aid unpredictability and economic growth in kenya · aid unpredictability and economic growth in...
TRANSCRIPT
No 226 – July 2015
Aid Unpredictability and Economic Growth in Kenya
Elphas Ojiambo, Jacob Oduor, Tom Mburu and Nelson Wawire
Editorial Committee
Steve Kayizzi-Mugerwa (Chair) Anyanwu, John C. Faye, Issa Ngaruko, Floribert Shimeles, Abebe Salami, Adeleke O. Verdier-Chouchane, Audrey
Coordinator
Salami, Adeleke O.
Copyright © 2015
African Development Bank
Immeuble du Centre de Commerce International d'
Abidjan (CCIA)
01 BP 1387, Abidjan 01
Côte d'Ivoire
E-mail: [email protected]
Rights and Permissions
All rights reserved.
The text and data in this publication may be
reproduced as long as the source is cited.
Reproduction for commercial purposes is
forbidden.
The Working Paper Series (WPS) is produced by
the Development Research Department of the
African Development Bank. The WPS disseminates
the findings of work in progress, preliminary
research results, and development experience and
lessons, to encourage the exchange of ideas and
innovative thinking among researchers,
development practitioners, policy makers, and
donors. The findings, interpretations, and
conclusions expressed in the Bank’s WPS are
entirely those of the author(s) and do not
necessarily represent the view of the African
Development Bank, its Board of Directors, or the
countries they represent.
Working Papers are available online at
http:/www.afdb.org/
Correct citation: Ojiambo,Elphas; Oduor, Jacob; Mburu, Tom and Wawire, Nelson (2015), Aid unpredictability and
Economic Growth in Kenya, Working Paper Series N° 226 African Development Bank, Abidjan, Côte d’Ivoire.
Aid unpredictability and Economic Growth in Kenya
Elphas Ojiambo1, Jacob Oduor2, Tom Mburu3 and Nelson Wawire4
1 Corresponding Author; [email protected], Department of Applied Economics, Kenyatta University 2 Research Department, African Development Bank 3 African Economic Research Consortium 4 Department of Applied Economics, Kenyatta University
AFRICAN DEVELOPMENT BANK GROUP
Working Paper No. 226
July 2015
Office of the Chief Economist
Abstract
Studies on the impacts of aid on growth
are not limited. Some have found
positive effects, others negative effects,
others no effect and others have found
positive impacts only under certain
conditions. A less examined aspect is the
role of aid unpredictability on growth.
While aid unpredictability may
negatively affect growth when it
disorientates the recipient country’s
consumption and investment planning, it
may also boost economic growth if it is
triggered by responses to economic
shocks. It may also induce strengthening
of systems of aid absorption in the
recipient country that improves aid
effectiveness and economic growth. This
paper assesses the heterogeneous
impacts of aid on growth in a low
income country with different aid
unpredictability episodes and finds that
increased aid unpredictability weakens
economic growth in Kenya. In addition,
aid unpredictability is found to improve
economic growth in an unstable
macroeconomic environment implying
that aid unpredictability forces weak
governments to be more prudent in
managing the limited uncertain
resources at their disposal during
periods of macro instability. We
however find no evidence of different
impacts of aid unpredictability during
periods of shocks.
Key words: Foreign Aid, Aid Unpredictability, Macroeconomic Policy, Economic
Growth
JEL Classification: C22, E02, E61, F35, F43
5
1. Introduction
Foreign aid forms one of the largest components of foreign capital flows to low-income
countries (Radelet, 2006). The question as to how foreign aid affects the economic growth of
developing countries has drawn the attention of many scholars over time. The results of their
studies have been varied.
Several studies (Rosenstein-Rodan, 1961; Chenery and Bruno, 1962; Chenery and
Strout, 1966; Mosley, 1980; Mosley, Hudson and Horrell, 1987; Karras, 2006; among others)
found that foreign aid positively affects economic growth. Using an Autoregressive
Distributed Lag (ARDL) model and various components of foreign aid including loans,
grants, technical cooperation, multilateral and bilateral aid flows to estimate a disaggregated
short-run and long-run relationship between foreign aid and economic growth, Gounder
(2001) found that total aid flows and its various forms had positive and significant impacts on
economic growth in Fiji.
Other studies (Singh, 1985; Snyder, 1993; Burnside and Dollar, 1997; World Bank,
1998; Morrissey, 2001; Bearce and Tirone, 2008; Salisu and Ogwumike, 2010; Herzer and
Morrissey, 2011) have found that foreign aid leads to growth but only under certain
conditions. Singh (1985) found that foreign aid had a strong positive impact on economic
growth in less developed countries (LDCs) for both periods 1960-1970 and 1970-1980, when
state intervention was not taken into account. When the state intervention variable was
included in the regression, the effect of foreign aid got statistically weak over time. Snyder
(1993) argued that when country size was not included, the effects of aid on economic growth
were small and insignificant, but when country size was taken into account, the coefficient of
aid became positive and significant. Burnside and Dollar (1997) and World Bank (1998)
emphasized that good-policy environment was essential for this positive relationship to occur.
Bearce and Tirone (2008) explored the relationship between economic growth and foreign aid
conditioned on the level of democracy in the potential recipient country. Herzer and
Morrissey (2011) found that the effect of aid on GDP depended on a trade-off that is country-
specific: aid had a direct positive effect through financing investment but could have an
indirect negative effect on aggregate productivity.
Yet other studies (Papanek, 1972, 1973; Newlyn, 1973; Knack, 2000; Gong and Heng-
fu, 2001; Boakye, 2008; Mallik, 2008) found a negative relationship between foreign aid and
growth. Knack (2000), for example, observed that high levels of aid had the potential to erode
institutional quality, increase rent-seeking and corruption, thus negatively affecting growth.
Easterly, Levine and Roodman (2003) re-examined works by Burnside and Dollar (1997)
using a larger sample size and found that the results were not as robust as before. Some of the
explanation for the negative relationship was that foreign aid is fungible (Boakye, 2008).
The last strand of studies (Pedersen, 1996; Ekanayake and Chatrna, 2009) noted the
inconclusiveness of the impact of foreign aid on economic growth. Ekanayake and Chatrna
(2009) found that foreign aid had a mixed impact on economic growth of developing
countries.
Amidst these contestations, a less discussed aspect of foreign aid which could have
serious implications on its effects on growth is aid unpredictability. The above studies assume
that foreign aid flows are predictable and therefore the recipient countries can effectively and
timely reflect them in their development planning process. It is also based on the premise that
aid commitments and disbursements are the same. While no country specific study to our
6
knowledge has attempted to determine the effect of aid unpredictability on growth,
unpredictability may have both positive and negative effects on growth depending on what
triggered it. Aid unpredictability may be triggered by the inability of recipient countries to
comply with the conditions of the aid. It may also be triggered by the inability of donors to
disburse in time due to administrative delays. Aid unpredictability may also occur in times of
emergency or shocks to the economy and donors are required to disburse more aid than they
committed at the beginning of the year.
