for their products, easy

access to supply of raw materials, and work place.

5

Table of Contents

ACKNOWLEDGMENT .............................................................................................................. 2 ABSTRACT ................................................................................................................................. 3 LIST OF TABLES ........................................................................................................................ 6 LIST OF FIGURES ...................................................................................................................... 7 1. INTRODUCTION .................................................................................................................... 8

1.1. Background of the Study .................................................................................................................................... 8 1.2. Objective of the Study ...................................................................................................................................... 15 1.3. Data Sources and Methodology ........................................................................................................................ 15 1.4. Significance and Scope of the Study ................................................................................................................ 16 1.5. Limitation of the Study ..................................................................................................................................... 16 1.6. Organization of the Paper ................................................................................................................................. 17

2. LITERATURE REVIEW ...................................................................................................... 18

2.1 Definition and Concepts of Unemployment ...................................................................................................... 18 2.2. Type of Unemployment .................................................................................................................................... 21 2.3. Theories of Unemployment .............................................................................................................................. 23 2.4. Causes of Unemployment ................................................................................................................................. 27

2.4.1. Supply Side Factors .................................................................................................................................. 28 2.4.2. Demand Side Factors ................................................................................................................................ 33

2.5. Active Labor Market Policies to Address Unemployment ............................................................................... 38 2.6. An Overview of Empirical Evidences on Unemployment in Ethiopia ............................................................. 41 2.7. Policy Responses to Address Unemployment in Ethiopia ................................................................................ 43

2.7.1. Expansion of Technical and Vocational Education and Training Programs ............................................. 43 2.7.2. Micro and Small Scale Enterprises (MSEs) Development ....................................................................... 45

3. METHODOLOGY ................................................................................................................. 49

This section presents a discussion of the specific steps used in conducting the research. It provides information on research methodology, data sources, sampling techniques, data collection instruments, methods of data analysis and specification of econometric models. ............................................................................................................... 49 3.1. Research Method .............................................................................................................................................. 49 3.2. Data Sources ..................................................................................................................................................... 49 3.3. Sampling Techniques and Procedures .............................................................................................................. 50 3.4. Data Collection Instruments ............................................................................................................................. 51 3.2. Data Analysis ................................................................................................................................................... 52 3.2.1. Pooled Cross-sectional Data Analysis ........................................................................................................... 52

3.2.2 Specification of Study Variables ............................................................................................................... 57

4. RESULTS AND DISCUSSION ......................................................................................... 57

4.1. Demographic Characteristics of Respondents .................................................................................................. 57 4.2. The Urban Labor Force Participation Trends ................................................................................................... 58 4.3. Urban Versus Rural Unemployment ................................................................................................................ 59 4.4. Urban Employment Trends .............................................................................................................................. 60 4.5. Urban Employment-to-Population Ratio .......................................................................................................... 61 4.6. Urban Unemployment Trends .......................................................................................................................... 62

6

4.7. Regional Unemployment Trends ...................................................................................................................... 65 4.8. Urban Unemployment and Education .............................................................................................................. 66 4.9. Unemployment Duration .................................................................................................................................. 69 4.10. Urban Unemployment and Training ............................................................................................................... 71 4.11. Training and Self-employment ....................................................................................................................... 73 4.12. School to Work Transition ............................................................................................................................. 74 4.13. Socioeconomic Causes of Urban Unemployment .......................................................................................... 75 4.14. Theories of Unemployment ............................................................................................................................ 79 4.15. Effect of Education and Training Polices on Labor Market Outcomes .......................................................... 83

4.15.1. Effect of Education Polices on Urban Unemployment ........................................................................... 84 4.15.2. The Effect of Training Polices on Urban Unemployment ...................................................................... 89 4.15.3. Effect of Education and Training Polices on Self-employment and School-to-Work Transition ........... 90

4.16. An Assessment of Strategies to Promote Employment in Ethiopia ................................................................ 91 4.16.1. Strategies to Increase Employment through TVET ................................................................................ 91 4.16.2. Employment Growth within Micro and Small Scale Enterprises ......................................................... 100

4.16.2.1. Characteristics of Micro and Small Scale Enterprises .................................................................. 101 4.16.2.2. Employment Contribution of MSEs .............................................................................................. 103 4.16.2.3. Startup Motives of MSEs .............................................................................................................. 105 4.16.2.4. Constraints of Micro and Small Scale Enterprises ........................................................................ 105 4.16.2.5. Market and Other Constraints to Expand Business ................................................................... 105 4.16.2.6. Source of Startup Capital and Capital Growth .............................................................................. 107 4.16.2.7. Cause of Job Interruption .............................................................................................................. 108 4.16.2.8. Assistance Needed from Government ........................................................................................... 109 4.16.2.9. Determinants Urban Employment Growth within MSEs .............................................................. 110

5. CONCLUSIONS AND RECOMMENDATIONS ........................................................... 112

5.1. Conclusions .................................................................................................................................................... 112 5.2. Recommendation..................................................................................................................................... 120

REFERENCES ......................................................................................................................... 123 ANNEX .................................................................................................................................... 127

List of Tables

Table: 2.1 Number of establishments and jobs created and amount of loan ............. Error! Bookmark not defined.

Table 4.1: Regional unemployment distribution (%) .............................................................................................. 66

Table 4.2: Unemployment rate by education ........................................................................................................... 68

Table 4.3: Distribution of respondents .................................................................................................................... 92

Table 4.4: Evaluation of the innovativeness of the program ................................................................................... 93

7

Table 4.5: Evaluation of the feasibility of the program ........................................................................................... 94

Table 4.5: Evaluation of the TVET program responsiveness ................................................................................. 95

Table 4.6: Evaluation of the relevance of the TVET program................................................................................. 97

Table 4.7: Evaluation of the relevance of the TVET program................................................................................. 98

Table 4.8: Evaluation of the efficiency and effectiveness of the program ............................................................. 98

Table 4.9: Up Scalability of the Program ................................................................................................................ 99

Table 4.10: Coordination of the TVET program ................................................................................................... 100

Table 4.12: causes of job interruption ................................................................................................................... 108

Table 4.13: Assistance needed from government .................................................................................................. 109

List of Figures Figure 2.1: The ILO’s Labor Force Framework ..................................................................................................... 20 Figure 4.1: urban labor force participation rate (%) ............................................................................................... 58 Figure 4.2: The trend of labor supply by years of schooling (%) ........................................................................... 59 Figure 4.3: Urban employment trends (%) .............................................................................................................. 60 Table 4.4: Urban employment-to-population ratio (%) ........................................................................................... 62 Figure 4.5: urban unemployment rate (%) ............................................................................................................... 63 Figure 4.6: Mean spell of unemployed (in year) ..................................................................................................... 69 Figure 4.7: The comparison of unemployment rate by training (%) ....................................................................... 71 Figure 4.8: Unemployment differential between female, youth and adult male with training ................................ 72 Source: UEUS 2003-11 ........................................................................................................................................... 72 Table 4.9: Unemployment differential between TVET and secondary school graduates ........................................ 73 Table 4.10: Self-employment by training ................................................................................................................ 74 Figure 4.11: Average time from school to work transition by education ............................................................... 75 Figure 4.12: Relationship between unemployment rate and GDP ........................................................................... 76 Figure 4.13: relationship between participation and employment ratio ................................................................... 77 Figure 4.14: Employment contribution of MSEs (%) ............................................................................................ 103 Figure 4.15: Employment by type of MSEs (%) ................................................................................................... 104

8

1. INTRODUCTION

1.1. Background of the Study

The developing economies of the world are characterized by a rapidly growing urban

population and urban work force combined with a much slower increase in employment

opportunities and, as a result, high urban unemployment and under-employment. Indeed, a

rising level of urban unemployment could be a great social evil as it is one of the prime sources

of urban poverty and political instability. Moreover, the presence of large numbers of poor and

jobless people in urban areas has depressing impact on tax revenues while putting a great deal

of pressure on government’s current expenditures to meet rising demands for basic urban

services and to create jobs for the unemployed. This will inevitably have a crowding effect on

resource allocation for growth enhancing sectors of the economy. For these and other reasons,

the general consensus among social scientists and policy makers is that the issue of urban

unemployment has to be wisely managed, particularly in developing countries where social

security services are nonexistent. Therefore, the study of unemployment is an area of

considerable importance which is of both theoretical and empirical interest.

Unemployment and underemployment are among the greatest challenges to the development of

African continent. Africa’s labor force, with over 368 million women and men predominantly

engaged in agriculture and rural non-farm activities, accounts for 11.9 per cent of the total

world labor force. The overall unemployment rate in sub-Saharan Africa was estimated at 9.8

per cent in 2006 (ILO, 2007) and stood at an estimated 7.9 per cent in 2008 (ILO, 2009a).

Although the official unemployment rates seem declining and relatively lower, when the

number of working poor reflected mainly in underemployment and vulnerable employment is

included, the employment situation looks even more desperate. As stated in ILO (2007)

concerning the decent work agenda in Africa, the total number of people worldwide living on

less than $1 a day declined from 1.45 billion in 1981 to 1.1 billion in 2001. In contrast, the

number in sub-Saharan Africa increased from 164 million to 314 million during the same

period, of which roughly 50 per cent are women and men of working age. Consequently,

Africa has the largest number of working poor in total employment of any region.

9

The fact that most African countries lack formal social insurance schemes make most poor

people to have no option other than being employed, underemployed or dependent on

employed people through informal social networks for their livelihood. Thus, even people

outside the labor market tend to be dependent on individuals in the labor market. In effect,

labor markets are central to the livelihoods of poor people in Africa both in and outside of the

labor force (ECA, 2005). Africa, like its higher rate of poverty, is also known for its higher

unemployment. The failure to create more and better paid jobs to meet the needs of the

growing labor force and reduce poverty remains a fundamental issue in many African

countries. A spatial perspective of Africa’s labor market outcome witnessed higher rates of

unemployment in urban areas than in rural ones. It is about 3 times higher in urban areas than

in rural areas (ADB, 2010).

According to international labor organization, despite the constraints of reliable and

comprehensive data, it is estimated that around three-quarters of activities in the urban

economies of Africa are informal in nature. This is why improving productivity and market

access for workers and producers in the informal economy should be at the heart of many

poverty reduction efforts in Africa. In the face of considerable improvement in macroeconomic

performance in recent years across the region, the resulting job opportunities are not sufficient

(ILO, 2007). The implication is that if the MDG of halving extreme poverty by 2015 is to be

realized in the region, an employment-centered growth strategy coupled with active population

policy is required.

Similar to other sub-Saharan Africa countries, employment in Ethiopia is characterized by a

heavily segmented labor market situation. It can be divided among different segments, with

significant distinction between formal and informal employment, private and public

employment, wage and self-employment, and urban and rural employment (EEA, 2007). From

a rural-urban perspective, the Ethiopian labor market exhibits a significant disparity. Generally,

the rural labor market is known by a pervasive problem of underemployment while the urban

one is characterized by a severe open (or official) unemployment.

10

As noted in Guarcello, Lyon and Rosati (2008), in rural areas, unemployment is lower but

with extremely low level of human capital, high underemployment or disguised

unemployment, and few chances to be employed in the formal sector. In urban areas, on the

other hand, although the labor force may face relatively better prospects in terms of income and

employment quality, finding a job is difficult and hence unemployment, especially youth

unemployment, is higher. Similarly, labor force surveys (LFS) by the Central Statistics Agency

(CSA) of Ethiopia indicate that the average unemployment rates for urban areas were 26.4

percent and 20.6 percent in 1999 and 2005, respectively while they were 5.1 and 2.6 percent

for rural areas in the same periods. The situation is rather worrisome in relatively bigger cities.

For instance, in Tegegn (2011), the overall unemployment rate in Addis Ababa was as high as

38.5 percent in 1999 and decreased to 31.7 percent in 2005, but elevated above very unpleasant

urban average rate (Tegegn, 2011).

The current government of Ethiopia has been implementing poverty and unemployment

reduction polices since the reform period 1991. Particularly, promoting micro and small scale

enterprises, expanding microfinance services, reforming the education and training system and

increasing its accessibility at all levels, encouraging inflow of FDI and promoting labor-

intensive technologies are among those worth mentioning. Yet, it is apparent that poverty

reduction and development policies and strategies of Ethiopia cannot bring the desired result

without creating gainful employment for the unemployed and underemployed population.

Despite the impressive economic growth in the past eight or so years and the various

development policy efforts, the incidence of urban unemployment is still higher and persisting.

According to the urban employment-unemployment surveys of CSA, the average urban

unemployment rates of Ethiopia for people aged between 10 and 64 years was 26.3 percent in

2003 and it stood at 18 percent in 2011. This means that the rates decreased only by 8

percentage points in the 8 year periods, implying a merely 1 percent average annual reduction.

Given the existing efforts, the annual reduction rate is slower and disappointing. Such

persistent and higher incidence of unemployment suggests the urgency of a deep and rigorous

examination of the root causes of the problem, which might be the key step towards the

solution.

11

There have been a number of empirical studies conducted on urban unemployment in Ethiopia.

For instance, Tegegn 2011; Guracello, Lyon and Rosati 2008; WB 2007; Seife 2006; Serneels

2008; Serneels 2007; Birhanu, Abraham, and van der Deijl 2005; Getinet 2003; Mulat et al.

2003; Krishnan, Gebreselassie and Dercon 1998 can be mentioned. Most of them did focus on

discussing either the demographic determinants of unemployment or duration of

unemployment using relatively older and single period cross-sectional data.

Tegegn (2011) assessed the socio-demographic determinants of urban unemployment in

Addis Ababa using data from 1999 and 2005 labor force surveys (LFS) of CSA. The

estimation results of the Logit model imply that a person’s sex, age, migration status, level of

education and training status are statistically significant and most important factors that

determine the unemployment probability of an urban worker. However, the scope of the study

is limited only to Addis Ababa and also didn’t explicitly discussed policy issues. Although he

used a relatively recent data, he estimated the two cross-section data sets separately and

didn’t link them and show the trend of unemployment in the model.

Guracello, Lyon and Rosati (2008a) also studied the challenges of child labor and youth

employment in Ethiopia using a 2001 LFS data. The estimation results of the Probit model

imply the employment chance of a young worker does significantly vary by sex, household

income and education. However, they used a single cross-section data. They did not clearly

indicate the reference education dummy in their discussion and also didn’t consider urban

location.

Seife (2006) examined the determinants of unemployment duration in urban Ethiopia using the

2000 Ethiopian Urban Socio-Economic Survey data and employed parametric and semi-

parametric models. The results of the regression analysis imply that age, marital status, level of

education, location of residence and support mechanism significantly affect the duration of

unemployment while ethnicity and gender do not. However, this duration study used only a

single cross-section data and also didn’t explicitly discuss the effect of policies meant for

addressing unemployment.

12

Serneels (2007) assessed the incidence and duration of unemployment among young men

(aged 15-30) in urban Ethiopia, He used the 1994 first round household data from the

Ethiopian Urban Socio-Economic Survey (EUSES) and analyzed by a probit and proportional

hazard duration models. He argues that male unemployment in urban Ethiopia does fit with

queuing model of unemployment. However, the study used a single cross-sectional data and

also its scope is too narrow and limited only to young males. Therefore, it is not

representative of the labor force and the current situation. Besides, the reference line of

education is not clearly indicated in the discussion.

Getinet (2003) studied the effect of individual characteristics on the incidence of youth

unemployment in urban Ethiopia using the first (1994) and fourth (2000) waves Urban Socio-

Economic Survey (EUSES) data. The findings of the multinomial logit analysis indicate that

young people who completed secondary education are more likely to be both unemployed

and active. On the other hand, those with at most elementary level education are more likely

to be in self-employment and casual/domestic types of activities as compared to those with

tertiary level education. Although he used two different cross-section data, he estimated the

two cross-section data sets separately and didn’t link them and show the trend in the model.

What remains to be explored, however, is how unemployment responding to education level

attained and training received and how it is changing overtime and variation in urban

location.

Promoting micro and small- scale enterprises (MSEs) was one of the strategies explicitly stated

in PASDEP (Plan for Accelerated and Sustained Development to End Poverty) to create

employment and generate income, primarily to reduce urban unemployment. Still the latest

five-year plan, the Growth and Transformation Plan 2010/10 – 2014/15 (FDRE, 2010), has

given particular attention to the expansion and development of micro and small-scale

enterprises. The sector is believed to be the major source of employment and income

generation for a wider group of the society. In this regard, identifying factors that affect

employment creating capacity of MSEs has policy relevance to take action in a way to enhance

employment potential of these enterprises in which many get employed and still a potential

source of employment for the unemployed.

13

Unfortunately, it is difficult to find empirical evidences on the employment effect of MSEs in

Ethiopia. Birhanu, Abraham, and van der Deijl (2005) did attempt to discuss the support given

to MSEs and the employment created before some 8 years relying mainly on the report of

FeMSEDA. Nevertheless, in recent years the government has given more emphasis to the

sector and significant changes would have been occurred. Recently, Rahel and Paul (2010)

assessed the growth determinants of women operated MSEs in four kebeles of Addis Ababa

city. However, firstly, the scope of the study is too limited and lacked strong and objective

analysis. Secondly, they didn’t adequately discuss the determinants of employment growth in

the MSEs. Nevertheless, there are enormous studies emphasized on causes of firm growth in

US, Canada and Europe and a few studies on causes of new firm growth in Latin American

countries (Capelleras and Rabetino, 2008). Even these studies already we have focused on

growth of new firms in general and but this work focus on exploring the factors that determine

average annual employment growth in MSEs.

Considering the drawbacks of the previous education system, a new education and training

policy has been designed and implemented since 1994. The new policy has given emphasis to

education and training that offer specific learning skills related to the market needs, i.e.

gainfully tradable skills based on demand driven and in response to the country’s development

approach. Consequently, considering the strategic importance of training, the first National

TVET strategy has been in effect since 2002. Furthermore, acknowledging the limitations of

former graduates of TVET in meeting the expectations and demand of the labor market, a

comprehensive development vision for the TVET sector has been outlined in the Education

Sector Development Program (ESDP) III (MoE, 2008). All these efforts are supposed to

improve the skill and employability of the trainees and thereby address the problem of urban

unemployment. Equally important, assessing whether these policy efforts are effective in

achieving the desired goals they are supposed to or not is necessary in order to take corrective

measures timely and to minimize the wastage of scarce resources. However, objective

assessments on the effectiveness of policies are uncommon in Africa in general and in Ethiopia

in particular. None of the so far empirical studies in Ethiopia did clearly and objectively

analyze the effect of the TVET program on unemployment, spell of unemployment and school-

to-work transition and self-employment by setting relevant referent group. Although Birhanu,

14

Abraham, and van der Deijl (2005) and Guracello, Lyon and Rosati (2008a) discussed the

existing education and training policies, they didn’t empirically examine their effects on

unemployment. For this reason, this study sheds light on the existing research gap by

attempting to explicitly examine the effect of the TVET program on labor market outcomes,

particularly on urban unemployment.

Evidently, the preceding discussions indicate that although there have been previous studies on

the issue of urban unemployment in Ethiopia, most of them focused mainly either on the

demographic determinants of unemployment or duration of unemployment. They used not only

older data but also a single cross-sectional data, except that two studies used two cross-section

data sets. Therefore, they didn’t empirically explain how unemployment changed overtime. In

addition, they didn’t adequately and explicitly analyzed the effect of policy responses meant

for reducing unemployment such as the TVET and MSEs sectors. Some of them are limited in

scope; and most of them, but two studies, didn’t consider urban location as important factor in

explaining urban unemployment.

Therefore, we argue that, relative to the persistent and severe unemployment problem in urban

Ethiopia, empirical studies conducted so far on the causes of urban unemployment are limited

in number and are not recent enough to explain the current situation. We also argue that effect

of policy interventions aimed at addressing the problem of urban unemployment is yet under

researched issue in Ethiopia. Previous studies didn’t duly consider the effects of policy

interventions, such as expansion of TVET and promotion of MSEs, on unemployment.

Accordingly, this study is timely and to some extent attempted to fill the research gaps

identified above. Unlike the other studies, we used five cross-sectional data sets ranging from

2003 to 2011 and combined to create pooled data that can better estimate population

parameters relative to a simple cross-section data. This helped us to better explain the trend and

the recent situation of urban unemployment. In doing so, we identified the following specific

research questions and tried to address them correspondingly.

1. What are the characteristics of urban unemployment in Ethiopia?

2. What are the socio-economic causes of unemployment in urban Ethiopia?

3. What are the effects of TVET program on unemployment?

15

4. What are the factors that determine the employment growth within MSEs?

1.2. Objective of the Study The general objective of the study is to examine the major socioeconomic causes of urban

unemployment and the effect of policy interventions, through expansion of TVET and

promotion of MSEs, on urban unemployment in Ethiopia. The specific objectives of the study

are to:

1. Describe the characteristics of urban unemployment in Ethiopia.

2. Investigate the socio-economic causes of urban unemployment.

3. Examine the effect of TVET program in reducing unemployment in urban Ethiopia.

4. Identify the factors that determine average employment growth within MSEs in urban

Ethiopia.

5. Suggest some policy implications

1.3. Data Sources and Methodology

In order to address the aforementioned objectives, we made use of both primary and secondary

data sources. The primary data were collected from three cities, namely Addis Ababa, Bahir

Dar and Hawassa. The secondary data were obtained from the labor force surveys (1999 and

2005) and urban employment unemployment surveys (2003, 2004, 2006, 2010 and 2011) of the

Central Statistical Agency of Ethiopia. In addition, data on some macroeconomic variables

were taken from the World Bank database.

The available data were analyzed by both descriptive and regression methods of analysis. The

descriptive analysis is used to describing the characteristics of urban unemployment. The

regression analysis involves econometric models to examine the effects of policy interventions.

We employed probit and duration (proportional hazard) models for analyzing the pooled cross-

sectional data and cross-sectional data.

16

1.4. Significance and Scope of the Study

The study is expected to provide some empirical overview on the socio-economic causes of

urban unemployment and on the role of TVET and MSEs in reducing urban unemployment.

First, understanding the relationships among education and training in general and TVET in

particular and unemployment can help to reveal underlying effects of improving human capital

on unemployment and can help concerned bodies to evaluate strategies. Hence, the research

report can be an input for concerned bodies at different levels who are interested in the issue.

Second, determining factors that increase employment size of MSEs are fundamental to

appropriate intervention to curb high urban unemployment. Therefore the study will be used to

reassess the development and implementation of employment policies and programs in

Ethiopia. Third, this work can supplement the existing empirical studies on urban

unemployment and serve as a reference material for teaching as well as for others who will

conduct related studies. Fourth, it may encourage interested researchers to undertake impact

evaluation to examine the effect of education and training on unemployment to fill the existing

gap in depth.

The scope of the study is limited in that its focuses only on the causes of unemployment

attributable to socio-economic factors (i.e. due to serious time series data shortage on

unemployment rate) and effects of public interventions on urban unemployment. Its spatial

coverage, as the title implies, is confined to urban Ethiopia. Nevertheless, the implications of

the findings are expected to be useful and applicable for rural parts of the country and for urban

areas of other sub-Saharan African countries as well.

1.5. Limitation of the Study

The major limitations of this study emanate from the obvious constraints of the availability of

employment-unemployment data in Ethiopia. The most important data this study made use of

are the labor force surveys and the urban employment unemployment surveys obtained from

the Central Statistical Agency of Ethiopia. Although the data set are comprehensive and cover

all regions of the country, they lack some important information supposed to be crucial for the

purpose of this study. For this reason, we had to collect primary data to supplement the

17

available secondary data. The primary data has also its own limitation pertaining to time and

other resource constraints. Hence, it covered only limited sample individuals and enterprises

located in three cities. However, an attempt was made to make the sample as representative as

possible so that the findings are believed to explain the same issue in other areas too.

1.6. Organization of the Paper

This paper is arranged in five sections. The next section reviews theoretical and empirical

literature while the third one describes the source and nature of the data and the method of

analysis. Section four presents the descriptive statistics and discusses the findings of the

regression analysis. Lastly, the fifth section concludes and put forth policy implications.

18

2. LITERATURE REVIEW

2.1 Definition and Concepts of Unemployment

Unemployment is usually viewed and defined from the human element point of view.

Although any factor of production can be unemployed, economists have put particular

emphasis on the human element –the unemployment of labor. According to Sapsford and

Tzannatos (1993), this is mainly due to the mental and sometimes physical sufferings and

hardships that the unemployed and their dependents experience. Thus unemployment

generally refers to a status in which individuals are without job and are seeking a job. For the

purpose of this paper, we make use of the Key Indicators of the Labor Market (KILM)

standard definitions of ILO, as adopted by the 13th International Conference of Labor

Statisticians (ICLS) in 1982 and 1998. Accordingly, the ensuing section presents the standard

definitions of the key indicators of a labor market such as activity rate, employment,

underemployment, unemployment and not currently active and then followed by an overview

of the labor force conceptual framework.

Activity rate or labor force participation rate refers to the share of the population aged

between 15-64 years and either engaged in, or available to undertake, productive activities.

Hence it captures the idea of labor supply for all productive activities according to the 1993

UN system of National Accounts. Employment is defined in terms of paid employment and

self employment. Paid employment covers persons who during the reference period

performed some work for wage or salary, in cash or in kind, as well as persons with a formal

attachment to their job but temporarily not at work. Self employment covers persons who

during the reference period performed some work for profit or family gain, in cash or in kind,

and persons with an enterprise but temporarily not at work. Hence employment rate is the

share of the employed over the labor force population aged from 15 years to 64, rather than

between 10 and 64 years as adopted by the CSA.

Underemployment is a concept that has been introduced for identifying the situations of

partial lack of work. According to the ILO, the “underemployed” comprise all persons in paid

or self-employment, involuntarily working less than the normal duration of work determined

19

for the economic activity, who were seeking or available for additional work during the

reference period. Thus the “underemployed” can be considered as a subgroup of the

“employed”.

On the other hand, the international standard definition of unemployment is based on three

criteria, which have to be met simultaneously. According to the definition, the unemployed

comprise all persons above the age specified for measuring the economically active

population who during the reference period were: (a) "without work", i.e. were not in paid

employment or self-employment as defined by the international definition of employment; (b)

"currently available for work", i.e. were available for paid employment or self-employment

during the reference period; and (c) "seeking work", i.e. had taken specific steps in a specified

recent period to seek paid employment or self-employment.

The aforementioned three criteria to define unemployment imply that merely joblessness per

se cannot qualify a person to be counted officially as an unemployed. A person without a job

is said to be involuntarily unemployed as long as he/she is available and willing to be

employed at the going wage rate; otherwise he/she is considered as voluntarily unemployed

and does not appear in the official statistics as he/she has dissociated himself from the labor

force. The unemployment rate is therefore, the share of the unemployed over the labor force

population aged between15 and 64 years. However, this standard definition is different from

Ethiopia’s official definition of unemployment by the CSA. The CSA definition, therefore,

relaxes the criterion of "seeking work" and adopts a relaxed definition which leads to higher

unemployment rates. The main rationale for relaxing the definition in Ethiopia is attributable

to the unorganized nature of the country’s labor market, in which job search media are not

well developed or quite limited and not accessible to majority of the job seekers.

The population not currently active (economically inactive populations) refers to the residual

category comprising those without work but were neither seeking nor available for work, such

as students, home keepers and the retired, as well as those below the minimum age specified

for measuring the economically active population.

20

In what follows, just to have a clear understanding of the statistical definitions and the

conceptual relations among them, the ILO's conceptual labor force framework is briefly

presented as follows. The labor force framework was developed according to the ILO

Resolution concerning statistics of the economically active population, employment,

unemployment and underemployment, adopted by the Thirteenth International Conference of

Labor Statisticians (October 1982). The employed and unemployed categories together make

up the labor force (or the currently active population), which gives a measure of the number of

persons furnishing the supply of labor at a given moment in time. The third category (not in the

labor force), to which persons neither seeking nor available for work plus those below the age

specified for measuring the economically active population are included, represents the

population not currently active. In short, these relationships may be expressed as:

UnemployedEmployedForceLabor

PopulationInactiveLaborPopulation+=+= Force

Figure 2.1: The ILO’s Labor Force Framework

Source: Prakash (2001)

Source: ILO

Total Population

Population above Specified Age Population below specified Age

Currently active population (the labor force)

Population Not Currently Active (NILF)

Employed Unemployed

Because of: school attendance, household duties,

retirement (old age), or other

reasons

21

2.2. Type of Unemployment

The theoretical literature identifies various types of unemployment categories on the basis of

their sources. Although there are more, the most frequently stated classifications are Demand

Deficient or Cyclical, Frictional, Structural, and Seasonal unemployment. However, it is worth

noting that the real-world unemployment may combine different types simultaneously, and

thus distinguishing clearly one from the other and measuring the magnitude of each of them is

difficult, partly because they overlap (EEA, 2007, Henderson, 1991).

Cyclical unemployment is involuntary unemployment arising from the business cycle effect

as a result of insufficient effective aggregate demand for goods and services. When there is a

recession or a severe slowdown in economic growth, economies face with a rising

unemployment because of plant closures, business failures and an increase in worker lay-offs

and redundancies. This is due to a fall in demand leading to a contraction in output across

many industries. According to Sapsford and Tzannatos (1993), this type of unemployment

coincides with unused industrial capacity; and as traditional Keynesian economics suggests, its

cure lies in policies that succeed in increasing the level of aggregate demand.

For Keynesian economists, unemployment is a situation in which the number of people who

are able and willing to work at prevailing wage exceeds the number of jobs available. When

the number of unemployed is significant, the demand in the product market will be negatively

affected, as a result, firms are unable to sell all the goods they would like. Businesses respond

to a declining demand for goods and services by cutting employment in order to control costs

and restore some of their lost profitability. Consequently, the higher unemployment will tend to

impede the growth of gross output, implying a vicious circle.

Frictional or Search unemployment is transitional and temporary unemployment that arises

because a person may take time to find a new job after losing or quitting a job, or after entering

or reentering the labor force following schooling, illness, or some other reason for being out of

the labor force. It usually occurs due to imperfect information in the labor market (Henderson,

1991, Mankiw, 2001). It is a consequence of the short run changes in the labor market that

constantly occur in a dynamic economy in response to changes in the product market. It arises

22

because the process of matching unfilled vacancies and unemployed workers is not

instantaneous (Sapsford, 1993).

In the context of developed economies, incentives such as unemployment benefits can also

cause some frictional unemployment as some people actively looking for a new job may opt

not to accept paid employment if they believe the tax and benefit system will reduce the net

increase in income from taking work. When this happens there are disincentives for the

unemployed to accept work. Normally, frictional unemployment may not pose much threat to

individual’s welfare as long as it is temporary and does not last long. There may be little that

can be done to reduce this type of unemployment, other than provide better information to

reduce the search time. This suggests that full employment is impossible at any one time

because some workers will always be in the process of changing jobs.

Structural unemployment refers to a mismatch of job vacancies with the supply of labor

available. It is caused by long-run changes in the structure of the economy, which give rise to

changes in the demand for labor in particular regions, industries or occupations. For instance,

technological progress may make an industry capital intensive from a purely labor intensive

one. The release in labor from such an industry gives rise to the problem of unemployment.

Although workers are available for employment, they may lack the skills that the available

vacancies required or they may be in the wrong location to take the available jobs (EEA, 2007,

Sapsford, 1993, Henderson, 1991). Increasing international competition due to

globalization leads to changes in the patterns of trade between countries over time; and hence

it could be one of the reasons for structural unemployment. Because structural unemployment

lasts longer, demand management instruments alone may not be effective remedies to the

problem. Besides, other instruments such as facilitating training programs and subsidizing

mobility of workers are required along with demand management policies so as to significantly

reduce its incidence (EEA, 2007).

Structural unemployment can also arise from the immobility of labor. In an economy,

industries that are growing and need labor are not necessarily able to employ the same workers

who have been displaced in the declining industries. This situation can be attributable to the

problem of labor immobility. Labor immobility includes geographical immobility, industrial

23

immobility, and occupational immobility. Geographical immobility occurs when workers are

not willing or able to move from region to region, or town to town. Industrial immobility

occurs when workers do not move between industries. Occupational immobility arises when

workers find it difficult to change jobs within an industry. Industrial and occupation immobility

are most likely to happen when skills are not transferable between industry and job.

Information failure also contributes to labor immobility because workers may be immobile

because they do not know where all the suitable jobs for them are. A resulting problem with

labor market immobility is that it can create regional unemployment, which is a type of

structural unemployment. This means that a change in the structure of industry leaves some

people unable to respond by changing location, industry, or job and as a result, they remain

temporarily or permanently unemployed.

Seasonal unemployment occurs as a result of normal and expected changes in the economic

activities over the season of a year. Seasonal unemployment exists because certain industries

only produce or distribute their products at certain times of the year. As noted in Sapsford &

Tzannatos (1993), workers in the agriculture and construction sectors as well as in the tourism

industry, who are often out of work during the winter months are typical examples of

seasonally unemployed people. Indeed, such phenomena are common in most Sub Saharan

African economies where seasonal unemployment following the end of harvesting season is

inherent in the agricultural sector.

2.3. Theories of Unemployment

In Classical economic theory, unemployment is seen as a sign that smooth labor market

functioning is being obstructed in some way. In a smoothly functioning market the equilibrium

wage and quantity of labor would be set by market forces. The Classical approach assumes that

markets behave as described by the idealized supply and demand model. The labor market is

seen as though it were a single, static market, characterized by perfect competition, in which it

is assumed that every unit of labor services is the same, and every worker in this market will

get exactly the same wage. Because such a Classical (idealized) market for labor is free to

adjust, there is no involuntary unemployment; everyone who wants a job at the going wage

24

gets one. Thus, the only thing that can cause true unemployment is something that interferes

with the adjustments of free markets, such as a legal minimum wage and other regulations.

Nevertheless, this seems far from the reality. As Solow (1980) puts, the labor market is

segmented in that not everyone in it is in competition with everyone else, among others, due to

the obvious differences in abilities, experience and skills.

The presence of a legal minimum wage is commonly considered as one such factor that can

distort the smooth functioning of the labor market. If employers are required to pay a

minimum wage that is above the equilibrium wage, this model predicts that they will hire fewer

workers and hence a fall in demand for labor. The market is, in this case, prevented from

adjusting to equilibrium by legal restrictions on employers. Now there are people who want a

job at the going wage, but can’t find one. That is, they are added to the unemployment pool,

letting the unemployment rate to rise. The empirical evidence, however, may not always

support the classical idea that minimum wages cause substantial unemployment. For example,

Card and Krueger (1993) found that a moderate increase in the minimum wage in New Jersey

did not cause low-wage employment to decline, and may even have increased it.

There are also other reasons that the economy might provide less than the optimal number of

jobs for the labor force. For instance, regulations on businesses often negatively affect their

demand for labor. Strong job protection, through employment protection legislation as well as

unionization, raises the cost of firing workers, which in turn causes firms to lower their demand

for labor (Pierluigi, 2008). Labor union activities and labor-related regulations such as safety

regulations, mandated benefits, or restrictions on layoffs and firings increase the cost of labor

to businesses. As a result, businesses tend to opt towards labor-saving technologies and thus

reducing job growth. Classical economists also argue against public safety net policies such as

disability insurance and unemployment insurance; they believe that such policies reduce

employment by causing people to become less willing to seek work. From a classical point of

view, labor-market recommendations tend to focus on getting rid of regulations and social

programs that are seen as obstructing proper market behavior. Like other classical proposals,

such labor market proposals assume that the economy works best under the principle of laissez-

faire.

25

The classical theory of labor markets depends on quick market adjustment, in particular, the

elimination of any labor surplus through falling wages and a resulting full-employment

equilibrium at a lower wage rate. But ‘to what extent is this realistic?’ is a natural question that

comes to everyone’s mind. According to the well known explanation of Keynes, based on the

experiences of the Great Depression, certain aspects of real world human psychology and

institutions make it unlikely that wages will fall quickly in response to a labor surplus. Thus,

Keynesian-oriented economists developed ‘sticky wage’ theories, which hypothesize that

wages may stay at a level above equilibrium for some time. Wages may eventually adjust in

the way shown in the Classical model, but too slowly to keep the labor market always in

equilibrium. In addition to psychological resistance to wage cuts, a minimum wage might also

make wages sticky. Wages may also become set at particular levels by long-term contracts,

such as many large employers negotiate with labor unions. Relatively in recent years, economists have also come up with two other theories: the insider-

outsider theory and the efficiency wage theory. The insider-outsider theory hypothesizes that

the efforts of insiders may contribute to keeping wages high. ‘Insiders’ are people who already

have jobs within an organization while ‘outsiders’ are workers who are not in the organization

but who are potentially competitors of the insiders and can be hired in the future by the

organization. Insiders may be able to keep their wages high by setting up various barriers that

prevent their employer from dismissing them and hiring lower-priced outsiders. Insiders may

have contracts that specify a high wage and that make them difficult to fire. Or they may refuse

to cooperate with new workers or harass them, reducing new workers’ productivity. In the

insider-outsider theory, employed workers use the power they derive from such labor turnover

costs to keep their wages artificially high.

According to the efficiency-wage theory, employers may find it to their advantage to pay

employees wages that are somewhat higher than would be strictly necessary to get them to

work. Employers must attract, train, and motivate workers if their enterprise is to be

productive. Efficiency wage theory suggests that paying higher-than-necessary wages may

improve employee productivity. Workers may be healthier and better nourished, and therefore

more able to do quality work, when they are better paid. Also, workers may quit less often if

26

they know they are getting ‘a really good deal’. A lower likelihood of quitting makes

employees more valuable to an employer because the employer saves on the costs of training

new workers. Workers may also work more efficiently if being caught shirking means

potentially losing their “really good deal.” If the higher-than-necessary efficiency wages

creates a pool of unemployed people, this only further reinforces employees’ incentives to

work hard because then they will be even more afraid of losing their good jobs.

In sum, in the Classical-Keynesian synthesis, legally or contractually-set wages, fear of worker

unrest, the power of insiders, and efficiency wages are thought sometimes to cause wages to be

"sticky." By making real world labor markets work differently than the market pictured in the

classical model, these phenomena mean that it is unrealistic to expect that labor markets can

adjust rapidly to maintain full employment.

The supply-demand analysis, whereby the classical model of the labor market is described, is

simply a way of thinking about a single and spot market in which a single, completely

standardized good is being traded. However, the economy as a whole is not just one smoothly-

functioning market in which prices move to equate quantity supplied and quantity demanded.

The economy is made up of several heterogeneous markets as well as a number of nonmarket

institutions and transfers of all sorts, which make it complex and difficult to explain by a

simple demand-supply analysis. For Keynesians, the classical theory, which assumes only an

idealized, abstract, and institutionless labor market, is fundamentally misleading and

unrealistic.

In the Keynesian model, aggregate employment depends on the level of aggregate demand in

the economy as a whole. If total spending is low and businesses cannot sell their goods, they

will tend to cut back on their investments and on the number of workers they employ. Prices as

well as wages may fall (as was observed during the Great Depression), keeping real wages

constant and thus giving employers no incentive to hire more workers. Low aggregate demand

for goods and services could lead to a vicious cycle of unemployment, low incomes, and low

spending in the economy as a whole. The Keynesians recommendation for fixing the problem

of unemployment in a recession or depression is stimulating aggregate demand in the economy,

and not just making labor markets work more smoothly.

27

According to Gunatilaka and Vodopivec (2010), the level of unemployment can also be

described by three hypotheses: the skill mismatch, the queuing, and the slow job creation

hypotheses. The skills mismatch hypothesis maintains that a mismatch between what the

education system teaches and what the labor market requires produces educated youth who

have few marketable job skills but who nonetheless aspire to ‘good’ jobs (jobs that are secure,

well-paid, and offer higher social status) and who spend a fair amount of time looking for such

jobs. The queuing hypothesis argues that the unemployed wait for an opportunity to take up

good jobs in the public sector and in the formal private sector. The public sector is often

characterized by job security, generous fringe benefits, low work effort, and high social status.

It is thus blamed for creating unemployment by encouraging job aspirants to queue for these

jobs.

The slow job creation hypothesis, also called the institutional hypothesis, argues that labor

market institutions raise the costs of formal job creation. In particular, highly restrictive

employment protection legislation and high wages resulting from strong bargaining power of

workers under conditions of virtually complete job security raise labor costs and impede job

creation. As a result, the job creation rate of the formal private sector is depressed and the

majority of workers are forced to opt to the unprotected informal employment (Gunatilaka,

2010).

2.4. Causes of Unemployment As stated in the preceding theoretical discussion on types and theories of unemployment, the

classification of unemployment is based on factors that result in unemployment. There are a

number of factors that may affect the level of unemployment; and hence identifying the major

causes is the leading step so as to treat it effectively. As Mankiw (2001) states, the main

rationale for studying unemployment is to identify its causes and thereby to help improve the

public policies in favor of the unemployed.

The issue of unemployment has always been a matter of great debate among the traditional as

well as contemporary economists. For the Keynesian economists, unemployment is generally

caused by insufficient aggregate demand in the economy, as a result of which individuals lose

28

their jobs and added to the unemployment pool. The views of the classical economists differ

from their Keynesian counterparts. Unemployment, termed as classical unemployment or real

wage unemployment, is caused when wages are too high. This explanation of unemployment

was the dominant theory, particularly before the great depression of the 1930s, when workers

themselves were blamed for not accepting lower wages, or for asking for too high wages. Yet,

advocates of classical economics strongly argue that the rigidities in the labor market, which

are mainly explained by taxes, minimum wage laws and the power of labor unions, are the

main reasons behind unemployment. Unemployment incidence from the classical perspective

is, however, less likely to be situated in most sub-Saharan African economies where a large

proportion of the labor force is working in unprotected and low paid jobs in the informal

sectors. Thus the major problem in these countries is more likely the inadequate capacity of the

economy to sustain the constant labor supply growth rather than the rigidity of wages and

prices.

The unemployment literature suggests that both supply and demand factors are to blame for

impacting unemployment, and hence, its magnitude is determined by the balance between the

demand for and the supply of labor. Whenever the supply of labor exceeds the demand for it at

the prevailing wage rate, unemployment arises. Hence, the causes of unemployment are

primarily explained by either factors that can increase the supply of labor and/ or factors that

can negatively affect the demand for labor. The factors that increase the supply of labor are

associated with the increase in population and labor market conditions that can either positively

or negatively affect the labor market participation decisions of working age population. The

demand for labor is a derived demand as it is demanded to meet the demand for goods and

services. The demand for labor is determined by the performance of an economy and the

choice of production techniques, which in turn are shaped by the existing economic policies

((Bakare, 2011);(ECA, 2010);(EEA, 2007); (Adebayo, 1999)).

2.4.1. Supply Side Factors

In the African context, among the important supply factors that can affect urban unemployment

are high population growth, rapid rural-urban migration, poor quality education and training,

and other demographic variables (ECA 2010; EEA 2007; (Okojie, 2003). Population growth

29

can be an opportunity for an economy because it is a source of potential labor and

entrepreneurs. On the other hand, it can also burden economies with saturated labor market that

is unable to provide decent employment opportunities for already employed labor and for new

entrants.

According to the classical economics view, an increase in labor supply will tend to raise

employment although it dampens productivity increases. The higher labor supply will lead to

lower average wages and consequently to an increase in demand for labor (Kapsos, 2005

(Walterskirchen, 1999). Empirical evidences also confirm the positive and significant

association between the labor supply and employment elasticity. A 1-percentage point increase

in the average annual growth rate of the working-age population is associated with an increase

in the employment elasticity by 0.24 (Kapsos, 2005).

But the situation is different in Africa, where the demographic transition is lowest and the

population growth rate is still around 2.4 per cent. Over the past 20 years, the economically

active population of Africa has grown at an average rate of 3 per cent, rising from 231 million

in 1990 to 403 million in 2009. This represents a 43 per cent increase just in two decades, one

of the highest increases among all regions of the world (ECA, 2010). Therefore, high

population growth and growing labor participation has rather resulted in excessive supply of

labor, which has continued to outstrip the demand for labor. In this regard, it is worth

mentioning the situation in Ethiopia as it can be a good instance for this fact. Between 1994

and 2005, in a decade, the Ethiopian labor force increased by 21.3 percent while the

employment creation increased by 18.7 percent (EEA, 2007). It implies that, despite lack of

evidence on the quality of employment generated, nearly 3 percent new jobless individuals are

added to the unemployed population during the period.

A key supply factor in urban labor market leading to urban unemployment, often cited in the

literature, is the high degree of geographical mobility of people, especially the youth, in the

form of rapid rural-urban migration. The well known classical analysis of rural-urban

migration and urban unemployment is attributable to the works of Harris-Todaro and Todaro

(1969). According to Todaro (1969), as long as rural-urban wage differential attracts rural

30

people, urban unemployment cannot be reduced regardless of creating more jobs through labor

intensive methods of production. The implication is that apart from creating employment in

urban areas, making rural areas more attractive is also equally important. Similarly, referring to

Harris-Todaro’s model, Bencivenga and Smith (1995) document that labor migrates to

wherever its expected income is highest; and hence in equilibrium expected incomes must be

equated between rural and urban employment. Since urban wages are invariably much higher

than rural wage rates, the equilibration occurs through the existence of unemployed or

underemployed urban labor. This is due to an institutionally fixed urban real wages mainly

attributable to minimum wage legislation and /or the power of labor unions.

Rural-urban migration, which is the important factor for the rapidly growing urban labor force,

can be explained in terms of push-pull factors. Even though there is high recorded employment

in rural areas of most African countries, this employment generates insufficient incomes for

rural workers mainly due to lower agricultural labor productivity. Rural to urban migration

occurs, to a large extent, because rural Africans are so desperate that they are willing to try

their chances in the unpromising urban labor market. This has resulted in a concentration of

youth in African cities where there are few jobs available in the formal sector (Leibbrandt,

2004). Among the push factors of rural-urban migration are the pressure resulting from the

diminishing land-man ratio in the rural areas and the existence of serious underemployment

arising from seasonal nature of most SSA rural economies (Adebayo, 1999).

Proponents of migration argue that rural to urban migration occurs because it is part of the

optimization strategy of rural households, where differences in returns in different markets

determine the allocation of labor (Leibbrandt, 2004). Indeed, in earlier economic development

literature, rural–urban migration was viewed favorably as a natural process in which surplus

labor gradually withdraws from the rural sector to provide needed manpower for the expanding

urban industrial sector. However, there are also arguments against this proposition as witnessed

by the insufficient absorptive capacity of the urban sector relative to the massive rural-urban

migration. As noted in Todaro (1997), the past three decades of African experience has made

clear that rates of rural–urban migration have greatly exceeded rates of urban job creation. One

of the major consequences of the rapid urbanization process has been the burgeoning supply of

31

job seekers in both the modern (formal) and traditional (informal) sectors of the urban

economy. In most African countries, the supply of workers far exceeds the demand, the result

being extremely high rates of unemployment and underemployment in urban areas. Thus he

argues that migration can no longer be casually viewed by economists as a beneficent process

necessary to solve problems of growing urban labor demand. On the contrary, migration today

remains a major factor contributing to the phenomenon of urban surplus labor; a force that

continues to exacerbate already serious urban unemployment problems caused by the growing

economic and structural imbalances between African urban and rural areas (Todaro, 1997).

Although labor market outcomes depend on several factors, education and relevant skills

remain the main determinants of good labor market outcomes for individuals. Education plays

a central role in preparing individuals to enter the labor force and in equipping them with the

skills needed to engage in lifelong learning experiences. The primacy of education stems not

only from its fundamental role in increasing individual earnings, but also from its noneconomic

benefits such as lower infant mortality, better participation in democracy, reduced crime, and

even the simple the joy of learning that enhance and enrich the quality of life and sustain

development (Fasih, 2008).

Evidences from a range of countries shows that education enhances opportunities in the labor

market, as those with the best qualifications enjoy superior job prospects. In the developed

countries, the differential chances of unemployment for qualified and unqualified young people

have been increasing. In a number of developing countries, however, many highly educated

young people remain unemployed. This problem arises from two key factors: an inappropriate

matching of university degrees with demand occupations and the insufficient demand for

skilled higher-wage labor in the formal economy. As most new job growth is in the informal

sector of the economy, there remain few opportunities for young graduates to find work that

corresponds to their level of educational attainment (UN, 2003).

African youth have obtained more formal education over the years. However, educational

systems in Africa have witnessed declines in quality and infrastructure at all levels since the

last decades. They are geared toward providing basic literacy and numeracy and not industrial

32

skills, and are yet to adjust to the changing demands for knowledge, skills and aptitudes

required in the labor market. Youth unemployment in Africa is concentrated among those who

have received some education, but who lack the industrial and other skills required in the labor

market, making them unattractive to employers of labor who prefer skilled and experienced

workers. Furthermore, educated youth prefer wage jobs in the formal sector and would prefer

to remain unemployed until they get the type of job they prefer, that is, they have high

reservation wages (Chigunta 2002; cited in (Okojie, 2003).

The conventional theoretical argument for education suggests that higher educational

attainment leads to better employment outcomes, such as higher wages and lower

unemployment. Empirical evidences indicate that the desirable effect of education on

unemployment is not always evident, particularly for youth. For instance, Guarcello et al

(2008b) analyzed the effect of education on school-to-work transitions for 13 Sub-Saharan

Africa countries based on World Bank Priority survey data. Their findings indicate that higher

educational attainment has not led to a decrease in the unemployment rate for youth in these

countries. Youth with secondary and tertiary education, particularly in Burundi, Cameroon,

Ivory Coast, Kenya, and Madagascar, have higher rates of unemployment than youth with

lower educational attainment.

Labor market outcomes also vary among individuals pertaining to demographic factors both in

rural and urban areas. There are significant differences in participation and unemployment

rates between older and younger cohorts as well as between males and females. Almost in all

countries, both in developed and underdeveloped, the probability of unemployment is strongly

dependent on age cohort of the labor force. Typically, low rates of unemployment for prime-

age workers coexist with high rates for young cohorts.

Gender is another important demographic factor that determines individuals’ position in the

labor market. In many economies, notably in the developing world, females tend to be far more

vulnerable than males. A review of youth unemployment in 97 countries confirms that more

young women than young men were unemployed in two-thirds of the countries. In a quarter of

these countries, female unemployment was more than 20 per cent higher than male

33

unemployment, In around half of the countries in Latin America and the Caribbean,

unemployment rates for female youth exceeded those for young males by more than 50 per

cent (UN, 2003). The situation is similar in Ethiopia too. In 2005, average unemployment rate

among urban females was about 27.2 percent compared to 13.7 percent among urban males;

and similarly, in rural areas, the rate was about 4.6 percent for females while it is only 0.9

percent for males (MoLSA, 2009). In Addis Ababa, it was 48.6 percent in 1999 and 40.4

percent in 2005 for women while it was 28.3 percent in 1999 and 22.7 percent in 2005 for men

(Tegegn, 2011).

2.4.2. Demand Side Factors

The demand side factors that are supposed to impact unemployment include economic

performance, production technology, and economic policies and regulations that can affect the

labor market demand. Slower economic growth arises from low economic activity and low

investment rates, which are unable to generate enough additional job opportunities. In

theoretical terms, as stated in Bakare (Bakare, 2011), when foreign direct investment and

domestic investment increase, unemployment will be minimized. Gross capital formation

including private domestic investment is expected to have a desirable impact on

unemployment. The greater the gross capital formation and private domestic investment, the

smaller is the level of unemployment. Capacity utilization and gross capital formation are

highly significant and negatively related to unemployment rates both in the short and long run

(Bakare, 2011).

Technological changes and inappropriate policies can explain the slow growth of employment

in Africa. If inappropriate technologies are employed, the employment-creating effects of a rise

in national income can be offset by the employment-saving effects of modern technology. In

his earlier article on urban unemployment in east Africa, Elkan (1970) argues that

inappropriate techniques of production are the result of not only technological factors but also

inappropriate policies. For instance, policies that encourage capital intensive techniques, failure

to give adequate inducements for training of skilled labor, and failure to manage rapid

increases in wages may lead to poor labor absorptive capacity of an economy.

34

In Ethiopia, the post 1991 period is characterized by a move to a market led system that

included the adoption of structural adjustment program and a range of other policy reforms. In

relation to these events, some evidences show that economic growth in Ethiopia following the

structural adjustment (after 1991) was less employment generating than that in the pre reform

period. According to Mulat et al. (2003), the post reform period arc elasticity employment was

-0.23 while it was 1.9 in the pre reform period. This means that as the economy was growing at

a rate of 1percent, employment rate was declining by 0.23 percent in the reform period until

1999. The implication is that the massive improvement in growth performance that the

Ethiopian economy experienced since 1991 had little effect in reducing urban unemployment.

There are some possible explanations that are suggested in relation to this fact. Among the

possible reasons, as stated in EEA (2007), are firstly, there might had been “overstaffing” in

the pre reform period and cutbacks for more efficient use of resources in the post reform

period. Secondly, the incentive structure of the reform period might encourage employers to

choose labor saving technology (EEA, 2007). Two other explanations are also forwarded. The

first one is that the private sector, including self-employment, has not yet overcome the effect

of the repression it had experienced in the pre-1991 period (Krishnan, 2001). The other

explanation is attributable to the fact that the post-1991 growth came dominantly from the

agricultural sector which is weakly linked to the urban sector (Alemayehu, 2005).

In recent years, Africa’s economy has witnessed relatively better performance and rapid

growth with most countries experiencing economic growth above their population growth

rates, thus leading to rises in per capita income. This rapid growth episode had, however,

insignificant impact on employment. For most African countries, unemployment rates

remained almost unchanged even during the recent growth upturn that ended in the second half

of 2008. The rates were estimated to have risen from 7.4 percent to 8.2 percent between 1998

and 2009 in Sub-Saharan Africa and from 12.8 percent to over 13 percent in North Africa in

the same period. Narrow-based economic growth combined with rapid population growth and

labor market imperfections mean that Africa’s growth rates consistently fall behind the growth

rate needed to create adequate employment and reduce poverty (ECA, 2010).

35

Indeed, growth with no employment is not an exception for Africa. The history of fast-growing

countries and their continued inability to cope with the problem of unemployment indicate that

something else besides rapid growth is required for a solution. Africa’s growth has relied

mainly on capital-intensive sectors rather than labor-intensive ones. The nature of growth is as

important as its quantity if Africa is to meet its employment and poverty reduction objectives

In labor abundant economies, as the factor endowment theory suggests, growth must occur by

investing in relatively labor-intensive activities rather than those which are capital-intensive.

The rationale is that not only will this result in more rapid growth because of the low

opportunity cost of labor relative to capital, but will increase the rate of growth of employment

for any given level of investment (Elhiraika, 2011).

Employment growth is a function of the sectoral composition of employment, sectoral growth

rates and the output elasticities of employment in the various sectors. This implies that

employment growth depends on the aggregate growth rate as well as the sectoral composition

of aggregate growth. This is the line of reasoning that Elhiraika’s (2011) explanation for the

poor labor absorptive capacity of Africa’s growth is based on. He contends that the major

source of the recent economic growth in several African economies has been the growth of

natural resource extraction sectors, which by their nature are capital intensive and, with a few

exceptions, have limited linkages to the domestic African economies. Value added in the

mining sector, which employs less than 10 percent of the labor force, grew at over 10 percent

per year, while agriculture, manufacturing and services with combined employment of over 80

percent of the labor force grew at less than 2.5 percent per year in the last two decades. The

combination of small size and low employment elasticities implies that growth based on rapid

expansion of the mining sector will not generate high-employment growth. In turn, this

suggests that a broad based employment strategy will not only have to rely on higher aggregate

growth but must also pay attention to sectoral composition.

In a well-functioning labor market, the demand of labor is inversely related to its price. The

higher the price of labor, the lower is its demand. The price of labor relative to that of other

inputs such as capital can also change the demand for labor by inspiring the more concentrated

use of the relatively cheapest input. In other words, relatively cheap capital will prompt firms

36

to be more capital-intensive, while relatively cheap labor will necessitate more labor-intensity

(Onwioduokit, 2009). In the same way, as Bakare (2011) argues, the level of minimum wage

and wage increases contribute to rising unemployment rates. When the wage rate increases,

there is tendency to substitute machine for labor. When this occurs, it will increase the

unemployment rate implying a positive relationship between wages and unemployment rates.

Labor market institutions that keep an appropriate balance between labor market flexibility and

worker protection can contribute positively to job creation and efficient labor allocation while

simultaneously protecting fundamental rights of workers. But if these institutions are

unbalanced and provide undue protection to certain groups, they may adversely affect labor

market outcomes (Gunatilaka, 2010). The increased labor market inflexibility raises the

indirect cost of labor for firms, since more time and money have to be spent negotiating with

unions, and an increasing amount of time and money is lost due to strikes. High indirect costs

may warrant a substitution of labor with capital, which means that demand for labor will grow

slower than output (Pierluigi, 2008).

In the context of Ethiopia, minimum wage is limited to public sector employment and to some

extent formal private sector employment. The higher wages for public employment leads to

queuing for it. Lack of employment services increase frictional unemployment and results in

long unemployment duration (EEA, 2007). Ethiopia’s labor law framework, outlined for the

private sector by Proclamation No. 377/2003 does provide a series of protections for workers.

However, as argued in WB (2007), labor regulations and labor relations in Ethiopia are not

seen by firms as significant impediments to doing business. This might be largely because

these provisions are not generally enforced outside of the public sector.