Aid unpredictability in general would jeopardize the recipient’s ability to plan and
effectively execute investments financed through such funds. Such occurrence would
necessitate a re-organization of the recipient’s consumption plans particularly if the activities
that were meant to be funded by donor funds cannot be postponed. This may negatively affect
growth. Lensink and Morrissey (2000) make a similar suggestion, that aid uncertainty may
affect growth negatively. Studies (Bulỉř and Hamann, 2001, 2003, 2005; Pallage and Robe,
2001; Rand and Tarp, 2002 and Chauvet and Guillaumont, 2008) have examined the issue of
aid volatility arguing that it may contribute to macroeconomic instability. According to
Kodama (2011), the effect of aid unpredictability on economic growth is significant and
results in a waste of one-fifth of the aid. Aid unpredictability could also be triggered by the
inability of the recipient country to comply with the conditions of the assistance including
inability to use aid effectively for the intended purposes. Where project implementation
systems are weak in case of project aid, cutting aid may force the recipient country to
strengthen their systems in a way that increases aid effectiveness. This undoubtedly would
have welcome positive impact on growth. In addition, aid unpredictability that is triggered by
donor’s response to emergencies including floods, famine and other natural disasters would
also have positive effects on growth. In this case, some level of aid unpredictability may be
necessary for growth.
The impact of aid unpredictability on growth may therefore be positive or negative
depending on what triggered the unpredictability and may differ from country to country
depending on the recipient country’s adherence to the aid conditions, economic environment
and strength of institutions and policies. The impact may also differ depending on the
economic and political situations in the donor country. An economic downturn or elections in
the donor country may for instance slow down disbursements of aid leading to aid
unpredictability. Unfortunately, poor countries have the weakest systems and therefore are
more likely to be most adversely affected by pull-backs of aid due to non-compliance with aid
conditions. The finding of Celasun and Walliser (2007) that aid unpredictability is more
prevalent in poor countries is therefore not surprising. Vargas (2005) also finds that on
average for sub-Saharan Africa, the difference between commitments and disbursements was
as much as +/- 20 per cent of commitments, and that on average over the period 1975–2002,
disbursements were less than commitments by up to 4.9 per cent. If such unpredictability
were to restrict growth, the poorer countries’ growth would be most affected.
From the preceding discussion, it is clear that at the aggregate level, the impact of aid
unpredictability on economic growth is not obvious. This paper contributes to the debate by
assessing the heterogeneous impact of aid unpredictability in the economy of a low income
country that experiences different aid unpredictability episodes. In particular, we assess the
impact of aid unpredictability on growth under different strengths of institutions,
macroeconomic environments, unexpected shocks and emergencies and sizes of the economy
in the recipient country and conditions in the donor country. We therefore assess whether
there are circumstances when aid unpredictability may be good for economic growth. We use
7
Kenya for this assessment because of the mixed experience that it has had with donors from
good to bad times.
The rest of the paper is organized as follows: Section 2 covers the foreign aid flows and
economic growth in Kenya while the model is presented in section 3. Section 4 presents the
Univariate Analysis while the data and sources are explained in section 5. The results are
presented in section 6 while section 7 presents the conclusions and policy issues.
2. Foreign Aid Flows and Economic Growth in Kenya
Historically, most aid has been given as bilateral assistance directly from one country to
another. Donors also provide aid indirectly as multilateral assistance, whereby resources are
pooled together from many donors. According to Mwega (2004), 78 per cent of Kenya’s aid
has been from the bilateral donors, and that the share of multilateral aid increased moderately
in the 1980s and early 90s, primarily due to the disbursement of the World Bank adjustment
lending under the Structural Adjustment Programme (SAPs). However, the bilateral aid share
rose again since then, with the decline in new adjustment lending after 1991. A summary of
the Official Development Assistance (ODA) disbursements to Kenya over the period 1968-
2011 is shown in Figure A1 in the appendix.
Appendix Figure A1 shows that average flows from multilateral donors were highest in
1990-2000 and 2001-2011 compared to the previous periods. On the bilateral flows, there was
a steady increase in the period 1979-89, a period that recorded the highest foreign aid flows to
Kenya. This was partly due to the flows arising from the need to cushion the economy from
the effects of the SAPs and the need to ensure sustenance of the reforms that the country was
undertaking during this period. This period also witnessed much of the bilateral support being
channeled outside of the Government to the Civil Society Organizations following a not so
good relationship between the government and the development partners during certain
periods. After a dip in the period 1990-2000, there was a noticeable increase in bilateral flows
for the period 2001-2011, possibly due to the regime change and Kenya's commitment to
reform.
The timely flow of aid is important as countries that are dependent on aid become
vulnerable when funds are committed and scheduled, but not disbursed on time, or when there
is insufficient information about donors’ intentions to disburse. In this respect, the Paris
Declaration of 2005 commits donors to “provide reliable indicative commitments of aid over
a multi-year framework, and disburse aid in a timely and predictable fashion according to
agreed schedules” (OECD, 2005, No. 26: 4).
According to Celasun and Walliser (2008), whereas aid predictability has been
highlighted as a key issue for aid effectiveness (see also OECD, 2005; IMF, 2007; and World
Bank, 2007), little systematic information is available on the magnitude of the predictability
problem and thus its potential impact on aid recipients. Zimmerman (2005) argued that donors
had not been able to make multi-year financial commitments that would enable the recipients
to rely on during their planning and budgeting process. They have continued to apply
complicated and inconsistent conditionalities, and have in a secretive and slow way been
determining how to allocate the resources. Thus, donors have continued to use different
disbursement mechanisms and monitoring procedures instead of coordinating their activities
(Gómez, 2005; Saasa, 2005). Indeed, creative initiatives such as the global funds run by
mixed groups of public, business and non-profit organizations, have but continued to add to
the confusion (Sagasti, Bezanson and Prada, 2005).
8
Since the 1980s, Kenya has experienced relatively unpredictable flows of international
aid (see also M’Amanja and Morrissey, 2005; Mwega, 2009; Oduor and Khainga, 2009).
According to OECD-DAC statistics, while Kenya experienced a dramatic build-up in nominal
aid flows in the 1980s, the 1990s witnessed a reduction in donor support (Mwega, 2009).
Nominal aid flows increased from US$55.6 million in 1963 to a peak of US$1.2 billion in
1990, before declining to a low of US$309.9 million in 1999, with some recovery thereafter in
response to a new government in December 2002.
Kenya has experienced major standoffs with the donor community, which has
sometimes led to aid freezes (Wawire, 2006). As a result, foreign aid disbursements were
short-lived due to the continued uneasiness amongst the donors with Kenya’s implementation
of aid conditionalities. For example, the World Bank in July 1982 did not release the second
tranche of US$50 million on the basis that Kenya was lax in undertaking policy reforms
(Njeru, 2003). The resumption of funding in 1984 was partly attributable to the humanitarian
gesture of providing large volumes of food aid in response to the devastating drought that
year. Further freezes were experienced in 1992 and 1997.
The extent of aid unpredictability in Kenya is demonstrated in Appendix Figure A2. The
Figure shows that commitments have been higher than disbursements over the years. The only
periods when disbursements were higher than commitments were 1968-70, 1981, 1985, 1992-
95, and 2001-2002. The situation obtaining between 1968 and 1970, up to 1981 could be due
to the effects of the oil shocks that necessitated the need for foreign aid to cushion the Kenyan
economy. Between 1992 and 1995 coincided with the funding of Structural Adjustment
Programmes (SAPs) by International Monetary Fund (IMF) and World Bank. The situation
seen between 2001 and 2002 coincided with the optimism in terms of possible change of
government in 2002 coupled with low commitments on the part of Kenya’s development
partners.