Regulations that promote competition in the product market have positive effect on

employment. Lower barrier to entry encourage new firms to enter in to the market and curbs

market power and monopoly profits. As a result, the expansion of economic activities tends to

increase labor demand. Particularly, lower monopoly profits reduce the scope for existing

workers to share in the rents generated by excessive prices. Reduced rent sharing between

employers and employees would then tend to shorten the length of unemployment spells as it

37

would become less attractive for the unemployed to limit their search for job opportunities in

high-wage sectors only (Pierluigi, 2008).

Inflation is among the macroeconomic variables that affect the level of employment through

its impact on economic performance. A reasonable inflation rate stimulates investment and

consequently raises the labor demand. The well known theoretical explanation on the

relationship between unemployment and inflation is attributable to the Phillips curve. There

are two possible explanations on the relationship between unemployment and inflation

depending on the time frame: one in the short term and another in the long term. In the short

term, there is an inverse correlation between unemployment and inflation explained by a

downward sloping curve. Put differently, the short term relation states that when the

unemployment rate is high, inflation is lower and the inverse is true as well, implying a

tradeoff between the two. The Phillips curve in the long term is different from the one in the

short term. As per the classical economics explanation, the long term Phillips curve is

basically vertical as inflation is not meant to have any relationship with unemployment in the

long term. It is therefore assumed that unemployment would stay at a fixed point, commonly

known as the natural rate of unemployment, irrespective of the status of inflation.

However, the empirical evidence on effect of inflation on unemployment seems ambiguous and

inconclusive. For instance, Bakare (2011) finds a negative relationship between inflation and

unemployment in both short and long run periods and is significant at 1% level, which is in

agreement with the Philip’s curve explanation. Similarly, the empirical study by Palley (2005)

in which he compares the European labor market with that of the United states confirm that

permanently lowering the inflation rate by 1 percent point increases unemployment by 0.4

percentage points. In contrast, the findings of Kapsos (2005) indicate that the average annual

rate of inflation is negatively associated with employment elasticity, implying a positive

relationship between inflation and unemployment.

38

2.5. Active Labor Market Policies to Address Unemployment A common way to look at the value of education and training for individuals is, as Becker’s

Human Capital Theory says, in terms of increased human capital based on the assumption that

the greater one’s human capital, the better are one’s labor market chances. Thus, human capital

accumulation from Active Labor Market Policies (ALMP)-training investments is expected to

increase the employability and labor market outcomes of the unemployed (Nordlund, 2010).

The faith in human capital has reshaped the way governments approach the problem of

stimulating growth and productivity, as has been shown by the emphasis on human capital in

both developed and developing countries.

Active labor market policies (ALMPs) are measures intended to improve the functioning of the

labor market that are directed towards the unemployed. The common active labor market

policies, through which governments intervene to deal with the problem of unemployment, can

be categorized in to three: i) labor market training in order to upgrade and adapt the skills of

job applicants; ii) direct job creation, which may take the form of either public-sector

employment or subsidization of private-sector work; and iii) employment services (or job

broking) with the purpose of making the matching process between vacancies and job seekers

more efficient (Boone, 2004, Calmfors, 1994) . The desired effect of ALMPs is a change in the

allocation of the labor force among sectors, skills, and regions. For instance, if there is full

employment among skilled workers, or in certain regions, or sectors, and if wages are flexible,

such programs intended to increase the employability of unskilled workers or workers

employed in regions with high unemployment and wage rigidity have a positive effect on

output and employment (Altavilla, 2006).

Training programs are on the supply side of the labor market aimed at providing job seekers

with marketable skills that potentially increase their employability as well as their earning

capacity. Training involves some form of public support such as direct provision of training,

financial support for trainees, or providing infrastructure services (Sanchez Puerta, 2010).

From the human capital theory point of view, such training programs primarily serve to

enhance the human capital of the participants, which, as a result, will have two desirable effects

39

on participants' labor market outcomes. The first is increased probability of employment, either

by enhancing the attractiveness of participants to potential employers or by enabling them

acquire the necessary skills to establish their own business. The second one is increased

employment earnings of participants resulted from improved productivity.

The role of TVET in furnishing skills required to improve productivity, raise income levels and

improve access to employment opportunities has been widely recognized (Bennell, 1999).

Developments in the last three decades have made the role of TVET more decisive; the

globalization process, technological change, and increased competition due to trade

liberalization necessitates requirements of higher skills and productivity among workers in

both modern sector firms and Micro and Small Enterprises (MSE). Skills development

encompasses a broad range of core skills (entrepreneurial, communication, financial and

leadership) so that individuals are equipped for productive activities and employment

opportunities (wage employment, self-employment and income generation activities). The

Bonn Declaration of October 2004 noted that TVET is the “Master Key” for alleviation of

poverty, promotion of peace, and conservation of the environment, in order to improve the

quality of human life and promote sustainable development (UNESCO, 2004).

In reviewing some empirical works on impact evaluation of training programs, Sanchez

Pauerta argues that although the impacts are not homogeneous and vary across age, gender and

region, the net impacts in Latin America and the Caribbean proved that the employment and

earnings prospects of participants have been improved; particularly the employment impacts

are more significant for women and the youngest (Sanchez Puerta, 2010). Similarly,

Betcherman et al (2007) assessed 49 evaluations of training programs primarily aimed at the

unemployed, of which 10 are from transition countries and 4 are from developing countries. In

the case of transition countries, almost all programs had positive employment impacts. On the

other hand, of the four developing countries evaluations, only one showed any gains in terms

of employment or earnings.

The impact of education, in particular technical and vocational training, on individuals' career

employment prospects is a crucial aspect of the current debate. As Psacharopoulos (1997) put,

“Vocational education and training has been in the past, is today, and will remain in the future

40

one of the hottest debated subjects in all countries of the world”. The persistently high level of

unemployment and the increasing amount of money spent on labor market programs have

brought issues regarding the effects and efficiency of labor market policies into the public

debate (Torp, 1994).

Critics of training for employment creation programs base their assertions on a series of

reasonable arguments. The first is the so-called substitution effect. Under this line of argument,

training may very well increase the chances of an individual to obtain a job; yet the number of

jobs at any moment is a given, determined by other variables, mostly at the macro level. The

implication is that training substitutes one job candidate for another, and often does so at high

costs to the public. Even if the employment rates of trainees increase, as compared to the

comparison groups, the substitution effect remains. In this regard, convincing evidence need to

be produced to differentiate between two independent issues. The first issue concerns

increasing the employability of trainees, i.e., graduates of training programs get more jobs than

they would in the absence of the programme. The second issue deals with the aggregate impact

on employment levels of such programs, i.e., the jobs created add to the total number of jobs,

rather than merely changing the distribution of jobs in favor of those who received training

(Castro, 2000).

Despite the empirical difficulties of substantiating their impact, the arguments for training

programs still make sense. When firms have vacancies or potential vacancies that remain

unfilled due to lack of skills on the part of candidates, training can make a significant

difference. In this case, there is no substitution effect but a net increase in employment. Indeed,

there might be ample evidences of job openings that remain unfilled due to lack of qualified

and suitable candidates, even in the presence of high unemployment. Nevertheless, there is

another question behind such seemingly surplus vacancies. As the conventional

microeconomics suggests, demand is a function of prices. There may be vacancies that remain

unfilled, but at what wage levels? The issue of reservation wage is another issue of concern to

be raised at this point. If sufficiently higher wages are offered, someone will appear with the

required qualifications (Castro, 2000). On the other hand, training can still be justified from

equity perspective. Even if substitution exists, as long as the beneficiaries are the most

41

vulnerable and disadvantaged groups, it may be regarded desirable as it will increase the social

equity of the system.

The most robust argument in favor of skills training is its strong impact on productivity and the

consequent benefits of increased productivity on growth and employment creation. The logic is

straight forward that a well skilled and trained labor force is effective and efficient and

produces more output. Thus, even if training does not increase employment immediately for

the graduates, it remains more than justified in the long run (Castro, 2000). Indeed, the long run

impact argument may provide strong justification for developing countries to invest in

education and training regardless of its controversial immediate and desirable outcomes on the

labor market.

2.6. An Overview of Empirical Evidences on Unemployment in Ethiopia

Despite some improvements in recent years, unemployment and underemployment in Ethiopia

continue to be serious social problems, especially in urban areas and among the youth.

According to the 2005 National Labor Force Survey, the national unemployment rate, based on

the population aged 10 years and above, is estimated at 5 percent of the total labor force. In the

same period, the unemployment rate in urban Ethiopia is estimated at 20.6 percent which is

about eight times higher than the 2.6 percent rates in rural areas ((MoLSA, 2009). Using the

international definition, based on the population aged 15 and above, measured urban

unemployment is still high at 14 percent with distinctive patterns by age cohort, gender and

education. Adult male unemployment fell by one percentage point (from 9.1 to 8.1 percent)

from 1999 to 2005, and stagnated around 13 percent for adult women. The median duration of

unemployment fell considerably, from 24 months in 1999 to 10 months in 2005, providing very

encouraging evidence of dynamism. Despite decreasing duration, the persistence of high urban

unemployment remains a major policy challenge (WB, 2007).

Both supply and demand side factors are responsible for the problem. The pressure on the labor

market primarily comes from the supply of labor, which is induced by the rapidly growing

population. On top of the high growth rate of the labor force, low productivity and low skills of

42

the working poor contribute to the high incidence of both poverty and unemployment. On the

other hand, the insufficient employment generation capacity of the modern industrial sector of

the economy is among the demand side factors for the persistent urban unemployment

(MoLSA, 2009).

Among the demographic factors, the rapidly increasing labor supply, which is incompatible

with the economic performance of the urban sector, is the most important reason behind the

persistent unemployment in urban Ethiopia. Although it is not the most important factor, rural-

urban migration does have a role in the excessively high level of youth unemployment in urban

areas (Getinet, 2003). The coefficients of migration status are statistically significant and

negatively related to the probability of unemployment in both the 1999 and 2005 data sets,

implying that a migrant is less likely to be unemployed than a non-migrant (Tegegn, 2011).

Age is also an important factor that is negatively related to the probability of unemployment.

Many empirical evidences also confirm the same. Age is statistically significant and negatively

related to the probability of unemployment (Tegegn, 2011); and for each 1-year increase in

age, there is about a 5.5 percent decrease in unemployment duration (Seife, 2006). In contrast,

Serneels (2007) found that age has strong positive effect on duration of unemployment among

young men aged 15 – 30. In terms of gender, females disproportionately suffer from

unemployment. As indicated in Guracello, Lyon and Rosati (2008a), the probability of a girl

being in employment is about 14 to 22 percent lower than that of a boy. Also in (Tegegn,

2011), a male worker is about 21.4 percent and 17.7 percent less likely to be unemployed than

a female in 1999 and 2005, respectively. However, Seife (2006) finds no variation in the

duration of unemployment by gender.

Previous studies also show that unemployment in urban Ethiopia does vary by level of

education and training status. According to (Tegegn, 2011), all levels of education, except for

first degree and above, are positively related to the probability of unemployment in the 1999

data set. A person with only primary education is 10.5 percent and with secondary education is

20.6 percent more likely to be unemployed than an illiterate person in 1999. However, in the

2005 data set all coefficients of the education dummies show negative signs and statistically

significant, except secondary level education. Training has desirable effect on unemployment

43

and statistically significant. A person who received some sort of training is 10.9 percent and

8.7 percent less likely to be unemployed in 1999 and 2005, respectively compared with a

person who did not. Similarly, the finding of Seife (2006) confirms “very high returns to higher

education, at least in terms of the probability of getting employment” (pp. 193). People with

vocational, college or university education have higher exit rates from unemployment than

secondary school graduates. Although less significant, the coefficients of primary education

imply shorter unemployment durations than secondary education. On the other hand, the

findings of Guracello, Lyon and Rosati (2008a) indicate that the probability of employment

decreases as the level of education increases, implying a positive relationship between

unemployment and level of education. Also as indicated in Serneels (2007), the probability of

being unemployed increases with education up to senior secondary level. Longer duration of

unemployment is associated with junior secondary education while it is shorter for senior

secondary education.

2.7. Policy Responses to Address Unemployment in Ethiopia

There have been a number of policy responses since the early 1970s introduced to create

employment and increase employability in Ethiopia. Nevertheless, for the purpose of this

study, among the various policy interventions made in recent years, only two - one from the

supply side and the other one from the demand side - are chosen for discussion. The supply

side policy response focuses on the expansion of Technical and Vocational Education and

Training (TVET) programs and the demand side policy response focuses on the development

of Micro and Small Scale Enterprises (MSEs). Expansion of TVET programs and

development of MSEs are among the major development strategies and policy priorities of the

current government of Ethiopia. Indeed, as a large body of the literature argues, such policies

are subcategories of ALMPs often said to have significant impact on employment opportunities

of working age population in general and of women and the youth in particular.

2.7.1. Expansion of Technical and Vocational Education and Training Programs

Expansion of education and training is among the active labor market policies that

governments adopt to increase the employability of the labor force. TVET is generally a kind

44

of education and training program mainly aimed at leading participants to acquire practical

skills, knowhow and understanding, and necessary for employment in a particular occupation,

trade or group of occupations (Atchoerena, 2002). The multidisciplinary nature of TVET and

its supposedly close links to the world of work make it one of the education sectors that

contributes greatly to the training of skilled labor (Rena, 2006). In many sub-Saharan African

countries, TVET is going through a stage of transition and reorientation in the region, as efforts

are being made to give students some basic skills and knowledge, as well as the tools they need

to play an active role in the production system (Atchoerena, 2002)

In 1994, a new education and training policy has been carried out considering the drawbacks of

the so far educational systems of Ethiopia. The new policy has given emphasis to education

and training that offer specific learning skills related to the specific needs or gainfully tradable

skills based on demand driven and in response to the country’s development approach. The

government considered Technical and Vocational Education and Training (TVET) as an

instrument for producing medium level technicians equipped with practical knowledge who

can create job rather than expecting employment opportunities to be offered by public. Also as

noted in the new GTP document, greater emphasis is given to TVET institutions so as to make

them serve as centers for technology transfer and accumulation for MSEs as well as to provide

employable skill to the youth.

TVET is expected to play a key role in this strategy by building the required motivated and

competent workforce. In the early strategic plan document, PASDEP, TVET is envisaged to

provide the necessary “relevant and demand-driven education and training that corresponds to

the needs of economic and social sectors for employment and self-employment” (MoE, 2008).

In reaction to the reform in the educational sector, enrollment has increased considerably in all

levels including TVET and higher education. Particularly, Technical and Vocational Education

and Training (TVET) enrolment reached 371,347 in 2010/11, showing a 5.1 percent increase

relative to 2009/10 and a 94.3 percent compared to 2006/07. In the same way, the number of

TVET institutions increased to 505 in 2010/11 from 388 in 2006/07 (NBE, 2011).

45

2.7.2. Micro and Small Scale Enterprises (MSEs) Development

The informal sector accounts for the majority of employment in Ethiopia. According to the

2005 LFS, it represented 71 percent of urban employment overall and 81 percent of youth

employment. Several sectors are almost exclusively informal. These include domestic work,

wholesale and retail trade, hotels and restaurants, and primary production. Overall,

manufacturing accounts for about 45 percent and trade/ hotels/ restaurants for about 38 percent

of informal firms (WB, 2007). Due to the growing labor supply and limited formal

employment opportunities, there is a lot of interest in building the capacity of the informal

economy that employs a significant portion of the labor force. On the other hand, small and

medium enterprises (SMEs) comprise the largest share of enterprises and employment in the

non-agricultural sector in Ethiopia. Therefore, SMEs have been a special focus of the

government and the promotion and development of SMEs was emphasized as one of the most

effective means for achieving faster development and creating job opportunities, especially for

women and the youth (MoLSA, 2009).

The Federal Micro and Small Enterprise Development Strategy Agency (FeMSEDA) oversees

the promotion of micro and small enterprises development, while the direct support and

promotional activities are carried out by institutions established at the Regional States

(ReMSEDA). According to Birhanu, Abraham, and van der Dejil (2005), the Regional

Governments have been promoting MSEs by providing training and counseling, finance and

credit facilities, organizational support, production and marketing space, market facilities and

raw material supplies. The results of the support provided to MSEs have been encouraging.

About 72,577 new jobs were created in micro and small enterprises, nearly 63 per cent in

Addis Ababa, in 2004.

MSEs constitute nearly 90 per cent of industrial employment. In recognition of their role the

Ministry of Trade and Industry (MoTI) formulated the micro and small enterprises

development strategy in 2004 with a major objective of creating long-term employment

opportunities. The strategy gives priority to enterprises operated by women. It also favors

enterprises operated by school dropouts, people with disabilities, and previously unemployed

46

youth. It outlines key limitations faced by micro- and small enterprises and sets out the goal of

providing comprehensive support (Guarcello, 2008a).

Many people are earning their livelihood from MSEs. However, these sectors are performing

below capacities and their growth has been severely constrained by many factors. Lack of

purchasing power of local people, lack of access to financial services, poor partnership and

networking, and inadequate entrepreneurial capacity are among the bottlenecks of the sector

(Mulat, 2006). Although measures are being taken to support them, most of the challenges that

MSEs face are yet to be tackled. Some of these challenges include: (i) unfavorable legal and

regulatory environments and, in some cases, discriminatory regulatory practices; (ii) lack of

access to markets, finance, business information; (iii) lack of business premises (at affordable

rent); (iv) low ability to acquire skills and managerial expertise; (v) Low access to appropriate

technology; and (vi) Poor access to quality business infrastructure (MoLSA, 2009).

Access to credit is one of the major constraints for the expansion and growth of MSEs mainly

due to the collateral requirements by commercial banks. In Ethiopia, beyond appreciating the

problem, the government’s effort to help MSEs get access to credit services through

microfinance institutes has been considerable. For instance, in 2004 Addis Ababa, Amhara,

Tigray, Oromia, South, and Dire Dawa Regions together provided more than 110 million ETB

loan to small and micro enterprises in 2004. Addis Ababa and Tigray States extended more

than Birr 44 million and Birr 33 million, respectively in the same period (Birhanu, 2005). This

being the case, a study by Rahel and Paul (2010) on the determinants of employment expansion

of women operated MSEs in Nifas Silk-Lafto and Kirkos sub cities of Addis Ababa suggests

that the problem is still a concern. According to them, although there are some efforts to help

women get access to credit, women in the survey area reported that the loan they received is

not enough to expand their enterprises.

Raw material is a fundamental component of inputs for the existence of an enterprise. The

types of raw materials demanded by MSEs and their sources vary with the nature of the

enterprises. The sources of raw materials for most MSEs are agricultural and/or industrial

products, which are domestically produced or imported. As noted in Rahel and Paul (2010), the

47

fact that the agricultural sector is the main source of raw materials for most MSEs signifies the

potential market and strong backward linkage that the MSEs create for the agricultural sector.

The higher cost of acquiring raw materials is reported to be the key problem pertaining to the

growth of enterprises in the survey area.

Working Premise is another important factor needed for a successful and sustainable growth of

enterprises. In this regard, the Regional States have made encouraging effort by preparing and

arranging working spaces for a number of enterprises. As noted in Birhanu, Abraham, and van der

Dejil (2005), the six Regional States, namely Addis Ababa, Amhara, Tigray, Oromia, South, and

Dire Dawa Regions supplied a total of 1,045,717 m2 of working space to micro and small

enterprises, as of 2004; and more than 62,417 operators of MSEs have benefited from such

arrangements. However, Rahel and Paul (2010) argue that lack of enough working space is still

the main problem for about 54 percent of women-operated MSEs in the survey area (Nifas

Silk-Lafto and Kirkos sub cities of Addis Ababa). At the same time, they pointed out that about

23 percent of the respondents do not face any problems related to the working place.

The five-year Growth and Transformation Plan (GTP, 2011-15) ambitiously targets to create a

total of three million micro and small-scale enterprises (MSE’s) at the end of the plan period.

The development of this sector is believed to be the major source of employment and income

generation for a wider group of the society in general and urban youth in particular. The major

objective of this program, which is creating and promoting MSEs in urban areas, envisages

primarily reducing urban unemployment rate. The government has continued and strengthened

its effort to support and promote MSEs. In recognition to their significant role in the total

economy, the greatest attention given to the development of MSEs as one of the priority policy

agendas can be an indication of the government’s commitment.

48

Table: 2.1 Number of establishments and jobs created and amount of loan

2008/ 09 2009/ 10 2010/ 11 No. of MSEs 73,062 176,543 51,983 Total Employment 530,417 666,192 541,883 Amount of loan (in millions of ETB)

662.7 814.1 983

Source: NBE, 2009/10 and 2010/11 Annual Reports

According to the annual report of the National Bank of Ethiopia, the total amount of loan

supplied to MSEs from micro finance institutions has been increasing over time. The amount

of loan supplied rose by 22.8 percent in 2009/10 relative to 2008/ 09; and consistently showed

a 20.7 percent increase in 2010/11 relative to 2009/10 fiscal year. With regard to

establishments and job creation, in 2009/10, a total of 176,543 MSEs were established

employing 666,192 people, which is 25.6 and 141.6 percent, respectively, higher compared to

2008/ 09. However, in 2010/11, the number of establishments and total employment

considerably went down by 70.6 percent and 18.7 percent respectively, compared to 2009/10

(NBE (2010); NBE (2011)).

49

3. METHODOLOGY This section presents a discussion of the specific steps used in conducting the research. It

provides information on research methodology, data sources, sampling techniques, data

collection instruments, methods of data analysis and specification of econometric models.

3.1. Research Method Obviously, any one approach by itself is not complete and perfect. As a research culture,

depending on the nature of data and the research problem, using both quantitative and

qualitative approaches can enhance the validity of the research output (Creswell, 2009). As it is

common in the unemployment literature and related empirical studies such as Serneels (2007)

and Seife (2006), a quantitative approach is frequently employed. Accordingly, given our data

type and research objectives, we mainly adopted a quantitative research method.

3.2. Data Sources We have used both primary and secondary data to see the trends of unemployment and to

examine socioeconomic causes of unemployment and effects of policy interventions. The

primary data were collected from three major urban areas, namely, Addis Ababa, Bahir Dar

and Hawassa. The data included information on labor market outcomes of individuals,

characteristics of MSEs and their employment generation capacity, and opinion of stakeholders

on the effectiveness of the TVET program.

The secondary data were taken from Urban Employment Unemployment Surveys (2003 to

2011) and National Labor Force Surveys (1999 and 2005) conducted by the Central Statistical

Agency (CSA) of Ethiopia. We constructed data set known as pooled cross-section from the

UEUS data collected for five years. The labor force surveys and urban employment

unemployment surveys consist of important variables to the study such as demographic

variables, educational qualification, employment status, training received and training type,

migration, duration of unemployment, hours of work in a week and other important variables.

50

In addition, to see the patterns and relationships between unemployment and some important

macroeconomic variables, we made use of the World Bank dataset.

3.3. Sampling Techniques and Procedures

The primary data collection was begun with identification of the study sites. We employed

purposive sampling to identify the three main urban areas (viz. Addis Ababa, Bahir Dar, and

Hawassa). Some of the features considered in selecting the three representative urban centers

and the respective sample sizes include the geographical location, the number of Technical and

Vocational Training institutions and size of TVET graduates, the level of unemployment rate

and the concentration of MSEs. Since the distribution of TVET institutions and MSEs are more

concentrated in Addis Ababa, a larger sample size was drawn from Addis Ababa relative to

Bahir Dar or Hawassa.

Taking into account the budget and time constraints, we employed some other sampling

strategies to finally arrive at the selection of representative sampling units. The desired

sampling units for the survey are an individual in the labor force aged between 15-64 years and

an enterprise in the MSEs category. In doing so, we first took list of sub cities and woredas in

each of the three cities. In collaboration with the concerned department of the respective city

administration, we stratified each of the three cities in to three strata, i.e., residential, business

and industrial zones. In Addis Ababa, one sub city per a stratum and from each sub city two

woredas and thus a total of six woredas were randomly selected. In the case of Bahir Dar and

Hawassa, one woreda (i.e. local administrative hierarchy one layer higher than keble) per a

stratum and a total of three woredas from each of the two cities were randomly selected. Again,

one kebele (the lowest local administrative tier in Ethiopia) from each woreda was randomly

selected, implying a total of six kebeles in Addis Ababa and three kebeles in each of the two

regional cities, Bahir Dar and Hawassa. Hence, these randomly selected twelve kebeles are the

specific survey areas where from the primary data were collected.

To select the target sample population in each randomly selected kebele, it was mandatory to

have a source list. For the MSEs survey, we made use of the list of MSEs obtained at each

selected woreda. For the survey of individual persons in the labor force, however, there had not

51

been a readily available list of the population in the desired format. Therefore, the research

team carried out registration of all economically active population in each selected kebele to

produce a source list.

Accordingly, data on labor market outcomes were gathered from 45 persons from each of the

three kebeles in Hawassa, four kebeles in Addis Ababa and one kebele in Bahir Dar as well as

54 and 52 persons, respectively from each of the remaining two kebeles in Addis Ababa and

Bahir Dar. Thus, in relation to labor market outcomes, 135 from Hawassa, 149 from Bahir Dar,

297 from Addis Ababa, and a total of 581 persons were interviewed. Information on MSEs

were collected from 36 randomly selected MSEs from each of the three kebeles in Bahir Dar

and Hawassa. In Addis Ababa, 36 MSEs from each of 5 kebeles as well as 49 MSEs from the

remaining one kebele were interviewed. Therefore, regarding the employment effect of MSEs,

a total of 445 MSEs were involved.

3.4. Data Collection Instruments To facilitate the data collection, eight interviewers and one supervisor per kebele and a total of

nine enumerators, with educational qualification of at least diploma and who reside in the

sample kebeles, were employed. The enumerators were given a half day training on the study

objective, how to manage the questionnaire, how to friendly communicate with the

interviewees, research ethics, and quality issues. Close monitoring and regular supervision

were among the strategies followed to correct errors on time and ensure the reliability of the

data.

Structured survey questionnaires were used to collect the primary data. Questionnaire-1 for

individual persons and Questionnaire-2 for owners of MSEs, were administered to the selected

interviewees via the trained interviewers. In addition, Questionnaire-3, a kind of Likert scale

was distributed to relevant and concerned groups such as TVET graduates, TVET instructors,

directors of TVET colleges, employers, public officials at regional TVET Bureau and parents

to assess their opinions on the relevance and strategic role of the TVET program in reducing

unemployment.

52

3.2. Data Analysis The available quantitative data have been analyzed using descriptive analysis to describe the

characteristics and trend of urban unemployment and econometric analysis to determine

socioeconomic causes of urban unemployment and to examine effect of policy interventions-

through TVET and MSEs- on unemployment. To this end, the data sets deployed were cross-

sectional, pooled cross-sectional. The econometric analysis consists of pooled cross-section

data analysis using probit and duration models, and logistic regression.

3.2.1. Pooled Cross-sectional Data Analysis

We used the pooled cross sectional data which is obtained by sampling randomly from two or

more points in time; for example, in our case UEUS was conducted for five periods. Therefore,

independently pooled cross-section is obtained by sampling randomly from a large population

at different points in time. It differs from a single random sample in that sampling from the

population at different points in time likely leads to observations that are not identically

distributed. For example, distributions of wages and unemployment have changed over time in

most countries (Wooldridge, 2000)

A standard regression model applied to a set of pooled cross-sectional and time series data take

the form of NT equations can be written as:

.sec)()(:

)1.3(,....,3,2,1,...,

,,...,1,...,,3,2,1...

21

211

111413122110

usedaredatationalcrosspooledwhenTperiodsdifferentforNnsobservatioofnumberdifferenthaveWeNote

TtNNN

NNNiUXXXXXY

TperiodT

Periodperioditkitititititit

−−−−−−−−−−−−−−−−−−−−−=+

++=+++++++=

44 344 21

44 344 214434421ββββββ

Suppose the true error structure does include a year component, but we ignore that and run

OLS on equation (2.1).This introduces serial correlation between observations within the same

time period and violates one of our assumptions for OLS. The OLS coefficient estimate is still

unbiased and consistent. However, the variance-covariance matrix is biased/ inconsistent which

53

leads to incorrect standard errors and incorrect inferences. This is similar to the problem of

serial correlation in time series models. Thus using pooled cross sections raises only minor

statistical complications. Typically, to reflect the fact that the population may have different

distributions in different time periods, we allow the intercept to differ across periods, years in

our case. This is easily accomplished by including dummy variables for all but one year, where

the earliest year in the sample is usually chosen as the base year. That is 2003 for

unemployment employment data and 1999 for labor force survey in our study.

Thus it makes sense to include time dummies (also known as year effects):

)2.3(...2

1413121110 −−−−−−−−−++++++++= ∑=

t

T

ttkitititititit XXXXXY εθββββββ

Where:

Yit, refers to the dependent variable for case i in period t

β0, the intercept or the base year and β1, β2, … , βk are slopes

θt, parameter for time dummies and t=2, 3, …..,T is time

εit, error term, εit/X~ (0, δε2) or conditional errors, are usually assumed to be

independent and identically distributed random variables with a mean of zero, a variance of δε2

that is constant across values of the Xs, and, in small samples, a normal distribution.

Each time dummy is the difference in the conditional expected value of dependent variable

between the base year (t=1) and the year t=T

Pooled cross sectional data has some advantages over other data categories (such as cross

sectional data). Pooling can lead to larger sample sizes. This leads to more precise estimators

and test statistics with more power. However, this is only true if the relationship between the

dependent variable and at least some of the explanatory variables remain constant over time

(Wooldridge, 2002).

If the explanatory variables are changing over time, it can also provide additional variation in

explanatory variables with which to estimate its effect on dependent variable. Alternatively, a

further benefit of these data is that we can explore changes in the coefficients over time.

54

In a two-period pooled cross section dataset with k explanatory variables expressed as:

021:

)3.3(...2

4433222110

otherwiseyearfromcomesnobservatiothetheiftakesdummyyeartheisDwhere

XXXDXXXY t

T

ttkitkititittititit −−−−−−−−−+++++++++= ∑

=

εθβββββββ

While changes in the coefficients may be interesting, one has to be very cautious in interpreting

the source of the changes (e.g., as the impact of a policy or changing economic structure).

We used pooled probit model to examine the impact of training and education on urban

unemployment and to determine trends of unemployment because the dependent variable is

dummy. Either logistic or probit regression can be used depending on the distribution assumed

for errors. If the standard normal distribution is assumed, probit analysis would be used. If

instead we assume a logistic distribution for εt (also a distribution symmetric about 0, but with

a variance of П2/3 instead of 1, where in this case П is the constant 3.14159), the analysis

becomes a logistic regression.

The pooled probit model below represents unemployment status where ∗itY denotes the

dependent variable, dummy for probability of unemployment. The dependent variable,

probability of unemployment is dichotomous such that: 1=∗itY if the worker comes from year t

is unemployed and 0=∗itY otherwise. We are interested in the probability that the labor is

unemployed, itit XYp /1( =∗ Where itX is used to denote the full set of explanatory variables

that come from year t, the factors that increase or decrease probability of involuntary

unemployment and Uit denotes the error term.

The general pooled probit regression model for the probability of being unemployed, using k

regressor variables can be described as:

55

)4.3(...81180680411

...04...12....8exp5.....1

1510

5432

0769

873210*

−−−++++•+•+•+

+++++++++++++=

it

it

somalitigryeducyeducyeducyy

yTVETeduceducermarmaragesexY

εββφφφφφβββ

ββββββ

In the above equation, the variable y11 is a dummy variable equal to one if the observation

comes from 2011 and zero if it comes from any other year. The base year is 2003. To show the

trend of unemployment, we analyze whether the coefficients on the year dummy variables

show a significant change or not in unemployment in the 2011.

The intercept for 2003 is 0β , and the intercept for 2011 is 20 φβ + . Whereas 9β is the discrete

effect of upper primary education on probability of unemployment in 2003 while 34 φβ + is the

discrete effect of upper primary education (educ8) on probability of unemployment in 2011.

Therefore, 5,43 , φφφ and measures how the probability of unemployment to another year of

education has changed over the one, three and eight year period.

The pooled cross-section data analysis is also used for cox-regression. This regression employs

proportional hazard models. The hazard rate for failure at time t is defined as [[

)()(Pr

)(ttimeafterfaillingofprobablityt

ttandttimesbetweenfailingofobablitytH

ΔΔ+

=

We model this hazard as a function the baseline hazard )(0 tH at time t and the effect of one or

more explanatory or X variables. Baseline hazard means the hazard for an observation while all

X variables equal to zero.

)...(exp)()( 443322110 kk XXXXXtHtH βββββ ++++++=

Or equivalently

56

[

)5.3....(...])(ln[)](ln[ 443322110 kk XXXXXtHtH βββββ ++++++=

)( tH is a survival time data contain, at a minimum, one variable measuring how much time

elapsed before certain event occurred to each observation. The literature often terms this event

of interest a “failure” a regardless of its substantive meaning. When a failure has not occurred

to an observation by the time when data collection ends, that observation is said to be

“censored”. Duration of unemployment (duration of weeks elapsed) before an individual get

employed or study ended is a time variable. Failure refers to a situation where a person is

employed before the end of the survey period. Censored is when he/she remains unemployed

during the survey period.

For all of the regression analysis we estimated, a log likelihood chi-square is an omnibus test

whether or not the model as a whole is significant. In our finding this test indicates that the

overall model is significant at below 1 percent. This means that the model that includes all the

independent variables included, for example; in annex table 4.22 including the constant fits the

data statistically significantly better than the model without at least one of these variables (or

i.e., a model with only the constant). We have used the link test whether or not the model is

correctly specified; one should be not able to find any additional predictors that are significant

except by chance. After the regression command link test uses the predicted value and the

predicted value squared as the predictors to rebuild the model. The predicted value should be a

significant predictor since it is the predicted value from the model. This will be the case unless

the model is completely miss-specified. On the other hand, if our model is properly specified,

the variable predicted value squared should not have much predictive power other than by

chance. Therefore if predicted value squared is significant, then the link test is significant. That

means either we have omitted relevant variable(s) or our link function is not correctly

specified. The link test result in the case of this study found to be not statistically significant.

To handle, the problems of heteroskedasticity, we estimated robust standard errors.

Transformation of the variables is the best remedy for multicollinearity when it works. We also

used logarithmic transformation of variables with the problem of strong collinearity such as

age. However variables generated through squaring other variables such as age and work

57

experience cannot be freed from multicollinearity problem. So we removed two variables

experience squared and age squared from our model.

3.2.2 Specification of Study Variables

The specification of study variables included in the regression analysis is summarized in a table

and located in the annex (Table 3.1).

4. RESULTS AND DISCUSSION

4.1. Demographic Characteristics of Respondents

The total sample size of respondents from secondary data for the five years is about 142, 547.

It is fairly distributed across time; nearly 18.9, 16.8, 18, 23, and 23 percent of the respondents

are from 2003, 2004, 2006, 2010, and 2011, respectively. The sex wise distribution is

reasonable; for instance, the proportion of males is estimated about 51.4 percent of the total

sample while that of female is 48.6 percent. Regarding the age wise distribution, adult women

and men account for nearly 28 and 21 percent while youth females and males are about 24 and

27 percent, respectively of the total sample. The literacy rate of urban labor force increased

from 76 percent in 2003 to 80 percent in 2011 and relatively males are more likely literate than

females. Literacy rate is relatively higher for youth than adults, but increasing overtime for

both categories. With regard to marital status, about 49 percent of respondents are found to be

married.

The sample size of respondents in the primary survey is 582 with a sex wise distribution of

fifty-fifty. Age wise adults account for 45 percent of the total sample size while the share of

youth is 55 percent. The share of youth female is 30 percent which makes the largest size of the

sample while the least 20 percent of total sample comes from women. The literacy rate of the

males is around 97 percent and that of females is about 89 percent. Almost half of the

58

respondents, about 47 and 55 percent of females and males respectively received training. Over

48 percent of the respondents are married and nearly 45 percent are single.

4.2. The Urban Labor Force Participation Trends

Growth of labor force is engine for economic growth when the labor force is gainfully

employed; otherwise it contributes more to social unrest as many working age people remain

unemployed, underemployed or work in the informal sector as working poor. The secondary

data shows that the trend of overall participation rate was decreasing between 2003 and 2006;

but thereafter, it has been slightly increasing (Figure 4.1). Male youth as defined in Ethiopia

youth policy (15-29 years old) expands however the rate of growth lacks uniformity. The

erratic trend of participation maintained for female youth, female and men for the period.

Marginal decline in participation rates for adult male categories of the labor force as compared

with female and female youth. On the other hand, Male youth participation showed

pronounced growth relative to other groups between 2003 and 2011. On the other hand, female

youth labor involvement decreased between 2003 and 2011 at least by one percentage point

while women and men rates are almost constant.

Figure 4.1: urban labor force participation rate (%)

0

20

40

60

80

100

2003 2004 2006 2010 2011

youth female 

youth male 

adult male 

female total 

total

Source: UEUS 2003-2011

The male youth labor force as the share of total increases by 4 percentage points from 2003 to

2011, allowing the overall participation rate to rise at least by 2 percentage points in the period.

59

The rising trend of youth participation increases with age. The participation rate of male youth

age between 15-19 years over the period decreased from nearly 23-22 percent and by far below

the rate of youth age between 20-24 years old that increased from nearly 71-72 percent and

also adult youth participation rate increased by one percentage point between 2003 and 2011.

The fall in participation rate of teenagers can be taken as positive, if enrollment for the age

increased and responded by better job offer after successful completion of ongoing education

or training.

The quality of urban labor supply has been increasing although the rate of its growth is not

attractive (Figure 4.2). By 2003, 87 percent of the urban labor force participants had at least

five years of schooling (above lower primary education), and the size of labor force with the

same level of schooling has increased to 90 percent in 2011. Similarly, the labor force with

more than eight year of schooling increased from 62 to 65 percent over 2003 to 2011. Figure 4.2: The trend of labor supply by years of schooling (%)

0102030405060708090

100

2003 2004 2006 2010 2011

above grade4

above grade8

above grade10

Source: UEUS 2003-11

Labor force size above grade ten educational attainment increased from 37 percent in 2003 to

50 percent in 2011.

4.3. Urban Versus Rural Unemployment

According CSA labor force survey, most urban dwellers aged between 10 and 64 years in

Ethiopian found hard to get job and faced peak unemployment rate of 26.4 percent in 1999

60

using current status approach, which is five times higher than rural areas (5.1 percent). The rate

is decreased to 20.6 percent in 2005 nevertheless the figure is still considerably above 2.6

percent rate of unemployment for rural population. The rural unemployment is almost threefold

lower than 8 percent rate of unemployment observed for sub-Saharan Africa age groups

between 15 and 64 for the period. The implication is that urban unemployment rate is not only

significantly higher than national unemployment rate but also continental wide average

unemployment rate between 1999 and 2005.

The rate of unemployment also varies throughout sex and age. According to CSA figures, 30

percent of unemployed in the period are women whereas the rate is 18.3 percent for male

workers. Women unemployment rate is higher than male rate for the period 2005 as well. For

instance, the rate is 27.2 and 13.7 percent for females and males respectively. Therefore,

unemployment rates of females are greater than that of males for all age cohorts between1999

and 2005.

4.4. Urban Employment Trends

Urban employment showed an increasing trend over the last eight years; however, the rate of

growth is slow relative to the high rate of urban unemployment. The employed labor force

increased from 74 percent in 2003 to 82 percent in 2011 (Figure 4.3). The growth of

employment varies across gender and age. The employment rate of female workers and young

workers increased by 9 and 12 percentage points, respectively while that of adult males grew

by only 3 percent over the period. However, the employment rates of females and the youth are

still significantly less than that of adult males, signifying their disadvantaged position in the

labor market. Despite the information gap about the quality of employment, the existing

information indicates that between 2003 and 2004 above 64 percent of the employed were

working in the informal sector. For the remaining periods, formal sector employment is at least

above 55 percent for each year and informal employment is decreasing over the time except

marginal increase for the period 2011.

Figure 4.3: Urban employment trends (%)

61

Source: UEUS 2003-11

When we compare urban employment and unemployment trend urban employment rate is the

mirror image of the total urban unemployment. As indicated in figure 4.3 when the rate of

employment is peak unemployment rate is at its lowest level and vice versa.

Evidences from three cities using primary data strongly oppose the findings from CSA survey

for five years. Female employment rate is as high as twice as male employment rate. Both

women and youth female employment rates are higher than adult male and youth male

employment rates. For instance employment rate of women and youth female are 84.87 and

76.7 percent respectively where as men and youth employment rates are 50.7 and 34 percent

respectively.

4.5. Urban Employment-to-Population Ratio

Despite the fact that employment-to-population ratio indicator is good measure of quantity than

quality of employment, it indicates the capacity of an economy to create job. Commonly, the

range of the indicator is between 50 and 75 percent, with a higher ratio implies that majority of

the population that could be working does work-involved in income generating activities (ILO,

2009b).

0

20

40

60

80

100

2003 2004 2006 2010 2011

year 

 rat (%)

total employment

adult male 

total female

total youth

youth female

total unemployment

62

Table 4.4: Urban employment-to-population ratio (%)

Source: UEUS 2003-11

The total urban employment-to-population ratio in Ethiopia increased from 51 percent in 2003

to 58 percent in 2011 but with significant variations across gender and age (Figure 4.4). It is 40

percent or less for youth females, wile the ratio it is over 80 percent for adult males in the same

period. The implication is that although the strong economic growth since 2003 has generated

more employment opportunity to the entire working age population, it has still been in favor of

the adult male category of the labor force.

4.6. Urban Unemployment Trends

In general, Ethiopia’s urban unemployment rate has been trending downward since 2003

(Figure 4.5), but remained at high level. The composite unemployment rate reached peak of

nearly 26 percent in 2003, and decreased to its lowest level 17 percent in 2006. This might be,

at least partly, attributed to the falling trend in participation rate observed during the same

period. Despite the sound economic growth, the urban unemployment rate is still higher and

stood around 18 percent in 2011. This translates to an average unemployment rate of about 21

percent for almost a decade and a reduction of 8 percentage points between 2003 and 2011.

Despite their lower participation rate and the decrease in unemployment rate over 2003 to 2006

unemployment is concentrated among youth and young females continued to be hardest hit by

unemployment in the period mentioned.

0

20

40

60

80

100

2003 2004 2006 2010 2011

year

ratio

total ratio

adult male 

total female

total yout

youth female 

63

Figure 4.5: urban unemployment rate (%)

5

15

25

35

45

2003 2004 2006 2010 2011

year

rate

adult men

 all femal 

all youth

youth female

total 

Source: UEUS

The youth and female unemployment rates are considerably higher than adult male rate and

well above the total urban unemployment in each period, which is consistent with the global

experience, and reflecting the relative disadvantaged position of youth and women in job

markets. The gender bias in unemployment is maintained across all cohorts in urban job

markets of Ethiopia. Women and young workers age between 15 and 29 years old (i.e. national

age standard for young workers) are more likely unemployed than adult male workers

consistent with global unemployment incident. The unemployment rate of the youth was more

than triple that of adult males between 2003 and 2006; and the gap increased to fourfold

between 2010 and 2011. Likewise, women unemployment rate was as triple as adult male

unemployment rate between 2003 and 2011 except for the initial period. The situation is the

worst for young females labor force. Nearly one out of every three unemployed persons (42

percent) are young female in 2011 where as one out of six unemployed persons are adult men

in the same year.

Unemployment is more challenge to youth and female workers than adult male for several

reasons. According to ILO (2010), job interruption is high among women due to maternity

leave; low educational qualification of women relative to their male counterparts expose them

to labor market discrimination and prejudice by employers. However our finding indicates that

adult male and female have almost equal job interruption rates. For example, nearly 76 percent

of adult women on average working on their main job daily and 77 percent of adult men

64

engaged in their main job each day between 2010 and 2011. Nevertheless obviously males

have more educational qualification than females and a labor market prejudice adversely

affects women employment opportunities.

Youth are more likely remain unemployed relative to adult male and consequently suffer more

from unemployment costs. On the supply side, youth extend job search until they secure better

job when they are lucky for family support during spell of unemployment. Family support in

2005 is substitute for unemployment benefit for 75 percent of unemployed youth in Ethiopia

whereas it is means of access to basic needs for 38 percent of adult male in urban areas hence

many youth are more likely to remain unemployed than adult males. The primary data from

three cities indicates that family income is source financial support during unemployment spell

for 72 percent of youth while 2 percent of adult depend on family support during spell of

unemployment.

Furthermore youth lack labor market information and job search experience. In many

developing economies, informal job search methods such as through family, relatives and

friends are dominant methods to find job for youth(ILO, 2010). However this does not hold for

urban higher youth unemployment rate relative to adults in Ethiopia. Annual average of adults

seeks assistance of relatives and friends to seek job are 27 percent and while youth use these

informal job search methods are 22 percent. Young workers who use formal job search options

like search through vacancy advertizing boards, media, direct application to employers and

hold unemployment card are 55.9 percent whereas annually average of male adults use these

options are 38.6 percent.

Youth also switch between job, school enrollment and unemployment as educational

institutions open and close leads to young students more likely enter and exit the labor force.

Not only supply-driven causes but also labor market partiality cause youth are more probably

unemployed than adults. Youth have less work experience, lower job specific training and

firms incur low investment cost to train youth relative to adult. Employment protection

legislation usually requires a minimum duration of employment before it operates and

reparation for job loss often elevate with tenure. The implication is that youth are more

65

exposed to redundancy than adult. According to the primary survey result from three cites,

average work experience of adult estimated about 8 years was significantly above youth labor

market experience of 2 years in Ethiopia.

However, the primary survey result from three cities indicates that both women and youth

female are experiencing lower rate of unemployment. Women unemployment rate is as low as

15 percent and youth female rate is 23 percent while male youth faced 65.8 percent highest rate

of unemployment and unexpectedly 49 percent of adult males are unemployed during survey

period using current unemployment status approach. Similarly, youth female and women are

more likely to be employed than youth male and adult male. The employment rate of women is

nearly 85 percent while the rate is 50 percent for men. Youth female employment rate is 76.7

percent which more than twice the employment rate of youth male, 34 percent. The significant

variation between the two data sources may be associated with the sample size and the survey

period.

4.7. Regional Unemployment Trends

Understanding spatial distribution of unemployment is important for area specific intervention.

The secondary data cover cities and towns in all the nine regions and two city administrations.

The two city administrations, Dire Dawa and Addis Ababa, are found to experience highest

rate of unemployment over the survey periods. The unemployment rates observed in these

cities are significantly above the national average and all other regions. On the other hand,

Gambella region is relatively lucky for having the lowest unemployment rate in the same

period (table 4.1).

The trend of unemployment across regions is almost synonymous with the national level. The

rate of unemployment decreased from 2003 to 2006 in all regions and increased after 2006; and

it again showed a decreasing trend since 2010. Relatively, the lowest regional unemployment

rates were observed in 2006 for Tigray, Amhara, oromia, Benishangul Gumz and Dire Dawa

regions while for remaining regions the lowest level of unemployment rate occurred either one

of the next two years.

66

Table 4.1: Regional unemployment distribution (%)

Source: UEUS 2003-11

The urban unemployment disparity observed in terms of gender and age in the regions is

similar to the national one. The unemployment rates for females and youth are significantly

higher than that of adult males in all regions. For example, in Tigray region in 2003, women

and youth unemployment rates were 35 and 39.7 percent, respectively while it was 12 percent

for adult males in the same period. The rates of unemployment decreased to 27percent for both

women and youth in 2011 while it reached down 4.6 percent for adult males. However,

unemployment rate for all categories of the labor force is the highest in Addis Ababa and Dire

Dawa compared with all other regions. For example, in Dire Dawa in 2003, women and youth

unemployment rates were 49 and 52 percent, respectively while that of adult males was 19

percent. The rates decreased to 34 and 29 percent for women and youth, respectively while it

fell to 13 percent for adult males in 2011.

4.8. Urban Unemployment and Education Education system of Ethiopia divided into lower primary comprises grade one to four, it is four

years of formal education, and upper primary includes education from grade five to eight.

Secondary education that is either from grade nine to ten for those educated in existing

curriculum or grade nine to twelve for those attended the old curriculum. After completing

secondary education, a student can apply either to TVET education or preparatory education

based on results achieved in national examination for leaving secondary general education.

Regions Year 2003 2004 2006 2010 2011

Tigray 28.81 20.95 14.31 18.85 19.08 Afar 24.35 18.16 18.22 10.9 15.17 Amhara 22 19.9 11.89 18.67 20.22 Oromia 26.2 23.09 14.9 19.49 16.79 Somali 21.74 22.53 25.43 16.41 20.22 B/G 15.26 12.2 8.02 12.12 9.24 SNNPRS 20.11 15.56 13.45 15.76 13.53 Gambella 12.73 - 10.95 12.2 8.9 Harrari 27.33 22.36 13.94 15.37 13.91 Addsa Ababa 34.5 31.32 29.29 27.27 24.82 Dire Dawa 39.57 34.26 22.92 30.54 23.77 National 26.29 23.28 17.61 19.44 18.08

67

Those who scored above certain point specified for the given year are eligible to apply for

preparatory education (i.e. grade 11 to 12 in the new curriculum), and then compete for

university education. Those who scored below the cut-off point for preparatory education can

compete to enroll in TVET program either at 10 + 1, 10+2, or 10+3 level based on their

performance.

Education explains the rate and spell of unemployment. Many international evidences are in

favor of the notion that higher educational qualification boosts employment outcomes such as

better earning and lower unemployment(Garcia and Fares, 2008). The effect of educational

attainment on urban unemployment in Ethiopia is mixed and different from the international

experience. To examine the association between educational achievement and unemployment,

we used lower primary education as a baseline education to compare effects of educational

attainment on unemployment. The overall urban unemployment rate for workers with lower

primary education decreased from 22 percent in 2003 to its lowest level 13 percent in 2006.

The rate for this category of education again increased to nearly 19 percent in 2010 and then

decreased to nearly 15 percent in 2011 (Table 4.2).

Most of higher educational attainments have not resulted in lower rate of unemployment in

Ethiopia. Only degree and above graduates have significantly lower rate of average

unemployment rate than labor force participants with baseline education between 2003 and

2011. Labor force with all other educational qualification such as secondary education,

certificate and diploma and degree not completed including vocational education not

completed face higher average rate of unemployment in the period. The average

unemployment peaks for preparatory education and followed by secondary education despite

preparatory education has short history in Ethiopia. However the average unemployment

variation between primary education and TVET graduates is not significant.

Primary data result also supports the evidences from secondary sources. Unemployment rises

proportionately with more education up the educational ladder including TVET except for

upper primary. Relatively lower unemployment rate is observed for those with at least diploma

education in the old curriculum. For example the unemployment figure is 37. 8 percent for

68

those with lower primary education and increased to 44 and 39.8 percent for those with grade

nine to ten, and some one completed TVET correspondingly. However it is estimated around

33 percent for those with at least university degree educational qualification.

Table 4.2: Unemployment rate by education

Both at higher and lower educational levels female and youth experience a higher level of

unemployment rate than adult male. Female and youth mean unemployment rate at lower

primary is nearly 25 and 20 percent respectively while the corresponding mean figure for adult

male with the same educational qualification was 8 percent from 2003 to 2011. For workers

with secondary education, the average annual unemployment rate of total youth and female

youth is fourfold the rate of adult male while for adult women it is three times adult male rate

from 2003-11.

The unemployment rate discrepancy between female, youth and adult is non-declining

overtime along the educational ladder relative to baseline educational qualification. For

example, unemployment rate of female with lower primary qualification was falling but

relatively higher rate between 32 and 23 percent from 2003 to 2011 whereas adult male with

equal educational achievement confronted comparatively low unemployment challenge with

rates between 12 to 4 percent over the same period. The level of unemployment rate for

Educational Qualification Annual unemployment rate (%) 2003 2004 2006 2010 2011

Non formal education 20.36 13.57 13.44 12.95 12.71 Lower primary education (grade 1-4) 22.34 19.17 13.37 18.75 14.98 Upper primary education (grade 5-8) 27.67 25.21 18.69 20.09 18.15 Secondary education (grade 9-10) 36.76 32.41 24.97 24.17 23.21 Preparatory education 35.68 27.68 30.23 26.77 33.57 Certificate 29.31 26 18.66 18.31 18.65 TVET completed (TVETc) 15.04 15.05 14.83 14.54 18.53 TVET not completed (TVETnc) 23.64 22.11 23.17 24.5 21.57 Degree and diploma not completed 25 21.74 17.65 14.71 17.3 Diploma completed - - - - 8.75 Degree completed 4.41 4.61 3.58 4.66 7.15 Source: UEUS 2003-2011

69

secondary qualification is varying from 50 percent in 2003 to 34 percent in 2011 for female

while it is slightly falling from 11 percent for year 2003 to 10 percent for 2011 for adult male

with equal educational achievement. The non-declining unemployment ratio along the

educational ladder is attributable to the fact that youth with better education tend to shop

around in the labor market and luck labor market experiences and additional education may fail

to reduce labor market discrimination of females.

The primary data survey results are consistent with urban employment and unemployment

survey. The rate is higher for all educational attainments than lower primary education except

for upper primary, diploma and degree education. For example, the unemployment rate of

degree and diploma graduates are 25 and 33 percent respectively while the rate is estimated

around 38 percent for lower primary education.

4.9. Unemployment Duration

Long unemployment spell permanently weaken an individual’s productive potential and human

capital and hence employment opportunities. Extended unemployment duration is barrier to

enter into employment particularly in the formal sector. One of the features of Ethiopia’s urban

unemployment is the lengthy spell of unemployment. The unemployed remained unemployed

for more than half year was 93 and 95 percent respectively for year 2003 and 2011. Over 90

percent of the unemployed people unable to find work for more than one year in 2003 and the

worst rate of unemployed unable to find job for more than one year sustained in 2011 as well.

The highest mean spell of unemployment figure was estimated around 2.5 years in 2004

followed by 2.4 years in the base period and the lowest level of spell nearly 1 year and six

months observed in 2010. However the average spell of unemployment increased to 2 years in

2011. In average, 37 percent of unemployed persons remain unemployed at least for two years

over the period signals prolonged unemployment permanently impairs the future employment

and productivity to the remarkable size of working age population. The rate of unemployment

in the base year is not only found to be the highest but also ends after long period while we are

linking the rate with mean spell of unemployment.

Figure 4.6: Mean spell of unemployed (in year)

70

0

0.5

1

1.5

2

2.5

3

2003 2004 2006 2010 2011

year

mean spell

youth 

adult male

female

total 

Source: ECSU 2003-11

The average spell of unemployed is not gender impartial over the period and statistically

significant (see figure 4.6). It supports the unpleasant situation of women. The more troubling

is the spells of unemployment increases with years of schooling. Even if, it decreases with

training received close to 90 percent of unemployed with training have more than 24 months

unemployment span between 2003 and 2011.

The primary data survey from three cities is in line with the results of the secondary data. The

average spell of unemployment is estimated about 1.4 years and the higher unemployment

differential across gender and age is statistically significant at 0.001 level. For instance,

average annual spell of unemployment for currently unemployed adult male, female and youth

are 0.6, 1.4 and 2.2 years respectively. Assuming lower primary education as reference line,

average spell of unemployment is significantly higher and inconsistently increasing up the

educational ladder until degree and above educational qualification. Currently unemployed

workers with base line education stay unemployed for half years in average and it is close to

1.6, 0.9 and 1.8 years to someone with upper primary, secondary and preparatory education.

TVET graduates keep unemployed at least for 1.1 years in average; diploma holders may

remain unemployed for 1.4 years however the spell decreased to almost one month for degree

and above educational qualification. The mean spell of unemployment variation along

educational ladder is significant at 1 percent while the variation disappears for those with

71

training. Furthermore people with training has less annual average spell of unemployment and

significant at 1 percent.

4.10. Urban Unemployment and Training

Labor force participants who have received training are increasing overtime, but the rate of

growth is quite slow. They were 20 percent in 2003, and after eight years increased to 31

percent. Unemployment challenge among labor force with training is lower and they have been

more likely to be employed than those fail to take an opportunity in each of the years since

2003 (figure 4.7). Unemployment among trained workers was 18 percent in 2003 whereas it

was over 28 percent for those without training in the same period. The rate decreased to 13

percent for the former category and 20 percent for those denied training in 2011.

Figure 4.7: The comparison of unemployment rate by training (%)

0

10

20

30

2003 2004 2006 2010 2011

year

rate with training

without training

UEUS 2003-11

Generally, the effect of training on labor market outcome is found to be significant over the

period. However, training failed to reduce unemployment differentials between female, youth

and adult male. Over 2003 to 2011 periods, in each year, average unemployment rates of

female and youth with training is approximately fourfold higher than adult male had training.

72

Figure 4.8: Unemployment differential between female, youth and adult male with training

0

5

10

15

20

25

30

35

2003 2004 2006 2010 2011

adult male 

youth

female

Source: UEUS 2003-11

The youth and female with training, for example; face unemployment challenge of 29 and 31

percent respectively in 2003 while the adult male with training face unemployment rate of 7

percent which is quite close to natural rate of unemployment. The adult male with training has

unemployment rate of 5 percent which is within the range of natural rate of unemployment

however the female and youth unemployment rate were 21 and 19 percent after eight years.