Turning to economic growth performance in Kenya, it is observed that Kenya has had a
chequered history on economic growth since independence. The first period, from 1963 to the
beginning of the 1980s, was characterized by strong economic performance and huge gains in
social outcomes. A second period, from the 1980s to 2002, was typified by slow or negative
growth, mounting macroeconomic imbalances and significant losses in social welfare, notably
rising poverty and falling life expectancy. Failure to reform and the increased role of politics
over policy were at the heart of this structural break (Legovini, 2002). The third period, from
2003 to 2011, was marked by the resurgence in economic growth following the 2002 elections
and regime change thereof. Kenya has also pursued macroeconomic policy reforms over the
period. Some of these reforms have been at the behest of the donor community. Whether these
reforms have led to the desired outcome is an issue worth examination.
Invariably, increased aid should lead to increased growth, but from the above
discussion, even if aid was increasing, this may not lead to increased economic growth as aid
is unpredictable. As has been mentioned earlier, aid flows to Kenya has increased sharply in
the last few years. However, economic growth has remained dismally low. Could aid
unpredictability explain this poor performance in economic growth? This resonates well with
the inherent belief that aid is supposed to promote growth, and therefore, any unpredictability
has a possibility to affect growth negatively. It is for this reason that this study found the
debate on aid unpredictability in Kenya an issue worth examining as most studies (M’Amanja
and Morrissey, 2005; Mwega, 2009; Oduor and Khainga, 2009) had just but mentioned it
without specifically factoring it in their models. The objective of this study is therefore to
9
examine the extent to which output growth is retarded as a result of the unpredictability in aid
flows to Kenya.
3. Model Specification
Lensink and Morrissey (2000) measured aid unpredictability by first estimating a
forecasting equation so as to be able to determine the expected component of the variable
under consideration. Secondly, the uncertainty (unpredictability) proxy was derived by
calculating the standard deviation of the residuals from the forecasting equation. This study
used a predictability indicator defined as the difference between commitments and
disbursements as a percentage of disbursements (also known as the predictability per a unit of
aid dollar) (see Celasun and Walliser, 2008). This measure of aid predictability is independent
of the scale of aid as a share of GDP.
100*)/)(( tttt DDCAid (1)
where χAidt is a predictability indicator, Ct is Commitments and Dt is disbursements.
An increase shows that aid is becoming more unpredictable. If the variable is decreasing
it means aid is becoming more predictable. A negative sign means that donors disbursed more
than they committed as may be the case during periods of economic shocks. Thus, an aid
predictability indicator (XAid) and an interactive variable linking predictability to
macroeconomic policy environment (POLICY) (see Appendix and Table A1 for computation
of the Policy variable) and other relevant factors were used in the regressions. The functional
relationship is given as:
,,,,,,,,, 11 tttttttttt PRIPAIPXAIDBPOLICYFdebtTaxAidPinvYpfYp (2)
where Yp is real per capita GDP, Tax is the total tax revenue as a share of GDP, Pinv is
private investment as a share of GDP, Fdebt is Foreign debt as a share of GDP, AIP is an
interactive variable of AID and Policy (see appendix for computation of the policy variable).
PRIP is the interaction term between aid predictability and the policy environment.
As a robustness check, we use H-P filter (Hodrick et Prescott, 1997) to extract the trend
and cycle components of foreign aid in line with Bulir and Hamann (2001, 2003, 2005),
Pallage and Robe (2001) and Rand and Tarp (2002). The H-P filter decomposes a series,
xt(where xt is the logarithm of the observed series Xt in a cycle, xtc and a trend xt
g by
minimizing the following equation:
(3)
Where ʎ is the smoothing parameter of xtg .The choice of the value of λ depends on the
frequency of observations. Pallage and Robe (2001) use λ equals 100 for annual data. Bulir
and Hamann (2001) use λ equals 7 while Ravn and Uhlig (2002) show that on annual data, λ
should be of the order of 6.25. We experimented with each of these values of λ to arrive at the
best results. Through this process, we opted to use λ equals 100 in line with Pallage and Robe
(2001) as it represented a better fit (see Appendix Figures A3 and A4).
10
Our choice of the HP filter is further driven by the conclusion from previous studies on
aid volatility that more often than contra-cyclical, aid is pro-cyclical, at best not correlated
with the cycles of national income or fiscal revenues (Bulir and Hamann, 2001, 2003, 2005;
Pallage and Robe, 2001). The foregoing argument resonates well with our concern about aid
unpredictability thus our resolve to use the filter as a robustness check.
4. Univariate Analysis
We use an autoregressive distributed lag (ARDL) model developed by Pesaran and Shin
(1995) to examine the direction of causation between the variables. The main advantage of
using this approach to examine long-run relationships among variables over other approaches
including Johansen cointegration test is that the ARDL approach does not necessarily require
pre-testing of the variables, implying that the test is possible even if the underlying regression
is purely I(0), purely I(1) or a mixture of the two.
The ARDL approach is conducted in two steps (Pesaran and Pesaran, 1997). In the first
stage, the hypothesis of no cointegration is tested. The null hypothesis in this case is that the
coefficients on the lagged regressors (in levels) in the error-correction form of the underlying
ARDL model are jointly zero. The approach uses the F-test, although the asymptotic
distribution of the F-statistic in this context is non-standard irrespective of whether the
variables are I(0) or I(1). We use the critical values provided by Narayan (2004), due to their
appropriateness for small samples (Boakye, 2008). Two sets of values are tabulated. The first
assumes that all the variables are I(1) and the second that they are I(0). This band allows for
the fact that variables may be stationary, integrated of order one, or even fractionally
integrated. In this respect, when the calculated F-statistic is above the upper value of this
band, the null hypothesis will be rejected, indicating cointegration between the variables
irrespective of whether they are I(1) or I(0). If the F-statistic falls below the band, then the
null hypothesis of no cointegration cannot be rejected. A value within the band implies the
test is inconclusive.
In the second stage, the lag orders of the variables are chosen using Schwarz Bayesian
Criteria (SBC). The importance of this step is that an appropriate lag selection enables us to
identify the true dynamics of the models. We further check on the performance of the
estimated model and present the associated diagnostic tests of serial correlation, functional
form and heteroscedasticity.
Further, we conducted two crucial stability tests - CUSUM (Cumulative Sum) and
CUSUMSQ (CUSUM of Squares) of recursive residuals – based on Brown et al. (1975). The
CUSUM test does not require the variables to be I(1) and is based on the cumulative sum of
recursive residuals based on the first set of n observations. The estimated coefficient is said to
be stable if the CUSUM statistic stays within 5 per cent significance level. The CUSUMSQ
on the other hand is based on the squared recursive residuals and is more suitable for
detecting systematic changes in the regression coefficients and is useful in situations where
the departure from the constancy of the regression coefficients is haphazard and sudden.
The long and short-run parameters are estimated using the ARDL method. We set the
initial lag length at two periods on all variables in the general ARDL equation. Once
cointegration is established, the conditional ARDL (p, q1, q2, q3, q4, q5, q6, q7, q8) long-run
model for Ypt in equation (2) is specified as:
11
,lnln
lnlnlnlnln
7 8654
321
0 0
98
1
7
0
6
0
5
0
4
0
3
0
2
1
10
t
q
i
it
q
i
iiti
q
i
iti
q
i
iti
q
i
iti
it
q
i
iit
q
i
iit
q
i
i
p
i
itit
PRIPAIPFdebtXAIDBPOLICY
AidPinvTaxYpYp
(4)
where p, q1, q2, q3, q4, q5, q6, q7, q8 are the lag lengths for each of the variables.
We obtained the short-run dynamic parameters by estimating an error correction model
associated with the long-run estimates. This was specified as follows:
,ln
lnlnlnlnlnln
1
9
1
8
1
7
1
6
5
1
4
1
3
1
2
1
10
tit
n
i
iti
n
i
iti
n
i
iti
n
i
iti
n
ii
itiit
n
i
iit
n
i
iit
n
i
i
n
i
itit
ecmPRIPAIPFdebtXAIDB
POLICYAidPinvTaxYpYp
(5)
where φ1, φ2, φ3, φ4, φ5, φ6, φ7, φ8, φ9 are the short-run dynamic coefficients of the
model’s convergence to equilibrium, and π is the speed of adjustment to long-run equilibrium
following a shock to the system.
We include a dummy variable to show the periods when the country experienced major
standoffs with the donor community (the country was out of the IMF programme). We are
cognizant of the fact that such period of standoff provided an opportunity for foreign aid to be
disbursed outside the regular government channels, the so called bypass. We do this to
account for such unexpected aid shocks.
To test for the robustness of the results as discussed above, we introduce two new
variables AIDRES and AIDRESP derived from the HP filter and an interaction of the filter
with the policy variable respectively. These two variables are used in the above two models in
place of XAIDB and PRIP respectively. A graphical representation of the foreign aid
disbursement and HP trend is shown in Figure A3 in the appendix. Figure A4 shows the aid
cycle as derived from the HP filter. This is a confirmation of the pro-cyclical nature of foreign
aid and could be an indication of aid unpredictability.
5. Data
The study used secondary sources of data covering the period 1966 – 2010. The data on
real GDP, Inflation, degree of openness, Final Government Consumption and private
investment were obtained from the World Bank, Africa Development Indicators database.
Other complimentary sources were Economic Survey, and Statistical Abstract. Foreign aid
data (commitments and disbursements) were from the Organization for Economic
Corporation and Development (OECD), OECD.Stat online database. This data was
supplemented with those from the World Bank, Africa Development Indicators. Data on tax,
external debt variables were from Economic Survey (various issues).
Real Per capita income is defined as real income divided by the population. Per capita
income is often considered an indicator of a country's standard of living. The economic
growth rate is measured in this study as the growth of real GDP per capita in constant (2000)
U.S. dollars. Private Investment is proxied by the gross capital formation less public
investment, and was measured as a share of GDP. Foreign Aid (Aid) is defined as Official
Development Assistance (ODA), which includes all loans with a grant component above 25
per cent measured as a share of GDP. Aid Predictability Indicator is defined as the difference
12
between commitments and disbursements as a percentage of disbursements (also known as
the unpredictability per a unit of aid dollar). Tax is defined as the financial charge or other
levy charged by the state to a taxpayer and is an important source for government revenue. In
this study the tax variable is measured as a share of GDP. External Debt is the stock of
resources borrowed externally by the Government and measured as a share of GDP. Openness
to trade refers to the degrees to which countries or economies permit or have trade with other
countries or economies. The trading activities include import and export, foreign direct
investment (FDI), borrowing and lending, and repatriation of funds abroad. It was measured
as a summation of exports and imports as a percentage of GDP. Inflation is defined as a
percentage change in consumer price index. Final Government Consumption Expenditure
includes all government current expenditures for purchases of goods and services (including
compensation of employees). It also includes most expenditure on national defense and
security, but excludes government military expenditures that are part of government capital
formation. It is measured as a share of GDP. Macroeconomic Policy Index is constructed
from selected macroeconomic variables such as inflation, degree of openness and final
government consumption. An increase in the policy index is an indication of unstable
macroeconomic policy environment.
6. Empirical Results
The cointegration results are shown in Table A2 in the Appendix. The results show that
the calculated F-statistic (4.948) was higher than the upper bound critical value at 1 per cent
level of significance (4.832) using a restricted intercept and no trend. The calculated F-
statistic was also higher than the upper bound critical value at 5 per cent level of significance
(4.004) using restricted intercept and trend. Thus, the null hypothesis of no long-run
relationship between the above variable was rejected irrespective of the order of their
integration implying that there was a long-run relationship between the variables.
We estimate Equation (4) and use SBC in selecting the lag length on each of the first
differenced variable. The estimates of the ARDL representation are summarised in Appendix
Table A3. The requisite diagnostic tests of the primary model and the robustness check model
are also provided.
The table shows that there was no evidence of autocorrelation in the disturbance of the
error term with p-values of 0.785. The study used Breuch-Godfrey Lagrange Multiplier (LM)
to test for serial correlation. The ARCH tests suggest the errors were homoskedastic and
independent of the regressors as evidenced by the p-values of 0.198. The model passed the
Jarque-Bera normality tests (p-value of 0.943), suggesting that the errors are normally
distributed. The RESET test indicated that the model was correctly specified with p-values of
0.192.
Appendix Figures A4 and A5 show the plots of both the CUSUM and CUSUMSQ for
the primary model. It can be seen from the figures that the plot of CUSUM stays within the
critical 5 per cent bound for all equations, and CUSUMSQ statistics does not exceed the
critical boundaries that confirms the long-run relationships between economic growth and the
variables. It also shows that the stability of co-efficient plots lie within the 5 per cent critical
bound, thus providing evidence that the parameters of the model do not suffer from any
structural instability over the period of study. From the foregoing diagnostic tests, it is clear
that the model passed all the required tests and thus paving way for interpretation of estimates
of both the long-run and short-run coefficients as required in an ARDL approach. It is
13
therefore on the basis of these tests that it was reasonable to claim that the model had a good
statistical fit.
The empirical results of the short run error-correction model (ECM) results are
presented in Appendix Table A4. The estimated long-run coefficients are presented in
Appendix Table A5. Appendix Table 4 shows that the coefficient of the adjustment term (ecm
(-1)) is significant at the 5per cent level and carries the expected negative sign. The
coefficient is found to be -0.390, suggesting that that the deviation from the long-term in
economic growth is corrected by 39 per cent in the coming year. This means that an
exogenous shock to the economy dissipates completely within two and half years. The
relatively slow speed of adjustment could be attributed to the structural rigidities - lack of
access to productive land, lack of capital, problem of enclave formal economies, export of raw
materials, policy distortions - that are inherent in developing countries such as Kenya, which
slows down the adjustment process as intimated by M'Amanja and Morrissey (2005).
The long-run coefficients for the equation are reported in Table A5. Results show that
the elasticity of foreign aid was positive and statistically significant at 5 per cent in the long-
run. The implication of this is that a 1 per cent increase in foreign aid would result in a 0.16
per cent increase in economic growth in the long-run. The study finding is in line with other
studies such as Mosley (1980), Singh (1985), Mosley et al. (1987), Snyder (1993), Murthy,
Ukpolo and Mbaku (1994), Gounder (2001), Lloyd, Morrissey and Osei (2001), Mavrotas
(2002), M’Amanja and Morrissey (2005), Karras (2006), Bhattarai (2009), and Kargbo
(2012).