The rates of unemployment among females without training are more than those with training

at least by 5 percentage points for all periods since 2003 except for 2006. Corresponding

average is 6 percentage points unemployment differential for youth and adult men in the

period.

Among people with the equal level of education except for vocational and technical training

those with vocational training (TVETc2)- who have successfully completed TVET training at

10+1, 10+2 and 10+3 level) are less likely to be unemployed across all the periods. For

example, grade ten graduates more likely to be unemployed than those take technical and

vocational training. The mean unemployment spell of grade ten graduates exceeds average rate

of TVET graduates over the period is statistically significant at 1percent. Labor force with

vocational training are less likely to be unemployed than those with only lower primary

education for some periods. The implication is vocational education and training is successful

in enhancing the employment chance of trainees relative to tenth grade graduates but not vis-à-

vis lower primary education. Hence effectiveness of vocational education on the labor market

outcome of trainees is not strong.

73

Table 4.9: Unemployment differential between TVET and secondary school graduates

0

10

20

30

40

50

2003 2004 2006 2010 2011

year 

unem

ploy

men

t rate ( %)

grade 10 graduates

TVETc2

total unemployment

lower primary

Source: UEUS 2003-11

Primary survey result is consistent with secondary source with regard to association between

training received and the level unemployment. Working age categories with some training are

less likely to be unemployed than those without training. Unemployment rat among trained

workers is 35.5 which is less than 42.4 percent rates among people do not received any

training. Unemployment variation between TVET graduates and secondary school graduates is

almost narrowed dawn to zero, therefore; the finding still questions the effectiveness of TVET

training.

4.11. Training and Self-employment

Government is the dominant training provider for working age population in Ethiopia and the

major objective of the state training is to enhance self-employment. To examine, to what extent

certified trainings promote self-employment in urban Ethiopia, we compared the trend of self-

employment between TVET graduates and secondary school graduates. The share of self-

employment by TVET graduates is less than the comparison groups who completed grade ten

in the new curriculum and grade 12 in the old curriculum. Vocational and technical education

graduate entrepreneurs were 8.1 percent in 2003 and increased to 8.6 percent in 2010 and it

marginally increased to 9.7 percent in 2011. The lowest 6.9 percent self-employment among

vocational graduates observed in 2004. The average self-employment rates of vocational

74

graduates are 8.6 percent over the period. However the average annual size of entrepreneurs

who completed secondary education is 27 percent over the period. The trend of self-

employment for both comparable groups is falling up to 2006. Table 4.10: Self-employment by training

Source: UEUS 2003-11

Regarding overall training provided, labor force without training are more likely to be self-

employed than those with formal training over 2003-2011 period. The average magnitudes of

entrepreneurs with training and without training are approximately constant at 14 and 50

percent over the period 2003 to 2011. The implication is that training has not increased

entrepreneurship in urban Ethiopia.

The primary survey result neither strongly support secondary source nor oppose it. Self-

employment rate of TVET graduates are marginally higher than secondary school graduates

(such as grade nine to ten completes). However consistent with secondary source workers

without training are more likely involves in to be self-employment business than those missed

the chance.

4.12. School to Work Transition

School to work transition of urban youth in Ethiopia is long and difficult. The primary data,

from three major cities Addis Ababa, Hawassa and Bahir Dar, indicates that the average time

to find their first job after completion of training or giving up schooling is nearly one year and

three months. The average time to find first job varies significantly by training status. Average

0

10

20

30

40

50

2003 2004 2006 2010 2011

grade 10 graduates

TVETc2

w ith training

Without Training

75

time wasted by those with training decreased to 10 months, while those without training take as

high as two years until they find their first job and statistically significant at 1percent level.

Figure 4.11: Average time from school to work transition by education

1.3

0.94

1.28

0

0.5

1

1.5

2

2.5

education

average time 

grade 9‐10

TVET

grade 11‐12

total

Source: Primary survey result

We examined the effect of technical and vocational education on average time wasted to find

first job in order to see whether or not vocational training could ease school to work transition

in urban Ethiopia. We compared average time taken to find first job by TVET graduates and

those who have grade 9 to 10 and grade 11 to 12 educational qualification. Average time spent

by those with technical and vocational training to find the first job is nearly eleven months

where as for those with grade 9-10 educational qualification it takes nearly one year and four

months and the mean school to work transition is significant at 1 percent level. Those who

have educational qualification of grade 11 to 12 need to waste labor hour equals two years to

find their first job after they enter into the labor market. Consistently those with general

training needs less school- to- work transition time relative to those without training and the

mean variation is significant at 1 percent.

4.13. Socioeconomic Causes of Urban Unemployment

We analyzed the relationship between trend of urban unemployment and some important

macroeconomic variables using World Bank database. Ethiopia is now one of the fastest

growing non-oil economies in the continent. However the outstanding economic performance

for a decade has not resulted in equivalent employment creation. For example, according to

World Bank figures in 1999 annual real GDP growth is estimated around 5 percent while the

76

urban unemployment rate is 21.8 percent for working age population. The rate of annual GDP

growth is negative 2 percent in 2003 while the urban unemployment rate is 26.3 percent. The

real GDP growth had already recovered significantly since 2004, but unemployment remained

at an all-time high. The average real GDP growth is estimated around 10.56 percent, double

digit, for the period between 2004 and 2011 while the average unemployment rate for the

period is 20 percent.

Figure 4.12: Relationship between unemployment rate and GDP

‐5

0

5

10

15

20

25

30

1999 2003 2004 2005 2006 2010 2011

GDP percpita growth

Urban Unemployment rate

Source: UEUS, LFS and WB

This downward rigidity of unemployment, that unemployment rate tend to remain elevated

regardless of promising improvement in real economic growth, raising concerns that

sustainable economic growth is less likely results in the corresponding employment

opportunity. Consistently an empirical study indicated that variation in economic growth has

not result in unemployment reduction in Turkey (Aktar and Ozturk, 2009).

Rapidly growing urban labor force arising from rural-urban migration is a cause of high urban

unemployment in Ethiopia. Annual growth of urban population is at least 1 percentage point

higher than rural population rate of growth for each year between 1999 and 2011. The

significant population growth differential between the urbanized and rural areas primarily

attributed to high net rural-urban migration rather than high net birth rate in urban areas since

CSA figures indicate that rural-urban migration is the second dominant form of migration

between 1984 and 1999. Rural urban migration is often explained in terms of push-full factors.

The push factors comprise the pressure resulting from labor to land ratio in the rural areas and

the existence of serious unemployment arising from the seasonal cycle of climate. Arable land

77

(hectare per person) marginally increased from its lowest level 0.15 to 0.17 between 2003 and

2010. On the other hand agricultural value added is highly fluctuating from 1999 to 2000. It

ranges from lowest annual growth rate of negative 10.5 in 003 to highest rate of 16 percent in

2004. The average growth rate was 2.72 for the period. For the remaining periods (2003 to

2011) annual agricultural value added is also highly fluctuating from its lowest level 10.5

percent in 2003 to 16.9 in 2004 and decreasing successively for the remaining periods and

stood at 5.2 percent in 2011 and the average growth rate for the period was 7.17 percent. The

undesirably fluctuating growth rate of agricultural value added may lead to high rural-urban

migration. The situation can be further exacerbated by poor infrastructure facilities in rural

areas relative to urban centers. The rural labor force moves to urban areas with the likelihood

of searching lucrative employment in the industries. The industry value added annual growth is

more stable than agriculture and increased from 6.4 percent lowest level in 2003 to 11 percent

in 2011 and the average annual growth rate for the period was 9.6 percent.

Figure 4.13: relationship between participation and employment ratio

60

65

70

75

80

85

90

1999 2003 2004 2005 2006 2010 2011

year

employment to population ratio

labor force participation rate

Source: WB, UEUS and LFS

The high population growth results in excess of labor force growth over employment-to-

population growth for the period between 2003 and 2011 can elevate unemployment growth in

urban areas. The national average labor force participation rate is 84 percent for the period

under consideration where as average employment-to-population ratio figure is 72 percent

which is almost 12 percentage points lower than the participation rate. For all the periods

identified the labor force participation rate exceeds employment- to-population ratio. The

implication is that the high population growth rate leads to rapid growth rate of labor force

78

participation rate, which is far exceeds the supply of jobs. The accelerated growth of

population on Ethiopia’s unemployment problem is multifaceted. It affects the supply side

through high and rapid rise in the labor force relative to the absorptive capacity of the

economy.

The industrial and service sectors are not vibrant enough to absorb unemployed and surplus

labor migrating from agricultural sector. The agricultural sector employment share is decreased

by 10 percentage points between 1994 and 2005 while the service and industrial sector

employment share increased by 5.4 and 4.3 percentage point respectively in the decade. The

net growth of employment over the decade is almost zero while the labor force participation

rate increased by 3.2 percentage points. Aggregate non-agricultural employment rates of

growth are negative and high growth rates of urban labor force in Africa have not been

matched by correspondingly high rates of growth of urban labor demand by the modern large

scale establishments(Frank, 1968).

The school curricula are not effective in providing of employable skills. As we discussed

before unemployment is found to increase along educational ladder at least up to college

education relative to primary education. Therefore, general school graduates do not acquire the

skills needed by employers of labor for a formal employment. Moreover low human-capital

that is characterized by low literacy rate over the past two decades and low educational

attainment leaves the labor force especially the young people ill-equipped to the work place

and entrepreneurship. The literacy rate of both youth and adult remain low for a decade. The

literacy rate for adult was 27 percent in 1994 and stood at 39 percent in 2007 while the average

literacy rate for women was 22 percent in the same period. Even if literate youth population is

higher than adults, it is as low as 55 percent for the specified span of time and the gender

disparity is maintained. Tertiary school enrollment increased significantly from its lowest level

0.6 percent in 1994 to 7.6 percent in 2011. Vocational education and training participation of

females increased from 27.3 percent in 1994 to 45.6 percent in 2011 signals commitment of the

government to increase entrepreneurship. However, much is left in both quantity and quality of

TVET to achieve the desired result.

79

Foreign direct investment (FDI) attraction is effective way to increase employment and FDI

can affect the economy in terms of technological externality, information and management

experience (Song, 2003). Moderate flow of foreign capital relative to other countries and its

extent of employment generation to local labor can be considered as a cause of high

unemployment rate. In 1988 Ethiopia attracted US$ 261 million FDI while in the same period

Song (2003) observed that China’s annual FDI estimated to US$ 45.5 billion and since 1992,

China becomes the largest recipient of FDI next to USA. Ethiopia’s FDI flow increased to US$

627 million in 2011, after two decades. However other evidences indicated FDI attraction may

not contribute remarkably to the reduction of the serious structural unemployment that some

regions faced due to the fall in traditional manufacturing (Driffeld and Taylor, 2000). They

argued that flow of foreign capital generate employment for already employed skilled workers

elsewhere rather than to the structurally unemployed. However unemployed workers substitute

workers who have moved to the modern sector, if they are equipped with the required skills.

Hence, even if FDI flow to Ethiopia is low relative to China but still can result in sound

employment generation.

However we strongly argue for the fact that FDI produce job opportunities for already skilled

workers rather than to the unemployed people. On the other hand, FDI in agriculture results in

employment generation. Chaudhuri and Banerjee (2010) theoretically proved importance of

drawing FDI in agriculture in the developing economies. They find that FDI in agriculture in

these economies curb the unemployment problem of a skilled and unskilled labor. Contrary to

their evidence Aktar and Ozturk (2009) empirically tested FDI do not have significant effect on

unemployment in Turkey. Impact of FDI on job opportunities is not significant in European

union (Seye, 2000). More over we have an important argument on the relationship between two

variables as follows: “An increase in foreign capital investment leaves social welfare intact

and reduces unemployment if foreign capital is specific to foreign firms, and it may increase

social welfare and reduce unemployment if foreign capital is also used in the domestic

manufacturing sector”(Yabuuchi, 2006 P. 360). [

4.14. Theories of Unemployment

80

Previous literatures have identified three main arguments to explain causes of unemployment.

The first one makes a case for “skill mismatch”- that is inequality in the supply and demand

side of the labor market-causes unemployment. The second is the “queuing” hypothesis which

posited unemployment is created when unemployed wait to find better paid and secure jobs

either in the public or formal private sector. The third theory of unemployment argues that

costly labor laws results in job destruction and slow job creation. That is employers find

restrictive labor laws difficult to expand employment. Employers may prefer employment

without written contract between two parties to avoid costs of job dismissal.

The skill mismatch is said to be the cause of unemployment when skills gained from education

system teaches fails to match well the skill demands of the labor market. When graduates face

extended spell of unemployment and lower wages when they find a job suggests existence of

skill mismatch in the labor market. We examined the association between years of schooling

and spell of unemployment as well as years of schooling and likelihood of unemployment to

determine whether skill mismatch is cause of unemployment in urban Ethiopia.

The Cox regression result (annex table 4.19), after controlling for demographic variables,

training received, location dummies, tends to support the skill mismatch theory. The worker

with more years of schooling is less likely to end unemployment spell over a short period. The

implication is that additional years of schooling results in larger total duration of

unemployment in the last six months during the survey period. Similar result was obtained

while we replace explanatory variable years of schooling by dummies of educational levels.

For example, someone with degree and above educational qualification has 84 percent

probability to exit out of unemployment latter than the one with lower primary education.

Similarly, the exit out of spell of unemployment does not improve with educational levels for

the rest of educational achievements up the educational ladder except for non-formal

education.

The result with respect to education variables, contradicts findings by Seife (2006)-college

diploma or degree have far shorter unemployment duration relative to secondary education.

Variation in results may be attributed to he used cross-sectional data for estimation that

81

collected some 10 years before, referent group used and variations in the local unemployment

levels in two periods. We argue that reservation wage increased with more education and the

ratio of graduates to labor demanded may be significantly increased in recent periods as

compared with the ratio we have ten years before. Equivalently study in US labor market finds

that education remarkably increases employment success of unemployed workers. For

example, graduating from high school increases reemployment probability by around 40

percentage points and an additional year of schooling increases this likelihood by 4.7

percentage points (Riddell and Song, 2011). However impact of continued education and

training for adult on labor market performance in Germany, Denmark and British varies due to

institutional differences, which may influence the quality of training supplied (Dieckhoff,

2007).

TVET education increases the spell of unemployment relative to lower primary education

however it has no significant effect on rate of reemployment Seife (2006). On the other hand,

in Western Germany participation in vocational training programs results in adverse effect on

the transition rate into employment (Hujer et al., 2006b). They put forward reasons for negative

effect of participation which seem relevant to the context of Ethiopia. Participants would have

obtained job anyway, the involvement into the program may increase the spell of

unemployment artificially. The content thought to the participants do not match with the

interests of the labor market. However the most important point is there may be unemployment

arise from shortage of labor demand in Ethiopia.

However any training received associated with fast job finding following spell of

unemployment. Subjects with training were 34 percent more likely to exit earlier from

unemployment spell than those without training. Participation in short-term training is effective

in shortening the unemployment duration of job seekers in West Germany (Hujer et al., 2006a).

Age is negatively associated with the probability of reemployment. Male are 42 percent more

likely transferred from being unemployed to employed state than women. Put differently,

female have more unemployment duration than their male counterparts.

82

More time has been required to terminate spell of unemployment over the years except for year

2010. Relative to the base year, time required for unemployed to escape the spell of

unemployment decreased by 4 percent in 2010. The result suggests that unemployed person in

2010 has less total unemployment duration than the one who had been unemployed in 2003.

We also examined the relationship between unemployment and two closely related variables

years of schooling and education further to test the validity of skill mismatch hypothesis. The

result is consistent with the duration model in both approaches. Additional year of schooling is

measured by years of schooling increases the probability of unemployment and statistically

significant at less than one percent (annex table 4.14). Relative to the reference category of

education (lower primary education), more educational qualification is likely to decrease

unemployment only after at least college diploma education. Educational qualification below

diploma education such as upper primary, secondary, and preparatory education and vocational

education are found to associate with higher probability of unemployment relative to lower

primary education (table 4.5). Therefore, in urban Ethiopia, skill mismatch theory holds up to

diploma education. However Glawwe (1987) and Dicken and Lang (1995) examined the

relationship between education and unemployment to test the validity of skill mismatch

hypothesis. Glawwe find support for the skill mismatch hypothesis among both women and

men in rural and urban sectors while Dicken and Lang fund supportive evidence only for rural

women.

The queuing hypothesis can be either shopping around for better paid job and good work

environment in the public sector or in the formal private sector. To test the validity of this

hypothesis, we examined the association between probability of unemployment and regional

share of both public and formal private sector employment in urban Ethiopia. The probit

regression result supports the queuing hypothesis. There is positive and statistically significant

relationship between the share of both public and the formal private sector job. That means a

person in the labor force more likely to be unemployed when either regional public or formal

private sector employment increases by one percentage point. Indeed the finding supports the

existence of queuing hypothesis as source of unemployment in urban Ethiopia. The result is

consistent with Tan and Chandrasiri (2004 ). They find support for queuing hypothesis by

83

showing positive correlation between the share of public sector job and probability of being

unemployed. The implication is that unemployed people are more likely to shop around for

public sector job and formal private employment opportunities.

We analyzed the validity of slow job creation hypothesis to determine whether or not job

destruction is cause of unemployment in urban Ethiopia as shown in annex table 4.20. To this

end, first, we examined the relationship between shares of either formal or informal private

sector job to probability of unemployment and second the association between unemployment

and the share of private sector job that has either written or verbal employment agreement

between employer and employee. The probit regression result supports job destruction

hypothesis. There is positive and significant relationship between regional share of private

formal sector employment and probability of unemployment. The likelihood of unemployment

increases by nearly 67 percent when formal private sector employment increases by one

percentage point. When regional share of informal sector employment increase by one

percentage, probability of unemployment increases by nearly 5 percentage point. This may be

attributable to restrictive labor market regulations, for example; high cost of dismissal

discourages employers from making investments that would expand size of the formal sector

job openings. Employers may choose to expand informal sector jobs that decreases

compensation costs when the business going dawn. On the other hand, unemployed people

may shop around rather than simply accepting precarious jobs. The regional share of private

sector employment that involves either written agreement or verbal agreement has no

significant effect on probability of unemployment.

4.15. Effect of Education and Training Polices on Labor Market Outcomes

This section of the study focuses on econometric evidences on the effect of policy

interventions, particularly through the expansion of education and training on labor market

outcomes. To evaluate the effect of the aforementioned policy interventions, we analyzed the

relationship between educational attainment and training against labor market outcomes with

especial emphasis to unemployment.

84

4.15.1. Effect of Education Polices on Urban Unemployment

Policy makers, almost across the globe, give much emphasis to measure the benefit of

education and training in terms of their labor market outcomes. A common way to look at the

value of education or training for individuals is, as Becker’s Human Capital Theory says, in

terms of increased human capital based on the assumption that the greater one’s human capital,

the better are one’s labor market chances. Much of up to now global evidences in favor of the

notion that more educational qualification results in better employment outcomes such as

higher wages and lower unemployment (Garcia and Fares, 2008). Accordingly, to explore the

effect of education on probability of urban unemployment we deployed probit regression on

pooled cross-sectional data and logistic regression on cross sectional primary data set. We give

more emphasis in this part to answer mainly the question ‘how does education affect likelihood

of urban unemployment?’ and latter the question is extended to training.

The dependent variable unemployment is dichotomous so that it is set to take one if the

individual is unemployed and zero if employed. The analysis of probability of unemployment

uses two alternative measures of unemployment defined by the CSA’s urban employment

unemployment survey. These are the current activity status approach and usual status approach.

Unemployed, measured using current status approach refers to a person who is available and

did not work in the last seven days prior to the day of interview. On the other hand, as per the

usual status approach, “usually unemployed” refers to a person available for work but did not

work during most of the last six months. The second measure, by averaging over a six months

period, may result in robust characterization of economic activity status relative to the

traditional unemployment measure, which has short span of time.

The independent variables included in this specific model are demographic variables such as

age, marital status, sex and education variables categorized in to different levels such as lower

and upper primary, secondary, and preparatory education. Moreover, the probit model includes

the year dummies, to see how probability of being unemployed has been changing overtime,

and urban location, to observe how unemployment differs across regions (see annex table

4.21).

85

The regression results obtained from the two approaches, the current status and the usual status,

are found fairly similar whether the length of time is last week or last six months except for the

magnitude of coefficients. For this reason, we indifferently used the current unemployment

status to interpret the result and explained differences. As expected, when the age increases, a

person is less likely to be unemployed. This is in line with the descriptive result of the study as

well as earlier studies in which age and unemployment exhibited an inverse relationship.

Regarding gender gap, being male is likely to reduce probability unemployment. The marginal

effect implies that the probability of women to be unemployed is 16.5 percent higher than that

of male. The gender gap for usual unemployment status appears to have decrease from 2003-11

by about one percentage points while the gender gap in probability of current unemployment

status remained constant even after eight years. It signals existence of some gender

discrimination though it is difficult to single out contribution of all other factors (such as

education gap) to the disparity. The result is consistent with the descriptive finding and with

previous works such as UN (2003), WB (2007), Guarcello, Lyon and Rosati (2008a), and

Tegegn (2011).

Controlling for other factors and considering never married labor market participants as a

referent group, the marginal effect indicates for all categories of marital status lower

probability of unemployment and statistically significant at below one percent significance

level. Divorced and widow/widower were 6.4 and 3.3 percent, respectively less likely to be

unemployed than single labor market participants. Separated person but available and ready to

take job offers is found to have 4.3 percent lower probability of unemployment than who is

never married. The implication is that, married, widowed, separated and divorced are more

responsible to the livelihood of family as bread winners so that they do not want to extend time

to shop around for better jobs, rather they would take any available job to discharge their

family responsibilities.

Regarding education, the marginal effect implies that educational qualification unable to

reduce likelihood of unemployment among the whole urban labor force as compared with the

reference line education, lower primary (grade 1 to 4). For instance, worker with upper primary

86

education (grade 5 to 8) confront 6.7 percent higher probability of unemployment. Having

secondary and preparatory education associated with 11.7 and 12 percent respectively higher

probability of unemployment. It appears to increase chance of being unemployed by nearly 12

percent relative to lower primary education. Likelihood of unemployment falls with any level

of non-formal and with no education relative to reference category of education.

Tertiary education in Ethiopia commonly concentrates on general long term training dominated

by theory oriented curriculum. The probit regression result indicates that higher education,

except diploma level and above education, including vocational training has no tendency to

bring salutary effect on probability of unemployment estimated using current status approach.

It contradicts Arum and Shavit (1995) that provides evidences in favor of possibility of

vocational education to decrease risk of unemployment and increase chance of students

employment as skilled workers. After eight years of practices TVET education decreases

probability of usual unemployment however it keeps on increasing likelihood of current

unemployment. For example, as indicated in table 4.5 a vocational graduate is 1.9 percent less

likely to fall into usual unemployment status in 2011 relative to 2003. However overtime the

employment effects of general education is promising. TVET drop outs have 7 percent higher

probability of unemployment relative to baseline education. Labor force participants who have

started university degree and diploma education but not completed face 6 percent higher

probability of current unemployment status. The worker with the same educational

qualification may less likely confront usual unemployment. Diploma graduates are 5.7 percent

less likely to be unemployed as compared with referent educational qualification. Other things

being equal, workers with first degree and above qualification have 9.6 percent lower

probability of unemployment relative to baseline education. The implication is that the

employment effect of tertiary education (diploma and above educational qualification) is

relatively highest. For instance, the WB (2007) indicates that general education in urban

Ethiopia, especially among the youth population, is associated with a 78 percent higher

likelihood of being unemployed. Similarly, the finding of Getinet (2003) and Kirishnan,

Gebreselassie and Dercon (1998) confirm that those that completed secondary education are

more likely to be unemployed. Across Europe an academic degree likely to decrease short term

unemployment more effectively than long term unemployment (Nùñez and Livanos, 2010).

87

This is the effect of employers prefer better educated people to less well-educated people for

positions that were held by less educated workers.

One question of interest is: After controlling for other observable factors, what has happened to

the likelihood of unemployment overtime? The factors we controlled for were sex, age; marital

status and variables related to different educational qualification including tertiary education

and TVET and urban location. The base year is 2003. The variable y04 is a dummy variable

equal to one, if the observation comes from 2004 and zero if it comes from other years. The

coefficient on the year dummy variables showed a sharp drop in probability of unemployment.

That is the likelihood of current unemployment decreases by 2.5 percent in 2004, 7 percent in

2006, and 5.6 percent in 2010 and it falls by 5.2 percent in 2011. The coefficients on year 2004

to 2011 suggest that there are drops in probability of unemployment for reasons that are not

captured in the explanatory variables. Since we controlled, for example, different levels of

educational qualification, this drop was separate from the decline in occurrence of

unemployment that is due to the change in average educational levels. The average years of

education increased from 8.3 in 2003 to 9.3 in 2011. However, probability of unemployment

decreases by 5.2 percent in 2011 relative to 2003 is not attributed to rise of this average

education.

Urban location is statistically important. Probability of unemployment appears to rise

proportionately with the level of urban size except for Jigjiga and other towns in Somali

regional state. Regions with relatively bigger towns have higher probability of unemployment.

We excluded Gambella regional state from this analysis as the region does not have survey for

2004. Relative to Benishangul Gumuz, unemployment probabilities were higher in all regions

and statistically significant at one significance level. The probability of unemployment was

found to be highest in Dire Dawa and Addis Ababa followed by Somali regional state. SNNP

has been experiencing least unemployment probability in the last eight years relative to all

other regions except reference region (Benishangul Gumuz Regional state). Therefore a person,

who lives in Dire Dawa, is about 23.4 percent more likely to experience statistically significant

at 1 percent level of unemployment relative to someone residing in Benishangul Gumuz

regional state. The probability of experiencing unemployment for those who reside in Addis

88

Ababa, Somali and SNNP regional states is higher by 21.4 percent, 14.8 percent, and 6.2

percent, respectively as compared to a person in Benishangul Gumuz regional state. The probit

regression result is in favor of the unemployment discrepancy among regions identified using

simple descriptive statistics.

To validate the pooled probit model regression result obtained from urban employment

unemployment survey between 2003 and 2011, we made pooled probit regression on different

data set (such as labor force surveys of 1999 and 2005). The results are the same for most of

the variables except for minimal variations due to inclusion/exclusion of few variables to avoid

multicollinearity and specification problems (see annex table 4.23). The dependent variable,

measured using the current status approach for both data sets, is equal to zero for employed and

one for unemployed. Demographic variables such as age, and sex and were likely to decrease

probability of unemployment and statistically significant at 1 percent significance level for

both data sets. Labor market experience is likely to decrease probability of unemployment.

Regarding educational qualifications represented by different dummy variables for most

variables the results are almost the same. For example, for both surveys (urban employment

unemployment surveys and labor force surveys), upper primary, secondary education,

preparatory education, and technical and vocational education have positive and significant

effect on probability of unemployment at below 1 percent significance level (see annex table

4.23). On the other hand, for both data sets diploma and above educational qualification is

likely to decrease probability of unemployment. However the difference was observed for one

year training obtained from teachers training institutions after completion of secondary

education and TVET not completed.

As far as regional unemployment distributions are concerned the same result is obtained for all

regional dummies. Relative to the reference region, Benishangul Gumz, all regions have higher

probability of unemployment. As presented in annex table 4.8 the probability of being

unemployed is highest for Dire Dawa followed by Addis Ababa in both urban employment

unemployment and labor force surveys and statistically significant at below 1 percent level.

89

Regarding migration variable, migrants with minimum of half year to ten and above are less

likely to be unemployed than non-migrants. This might be attributable to lower wages of rural

areas from where the migrants moved to urban areas for better earning opportunities so that

their reservation wages are low relative to the non- migrants Moreover, migrants may have no

option other than their labor income for their subsistence, so that they are less likely to shop

around for better job than non-migrants. The result is consistent with descriptive analysis

where migrants are less likely to be unemployed. It is also consistent with Tegegn (2011) and

Birhanu, Abraham and van der Dejil (2005).

The result from primary survey using logistic regression is not consistent with the result from

secondary survey. This might be attributed to the variation in spatial coverage of surveys,

sample size, explanatory variables controlled, survey period and other potential reasons. The

paradox is females are less likely to be unemployed than males (see annex table 4.17). The

association between education and probability of unemployment is varied to some extent.

Upper primary, secondary and preparatory schooling do not have any effect on probability of

unemployment relative to lower primary education. TVET, diploma and degree and above

graduates are less likely to be unemployed as compared with lower primary education.

Alternatively after considering years of schooling as explanatory variable we found that more

years of education decreases likelihood of unemployment. Access to any form of credit (formal

and informal) has also negative effect on probability of unemployment. However over 76

percent of respondents borrowed money from formal financial institutions in three cities. It

suggests increasing access to formal financial institutions as sustainability of informal sources

are uncertain and using this sources are relatively costly.