From the results reported, the coefficient of the aid predictability indicator is negative
and statistically significant at 1 per cent. This implies that increased aid unpredictability
weakens economic growth in Kenya. A one per cent increase in aid unpredictability decreases
economic growth by 0.7 per cent. This is expected and is in line with the general hypothesis in
the literature that increased aid unpredictability reduces economic growth. This is also
consistent with the findings of Lensink and Morrissey (2000) and Kodama (2011) who found
that aid uncertainty negatively affects economic growth.
Results further show that coefficient of the log of macroeconomic policy environment
was negative and statistically significant at 1 per cent. This implies that a 1 per cent increase
in the macroeconomic policy environment (instability) leads to a decrease in economic
growth by 1.2 per cent and points to the fact that Kenya has experienced macroeconomic
instability for a greater part of the study period.
The coefficient of the interaction term between aid unpredictability and macroeconomic
policy environment is found to be positive and statistically significant at 1 per cent level in
both the short and long-run. Thus, a one per cent increase in aid unpredictability could
increase economic growth by 0.26 per cent in the long run under unstable macroeconomic
policy environment. This implies that aid unpredictability could be beneficial for economic
growth in both the short run and long run during periods of unstable macroeconomic policy
environment. Uncertain aid inflow has the potential to increase prudence in the use of limited
resources a country has thus improving economic growth.
The foregoing findings are confirmed when a robustness check is conducted. The results
of this model are presented in Appendix Table A6 and A7 for the Short run and Long run
respectively.
14
We assess the impact of aid unpredictability on growth when a country is experiencing
some shock emanating from the donor community. Kenya has experienced episodes of frosty
relationship with her development partners, particularly the World Bank and IMF resulting in
aid freeze and some episodes of abundant aid flows. For example, the World Bank in July
1982 did not release the second tranche of US$50 million on the basis that Kenya was lax in
undertaking policy reforms (Njeru, 2003). Further freezes were experienced in 1992 and
1997. To assess the impact during these periods, we introduce a dummy variable to account
for periods when Kenya was not on the IMF/World Bank programme (periods of standoff).
The additional set of models in which the periods are disaggregated to take care of IMF=1 and
0 otherwise and vice versa was constructed and interacted with the aid predictability indicator.
The results reported in Appendix Table A8 and A9 show that there was no statistically
significant effect when this variable was taken into consideration. This could be attributed to
the fact that while there was a standoff between the Government of Kenya and the Bretton
Woods institutions, the bilateral donors opted to channel their funds through the Non-
Governmental Organizations (NGOs) or established Project Management Units (PMUs)
through which they were able to ensure that their aid flows continued. In addition, it could be
argued that the periods of standoff where short and therefore the magnitude of their effects
was not drastically felt given that foreign aid from the bilateral donors was flowing albeit
through a bypass means.
7. Conclusion and Policy Issues
Aid-Growth literature is awash with studies that either support or find no evidence for
the importance of aid as a driver of growth. This study examines a related but different aspect
of this debate; whether aid unpredictability has any positive or negative effects on economic
growth. Proponents have argued that some level of aid unpredictability may force recipient
countries to clean up their on systems and this may improve aid effectiveness and economic
growth. Critics argue that aid unpredictability disorientates the recipient country’s
consumption and investment plans and may adversely affect economic growth. They also
argue that unpredictability of aid may increase pro-cyclicality of growth as aid may be
disbursed during good times and slowed down during hard times.
Using an autoregressive distributed lag model, the findings show that increased aid
unpredictability reduces economic growth in Kenya. In addition, aid unpredictability is found
to improve economic growth during periods of macroeconomic instability implying that aid
unpredictability forces governments to be more prudent in managing the limited uncertain
resources at their disposal during periods of macro instability.
On the part of donors, the finding that aid unpredictability reduces the pace of economic
growth should be a pointer to the need for the trade-off between strict enforcement of
conditionalities and the need to ensure that aid is useful for growth particularly during periods
of macro stability. However, the donors need to be strict and withdraw aid if the
macroeconomic environment is weak since this is found to instill prudence among the weak
economies that increases economic growth.
While appreciating the number of studies that have been done on the foreign aid debate,
there are still new areas that have not been examined. This study is among the few that have
examined the issue of aid unpredictability. The emerging interest on the nexus between
foreign aid and trade (aid for trade) on one hand and aid and private sector development are
other areas that still require examination.
15
8. References
Bearce, D. H., and Tirone, D. C. (2008). Foreign Aid, Recipient Growth, and the Strategic
Goals of Donor Governments. Unpublished manuscript, University of Pittsburgh.
Retrieved on 2011-04-05 at http://www.allacademic.com//meta/p_mla_apa_research_
citation/2/0/9/5/5/pages209559/p209559-1.php.
Boakye, P. F. (2008). Foreign Aid and Economic Growth in Ghana (1970-2005).
(Unpublished Masters dissertation), University of Science and Technology, Kumasi,
Ghana.
Brown R. L., Durbin J., Evans J. M. (1975). Techniques for Testing the Constancy of
Regression Relationships over Time. Journal of the Royal Statistical Society. Series B
(Methodological), 37: 149–92.
Bulỉř, A., and Hamann, A. J. ( 2001). How Volatile and Unpredictable Are Aid Flows,and
What Are the Policy Implications? IMF Working Paper 167, October.
Bulỉř, A., and Hamann, A. J. ( 2003). Aid Volatility : An Empirical Assessment. IMF Staff
Papers 50(1), p. 64-89.
Bulỉř, A., and Hamann, A. J. (2005). Volatility of Development Aid: From the Frying Pan into
the Fire ? IMF, mimeo.
Burnside, C., and Dollar, D. (1997). Aid Spurs Growth – in a Sound Policy Environment,
Finance and Development, Vol. 34 (4), pp. 4-7.
Burnside, C., and Dollar, D. (2000). Aid, Policies, and Growth. American Economic Review,
90(4), pp. 847–868.
Celasun, O., and Walliser, J. (2008). Predictability of Aid: Do Fickle Donors Undermine the
Predictability of Aid? Economic Policy, 23(55), pp. 545-86.
Chauvet, L., and Guillaumont, P. (2008). Aid, Volatility and Growth Again: When Aid
Volatility Matters and When It Does Not. UNU-Wider Working Paper No. 2008/78.
Chenery, H. B., and Bruno, M. (1962). Development Alternatives in an Open Economy: The
Case of Israel. Economic Journal, 72, pp. 79-103.
Chenery, H., and Strout, A. (1966). Foreign Assistance and Economic Development.
American Economic Review, Vol. 66. No.4, pp. 679-733.
Easterly, W., Levine, R., and Roodman, D. (2003). New Data, New Doubts: A Comment on
Burnside and Dollar's Aid, policies and growth, NBER Working Paper No. 9846,
Cambridge, MA.
Ekanayake, E. M., and Chatrna, D. (2009). The Effect of Foreign Aid on Economic Growth in
Developing Countries. Journal of International Business and Cultural Studies.
Feeny, S. (2005). The Impact of Foreign Aid on Economic Growth in Papua New Guinea. The
Journal of Development Studies, Vol.41, No.6, August 2005, pp.1092 – 11.
Gómez, L. M. (2005). Nicaragua: Harmonisation and Alignment Process, presented at
Workshop on “The International Aid System: What’s Next?” Accessed on 2011-04-05
at www.oecd.org/dev/meetings/aidworkshop.
Gong, L., and Heng-fu, Z. (2001). Foreign Aid Reduces Labour Supply and Capital
Accumulation. Review of Development Economics, Vol. 5, No. 1, pp. 105 – 118.
16
Gounder, R. (2001). Aid-Growth Nexus: Empirical evidence from Fiji. Applied Economics.
Vol. 33, pp. 1009- 1019.
Hansen B. E. (1992). ‘Tests for Parameter Instability in Regressions with I(1) Processes’,
Journal of Business and Economic Statistics, 10: 321–35.
Hansen H., Johansen S. (1999). Some Tests for Parameter Constancy in Cointegrated VAR-
Models. Econometrics Journal, 2: 306–33.
Herzer, D., and Morrissey, O. (2011). Long-Run Aid Effectiveness. CSAE Conference 2011,
Economic Development in Africa, Oxford, March 20-22, Paper No. 144, Retrieved on
2012-04-12 at www.csae.ox.ac.uk/conferences/2011-EDiA/papers/144-Morrissey.pdf.
Hodrick, R., and Prescott, E. C. (1997). Postwar U.S. Business Cycles: An Empirical
Investigation. Journal of Money, Credit, and Banking 29(1), p. 1-16.
International Monetary Fund. (2007). Fiscal Policy Response to Scaled-Up Aid, Washington,
DC: International Monetary Fund.
Javid, M., and Qayyum, A. (2011). Foreign Aid-Growth Nexus in Pakistan: Role of
Macroeconomic Policies. Retrieved on 2012-04-12 at http://mpra.ub.uni-
muenchen.de/29498/.
Kargbo, P. M. (2012). Impact of Foreign Aid on Economic Growth in Sierra Leone Empirical
Analysis. UNU-Wider Working Paper No. 2012/07.
Karras, G. (2006). Foreign Aid and Long-run Economic Growth: Empirical Evidence for a
Panel of Developing Countries. Journal of International Development, vol.18, no.7,
pp.15–28.
Knack, S. (2000). Aid Dependence and the Quality of Governance: A Cross-Country
Empirical Analysis. World Bank Policy Research Paper.
Kodama, M. (2011). Aid Unpredictability and Economic Growth. World Development Vol.
40(2), pp. 266-272.
Legovini, A. (2002). Kenya: Macro Economic Evolution Since Independence. Retrieved on
2011-4-12 at http://ebookbrowse.com/kenya-macro-economic-evolution-since-indepen
dence-doc-d28496277.
Lensink, R., and Morrissey, O. (2000). Aid Instability as a Measure of Uncertainty and the
Positive Impact of Aid on Growth. Journal of Development Studies, 36(3), pp. 31-49.
M’Amanja, D., and Morrissey, O. (2005). Foreign Aid, Investment, and Economic Growth in
Kenya: A Time Series Approach. Credit Research Paper, Centre for Research in
Economic Development and International Trade. University of Nottingham.
Mallik, G. (2008). Foreign Aid and Economic Growth: A Cointegration Analysis of the Six
Poorest African Countries. Economic Analysis and Policy, Vol. 38 No. 2, September.
Morrissey, O. (2001). Does Aid Increase Growth in Cameroon? Progress in Development
Studies, Vol. 1, No. 1, pp. 37 – 50.
Mosley, P. (1980). Aid, Savings and Growth Revisited. Oxford Bulletin of Economics and
Statistics, Vol. 42 (2), pp. 79-95.
Mosley, P., Hudson, J., and Horrell, S. (1987). Aid, the Public Sector and the Market in Less
Developed Countries. Economic Journal, Vol. 97 No. 388, pp. 616-41.
17
Mwega, F. M. (2004). Foreign Aid, Monetary Policy and Economic Growth in Kenya, paper
prepared for Kenya Institute for Public Policy Research and Analysis (KIPPRA).
Mwega, F. M. (2009). A Case study of aid effectiveness in Kenya: Volatility and
fragmentation of foreign aid with a focus on health. A paper commissioned by the
Wolfensohn Center for Development at the Brookings Institution.
Narayan, P. K. (2004). Reformulating the Critical Values for the Bounds R statistics
Approach to Cointegration: An Application to the Tourism Demand Model for Fiji.
Discussion Papers, Department of Economics, Monash University, Australia.
Newlyn, W. T. (1973). The Effect of Aid and Other Resource Transfers on Savings and
Growth in Less Developed Countries: A Comment. Economic Journal 83(331), pp.
863-69.
Njeru, J. (2003). The Impact of Foreign Aid on Public Expenditure: The Case of Kenya.
AERC Research Paper No. 135, November.
Oduor, J., and Khainga, D. (2009). Effectiveness of Foreign Aid on Poverty Reduction in
Kenya, Global Development Network Working Paper No. 34, November.
OECD. (2005). Paris Declaration on Aid Effectiveness: Ownership, Harmonisation,
Alignment, Results and Mutual Accountability. Paris.
Pallage, S., and Robe, M. (2001). Foreign Aid and the Business Cycle. Review of
International Economics 9(4), p.641-672.
Papanek, G. (1972). The Effects of Aid and Other Resource Transfers on Savings and Growth
in Less Developed Countries. Economic Journal, Vol.82, No. 327, pp. 934-50.
Papanek, G. (1973). Aid, Foreign Private Investment, Saving and Growth in Less Developed
Countries. Journal of Political Economy, Vol. 81, No. 1, pp. 137-138.
Pedersen, K. R. (1996). Aid, Investment and Incentives. Scandinavian Journal of Economics,
Vol. 98, No. 3, pp. 423-438.
Pesaran, M. H. (1997). The role of Economic Theory in Modeling the Long-run. Economic
Journal, Vol. 107, pp. 178-191.
Pesaran, H. M., and Pesaran, B. (1997). Microfit 4.0, Oxford: Oxford University Press.
Pesaran, H. M., and Shin, Y. (1995). An Autoregressive Distributed Lag Modelling Approach
to Cointegration Analysis. DAE Working Paper Series, No.9514, Department of
Applied Economics, University of Cambridge.
Pesaran, M. H., and Shin, Y. (1999). An Autoregressive Distributed Lag Modelling Approach
to Cointegration Analysis, Chapter 11, in Storm, S., (ed.), Econometrics and Economic
Theory in the 20th Century: the Ragnar Frisch Centennial Symposium, Cambridge:
Cambridge University Press.
Pesaran, H. M., Shin, Y., and Smith, R. (1996). Testing the Existence of a Long-run
Relationship. DAE Working Paper Series, No.9622, Department of Applied
Economics, University of Cambridge.
Radelet, S. (2006). A Primer on Foreign Aid. Working Paper No 92. Washington, Center for
Global Development.
Rand, J., and Tarp, F. (2002). Business Cycles in Developing Countries : Are They Different?
World Development 30(12), p. 2071-2088.
18
Ravn, M. O., and Uhlig, H. (2002). On Adjusting the Hodrick-Prescott Filter for the
Frequency of Observations. The Review of Economics and Statistics 84(2), p. 371-380.
Rosenstein-Rodan, P. N. (1961). International Aid for Underdeveloped Countries. Review of
Economics and Statistics, 43 (May), pp. 107-48.