4.15.2. The Effect of Training Polices on Urban Unemployment

The government of Ethiopia has emphasized on technical and vocation education as a strategy

to enhance employment and employability of youth. Labor market participants who received

any training are at most 9.9 percent less likely to be unemployed than those without training

and statistically significant at below one percent. So any training received is preparing people

to meet the skill demands of the labor market overtime.

90

We also compared the effect of TVET and grade ten completed general education on

unemployment. Surprisingly TVET graduates are more likely to be unemployed in 2011

relative 2003. One additional year of TVET education (10+1, 10+2 and 10+3) has positive

effect on probability of unemployment after eight years. However grade 10 graduates

probability of entering into unemployment is decreasing (annex table 4.14). Suggesting even if

employment effect of TVET is inconsistent it does not have desirable effect on labor market

outcomes after more than eight years of implementation experiences. Hence it calls upon to

reconsider the implementation and the curriculum of technical and vocational education in line

with the demand of industries.

4.15.3. Effect of Education and Training Polices on Self-employment and School-

to-Work Transition

The entrepreneurial effect of education and training in general and TVET in particular is not

attractive. The regression result suggests that a self-employment condition of workforce is

likely to decrease significantly with access training and better education. For example, degree

and above graduates are 28.8 percent less likely to be self-employed than worker with not more

than lower primary educational qualification. TVET graduates are less likely to be self-

employed than lower primary graduates. One additional years of schooling decreases

probability of self-employment by 1.7 percent (annex table 4.15). However completed grade

ten education and technical and vocational education has salutary effect on self-employment

status of labor force after eight years.

As indicated in annex table 4.16 work experiences, upper primary education, technical and

vocational training, any training received, access to any form of credit received and university

education are associated with shorter school-to-work transition period. The implication is that

TVET, university level education and access to start up capital are important policy variables

to reduce high unemployment rate and extended duration of unemployment in urban areas of

Ethiopia.

91

4.16. An Assessment of Strategies to Promote Employment in Ethiopia Vocational education and training and expansion of micro and small enterprises commonly

used as a remedy to keep youth out of streets and unemployment and to raise the income of the

poor. Furthermore trainings in Ethiopia are mainly provided to increase employability and

encourage self-employment of dropouts and graduates. For example, PASSDEP envisages

TVET to provide relevant and demand driven and training that increases employment and self-

employment. Important reform measures have been introduced after the adoption of the

National TVET Strategy of 2002 and the TVET Proclamation of 2004. However existing

evidences assure that share of self-employment by TVET graduates is less than their

counterparts who are secondary school graduates and employment growth within MSEs is

slow. In the next section, we evaluate the implementation and effectiveness of some important

strategies to increase employment in Ethiopia. Our focus will be on the role of technical and

vocational education and training (TVET) and micro and small scale enterprises (MSEs) in

enhancing employment.

4.16.1. Strategies to Increase Employment through TVET

Strategies to increase employment through effective implementation of technical and

vocational education and training need to begin from participatory curriculum development to

linking graduates to the labor market. To evaluate the implementation and effectiveness of

technical and vocational education and trainings in Ethiopia, we adapted good training

practices for disadvantaged youth identified in Brewer (2004). The evaluation result helps us to

identify the bottlenecks to job creation potentials of the program and its implementations and to

measure outcomes of the program in relation to 2008 National TVET strategy document. We

collected relevant data to undertake this assessment from different TVET respondents such as

employers, parents, TVET students, TVET teachers and experts who guide and oversee the

implementation of the TVET program. When distributing the questionnaire the sex

composition of respondents is taken in to account. As shown in table 4.3, over 26 percent of the

respondents are either teaching in the government technical and vocational collages or working

as directors, experts and practitioners.

92

Table 4.3: Distribution of respondents

Stakeholder category Frequency Percent

Government TVET college (teachers, experts and directors) 59 26.46

Private TVET college (teachers, experts and directors) 24 10.76

Employers (both government and private employers) 41 18.39

Parents, above 1st year TVET student and TVET graduates 99 44.39

Total 223 100

Source: primary survey

Share of teaching staffs in private colleges are 10.8 percent and the share of employers who are

the major beneficiaries from effective implementation of TVET programs are nearly 18

percent. Parents of the technical and vocational students, students who attended at least one

year course in the college and TVET graduates constitute nearly 44 percent of the respondents.

The good training practices relevant to the current TVET environment of Ethiopia against

which TVET implementations are evaluated comprises innovativeness, feasibility,

responsiveness, relevance, flexibility, upscale and coordination.

Innovativeness refers to the unique quality of the TVET program, which deals with the

drawbacks in other training practices in enabling the disadvantaged youth; appeals to the

interest of all respondents. The Likert scale inquiry is used to gather information from

respondents about innovativeness of the program. We find that nearly three fourth of

respondents (75 percent) of the respondents agree that the training program addresses the

limitations in other training programs. While 17 percent are uncertain about the innovativeness

of the program and the remaining 8 percent at least disagree with the innovativeness of the

training.

93

Table 4.4: Evaluation of the innovativeness of the program

Criteria to measure innovativeness of the program

(number of the respondents= 219)

percentage of respondents

Disagree and not

certain

Agree and strongly

agree

The TVET program provides training to women and youth

marginalized by other training programs and prepares them for

better job

25.57 74.43

Source: Primary survey

The result suggests that majority of the respondents are in favor of the innovativeness of the

program. Specifically, the program addresses training gaps of youth and women who are

disadvantaged in other training programs and prepares these groups for good job relative to

other trainings. Moreover, the outcome is inline with 2008 National TVET strategy developed

by involving various interest groups. The TVET strategy makes every effort for social

inclusion, equal access and opportunity by increasing the access regardless of level of

educational attainment, gender, location, ethnic and religious membership. Gender sensitivity,

particularly promoting TVET institutions to develop gender sensitive polices to boost fair

access and to avoid any bias against female trainees and staffs. This does not mean further

improvement is not required, but the achievement is salutary thus the respondents can achieve

more on fair access to the training and can improve labor market out comes of marginalized.

Feasibility of the training is a standard that measures ‘the program can, realistically, be

implemented; there is sufficient support and funding capacity’. We have set specific criteria to

evaluate the feasibility of the program as indicated in the table 4.8. The average result indicates

that nearly 59 percent of the respondents are uncertain and disagree. They question the

feasibility of the TVET program. The survey result depicts that about 69 percent of the

respondents believe that the program does not have sufficient support and budget; and nearly

67 percent of respondents pointed lack of qualified staff and inadequate laboratory equipments.

Poor access to credit for business startups by technical and vocational graduates is reported by

67 percent of respondents.

94

Table 4.5: Evaluation of the feasibility of the program

Items to evaluate feasibility of the program

(number of respondents= 221)

Percentage of respondents

Disagree and

not certain

Agree and

strongly agree

The program is realistic and practical 26.24 73.76 The TVET program has enough support and budget 69.48 30.52 The program has qualified and experienced teachers, enough lab 67.28 32.72 TVET graduate access to sufficient credit for business start ups 66.97 33.07 To fill shortage of experienced teachers the program employees

foreign staffs to facilitate technology transfer

64.38 35.62

Average 58.87 41.13 Source: Primary survey

[

Only 35 percent of the respondents agree that the program employs expatriates to fill shortage

of experienced teachers and to ease the technology transfer. However the strategy document

identified set of alternatives to increase qualified staffs, employment of foreign staffs to closing

teachers and trainers competences, collaborative TVET schemes to decrease cost of training

and increasing income sources of TVET schools ( such as sale of products produced by

students). The policy strategy is silent about availability of credit to graduates for business

startup.

The resource available for every economic activity might be scarce but the way we utilize the

resource can matter the implementation of the program. Graduates need soft loan for business

start ups and colleges need money to acquire lab equipments and to employ qualified staffs.

Government may lack sufficient budget for all these activities, however, the labor of graduates

can substitute part of the budget deficit. Similarly, locally available materials, which cost less,

can be used to furnish some of the lab instruments. Nearby private and public organizations can

be source of finance for lab equipments as well as teaching staffs on part time basis. Therefore,

close collaboration of government, private sector and NGOs are important to alleviate these

problems.

Responsiveness refers to whether or not the practice of the program is consistent with the

needs identified by young women and men; it has involved a consensus-building approach; it is

responsive to the interests and desires of the participants and others. We identified specific

95

points summarized in table 4.5 to measure whether or not the program is responsive to the

interests of respondents.

Table 4.5: Evaluation of the TVET program responsiveness

Criteria to evaluate responsiveness of the program

(number of respondents= 222)

Percent of respondents

Disagree and not

certain

Agree and strongly

agree

The program enhances entrepreneurship and has played role to reduce

youth unemployment

31.08 68.92

Government polices and strategies have been encouraging business

startups by TVET graduates

44.50 55.5

TVET program has meaningfully contributing to better productivity of

MSEs through technology transfer

37.10 62.9

TVET program gives more support and more attention to youth

females relative to other programs

49.08 50.92

Average 40 60

Source: primary survey

Over 68 percent approve that the program enhances entrepreneurship and has played role to

reduce youth unemployment, 55 percent noted that government policies enhances business

startups of TVET graduates, 63 percent of respondents consent that the program contributes to

productivity of MSEs through technology transfer and half of the respondents argue for the

program that it gives more attention to females than other programs. The average value

indicates that 60 percent of respondents believe that the program is responsive while the

remaining 40 percent disagree or uncertain about the responsiveness of the program.

Apparently, the opinion of the respondents suggests, the need to invest more in these institutes

so as to increase their responsiveness to participants. In fact, the opinion of beneficiaries

supports the realization of expectation of TVET strategy. The strategy expects TVET

institutions to replicate and transfer selected technologies to industries and MSEs. It also

underlines importance of entrepreneurial skill development training, internship and connecting

TVET to MSEs.

96

Relevance of the program refers to the contribution of the program, either directly or

indirectly, to demands of the market and the needs of the participants. We set some points to

evaluate the relevance of the program as indicated in table 4.6. To majority of the respondents,

the program has been student centered and practice oriented relative to other trainings, over 57

percent of respondents’ consensus about vocational training provides skills required by the

labor market.

However, the progress of the program in terms of enhancing competences of trainees’ overtime

found to be weak and moreover over 50 percent of respondents agreed TVET less likely equip

new labor market entrants and youth with skills required by local and international markets.

Thus, the need assessment efforts of the TVET institutions to identify current skill needs of the

labor markets, and participating employers and trainees in the need assessment are found to be

important areas of focus for further improvement.

Over half of the respondents agree that the program facilitates school to work transition.

Students from higher income families struggle for tertiary education and farmers refused TVET

because they believed that it limits the chance of their off springs to compete for better jobs in

urban areas(Psacharopoulos, 1997). Furthermore, the TVET strategy document (MoE, 2008)

pointed out that poor awareness of the community on the program, for instance, the perception

that those failed to join tertiary education are admitted to the program and earnings for the

graduates is low.

To address such prejudice, the strategy paper suggested various mechanisms such as discussion

with stakeholders on the fact that the program has clear educational qualification and

promotion structure, participating private employers as stakeholders, arrangement of different

competition for entrepreneurs and involving them in international competitions.

97

Table 4.6: Evaluation of the relevance of the TVET program

Items to evaluate the relevance of the program

(number of respondents= 217)

Percent of respondents

Disagree and not

certain

Agree and strongly

agree

TVET program ease school to work transition and reduced

unemployment spell of youth

46.19 53.81

It is more student centered and practice oriented relative to other

programs

32.72 67.28

TVET provides skills and knowledge required by local labor market 42.99 57.01

TVET program enhances the skill and competences of job seekers and

entrepreneurs more than before so it becomes more fitting to the skill

needs of the labor market

59.36 40.64

TVET institutions make need assessment to identify current skill and

technical skill demands of the labor market

63.96 36.04

TVET allows private and government employers and trainees sufficiently

participate in the need assessment

62.50 37.5

TVET equip new labor market entrants and youth with skills required by

local and international markets

56.88 43.12

TVET facilitates employability of youth 42.13 57.88

To enhance local people awareness about TVET sufficient effort was

made

64.38 35.68

Average 52.35 47.65

Source: primary survey

However, the survey result indicated that sufficient effort has not been made to increase

awareness of the beneficiaries. The average measure of the overall relevance of the program

shows that almost half of the stakeholders were uncertain or disagreed with the relevance of the

program. The implication is that concerned bodies need to give more attention to improve the

relevance of the TVET program so as to meet the technical skill demand of the labor market.

Flexibility of the program evaluates the capacity of the program to provide training programs

that meets the changing labor market and international economic environment. As can be seen

in Table 4.7, about 57 percent of the respondents do not agree that TVET institutions are

flexible and capable to make timely adjustment of training programs to meet the changing skill

demands of the labor markets. Hence, the most likely inference would be that TVET

98

institutions must be aware of the low flexibility of their programs. Furthermore, the result less

likely meets the 2008 national TVET strategy that advocates flexibility to respond to the

changing occupational requirements.

Table 4.7: Evaluation of the relevance of the TVET program

Items to evaluate flexibility of the program

(number of respondents= 218)

Share of respondents

Disagree and not

certain

Agree and

strongly agree

TVET institutions have enough capacity to make timely adjustment of

training programs to meet changing labor market skill demands

58.26 41.74

The TVET program make quick adjustment to meet dynamic labor

market needs that vary with international conditions and national

economic growth

55.87 44.13

Average 57.07 42.93

Source: primary survey

Hence, the most likely inference would be that TVET institutions must be aware of the low

flexibility of their programs. Further the result less likely meets the 2008 national TVET

strategy that advocates flexibility to respond to the changing occupational requirements.

Efficiency and Effectiveness refers to use of resources (human, financial, and material) in

such a way that maximizes desired impact. The result in table 4.8 indicates that training

programs are cost efficient.

Table 4.8: Evaluation of the efficiency and effectiveness of the program

Items to evaluate Efficiency and Effectiveness

(number of respondents=213 )

Share of respondents

Disagree and not

certain

Agree and

strongly agree

TVET reduces duration of unemployment and job search time and

efforts

35.51 64.49

TVET program is achieves the desired result at least cost 49.30 50.70

Average 42.40 57.60

Source: primary survey

Almost 65 percent of respondents do agree that the TVET program is effective in reducing

duration of unemployment and job search time. Regarding the efficiency of the program, about

half of the respondents agreed that the TVET program is achieving the desired results at least

99

cost. The remaining 50 percent of the respondents did not believe that the program is efficient,

implying that the program is still required to exert more efforts to improve its cost

effectiveness. Furthermore the figures contradict with the emphasis of strategy document to

increase efficiency of human and financial resource utilization and cost effective TVET

delivery.

Up scalability of the program evaluates whether or not the practice of the training can be

expanded to operate on a wider level (e.g. from community level to national level). The result

indicates that the practices are not significantly expanded in different areas and circumstances.

Table 4.9: Up Scalability of the Program

Items to evaluate Up scalability

(number of respondents= 218)

Share of respondents

Disagree and not

certain

Agree and

strongly agree

TVET’s best practices were expanded in different areas and

circumstances

49.54 50.46

Source: primary survey

Coordination, cooperation and commitment among respondents is vital for effective

implementation of the TVET program (Brewer, 2004). However the evaluation result indicated

that above 56 percent of the respondents are disagree or uncertain about the coordination

between respondents. For instance, participation of respondents in the TVET program from

inception to implementation stage is crucial for successful outcome, but majority of the

respondents agree about lack of involvement of all participants.

Linking vocational trainings to the industries is important to ease school to work transition of

the youth. Nevertheless, the achievement in TVET-industry partnership is not much attractive

and needs further effort to link training programs with industries. Although the private sector is

among the dominant employer of graduates, its participation in curriculum development and

training provision is found to be weak.

100

Table 4.10: Coordination of the TVET program

Items to evaluate coordination

(number of respondents= 216)

Share of respondents

Disagree and not

certain

Agree and

strongly agree

Respondents participates in the evaluation of TVET programs

from inception to implementation phase

66.67 33.33

TVET program established close link with industries by

participating firms in curriculum development and achieved better

result from labor market outcomes

49.54 50.46

TVET participates Private sector in curriculum development and

training provision as a major beneficiary from the program

59.91 40.09

Government, civil society organizations, private sector, local

community and other respondents has been implementing the

TVET program in collaboration

52.73 47.27

Government organizations which closely oversea the

implementation of TVET programs encourage private sector to

involve in curriculum development and training provision

55.25 47.75

Average 56.82 43.78

Source: primary survey

The result does not support 2008 national TVET strategy. The TVET strategy document

stipulates interference of a broad stakeholder groups particularly education sector, the

employers, industry and MSE sectors. It also promotes public private partnership through

facilitation, regulation of the system through proclamation, licensing accreditation, and

apprenticeship.

4.16.2. Employment Growth within Micro and Small Scale Enterprises

Despite government policy interventions, for example; short term trainings to MSEs and other

business development services, assign appropriate organs and experts to support and to oversee

activities of MSEs, and credit supply through financial institutions to increase employment,

employment contribution of the sector is low. To point out the puzzle to be solved, this section

explains characteristics of micro and small scale enterprises and constraints and opportunities

and factors that determine employment growth within MSEs.

101

4.16.2.1. Characteristics of Micro and Small Scale Enterprises

Among the surveyed sample MSEs from three cities namely, Addis Ababa, Hawassa, and

Bahir Dar, the average number of male and female operators are nearly four and three,

respectively per an enterprise. It suggests that relatively male are more likely to undertake

small business activities than women in the three survey areas. Majority of the business

operators are non-teenagers; the share of teenagers is only 16 percent. The size of middle level-

skilled operators is more than those with non-formal education. Only 6.8 percent of the sample

operators have non-formal education. Those with primary education (grades 1-8) and with

secondary education (grade 9-12) constitute about 39.6 and 30 percent, respectively, of the

sample operators. Those with educational qualification beyond secondary education but below

first degree are 9.9 percent while operators attended first degree and above degree education

are more than 1.6 percent. There are at least 6 literate operators per small business and there is

one illiterate operator per business firm.

Regarding the training status of operators, about 55 percent of them have taken some sort of

training; and at an enterprise level, on average, there are at least four persons who received

training. However, out of those with training only a few of them took long term training

provided by colleges or universities. Nearly 74 percent of the owners with training have short

term training provided by organizations and experts that facilitate and oversee the performance

of MSEs. Among those with training, only 15 percent of owners of MSEs in the sample survey

are diploma holders who graduated from TVET colleges. Among which, about 4 percent of

them earned 10+1, 3 percent of them earned 10+2, and about 8 percent of them earned 10+3

levels of diploma in various technical and vocational education and training institutes.

Out of 55 percent of the sample operators that have attained some sort of training, including

short term training, almost 89.6 percent of them are found using the training they received to

run their current businesses. They are involving in businesses related to their field of training.

This might be because most of the trainings received are short term trainings provided by

experts working for MSE development. These trainings are compulsory to license enterprises

and for business startups. However, many of those with long term training are involved in

102

businesses other than their field of study. For instance, out of fourteen operators with

agriculture related professional science (bachelor degree), only one person is found engaged in

urban agriculture activity; and out of forty with health training at degree level, nobody is found

involved in businesses related to his/her study. The implication is that the ongoing long term

training programs are less likely contributing to entrepreneurship relative to short term

trainings. Above all, it suggests the mismanagement of scarce public resources and the weak

linkages and coordination between training institutions and the industry; and hence, reminds all

concerned bodies to further evaluate the ongoing efforts and to take appropriate measures.

A. Type and composition of activities

The major activities undertaken by micro and small scale enterprises in the survey area are

classified in to four categories of enterprises: manufacturing, trade and service, urban

agriculture and construction. Manufacturing related enterprises constitute 33 percent, trade and

service related enterprises constitute 13 percent and 27 percent respectively, urban agriculture

estimated about 11 percent and construction enterprises comprise 15 percent of the sample

MSEs. Commonly, micro and small enterprises are perceived as traders and vendors. As

opposed to the common perception, the survey result suggests that among micro and small

scale enterprises three types of activities are identified as the most important categories. Food

and beverage activities account for 17 percent, metal work and wildering accounts for 15

percent and production and supply of construction materials accounts for 15 of the MSEs

production.

B. Location and ownership of enterprises

Proportionate to enterprises concentration in Addis Ababa, over half of the MSEs included in

this study are taken from Addis Ababa while the remaining 48 percent of the sample MSEs are

equally shared between Awassa and Bahir Dar. Regarding ownership status, nearly 36 percent

of owners are individuals, 36 percent are owned by proprietors consisting of group of people

between two and ten, and the remaining 28 percent are owned by cooperatives. Most of the

cooperatives are involved in construction and manufacturing activities. Service and

103

manufacturing activities are dominated by proprietors and individuals. Urban agriculture is

controlled by proprietors while most of the trade activities are run by individual owners.

4.16.2.2. Employment Contribution of MSEs Micro and small scale enterprises are fundamentals for the emergence of entrepreneurs.

Because they have considerable potential to generate large employment opportunities and to

contribute to the entire economy, they deserve special attention and support for their growth

and development. In this regard, those agencies established at various levels, from federal to

local levels, are expected to play a pivotal role in promoting and strengthening these

companies.

Micro and small scale enterprises (MSEs) have been dominant source of employment and

income in many countries of the third world. A quarter of working age people are get

employed in MSE activities. Even in USA, some scholars make a case that eight out of every

ten new jobs opportunities in recent years have been created by small businesses (Mead and

Liedholm, 1998). Employment contribution of MSEs is remarkable in urban Ethiopia while we

consider a firm as MSE when its employment size is not more than. According to CSA survey

result, in 2006, almost 76 percent of the work force had been employed in Micro and small

scale enterprises (MSEs). The employment share of MSEs increased to 83 percent in 2010-11.

Figure 4.14: Employment contribution of MSEs (%)

0

20

40

60

80

100

120

1999 2003 2004 2005 2006 2010 2011

MSE employment> 10 workers

Source: UEUS 2003-11

104

The primary data survey from the three cities (Addis Ababa, Bahir Dar and Hawassa) indicated

that out of the total 4,554 people employed in the sample MSEs in 2011, the employment

opportunities are largely contributed by MSEs engaged in manufacturing related activities

followed by the construction sector. As can be seen from figure 4.14 the employment share of

manufacturing enterprises accounted for 29.6 percent of the total employment followed by 27

percent employment contribution by construction sector while that of MSEs engaged in urban

agriculture accounted only for 12.6 percent. The relative employment contribution rate of trade

is lowest.

Figure 4.15: Employment by type of MSEs (%)

0

5

10

15

20

25

30

currentperiod

trade 

service 

manufacturing

construction

agriculture 

Source: survey data

Growth of employment is measured using the formula proposed by (Evans, 1987) calculated as

the ratio of change in natural logarithm of employment differential between initial and final

period divided by age of a firm. The initial period refers to year of establishment and the final

period is the survey period, October 2011. The result indicates that the employment growth is

positive for nearly 29 percent of MSEs, 21 percent of enterprises have negative employment

growth rate and 50 percent have zero growth rate since establishment. Highest negative

employment growth rate has observed for manufacturing enterprises estimated about 3 percent

followed by service sector accounts for 2.85 percent negative growth rate. However highest

positive growth rate of 4.5 percent observed for manufacturing enterprises followed by trade

accounts for 2.4 percent positive employment growth rate.

105

4.16.2.3. Startup Motives of MSEs

Majority of business operators, about 76 percent, started their own business because they

believe that self-employment has better financial returns than paid employment. On the other

hand, 18 percent of the operators responded that they started their own enterprise because they

had no another choice to get out of unemployment. Regarding the effect of training on

entrepreneurship, only 14 percent of the operators appreciated that the training they received

inspired them to be entrepreneur. The implication is that trainings provided are almost

ineffective in stimulating trainees, especially young people, to be risk takers and innovators. [

4.16.2.4. Constraints of Micro and Small Scale Enterprises

The operators’ entrepreneurial skills and supportive business environment are two important

determinants of success for business enterprises, where success is estimated in terms of

employment size and growth, income and profitability. In this part, we look at different

opportunities and constraints strongly linked to size and growth of employment within MSEs.

The opportunities and constraints comprise: source of startup capital, access to market, lack of

favorable government rules and regulations, lack of work place, shortage of and high prices of

inputs.

4.16.2.5. Market and Other Constraints to Expand Business

Constraints to expand business and then employment by micro and small scale enterprises are

indicated in table 4.15. Only 3 percent have responded that they have no problem to expand

their business. However for more than 54 percent of the enterprises lack of market is the major

barrier to increase the scale of their business. Relatively a moderate size of operators, about 29

percent, reported that lack of credit supply is the main challenge to expand their business and

hence employment. As noted earlier, own saving, usually a small amount, is the dominant

source of startup capital. The survey result indicated that only one enterprise in ten has

received loan from informal financial institutions and 3 enterprises in ten has borrowed formal

financial institutions. On the other hand, 45 percent of operators reported lack of capital as a

major barrier to expand their business and employment. In addition to this, 28 percent noted

that lack of equipment is challenge to expand operation. However, there are microfinance

106

institutions in each woreda that can lend money to new startups and existing ones. The overall

results suggest two important issues. First, the local communities may lack awareness and

skills on possibilities to finance their business through borrowing. Second, the opportunity cost

of borrowing money may be high at ongoing interest rate.

Table 4.11: Problems to expand business (%)

Problems to expand business Frequency Percent No problem 14 3.20 Lack of market 238 54.46 Lack of credit 125 28.54 Shortage and high price of inputs 203 46.35 Lack of information 87 19.86 Lack of equipment 121 27.63 Lack of capital 200 45.77 Higher business taxes 65 14.84 Government rules and regulations 77 17.58 Competition with other firms 75 17.16 Lack of work place 170 38.90 Lack of training 83 18.95 Shortage and high wage of labor 64 7.53 Other problems 33 7.53

Source: Primary survey

Shortage and high prices of inputs is impediment to 46 percent of enterprises to increase scale

of their production.

Most of enterprises supply their products to consumers. Over 82 percent of entrepreneurs sell

their products to consumers, followed by 24.5 percent of small firms supply their product to the

government organizations. Enterprises estimated around 26 percent trade their products with

retailers and whole sellers, manufacturers, and exporters. Hence to overcome the shortage of

market there might be possibilities to increase income of consumers or introducing strategies to

increase demand by manufacturers and traders. Most of the input suppliers are traders. They

supply inputs to 80 percent of entrepreneurs while producers, government organizations and all

others supply inputs to only 36 percent of small firms. A few enterprises buy inputs from

multiple suppliers. Therefore it is important to increase supply share of suppliers other than

traders to realize approximately competitive price to overcome production interruption

107

associated with shortage and rise of input prices. The source of input is dominated by domestic

industrial products for most producers (52%) followed by agricultural outputs (39%). Over 81

percent of producers depend on domestic agricultural products, imported inputs and natural

resources. The implication is that the row material supply is most likely exposed to weather

shocks and deficits since agricultural export is dominant source for foreign earning. On the

other hand there is an opportunity for strong linkages between agriculture, industry and MSEs.

There is also a possibility to decrease seasonal causes of job interruption by intensifying the

use of domestic industrial outputs as sources of raw-materials.

4.16.2.6. Source of Startup Capital and Capital Growth

The major source of startup capital for most entrepreneurs is personal savings and / or ‘equeb’.

Personal savings is a source of startup capital for more than 69.6 percent of operators, followed

by borrowing from financial institutions, which is used as source of fund by 31 percent. From

each of the remaining sources such as inheritance, donation from government and NGOs,

borrowing from village lenders and others means not more than 10 percent of firms generate

start up capitals. The average initial capital was Birr 55,426.37 while the current capital

significantly increased to Birr 194, 012. 8. The average growth of capital nearly 75 cents per

enterprise is considerably higher than the average growth of employment per enterprise

estimated about 0.04. However, overall growth rate of both capital and employment are not

satisfactory and the mean capital growth rate of firms with positive employment growth is not

significantly different from firms with negative employment growth rate.

Own saving is a source of startup capital to 73 percent of firms that have positive employment

growth rate. Where as borrowing from microfinance institutions is source of finance to 35

percent of micro and small scale enterprises with intended employment expansion rate.

Similarly majority of firms with negative employment growth rate, for example 68 and 29

percent use own saving and borrowing from micro finance institutions respectively to generate

funds. Each of the remaining means of finance are used to raise funds not more than 12 percent

of firms that have either positive or negative employment growth rate. The implication is that

formal financial institutions are almost equally accessible to firms that have positive or

negative employment growth rate and less likely used by both forms of firms to raise startup

108

fund relative to own saving /equib. This may be attributed to enterprises access to formal

financial institutions may be limited due to collateral and other requirements and hence

employment creation.