Saasa, O , and 10 . (2005). Features and Performance of Local Aid Systems, presented at
Workshop on “The International Aid System: What’s Next?” Retrieved on 2011-05-12
at www.oecd.org/dev/meetings/aidworkshop.
Sagasti, F., Bezanson, K., and Prada, F. (2005). The Future of Development Financing:
Challenges, Scenarios and Strategic Choices. Expert Group on Development Issues
(EGDI) Secretariat, Ministry for Foreign Affairs, Government of Sweden. Institute of
Development Studies. University of Sussex. Retrieved on 2011-06-05 at: www.new-
rules.org/storage/documents/ffd/segasti-bezanson-prada04.pdf.
Salisu, A. A., and Ogwumike, F. O. (2010). Aid, Macroeconomic Policy Environment and
Growth: Evidence from Sub-Saharan Africa. Journal of Economics Theory, Vol. 4,
Issue 2, pp. 59-64.
Singh, R. D. (1985). State Intervention, Foreign Economic Aid, Savings and Growth in LDCs:
Some Recent Evidence. KYKLOS, Vol.38, No.2, pp. 216-232.
Snyder, D. W. (1993). Donor Bias towards Small Countries: An Overlooked Factor in the
Analysis of Foreign Aid and Economic Growth. Applied Economics, Vol. 25, pp. 481-
488.
Sricharoen, T., and Buchenrieder, G. (2005). ‘Principal Component Analysis of Poverty in
Northern Thailand.’ Paper presented at the Conference on International Agricultural
Research for Development, 11–13 October. Stuttgart-Hohenheim.
Vargas, H. (2005). Assessing Rhetoric and Reality in the Predictability of Aid. UNDP Human
Development Report Office Occasional paper 2005/25, UNDP.
Wawire, N. H. W. (2006). The Determinants of Tax Revenue in Kenya (Unpublished PhD
Thesis), Nairobi, Kenyatta University.
World Bank. (1998). Assessing Aid: What Works, What Doesn’t and Why. New York: Oxford
University Press.
World Bank. (2007). Global Monitoring Report 2007, Millennium Development Goals:
Confronting the Challenges of Gender Equality and Fragile States, Washington, DC:
The World Bank.
Zimmermann, F. (2005). The International Aid System: A Question of Perspective. OECD
Policy Insight, No. 12. Accessed on 2011-07-07 at www.oecd.org/dataoecd/25
/23/35275333.pdf.
19
Appendix
Source: Based on data from OECD/DAC, March, 2013
Figure A1: Average ODA flows to Kenya from multilateral and bilateral donors
Source: Based on data from OECD/DAC March, 2013
Figure A2: Foreign aid commitments and disbursements (1966-2011)
Constructing the Macroeconomic Policy Index
Burnside and Dollar (1997, 2000), Feeny (2005) and Javid and Qayyum (2011) all
assumed that distortions affect growth that will determine the effectiveness of aid. Thus, they
assigned the weights to the policy variables according to their correlation with growth.
20
The macroeconomic policy index is constructed using the principal component analysis
with SPSS 17.0. The advantage of this construction is that it helps to conserve the degrees of
freedom arising from the reduction in the number of variables in the model while at the same
time helping to avoid the possibilities of high correlation among the macroeconomic variables
(Kargbo, 2012). Sricharoen and Buchenrieder (2005:2) consider this approach as 'indicator
reduction procedure to analyze observed variables that would result in a relatively small
number of interpretable components (group of variables), which account for most of the
variance in a set of observed variables’. The variables used in this study were inflation (INF)
as a proxy for monetary policy, final government consumption expenditure (FGC) as a proxy
for fiscal policy and degree of openness (OPEN) as a proxy for trade policy.
The Principal Components is a method that enables identification of patterns in data and
expressing them in a way that takes into consideration their similarities and differences. In
other words, it is a technique which involves a data reduction process in which the variables
are scored without much loss of information. The Eigen values for shows that 46 per cent of
the variance is explained by the first principal component, while the second and third account
for the remaining 54 per cent. The implication of this is that the first principal component
alone explains the variation of the dependent variable better than any other combination of the
three variables used.
We therefore considered the first principal component as an appropriate macroeconomic
policy index measure. The scores obtained in the construction of the variable are -0.160 for
inflation, 0.322 for final government consumption expenditure and 0.122 for degree of
openness. The respective contribution of the components of the macroeconomic policy index
is shown as:
where α1, α2 and α3 are the weights for inflation, final government consumption and
openness. The signs attached to the weights are essential in the construction of the policy
index such that α1< 0, α2> 0 and α3>0.
Policy index = -0.160INF+ 0.322FGC+0.122OPEN.
Index Policy 3 2 1 INF + OPEN FGC
21
Table A1: Principal Component Analysis for the Generation of the Policy Index
Variable
Principal component Eigenvalues % of variance Cumulative %
1 1.384 46.121 46.121
2 1.005 33.489 79.610
3 0.612 20.390 100.000
Variable Component loadings Component score
INF -0.160 0.588
FGC 0. 322 -0.116
OPEN 0.122 -0.026
Figure A3: Foreign aid disbursements and HP Trend(1966-2011)
22
Figure A4: HP Aid Cycle (1966-2011)
Table A2: The F-Statistic of Cointegration Relationship
Test Statistic Value Lag Significance
Level
Bound Critical
Values* (restricted
intercept and no
trend)
Bound Critical
Values* (restricted
intercept and trend)
F-statistic 4.948 1
1%
5%
10%
I(0) I(1) I(0) I(1)
3.383
2.504
2.131
4.832
3.723
3.223
3.595
2.643
2.238
5.225
4.004
3.461
Note: * Based on Narayan (2004)
23
Table A3: ARDL Estimations
Primary model
Robustness Check
Coefficient t-ratio p-value Coefficient t-ratio p-value
Constant 1.541** 4.379 0.000 1.142** 3.1599 0.004
Log Real per capita GDP (-1) 1.106** 9.640 0.000 1.190** 10.350 0.000
Log Real per capita GDP (-2) -0.496** -3.850 0.001 -0.485** -3.953 0.001
Log Private investment 0.027 0.658 0.516 0.042 1.203 0.240
Log Private investment (-1) 0.087 1.502 0.145
Log Private investment(-2) -0.096* -2.313 0.029
Log Foreign aid -0.011 -0.499 0.657 -0.058 -1.423 0.167
Log Foreign aid(-1) 0.074** 2.855 0.008 0.133** 3.774 0.001
Macro policy -0.199** -5.488 0.000 -0.170** -4.629 0.000
Macro policy(-1) -0.27* -1.934 0.064
Aid interacted with policy 0.001 1.666 0.107 0.003* 2.457 0.021
Aid interacted with policy(-1) -0.004** -5.381 0.000 -0.007** -5.801 0.000
Aid interacted with policy(-2)
0.002* 2.649 0.014
Aid predictability indicator -0.263** -3.442 0.002 1.94E-04 1.638 0.114
Aid predictability indicator (-1)
-4.62E-04** -4.386 0.000
Aid predictability indicator (-2)
-9.01E-05* -2.342 0.027
Predictability interacted with
policy 0.100**
3.223 0.003 -8.92E-06 -1.609 0.120
Predictability interacted with
policy
2.28E-05** 4.325 0.000
Log External debt (-1) 0.607** 3.992 0.000 0.390** 7.643 0.000
Log External debt(-2) -0.112** -4.379 0.000 -0.171** -4.262 0.000
R -Bar Squared 0.959
0.94593
SE of Regression 0.020
0.020911
F-Stat (15, 27) = 66.909**
(16,25) = 45.8274**
Diagnostic Tests
Serial correlation
F[1,26] = 0.045
(0.834)
F(1,24) = 0.055 (0.817)
Heteroscedasticity
F[1,41] = 1.642
(0.207)
F(1,40) = 3.410(0.072)
Functional form
F[1,26] = 1.071
(0.310)
F(1,24) = 2.676(0.115)
Normality (CHSQ 2)
CHSQ[2] =
0.118(0.943)
CHSQ(2) =
1.392(0.498)
Testing for existence of a level relationship among the variables in the ARDL model
F-statistics 2.537
3.854
95% Bound (Lower, Upper) (3.1065, 4.4903)
(3.122, 4.500)
90% Bound (Lower, Upper) (2.6196, 3.8401)
(2.612, 3.857)
* Significant at 5 per cent and ** indicate significant at 1 per cent.