4.16.2.7. Cause of Job Interruption

To determine the level and causes of job interruption, we analyzed months that operators were

producing for last 12 months. Over half of the firms operated for the last 12 months, 84 percent

of enterprises operated at least for last six months. Estimated figures of 5 percent of firms

operated only for a month. The mean months of operation of firms with desirable employment

growth rate are not remarkably different from firms with undesirable rate within 5 percent

significance level. The average interruption is nearly 3 months and it needs strong attention to

reduce its adverse impacts on employment. There are ample of causes for disrupting operation

for three months in average in the last 12 months before the start of the survey.

Lack of demand or market is a dominant cause of ceasing operation followed by shortage of

capital. Shortage and/or rise of price of a raw-materials and seasonal nature of business are also

among the causes forcing firms to halt operation. Sufficient supply of Power, water supply and

electricity are crucial for success of MSEs. However a few operators, about 3 enterprises in 50

reported that shortage of electricity is challenge for their success. Polices may

disproportionately affect MSEs. First polices tend to favor larger businesses-export oriented

businesses.

Table 4.12: causes of job interruption

Causes of interruption Frequency Percentage

Shortage or a rise of raw material prices 68 15.3

Lack of demand or market 106 23.8

Shortage of electricity and water 27 6.1

Unfavorable government rules and regulations 35 8.0

Seasonal nature of the business 66 14.83

Shortage of capital

Others

76

59

17.1

13.3

Source: Primary survey

109

Second start up and transaction costs for doing business are high due to complex administrative

systems and often due to corruption. However the survey result has shown that unfavorable

government rules have least impact on job interruption. Only 2 enterprises in the 25 reported

that government policies are reasons for job interruption. It suggests that market creation and

competitive supply of raw materials are important interventions to support micro and small

scale enterprises. Supply of sufficient credit is also worth mentioning to increase employment

and the productivity of labor and capital.

4.16.2.8. Assistance Needed from Government

MSEs are in need of different assistances to expand their capacity and hence employment.

However the survey result identified six major supports required from government. Nearly 75

percent of firms insist on government access to work place and over 74 percent of operators

responded that market creation and networking is vital for expansion.

Table 4.13: Assistance needed from government

Assistance needed Frequency Percentage

Access to working place 330 74.83

Access to building in rent 123 28

Market access 327 74.32

Access to raw materials 233 53.2

Access to technical training 246 56.29

Better access to bank loans 258 58.77

Favorable government rules 164 37.44

Safety and operation rights 137 31.42

Access to business information, advice

and account keeping

258 58.64

Other assistances 28 6.36

Source: primary survey

Nearly 58 percent requires supports like better access to bank loans, and equal size of

respondents needed supports such as access to business information, advice and account

110

keeping from government bodies. Supports for instance access to technical training is strongly

demanded by 56 percent of respondents and 53 percent of operators insist on local government

to facilitate access to raw-materials (Table 4:13).

The remaining supports required by operators from the government are favorable government

rules (37 percent), access to building in rent (28 percent), safety and operation rights (31

percent), and other assistances (only 6 percent).

4.16.2.9. Determinants Urban Employment Growth within MSEs

The logistic regression analysis is used to identify factors that determine urban employment

growth within MSEs. The dependent variable is the average annual growth rate of employment

equals one when the rate is positive since startup of the firm and zero otherwise. Estimated

around 29 percent of MSEs have positive average annual employment growth rate. We

employed different regression diagnostic tests such as multicollinearity, specification test and

estimated robust standard errors. We dropped some variables for specification and

multicollinearity problems. However for the remaining controlled explanatory variables we

estimated robust standard errors. The results from logistic regression analysis used to

determine employment growth indicated that business interruption, type and ownership of

MSE, firm size, location, and motive to start up MSE have important effect on employment

growth, however, human capital endowment do not have any significant effect on

unemployment.

The logistic regression result indicates that the model is statistically significant because the p

value is below one percent. While we are interpreting odds ratio of a statistically significant

variable it refers that odds ratios are equal to 1 if there is no effect, smaller than 1 if the effect

is negative and greater than 1 if it is positive. Average annual employment growth of firms

located in Bahir Dar is 2.68 times higher than those located in Addis Ababa. Employment at

start up leads to lower probability of employment growth. Enterprises with relatively more

initial employment size have lower probability of employment growth (annex table 4.18).

111

Relative to firms involving in trade activities agricultural and manufacturing enterprises have

higher probability of employment growth. For example, the odds of employment growth for

manufacturing firm are 3.4 times higher than trade enterprise. The odds of employment growth

for urban agriculture are 3.9 times higher than trade enterprises. Average magnitude of work

hours has positive effect on the likelihood of employment growth. For every one unit increase

in work hours leads to, an enterprise’s odds of employment growth 1.02 times higher. New

ventures established by cooperatives are more likely to grow faster than those funded by single

owners. Motivations have been found to influence new firm growth. Self-employment

preferred to paid employment as a motivation for starting a business positively affect firm

growth however unemployment and /or training as a motivation to start a business has no

significant effect.

On the other hand, shortage of electricity and water services, lack of good government rules

and lack of capital collectively used as proxy for policy shocks lead to 0.51 times lower odds of

employment growth. The entrepreneurs human capital acquired often considered as good to

his/her likely success. However the index of human capital in the firm’s estimated in terms of

proportion of owners with training, TVET and other different educational qualifications has no

significant effect on probability of employment growth within MSEs. Moreover social ties

(support from families) and access to credit supply are unexpectedly not important for growth

of new ventures.

112

5. CONCLUSIONS AND RECOMMENDATIONS

5.1. Conclusions

Ethiopia’s urban unemployment rate is significantly higher than national unemployment rate

between 1999 and 2005. These considerable rates of unemployment differential maintained for

youth, females and adults as well over the period. The rate trended downward for all categories

of labor force since 2003, but remained at high level. The composite unemployment rate reach

peak of nearly 26 percent in 2003, but decreased to its lowest level 17 percent in 2006. This

might be attributed to decrease in overall labor force participation rate in the period. Despite

the sound economic growth, the urban unemployment rate is still higher and stood around 18

percent in 2011.

The youth and female unemployment rates in urban Ethiopia are considerably higher than adult

male unemployment rate and well above the total urban unemployment in each period from

2003 to 2011, which is consistent with the global experience and reflecting the relative

disadvantaged position of youth and female in labor markets. The age and sex disparity in

unemployment is maintained across all cohorts in urban job markets of Ethiopia. For example,

youth and female are threefold more likely to be unemployed than adult male between 2003

and 2011. The situation is the wrest for young female labor force.

There are several factors contributed to the high unemployment disparity. High job interruption

of women due to maternity leave and childcare; low educational qualification of women

relative to their male counterparts and labor market discrimination and prejudice are commonly

cited ones. However existing evidences indicate that adult male and female have almost equal

job interruption rates. Nevertheless obviously males have more educational qualification than

females and a labor market prejudice may adversely affect women employment opportunities.

On the supply side, youth extend job search until they secure better job as they are lucky for

family support during spell of unemployment. Family support in 2005 is substitute for

unemployment benefit for 75 percent of unemployed youth in Ethiopia whereas it is means of

access to basic needs for 38 percent of adult male in urban areas hence many youth are more

113

likely to remain unemployed and shop around for better job than adult males. Contrary to many

developing economies, lack of labor market information and poor job search experience of

youth relative to adult does not result in higher youth unemployment rate in Ethiopia. Youth

also switch between job, school enrollment and unemployment as educational institutions open

and close leads to young students more likely enter and exit the labor force. Not only supply-

driven causes, but also labor market partiality causes youth to face higher unemployment rates

than adults. However the primary survey result from three cities contradicts with the forgoing

result. This may be attributed to sample size, coverage of survey and survey period.

As far as regional unemployment differences are concerned the two city administrations Dire

Dawa and Addis Ababa are hit hard by extremely unpleasant rate of unemployment over the

survey period. The unemployment rates observed in these cities are significantly above the

national average and all other regions. Gambella region is relatively lucky for having the

lowest relative unemployment rate in all periods. The trend of unemployment across regions is

almost synonymous with the national trend. Like national rate regional female and youth

unemployment rates are higher than adult male.

Many international facts proof the notion that additional education boosts labor market

outcomes such as better earning and lower unemployment. However the effect of educational

attainment on urban unemployment in Ethiopia is going wrong except for degree and above

educational achievement as opposed to the international experience. Considering lower

primary education (grade 1 to 4) as a baseline only degree and above graduates and non-formal

education have significantly lower rate of average unemployment rate than labor force

participants with baseline education between 2003 and 2011. Labor force with all other

educational qualification such as upper primary, secondary education, certificate and diploma

and degree not completed including vocational education not completed results in higher

average rate of unemployment in the typical period than labor force with baseline educational

qualification. Unemployment positions of TVET graduates are almost as equal as base level

category. Unemployment peaks through secondary to preparatory education.

114

At all educational levels female and youth experience a higher level of unemployment rate than

adult male. For example, youth females mean annual unemployment rate is estimated to be

more than threefold of adult men. It implies that the unemployment rate discrepancy between

female, youth and adult is non-declining up the educational ladder relative to baseline

education.

Extended unemployment spell permanently weaken an individual’s productive potential and

human capital and hence employment opportunities however long spell of unemployment is

one of the features of Ethiopia’s urban unemployment. The average duration of unemployment

keeps at high level between 2003 and 2011. The unemployed remain jobless for nearly 2.4

years in the initial period and elevated to 2.5 years in 2004 and the lowest level of spell nearly

1 year and six months observed in 2010. The unemployment rate in 2003 is not only found to

be the highest but also ends for long duration relative to other periods while someone linking

the rate with the spell of unemployment. The average spell of unemployment is not gender

impartial over the period. The most challenging fact is that the spells of unemployment

increases with years of schooling and statistically significant. Even if, those with training have

significantly lower rate of spell than those without training considerable size of workers who

has received training also remains unemployed for more than 2 years based on secondary data.

However the mean spell of unemployment variation along educational ladder is significant at 1

percent while the variation disappears for those with training in three cites where primary

survey is conducted.

The effect of training on labor market outcome is significant over the 2003 to 2011 period.

Working age population with training face relatively lower unemployment challenge and more

likely to be employed than those who lacked the opportunity. Nevertheless, training failed to

reduce unemployment differentials between female, youth and adult male.

Combining the results from primary and secondary survey among people with the equal level

of education except for vocational and technical training received we do not have sufficient

evidence to generalize that those with vocational training are less likely to be unemployed in

all periods.

115

The share of self-employment by TVET graduates is less than the comparison groups known as

secondary school graduates. Furthermore the trend of self-employment for technical and

vocational graduates is falling overtime while constant for secondary school graduates.

Regarding overall formal training, labor force without training are more likely to be self-

employed than those with training between 2003 and 2011.

School to work transition of urban youth in Ethiopia is long and difficult. The average time to

find first job varies significantly by training status. Mean time wasted to find first job after

quitting schooling by those with training decreased to ten and half months, while it takes

twenty two months to those without training. Average time spent for those with technical and

vocational training is nearly eleven months where as for those with grade 9-10 educational

qualification it takes nearly one year and four months. The mean school-to-work transition

differential is significant at 1 percent level. Those who have educational qualification of grade

11 to 12 need to wait two years to find their first job after they enter into the labor market.

Ethiopia’s fastest growth among non-oil economies for a decade does not result in equivalent

employment creation. Rapidly growing urban population arising from rural-urban migration,

lack of vibrant non-agricultural sector to absorbed surplus labor migrated from agricultural

sector are threats of unemployment. What school system thoughts does not required by labor

market and low level of human capital exacerbate the situation however effects of FDI

attraction on unemployment is mixed.

Existing evidences prove that skill mismatch, job destruction and queuing theories of

unemployment contributes to higher probability of unemployment in urban Ethiopia. The Cox

regression, after controlling for set of explanatory variables, consistent with the skill mismatch

theory. The worker with more years of schooling is less likely to end unemployment spell over

a short period. The implication is that additional years of schooling results in more total

duration of unemployment in the last six months before the survey period. Similar result was

obtained while we replace explanatory variables years of schooling by dummies of educational

levels. We also examined the relationship between probability of unemployment and years of

schooling and education further to test the validity of skill mismatch hypothesis using probit

116

regression. The result is synonymous with results find from the duration analysis. The probit

regression result supports both the queuing and job destruction hypothesis. The positive and

statistically significant relationship between the share of both public and formal private sector

job and the probability of being unemployed is found to strongly support the queuing

hypothesis. The positive and significant relationship between regional share of private formal

sector employment and probability of unemployment is support for job destruction.

Efforts are made to evaluate the implementation of the current TVET strategies in terms of

innovativeness, feasibility, responsiveness, relevance, flexibility, upscale and coordination

through the opinion of relevant respondents using a Likert scale inquiry. About 75 percent of

the respondents agreed that the TVET program is innovative in that it addresses the limitations

of other training programs. On the other hand, the respondents opinion casts doubt on the

feasibility of the program. The average result indicates that nearly 59 percent of the

respondents are uncertain and disagree regarding the feasibility of the TVET program. Lack of

qualified staff and shortage of budget to supply equipments and furnish laboratories are the

major factors behind the problem.

In terms of responsiveness of the program, the overall evaluation depicts that 60 percent of

respondents agreed that the program is responsive while the remaining 40 percent disagreed or

were uncertain about the responsiveness of the program. Apparently, the opinion of the

respondents suggests, the need to invest more in these institutes so as to increase their

responsiveness to participants that marginalized in other programs. With regard to the overall

relevance, the average measure score shows that more than half of the respondents are

uncertain or disagreed with the relevance of the TVET program. This has also been reflected in

one way or another in the descriptive and regression analysis of the effect of training on

unemployment. Hence, this could be an important message that reminds concerned bodies to

give attention and work hard to increase the relevance of the TVET program in order to meet

the technical skill demand of the labor market.

Concerning the flexibility of the program, about 57 percent of the respondents did not agree

that TVET institutions are flexible and capable to make timely adjustment of training programs

117

to meet the changing skill demands of the labor market. This suggests that TVET institutions

must be aware of the low flexibility of their programs and should work toward its

improvement. Similarly, the average result regarding the efficiency and effectiveness of the

program suggests that about 43 percent of the respondents do not agree that the TVET program

is efficient and effective. It implies that more effort is needed to improve the situation.

Another important area of concern is coordination among relevant respondents. Linking

vocational training to the market demand is crucial to ease school to work transition of the

youth. However, the evaluation result indicates the disagreement of over 56 percent of the

respondents on this issue; and also the participation of the private sector, which is the dominant

employer of the graduates, is found to be weak.

The probit regression analysis on the effect of education on unemployment shows additional

education starts to decrease the likelihood of unemployment when at least a person has college

degree. That is, education qualifications above upper-primary (grade 5-8) and below first

degree education level are positively related with the probability of unemployment.

Particularly, junior secondary education (grade 9-10) and preparatory education (grade11-12)

appear to increase chance of being unemployed by 11.7 percent and 12 percent, respectively

relative to lower primary education (grade 1-4). The same is true for other categories of

education, above secondary and below university degree, in which the probability

unemployment rate is positive. It is only for those with degree and above that the probability of

unemployment is negative. The implication is that the employment effect of education at

individual level is more pronounced at tertiary level of education. The result is almost similar

to earlier studies on the same area. Probability of unemployment is decreasing overtime

relative to base year and migrants are less likely to be unemployed than non-migrants. Location

is also statistically important determinant of occurrence of unemployment. Access to credit is

likely to decrease unemployment.

The survey result asserts the desirable effect of general training. Labor market participants who

received formal training are 9.9 percent less likely to be unemployed and statistically

significant at one percent level compared with those without training. The employment effect

of TVET is not consistent and decreasing after sufficient experiences of technical and

118

vocational training implementation. Its contribution to employability is not improving even as

completion of grade ten general schooling.

The entrepreneurial effects of education and training in general and TVET in particular are not

promising. University and TVET graduates are less likely involving in self-employment

activities than people with primary education. Additional years of schooling decreases

likelihood of individual’s involvement in self-employment businesses. However after eight

years of practices entrepreneurial effects of TVET and completed grade ten general schooling

is promising. Accesses to credit, TVET and university education make shorter school to work

transition period.

Indeed, it is undeniable that training has desirable effects on the individuals’ performance and

productivity. Training would make people more successful and more productive if they apply it

in their day to day business, otherwise it would depreciate and become obsolete. What is

observed from the survey of the three cities is that most of MSE operators take some sort of

short term training, 89.6 percent of them are found to apply it to run their current businesses.

However many of those with long term training are involving in business other than their field

of study. This can be indication of the weak linkages and poor coordination between training

institutions and the industry in general for long term trainings. Besides, regarding the effect of

training on entrepreneurship, the survey result shows that only 14 percent of MSE operators

appreciated that the training they received inspired them to be entrepreneur. This suggests that

training providers are ineffective in inculcating the spirit of entrepreneurship among trainees.

According to CSA’s employment unemployment survey, the employment contributions of

MSEs are significantly larger than employment opportunities created in large-scale enterprises

and in public sector in 2006 while the latter sectors employment contribution exceeds the

contribution of small enterprises for periods 2010 and 2011. This suggests that the relative

employment contributions of MSEs are declining. The sample survey in the three cities

indicates that MSEs involved in manufacturing and construction related activities constitute the

largest share of employment opportunities for the urban labor force. The other types of MSEs

engaged in service, urban agriculture and trade contribute 20.8 percent, 12.6 percent and 10

119

percent of the total employment, respectively. The overall result indicates that the employment

growth is positive for nearly 29 percent of MSEs, 21 percent of enterprises have negative

employment growth rate and only 50 percent have zero growth rate since establishment. The

average growth of capital nearly 75 cents per enterprise is considerably higher than the average

growth of employment per enterprise estimated about 0.04. However, overall growth rate of

both capital and employment are not satisfactory.

Even if own saving and borrowing from formal financial institutions are dominant source of

startup capital, rate of positive employment growth in these firms are almost as equal as

negative growth. The implication is that access to formal financial institutions does not result

in significant contribution to employment generation. This may be attributed to enterprises

access to formal financial institutions are limited due to collateral and other requirements.

The expansion of scale of production and hence growth of employment in the sample MSEs

has been lower. A number of barriers are identified as impeding the expansion of scale of

production and growth of employment with in MSEs. Insufficient market, shortage and high

prices of raw materials, lack of capital, lack of working place, and lack of credit are among the

major constraints responsible for the poor performance of MSEs in expanding employment.

Furthermore, enterprises did not fully utilize their capacity, and thus experienced job

interruption in the last 12 months just before the survey. They ceased operation on average for

3 months in the last 12 months. About 24 and 15 percent of MSE operators report that lack of

demand and shortage and/ or high prices of inputs are among the major reasons for ceasing

operation and hence job interruption. Interestingly, shortage of electricity and water as well as

unfavorable government rules and regulations are the least frequently cited reasons. Only less

than 6 percent of operators report them as challenges for their business expansion. This may

witness the ongoing effort of the government in supporting and promoting the development of

MSEs through its pro-MSE policies and strategies.

The logistic regression analysis on the determinants of employment growth within MSEs

indicates that employment growth within MSEs is found to vary with location, type and

120

ownership of enterprise, initial employment size, and average hours of work and motive for

business start up. For example, the probability of employment growth within MSEs located in

Addis Ababa is lower than those located in Bahir Dar. In addition, enterprises with relatively

large initial employment size have lower probability of annual average employment growth.

On the other hand, enterprises with relatively more hours of work have higher likelihood of

experiencing employment growth. Unexpectedly, the human capital endowment of new firms,

social ties and access to credit do not have significant effect on the growth of the firm.

5.2. Recommendation In the ensuing section, policy implications that are supposed to be relevant to addressing the

problem of urban unemployment based on the findings. Ethiopian urban labor market is

characterized by high and persistent unemployment. Although the rate declined from 26

percent in 2003 to 18 percent in 2011, it is still a cause for concern.

The skill-mismatch, queuing, and job destruction tendencies to increase unemployment can be

addressed by aligning the education and training polices to the needs of the labor market and

also by developing entrepreneurship oriented curriculum.

The effect of education on the labor market outcomes of individuals is not straightforward. The

real employment effect of education at individual level is more pronounced at tertiary level of

education. Therefore we suggest not only introduction of area specific demand driven nation

wide training to unemployed in general and school dropouts in particular that may be financed

by contribution from local community, government and private sector, but also deliberate

employment creation by government in service and industrial sectors. The government needs to

introduce training to school dropouts and unemployed that focus on development of tradable

skills and admission of dropouts to short term training at different levels regardless of

educational qualification.

Training in general has desirable effect on some of labor market outcomes of individuals.

However, it makes no significant difference in reducing gender and age disparity of

121

unemployment and in encouraging self-employment. We suggest expansion of training

programs that targeted the youth and females to facilitate school- to-work transition.

Furthermore, the higher unemployment rate and lower self- employment tendency among

TVET graduates is another issue of concern albeit such trainings result in earlier school-to-

work transition period. On the other hand, the 2008 national TVET strategy document

identified important limitations and strategies to improve the performance of the sector.

However majority of respondents are blamed the program for its feasibility, flexibility,

coordination, scale up and relevance. Most of the beneficiaries have positive attitude towards

the innovativeness, responsiveness, efficiency and effectiveness of the program; nevertheless

evaluation scores are average for these indicators as well. The commitment of bodies in charge

for effective implementation of strategies identified is sufficiently enough to enhance the

employment effect of TVET. Since the ongoing practices are not sufficiently employing

strategies in place. Increasing access to credit at reasonable interest rate is also important for

new business startups by graduates and school drop outs at any level for unemployment

reduction. The employment effect of MSEs is found to be insignificant and only one third of them

registered positive employment growth since start up. As long as the objective of promoting

and supporting the development of MSEs is to make them expand demand for the growing

labor force, therefore their success and performance should be evaluated in terms of the

employment growth they achieve. In this regard, as the finding suggests, support in terms of

market for their products and easy access to supply of raw materials, access to work place and

bank loans are required to help them expand their business and increase their demand for labor.

The government can also devise encouraging mechanisms such as rewarding private firms

based on their size of employment and intensive use of labor intensive technologies.

Furthermore, it could also be possible to consider employment creation of a firm as a criteria

item for increased access to government credits to private firms.

Encouraging ventures established by cooperative are more important than individual

enterprises, and introduction of polices that lead to expansion of manufacturing and urban

122

agricultural enterprises. Concerned bodies must be cautious about nil employment effects of

human capital endowments acquired by MSEs, social ties and access to credit.

Experiences of active labor market polices in Ethiopia are mostly limited to subsidized training

and occasional government employment services. The Government should introduce official

government employment program and other active labor market polices too.

123

References ADB 2010. Ethiopia Economic Growth Performance: Current Situation and Challenges. Economic Brief 

vol 1(5). Addis Ababa,  African Development Bank. ADEBAYO,  A.  1999.  Youth  Unemployment  and  National  Directorate  of  Employment  and  Self 

Employment Programs. Nigerian Journal of Economics and Social Studies 41, 81‐102. AKTAR,  I. & OZTURK, L. 2009. Can Unemployment be Cured by Economic Growth and Foreign Direct 

Investment in Turkey? International Research Journal of Finance and Economics  ALEMAYEHU, G., BEFEKADU, D. 2005. Explaining African Economic Growth Performance: The Case of 

Ethiopia. AERC Working Paper. ALTAVILLA,  C.  A.  C.,  F.  E.  2006.  Evaluating  the  Dynamic  Effects  of  Active  Labour  Policies  in  Italy. 

LABOUR 20  349‐382. ARUM, R. & SHAVIT, Y. 1995. Secondary Vocational Education and the Transition from School to Work. 

Sociology of Education, 68. ATCHOERENA, D. A. D., A. 2002. Revisiting Technical and Vocational Education in sub‐ Saharan Africa: 

An update on trends, innovations and challenges. Paris: UNESCO‐IIEP. BAKARE,  A.  S.  2011.  The  Determinants  of Urban Unemployment  Crisis  in Nigeria:  An  Econometric 

Analysis.  Journal  of  Emerging  Trends  in  Economics  and Management  Sciences  (JETEMS),  2, 184‐192. 

BENCIVENGA, V. R. A. S., B. D. 1995. Unemployment, Migration, and Growth. Working Paper No 561, Federal Reserve Bank of Minneapolis. 

BENNELL,  P.  1999.  Learning  to  Change:  Skills  Development  Among  the  Vulnerable  and  Socially Excluded in Developing Countries. Employment and Training Department. Geneva: ILO. 

BETCHERMAN, G., GODFREY, M., PUERTO, S., ROTHER, F. AND STAVRESKA, A. 2007. Global Inventory of Interventions to Support Young Workers. Synthesis Report. The World Bank. 

BIRHANU, D., ABRAHAM, T. AND VAN DER DEIJL,  H. 2005. Characteristics and Determinants of Youth Unemployment,  Underemployment  and  Inadequate  Employment  in  Ethiopia.  Employment  Strategy Papers. Geneva: ILO. 

BOONE, J. A. V. O., JAN C. 2004. Effective Active Labour Market Policies. Discussion Paper No. 1335, October 2004.  Institute for the Study of Labor (IZA) 

BREWER, L. 2004. Youth at Risk: The Role of Skills Development in Facilitating the Transition to Work. Geneva, Switzerland. 

CALMFORS, L. 1994. Active Labour Market Policy and Unemployment ‐ A Framework for the Analysis of Crucial Design Features. OECD Economic Studies, 7‐47. 

CAPELLERAS, J.‐L. & RABETINO, R. 2008. Individual, Organizational and Environmental Determinants of New Firm Employment Growth:Evidence from Latin America. Int Entrep Manag J, 4, 79‐99. 

CARD, D. A. K., A. B. 1993. Minimum Wages and Employment: A Case Study of the Fast Food Industry in  New  Jersey  and  Pennsylvania. Working  Paper  No.  4509.  National  Bureau  of  Economic Research, 1050 Massachusetts Avenue, Cambridge, MA 02138, October, 1993 

CASTRO,  C.  A.  V.,  A.  2000.  Training  Unemployed  Youth  in  Latin  America:  Same  old  sad  story?  In: ATCHOARENA, D. (ed.) The Transition of Youth from School to Work: Issues and Policies. Paris: International Institute for Educational Planning (IIEP). 

CHAUDHURI,  S.  &  BANERJEE,  D.  2010.  FDI  in  Agricultural  Land, Welfare  and  Unemployment  in  a Developing Economy. Research in Economics, 64, 229‐239. 

124

CRESWELL,  J. W. 2009. Research Design: Qualitative, Quantitative, and Mixed Methods Approaches, SAGE Publications, Inc. 

DICKENS, W. T. & LANG, K. 1995. An Analysis of the Nature of Unemployment  in Sirlanka. Journal of Development Studies 31, 620‐636. 

DIECKHOFF, M. 2007. Does it Work? The Effect of Continuing Training on Labour Market Outcomes: A Comparative  Study  of Germany, Denmark,  and  the United  Kingdom.  European  Sociological Review, 23, 295‐308. 

DRIFFELD, N. &  TAYLOR,  K.  2000.  FDI  and  the  Labor Market: A Review  of  the  Evidence  and  Policy Implications. Oxford Review of Economic Policy, 16. 

ECA 2005. Economic Report on Africa 2005: Meeting the Challenges of Unemployment and Poverty in Africa Addis Ababa: United Nations Economic Commission for Africa. 

ECA  2010.  Economic  Report  on  Africa  2010:  Promoting  High‐Level  Sustainable  Growth  to  Reduce Unemployment in Africa. Economic Commission for Africa, Addis Ababa 

EEA 2007. Report on the Ethiopian Economy. volume V, 2005/06. Addis Ababa:: Ethiopian Economic Association. 

ELHIRAIKA, A. B. 2011. Promoting High‐Level Sustainable Growth to Reduce Unemployment in Africa. Policy Brief No. 002, 2011. Addis Ababa: Economic Commission for Africa. 

ELKAN, W. 1970. Urban Unemployment in East Africa   Royal Institute of International Affairs, 46, 517‐528. 

EVANS, D. S. 1987. Tests of Alternative Theories of Firm Growth, . Journal of Political Economy 95, 657‐674. 

FASIH, T. 2008. Linking Education Policy to Labor Market Outcomes. The World Bank. FDRE  2010.  The  Federal  Democratic  Republic  of  Ethiopia  Growth  and  Transformation  Plan  (GTP), 

2010/11‐2014/15.  November  2010,  Addis  Ababa.  Ministry  of  Finance  and  Economic Development. . 

FRANK, C. R. 1968. Urban Unemployment and Economic Growth  in Africa. Oxford Economic Papers, New Series, 20, 250‐274. 

GARCIA, M. & FARES, J. 2008. The Effect of Education on Income and Employment. In: GARCIA, M. & FARES, J. (eds.) Youth in Africa’s Labor Market 

Washington, DC World Bank. GETINET, H. A. 2003. The Incidence of Youth Unemployment in Urban Ethiopia. Paper presented at the 

2nd EAF Conference, July 2003, Addis Ababa. GLEWWE, P. 1987. Unemployment  in Developing Countries: Economist's Models  in Light of Evidence 

from Sir lanka. International Economic Journal, 1, 1‐17. GUARCELLO,  L.,  LYON, S. AND ROSATI, F. 2008a. Child  Labor and Youth Employment  in Ethiopia  In: 

GARCIA, M., FARES, J. (ed.) Youth in Africa's Labor Market. Washington D.C.: World Bank. GUARCELLO, L., MANACORDA , M., ROSATI, F., FARES, J., LYON, S., AND VALDIVIA , C. 2008b. School‐

to‐Work Transitions: Regional Overview  In: GARCIA, M., FARES,  J.  (ed.) Youth  in Africa Labor Market. Washington D.C: The World Bank. 

GUNATILAKA, R. A. V., M 2010. Labor Market Institutions and Labor Market Segmentation in Sri Lanka. In:  GUNATILAKA,  R.,  MAYER,  M.  AND  VODOPIVEC,  M.  (ed.)  The  Challenge  of  Youth Employment in Sri Lanka. Washington D.C.: The World Bank. 

HENDERSON, V. A. P., W. 1991. Principles of Economics  D.C., Heath  and Company. HUJER, R., THOMSEN,  S.  L. &  ZEISS, C. 2006a. The Effects of  Short‐Term Training Measures on  the 

Individual Unemployment Duration in West Germany. Bonn. HUJER, R., THOMSEN, S. L. & ZEISS, C. 2006b. The Effects of Vocational Training Programmes on the 

Duration of Unemployment in Eastern Germany. Allgemeines Statistisches Archiv 90, 299‐321. 

125

ILO 2007. The Decent Work Agenda  in Africa: 2007‐2015. Eleventh African Regional Meeting, Addis Ababa, April 2007. Report of the Director‐General. Geneva, International Labor Office 

  ILO 2009a. Global Employment Trends: January 2009. Geneva. International Labour Office. ILO 2009b. Guide to the new Millennium Development Goals Employment Indicators: including the full 

set of Decent Work Indicators / International Labour Office. Geneva: ILO. ILO 2010. Global Employment Trends for Youth: Special Issues on the Impact of the Global economic 

Crises on Youth. Geneva. KAPSOS,  S.  2005.  The  Employment  Intensity of Growth:  Trends  and Macroeconomic Determinants. 

Employment Strategy Department, International Labour Office. KRISHNAN, P. 2001. Public sector pay and private sector wage premiums: Testing alternative models 

of wage determination. University of Cambridge. KRISHNAN, P., T. GEBRESELASSIE AND S. DERCON 1998. The Urban Labor Market   During Structural 

Adjustment: Ethiopia 1990 – 1997. Centre  for  the  Study of   African Economies,  Institute of Economics and Statistics, University of Oxford 

LEIBBRANDT, M.  A. M.,  C.  2004.  Youth  in  sub‐Saharan  Labour Markets.  Paper  for  the  DPRU/TIPS Conference:  “African  Development  and  Poverty  Reduction:  The  Macro‐micro  linkage”, Southern  Africa  Labour  and  Development    Research  Unit,  University  of  Cape  Town,  South Africa. 

MANKIW, G. 2001. Macroeconomics, USA, Worth Publishers. MEAD, D.  C. &  LIEDHOLM,  C.  1998.  Dynamics  of Micro  and  Small  Scale  Enterprises  in Developing 

Countries World Development 26, 61‐74. MOE 2008. National Technical and Vocational Education and Training  (TVET) Strategy. Addis Ababa: 

Ministry of Education. MOLSA 2009. Decent Work Country Programme (DWCP), Ethiopia (2009‐2012):  A Joint Government, 

Workers’ and Employers Organizations’ Programme. Supported by the ILO within the context of PASDEP, Addis Ababa, Ministry of Labor and Social Affairs. 

MULAT, D., FANTU, G. AND TADELE, F. 2003. Growth, Employment, Poverty and Polices in Ethiopia: An Empirical  Investigation.  Issues  in  Employment  and  Poverty,  Discussion  Paper  12.    Geneva, International Labor Office  

MULAT, D., FANTU, G. AND TADELE, F. 2006. Towards a More Employment  Intensive and Pro‐ Poor Economic Growth in Ethiopia: Issues and Polices. Addis Ababa, Ethiopia.   . 

NBE 2010. Annual Report of 2009/10 fiscal year.  Addis Ababa, National Bank of Ethiopia. . NBE 2011. Annual Report of 2010/ 11 fiscal year. Addis Ababa, National Bank of Ethiopia. NORDLUND,  M.  2010.  Who  are  the  lucky  ones?  Heterogeneity  in  Active  Labour  Market  Policy 

Outcomes   International Journal of Social Welfare, 1‐13. NÙÑEZ,  I. &  LIVANOS,  I.  2010. Higher  Education  and Unemployment  in  Europe: An Analysis of  the 

Academic Subject and National Effects. High Educ, 59, 475‐487. OKOJIE, C. E. 2003. Employment Creation for the Youths in Africa. International Journal of Adolescence 

and Youth, 1, 1‐13. ONWIODUOKIT, E. A., ADAMGBE, E.T. AND BUNO, E. N. 2009. Macroeconomic Determinants of  the 

Labour Market in Nigeria. Journal of Monetary and Economic Integration, 9, 62‐87. PALLEY, T. I. 2005. The Causes of High Unemployment: Labor Market Sclerosis versus Macroeconomic 

Policy   In: HEIN, H., AND TRUGER (ed.) Wages, Employment, Distribution and Growth. London: Palgrave/Macmillan. 

PIERLUIGI,  B.  A.  R., MORENO  2008.  Labour  Cost  and  Employment  across  Euro  area  Countries  and Sectors. Working Paper Series No. 912 / 2008,  European Central Bank. 

126

PRAKASH, A. 2001. Urban Unemployment: A study of Kochi City. Department of Economics, University of Kerala. 

PSACHAROPOULOS,  G.  1997.  Vocational  Education  and  Training  Today:  Challenges  and  Responses Journal of Vocational Education & Training, 49, 385‐393. 

RAHEL, W. A. P.,  I. 2010. Growth Determinants of Women‐operated Micro and Small Enterprises  in Addis Ababa. Journal of Sustainable Development in Africa, 12, 233‐246. 

RENA, R. A. B. K. 2006. Labour Market Needs and Development of Technical and Vocational Education in Eritrea. Manpower Journal, 41, 1‐15. 

RIDDELL, W.  C.  &  SONG,  X.  2011.  The  Impact  of  Education  on  Unemployment  Incidence  and  Re‐employment Success: 

Evidence from the U.S. Labour Market. Labour Economics, 18, 453‐463. SANCHEZ PUERTA, L. 2010. Labor Market Policy Research for Developing Countries: Recent Examples 

from  the Literature. What do we Know and What Should we know?SP Discussion Paper No. 1001. Social Protection and Labor, The World Bank. . 

SAPSFORD, D. A. T., Z. 1993. The Economics of the Labor Market., England, The Macmilan Press Ltd. SEIFE,  D.  2006.  Unemployment  Duration  in  Poor  Developing  Economies:  Evidence  from  Urban 

EthiopiaAuthor(s): . The Journal of Developing Areas, 40, 181‐201. SERNEELS, P. 2007. The Nature of Unemployment among  Young Men  in Urban Ethiopia. Review of 

Development Economics, 11, 170‐186. SEYE, A. 2000. Can More FDI Solve the Problem of Unemployment  in the EU? A Short Note. Applied 

Economics L etters, 7, 125‐128. SOLOW, R. M. 1980. On Theories of Unemployment. The American Economic Review, 70, 1‐11. SONG,  S. 2003. Polic  y  Issues of China’s Urban Unemployment. Contemporary Economic Policy, 21, 

258‐269. TAN, H. & CHANDRASIRI, S. 2004 Training and Labor Market Outcomes in Sri Lanka. TEGEGN, G. W. 2011. Socio‐Demographic Determinants of Urban Unemployment: The Case of Addis 

Ababa  Ethiopian Journal of Development Research, 33, 79‐124. TODARO, M.  P.  1969.  A Model  of  Labour Migration  and Urban Unemployment  in  Less Developed  

Countries The American Economic Review, LIX, 138‐148. TODARO, M. P. 1997. Urbanization, Unemployment, and Migration  in Africa:Theory and Policy. No. 

104, Paper prepared for Reviewing Social and Economic Progress in Africa. TORP, H. 1994. The Impact of Training on Employment: Assessing a Norwegian Labor Market Program 

The Scandinavian Journal of Economics, 96, 531‐550. UN 2003. World YOUTH Report 2003: The Global Situation of Young People. Department of Economic 

and Social Affairs, United Nations. UNESCO 2004. A UNESCO  International Meeting of Technical and Vocational Education and Training 

Experts. The Bonn Declaration, 28 October 2004 Bonn, Germany WALTERSKIRCHEN, E. 1999. The Relationship Between Growth, Employment  and Unemployment  in 

the EU. European Economists for an Alternative Economic Policy. WB 2007. Urban Labor Markets  in Ethiopia: Challenges and Prospects. vol.  I and  II, Synthesis Report, 

Report No. 38665‐ET. Poverty Reduction and Economic Management Unit (Africa Region), The World Bank. 

WOOLDRIDGE, J. 2000. Introductory Econometrics A Modern Approach 2e. YABUUCHI, S. 2006. Foreign Direct  Investment, Urban Unemployment and Welfare. J.  Int. Trade and 

Economic Development 8, 359‐371.  

127

Annex

Table 3.1: specification of variables (both dependent and independent variables)

Variable name Variable label Sex Sex: 1 if male ; 0 if female Age Age of respondent in years Agesq Age squared marr1 single: 1 if single; 0 if otherwise marr2 Married: 1 if married; 0 if otherwise marr3 Divorced :1 if divorced; 0 if otherwise marr4 Widow:1 if widow; 0 if otherwise marr5 Separated: 1 if separated; 0 if otherwise Exper Experience : age minus years of schooling minus 6 Expersq Experience square Trng Training: 1 if a respondent received formal training educ4 Grade 1-4 formal education: 1 if grade 1-4; 0 otherwise educ8 Grade 5-8 formal education: 1 if grade 5-8; 0 otherwise educ10 Secondary education; 1 if Grade 9-10 (new) and 11-12 (old ) Educ10c Grade 10 completed : 1 if grade 10 completed (new curr.) and grade 12 completed (old curr.) educ12 Preparatory: 1 if grade 11-12 in new curriculum; 0 otherwise Educnf Non formal education: 1 if non-formal education; 0 otherwise educno Illiterate Certfct Certificate ; 1 if had 1 year training after secondary education TVET TVET completed: 1 if 10+1, 10+2 and 10+3 completed TVETnc TVET not completed: 1 if TVET not completed Dipc Diploma, 12 +2, (in old curriculum Degdipnc Diploma/degree not completed: 1 if diploma/degree not completed Degac Degree and above completed: 1 if degree and above completed Tigr Tigray region: 1 if the location of respondent is Tigray region Afar Afar region: 1 if the location of respondent is Afar region Amhra Amhara region: 1 if the location of respondent is Amhara region Oromo Afar region: 1 if the location of respondent is Afar region Somal Somali region: 1 if the location of respondent is Somali region Snnpr SNNP region: 1 if the location of respondent is SNNP region Harar Harari region: 1 if the location of respondent is Harari region Addis Addis Ababa City A/min: 1 if the location of respondent is Addis Ababa Dire Dire Dawa region City A/min: 1 if the location of respondent is Dire Dawa y04 If the survey data comes from 2004 y06 If the survey data comes from 2006 y10 If the survey data comes from 2010 y11 Data comes from 2011 survey informal Share of regional informal sector employment Wragre Share of workers who have written agreement with employer Oragre Share of workers who have oral agreement with employer pubr Share of regional public sector employment Prvr Share of regional private formal sector employment Ss Natural logarithm of regional labor supply/labor force Migr1 Migration : 1 if a respondent is migrated at most a year before and zero if otherwise, or (0<migr1<1) Migr2 Migration : 1 if a respondent is migrated before 2 years and zero if otherwise, or (1<migr1<=2) Migr3 Migration : 1 if a respondent is migrated before 3 years and zero if otherwise, or (2<migr1<=3) Migr4 Migration : 1 if a respondent is migrated before 4 years and zero if otherwise, or (3<migr1<=4) Migr5 Migration : 1 if a respondent is migrated before 5-6 years and zero if otherwise, or (4<migr1<=6) Migr6 Migration : 1 if a respondent is migrated before 7-9 years and zero if otherwise, or (6 <migr1<=9) Migr7 Migration : 1 if a respondent is migrated before 10 years and above, or (10> migr1)

128

Table 4.14: effect of training on likelihood of unemployment- dependent variable 1 if unemployed & 0 if employed Independent variables

Coefficients Coefficients

Lage -0.135*** -0.131*** (0.00414) (0.00414) Sex -0.156*** -0.154*** (0.00224) (0.00224) marr2 0.00183 0.00588** (0.00273) (0.00274) marr3 -0.0733*** -0.0696*** (0.00397) (0.00403) marr4 -0.0373*** -0.0338*** (0.00512) (0.00519) marr5 -0.0495*** -0.0454*** (0.00690) (0.00702) Trng -0.0990*** -0.0911*** (0.00263) (0.00290) TVET -0.00509 (0.00560) educ10c 0.0849*** (0.00429) TVETy11 0.0462*** (0.0108) educ10cy11 -0.0167** (0.00761) Yrsch 0.00837*** 0.00626*** (0.000282) (0.000299) Tigr 0.121*** 0.120*** (0.00826) (0.00825) Afar 0.0979*** 0.0984*** (0.00927) (0.00927) Amhra 0.0949*** 0.0934*** (0.00658) (0.00656) Oromo 0.100*** 0.101*** (0.00611) (0.00611) Somal 0.158*** 0.159*** (0.00952) (0.00952) Snnpr 0.0489*** 0.0501*** (0.00654) (0.00655) Harar 0.0955*** 0.0968*** (0.00917) (0.00919) Addis 0.208*** 0.211*** (0.00721) (0.00723) Dire 0.245*** 0.248*** (0.00961) (0.00963) y04 -0.0267*** -0.0271*** (0.00316) (0.00317) y06 -0.0728*** -0.0612***

129

Source: UEUS 2003-11

(0.00283) (0.00298) y10 -0.0508*** -0.0404*** (0.00287) (0.00297) y11 -0.0620*** -0.0539*** (0.00283) (0.00313) Observations 142,540 142,540 Wald chi2 11969.08 12493.75 Prob > chi2 0.0000 0.0000 Pseudo R2 0.0920 0.0956

Table 4.15.: The effect of training and education on self-employment- dependent variable is 1 if a person is self employed & 0 if any other form of employment status Explanatory variables Coefficients Coefficients Lage 0.140*** 0.164*** (0.00576) (0.00582) Sex 0.0125*** 0.0419*** (0.00335) (0.00339) marr2 0.104*** 0.114*** (0.00421) (0.00428) marr3 0.132*** 0.124*** (0.00714) (0.00730) marr4 0.236*** 0.222*** (0.00809) (0.00829) marr5 0.150*** 0.148*** (0.0116) (0.0117) educ8 -0.0449*** (0.00428) educ10 -0.0775*** (0.00475) educ12 0.0108 (0.0210) Educnf 0.229*** (0.0119) Certfct -0.184*** (0.00481) TVET -0.271*** -0.106*** (0.00614) (0.00884) TVETnc -0.194*** (0.0143) Degdipnc -0.256*** (0.0161) Dipc -0.272*** (0.0107) Degac -0.311*** (0.00558) y11TVET 0.0409** (0.0182) y11educ8 -0.0171** (0.00864) y11educ10 -0.0120 (0.00941) y11educ12 0.0100 (0.0369) Tigr 0.00179 0.00669 (0.00822) (0.00828) Afar -0.0786*** -0.0944*** (0.00861) (0.00848) Amhra -0.0114* -0.0104 (0.00663) (0.00666) Oromo 0.0126** 0.0222***

130

Source: UEUS 2003-11

Table 4.16 : effect of TVET on school –to- work transition Explanatory variables Exit to employment Exit to employment Sex 0.978 1.013 (0.143) (0.141) Marr 0.962 0.960 (0.145) (0.145) Age 0.973** 0.972** (0.0111) (0.0110) Exper 1.052*** 1.055*** (0.0154) (0.0138) Migr 0.773 0.786 (0.130) (0.132) educ8 2.006** (0.577) educ10 1.436 (0.430) educ12 0.867 (0.174) TVET 1.747** (0.447) Dipc 1.502

(0.00638) (0.00642) Somal 0.0586*** 0.0522*** (0.00925) (0.00931) Snnpr 0.0296*** 0.0365*** (0.00709) (0.00713) Harar -0.000145 0.0326*** (0.00945) (0.00972) Addis -0.104*** -0.0816*** (0.00649) (0.00666) Dire -0.0381*** -0.0178* (0.00901) (0.00932) y04 -0.0101** -0.0117** (0.00510) (0.00514) y06 -0.0280*** 0.00506 (0.00492) (0.00508) y10 -0.0411*** -0.0145*** (0.00475) (0.00483) y11 -0.0169*** -0.0113** (0.00633) (0.00501) Trng -0.261*** (0.00417) educ10c -0.0314*** (0.00632) TVETy11 0.0632*** (0.0180) educ10cy11 0.0459*** (0.0143) Yrsch -0.0168*** (0.000427) Observations 112,981 112,981 Wald chi2 14055.02 Prob > chi2 0.0000 0.0000 Pseudo R2 0.1264

131

(0.556) Deg 4.184*** (1.457) Crdt 1.614* 1.420 (0.437) (0.331) Yrsch 0.986 (0.0259) Trng 1.684*** (0.301) Observations 414 414 ld chi2 43.09 31.14 Prob > chi2 0.0000 0.0001

Source: primary survey

4.17: Effect of education and training on unemployment (primary data result)- dependent variable is 1if a person is unemployed and 0 if employed

Explanatory variables Current unemployment status

Current unemployment status

Sex 1.950*** 1.967*** (0.199) (0.201)

Marr -0.204 -0.242 (0.223) (0.222)

Age 0.00304 0.00130 (0.0122) (0.0122)

Exper -0.0569*** -0.0571*** (0.0199) (0.0193)

Migr -0.266 -0.389* (0.221) (0.221)

educ8 -0.426 (0.317)

educ10 -0.235 (0.322)

educ12 0.231 (0.288)

TVET -0.539* (0.297)

Dipo -1.176** (0.538)

Deg -1.143* (0.644)

Crdt -1.628** -1.650** (0.687) (0.693)

Yrsch -0.0732*** (0.0265)

Constant -0.780* -0.325 (0.434) (0.452)

Observations 565 565 Source: primary survey

132

Table 4.18: Determinants of annual average employment growth within MSEs

Variable specification Variable name Model 1 Model 2 Logarithm of MSE age (year) lage 1.225 1.219

(0.185) (0.181) 1 if service providing MSE; 0 otherwise Mset2 0.647 0.646

(0.305) (0.300) 1 if Manufacturing MSE; 0 otherwise mset3 3.413*** 3.502***

(1.498) (1.521) 1 if Construction MSE; 0 otherwise Mset4 2.001 2.096

(1.114) (1.153) 1if Urban agriculture type MSE; 0 otherwise mset5 3.908*** 3.704***

(1.900) (1.752) 1 If MSE owner is proprietor mseo2 1.182 1.256

(0.533) (0.560) 1 If MSE owner is cooperatives mseo3 2.574* 2.511*

(1.411) (1.371) 1 if the MSEs is located in Hawassa ; 0 or else Town2 1.305 1.333

(0.430) (0.434) 1 if the MSEs is located in Bahir- Dar ; 0 or else Town3 2.688*** 2.655***

(0.843) (0.828) Average hours of work (MSE) lhrsw 1.017** 1.016**

(0.00771) (0.00766) Size of MSE (initial employment size) labor1 0.955* 0.952*

(0.0255) (0.0261) 1 if access to credit cr 1.148 1.152

(0.289) (0.289) Proportion of owners received training trngo 1.193 1.288

(0.367) (0.350) Average hrs of family labor hwfl 1.000 1.000

(0.00603) (0.00601) Proportion of owners with preparatory education educ12 1.272

(0.416) Proportion of owners with degree/diploma completed

degdip 1.445 (0.643)

Proportion of owners with TVET education TVET 1.009 (0.355)

Price shocks pshocks1 1.252 1.264 (0.481) (0.485)

Demand shocks dshocks1 0.787 0.791 (0.264) (0.266)

Policy shocks plshocks1 0.511* 0.509* (0.186) (0.183)

Motive to start MSE is it is only option to be employed

motv11 1.957 2.030

(1.287) (1.287) Self-employment is better than paid employment motv21 2.895** 2.918**

(1.472) (1.472) Training received aspired motv31 1.176 1.244

(0.598) (0.622) Proportion of owners with certificate to degree certificate

certab 1.059 (0.187)

Constant term Constant 0.0293*** 0.0338*** (0.0224) (0.0245)

Observations 437 437 Wald chi2 72.96 70.78

Prob > chi2 0.0000 0.0000Pseudo R2 0.1601 0.1582

Source: primary survey

133

Table 4.19: Cox regression result on pooled cross-section data

Independent variables Exit to employment Exit to employment Sex 2.417*** 2.376*** (0.0446) (0.0444) Age 0.998** 0.998** (0.000923) (0.000938) marr2 1.363*** 1.323*** (0.0309) (0.0302) marr3 1.988*** 1.943*** (0.0737) (0.0721) marr4 1.796*** 1.758*** (0.0815) (0.0797) marr5 1.713*** 1.658*** (0.105) (0.101) Trng 1.343*** 1.228*** (0.0321) (0.0313) educ8 0.801*** (0.0221) educ10 0.533*** (0.0166) educ12 0.578*** (0.0577) Educnf 1.219*** (0.0852) Educno 1.117*** (0.0328) Certfct 0.782*** (0.0253) TVET 0.562*** (0.0314) TVETnc 0.602*** (0.0628) Dipc 0.657*** (0.101) Degdipnc 0.511*** (0.1000) Degac 0.845** (0.0725) Tigr 0.521*** 0.523*** (0.0239) (0.0240) Afar 0.482*** 0.488*** (0.0287) (0.0291) Amhra 0.617*** 0.622*** (0.0222) (0.0224) Oromo 0.613*** 0.611*** (0.0211) (0.0211) Somal 0.450*** 0.457*** (0.0226) (0.0230) Snnpr 0.783*** 0.776*** (0.0293) (0.0291) Harar 0.677*** 0.677*** (0.0355) (0.0355) Addis 0.463*** 0.455*** (0.0167) (0.0164) Dire 0.403*** 0.398*** (0.0199) (0.0197)

134

y04 0.912*** 0.921*** (0.0252) (0.0255) y06 0.971 0.934** (0.0262) (0.0257) y10 1.042* 1.022 (0.0261) (0.0261) y11 0.936*** 0.914*** (0.0235) (0.0234) Yrsch 0.927*** (0.00214) Observations 36,879 36,878 LR chi2 4954.04 5217.68 Prob > chi2 0.0000 0.0000

Source: 2003-11

Table 4 .20: Probit model on theories of unemployment Variable Marginal effect Z- value P- value Age -0.0062 -5.93 0.000 Agesq -0.0003 -11.80 0.000 sex* -0.1684 -43.75 0.000 marr2* -0.0054 -1.72 0.085 marr3* -0.0574 -10.65 0.000 marr4* -0.0432 -6.49 0.000 marr5* -0.0356 -3.97 0.000 Expersq 0.0005 27.43 0.000 Yrsch 0.0698 38.17 0.000 Yrschsq -0.0032 -34.85 0.000 Informal 0.0477 2.99 0.003 Wragre 0.0172 1.36 0.175 Oragre -0.0074 -0.59 0.558 Pubr 0.0182 3.94 0.000 Prvr 0.6675 20.91 0.000 Ss -0.0069 -3.31 0.001 y04* -0.0324 -8.67 0.000 y06* -0.1132 -32.70 0.000 y10* -0.0944 -20.24 0.000 y11* -0.1064 -22.44 0.000

Number of obs = 111201 Wald chi2 (20) = 10868.36 Prob > chi2 = 0.0000 Pseudo R2 = 0.1095

Source: UEUS 2003-11

Table 4.21: Probability of unemployment last week and usually unemployed Urban employment unemployment survey Variables Current unemployment status Usual unemployment status Lage -0.172*** -0.101*** (0.00513) (0.00389) Sex -0.165*** -0.114*** (0.00298) (0.00242) marr2 -0.0118*** -0.0151*** (0.00306) (0.00250) marr3 -0.0639*** -0.0583*** (0.00522) (0.00349) marr4 -0.0331*** -0.0346*** (0.00697) (0.00451)

135

marr5 -0.0425*** -0.0416*** (0.00870) (0.00608) educ8 0.0671*** 0.0655*** (0.00466) (0.00446) educ10 0.117*** 0.111*** (0.00481) (0.00464) educ12 0.120*** 0.0702*** (0.0163) (0.0151) Educnf 0.0498*** -0.00282 (0.0112) (0.00878) Educno 0.0105*** (0.00402) Certfct 0.00991*** 0.0362*** (0.00372) (0.00340) TVET -0.00528 -0.00278 (0.00673) (0.00579) TVETnc 0.0610*** 0.0401*** (0.0132) (0.0116) Degdipnc 0.0309* -0.0243* (0.0180) (0.0133) Dipc -0.0570*** -0.0732*** (0.0131) (0.00785) Degac -0.0963*** -0.0866*** (0.00645) (0.00451) Tigr 0.119*** 0.0977*** (0.00959) (0.00816) Afar 0.0724*** 0.0706*** (0.0110) (0.00905) Amhra 0.111*** 0.0830*** (0.00794) (0.00646) Oromo 0.110*** 0.0812*** (0.00719) (0.00595) Somal 0.148*** 0.144*** (0.0123) (0.00978) Snnpr 0.0624*** 0.0556*** (0.00779) (0.00653) Harar 0.101*** 0.0722*** (0.0105) (0.00886) Addis 0.214*** 0.192*** (0.00829) (0.00743) Dire 0.234*** 0.202*** (0.0113) (0.0101) y04 -0.0253*** -0.0125*** (0.00365) (0.00313) y06 -0.0705*** 0.00783** (0.00338) (0.00328) y10 -0.0564*** 0.00369 (0.00336) (0.00312) y11 -0.0516*** 0.0544*** (0.00703) (0.00522) y11TVET 0.0225* -0.0178** (0.0123) (0.00800) y11sex 0.00574 -0.0104** (0.00587) (0.00441) y11educ8 -0.0245*** -0.0329*** (0.00797) (0.00507) y11educ10 -0.0269*** -0.0478*** (0.00775) (0.00448) y11educ12 0.0385 0.0168

136

(0.0245) (0.0206) Observations 111,201 133,377 Wald chi2(37) 11167.08 9556.18 Prob > chi2 0.0000 0.0000 Pseudo R2 0.1090 0.0899

Source: UEUS 2003-2011

Table 4.23: effect of education on unemployment (LFS) Variables coefficients

Lage -0.177*** (0.0294)

Exper -0.0111*** (0.00124)

Sex -0.587*** (0.0123)

migr1 -0.130*** (0.0208)

migr2 -0.194*** (0.0271)

Migr3 -0.195*** (0.0280)

migr4 -0.157*** (0.0295)

migr5 -0.173*** (0.0269)

migr6 -0.118*** (0.0283)

migr7 -0.143*** (0.0163)

Educno -0.134*** (0.0264)

educ8 0.368*** (0.0348)

educ10 0.386*** (0.0361)

educ12 0.520*** (0.110)

Educnf 0.295*** (0.0590)

Certfct -0.228*** (0.0482)

TVET 0.502*** (0.0698)

TVETnc -0.172*** (0.0395)

degdipnc 0.0893 (0.0618)

dipc -0.258*** (0.0444)

degac -0.841*** (0.0943)

tigr 0.141*** (0.0361)

afar 0.161*** (0.0439)

amhra 0.149***

137

Source: LFS 1999 and 2005

For all regression results:

• Robust standard errors in parentheses • *** p<0.01, ** p<0.05, * p<0.1

(0.0312) oromo 0.0912***

(0.0306) somal 0.558***

(0.0405) snnpr 0.0867***

(0.0314) harar 0.349***

(0.0471) addis 0.460***

(0.0305) dire 0.534***

(0.0434) y05educ8 -0.282***

(0.0329) y05educ10 -0.178***

(0.0326) Constant 0.131

(0.0879) Observations 63,456

Wald chi2(32) 5382.67 Prob > chi2 0.0000

Pseudo R2 0.0906