24
Table A4: The Error Correction Representation for the Selected ARDL model
Dependent variable is Δ Log real per capita GDP Coefficient p-value
Δ Log real per capita GDP1 0.496**
(3.850)
0.001
Δ Log private investment 0.027
(0.658)
0.516
Δ Log foreign aid -0.011
(-0.449)
0.657
Δ Log macro policy -0.199**
(-5.488)
0.000
Δ Aid interacted with policy 0.001
(1.667)
0.106
Δ Aid predictability indicator -0.262**
(-3.442)
0.002
Δ Predictability interacted with policy 0.100**
(3.223)
0.003
Δ Log foreign debt(-1) 0.607**
(3.992)
0.000
Δ Log foreign debt(-2) -0.112
(-4.286)
0.000
ecm(-1) -0.390**
(-4.477)
0.000
R-Bar-Squared 0.752
Akaike Info. Criterion 100.714
Schwarz Bayesian Criterion 86.625
DW-statistic 2.025
F-statistic F(11, 31) 12.929** 0.000
Note: (1) * Significant at 5 per cent and ** indicate significant at 1 per cent. (2) Δ Log Real per capita GDP1 = Log Real per
capita GDP(-1)-log Real per capita GDP (-2). (3) Δ is first difference estimator. (4) t-statistics in parenthesis.
25
Table A5: Long-Run Estimates
Dependent variable is log real per
capita GDP
Coefficient p-value
Constant 3.956**
(7.017)
0.000
Log private investment 0.046
(0.388)
0.701
Log foreign aid 0.161*
(2.016)
0.054
Log macro policy -1.204**
(-5.162)
0.000
Aid interacted with policy -0.007**
(-2.875)
0.008
Aid predictability indicator -0.674**
(-2.620)
0.009
Predictability interacted
with policy
0.258**
(2.483)
0.020
Log foreign debt(-1) 1.558**
(6.256)
0.000
Log foreign debt(-2) -0.289**
(-2.983)
0.006
Note: * Significant at 5 per cent and ** indicate significant at 1 per cent. t-statistic in parenthesis.
26
Table A6: The Error Correction Representation for the Selected ARDL model (Robustness
Check Model)
Dependent variable is Δ Log real per capita GDP Coefficient p-value
Δ Log real per capita GDP1 0.485**
(3.953)
0.000
Δ Log private investment 0.042
(1.203)
0.239
Δ Log foreign aid -0.058
(-1.423)
0.165
Δ Log macro policy -0.170**
(-4.629)
0.000
Δ Aid interacted with policy 0.003**
(2.457)
0.020
Δ Aid interacted with policy1 -0.002**
(-2.649)
0.013
Δ Aid predictability indicator 0.193E-3**
(1.638)
0.112
Δ Aid predictability indicator1 0.901E-4**
(2.342)
0.026
Δ Predictability interacted with policy -0.892E-5**
(-1.609)
0.118
Δ Log foreign debt(-1) 0.390**
(7.643)
0.000
Δ Log foreign debt(-2) -0.171
(-4.262)
0.000
ecm(-1) -0.295**
(-4.271)
0.000
R-Bar-Squared 0.840
Akaike Info. Criterion 96.733
Schwarz Bayesian Criterion 81.963
DW-statistic 1.881
F-statistic F(12, 29) 10.899 0.000
Note: (1) * Significant at 5 per cent and ** indicate significant at 1 per cent. (2) Δ Log Real per capita GDP1 = Log Real per
capita GDP(-1)-log Real per capita GDP (-2). (3) Δ is first difference estimator. (4) t-statistics in parenthesis.
27
Table A7: Long-Run Estimates (Robustness Check Model)
Dependent variable is log real per
capita GDP
Coefficient p-value
Constant 3.870**
(8.289)
0.000
Log private investment 0.143
(1.040)
0.308
Log foreign aid 0.255**
(2.747)
0.011
Log macro policy -0.575**
(-3.074)
0.005
Aid interacted with policy -0.007**
(-2.458)
0.021
Aid predictability indicator (HP Residual
Filter)
-0.001**
(-2.222)
0.036
Predictability interacted
with policy
0.4715E-4*
(1.994)
0.057
Log foreign debt(-1) 1.321**
(4.289)
0.000
Log foreign debt(-2) -0.578**
(-2.774)
0.010
Note: * Significant at 5 per cent and ** indicate significant at 1 per cent. t-statistic in parenthesis.
28
Table A8: The Error Correction Representation for the Selected ARDL model (Accounting for
external shocks)
Dependent variable is Δ Log real per capita GDP Coefficient p-value
Δ Log real per capita GDP1 0.616**
(3.807)
0.000
Δ Log foreign aid 0.0124
(0.442)
0.661
Δ Log macro policy -0.134**
(-3.661)
0.001
Δ Aid interacted with policy 0.001
(1.265)
0.215
Δ Aid predictability indicator -0.056
(-0.739)
0.465
Δ IMF Dummy -0.018
(-0.290)
0.774
Δ Predictability interacted with IMF dummy 0.034
(0.451)
0.655
Δ Log foreign debt(-1) 0.684**
(3.668)
0.001
Δ Log foreign debt(-2) -0.099
(-3.115)
0.004
ecm(-1) -0.465**
(-4.587)
0.000
R-Bar-Squared 0.623
Akaike Info. Criterion 89.453
Schwarz Bayesian Criterion 77.289
DW-statistic 2.147
F-statistic F(10, 31) 8.072 0.000
Note: (1) * Significant at 5 per cent and ** indicate significant at 1 per cent. (2) Δ Log Real per capita GDP1 = Log Real per
capita GDP(-1)-log Real per capita GDP (-2). (3) Δ is first difference estimator. (4) t-statistics in parenthesis.
29
Table A9: Long-Run Estimates (accounting for external shocks)
Dependent variable is log real per
capita GDP
Coefficient p-value
Constant 3.711**
(11.711)
0.000
Log foreign aid 0.157**
(2.498)
0.019
Log macro policy -1.031**
(-4.831)
0.000
Aid predictability indicator -0.120
(-0.736)
0.468
Aid interacted with policy -0.005**
(-2.850)
0.008
IMF Dummy -0.038
(-0.291)
0.773
Predictability interacted with IMF
Dummy
0.074
(0.450)
0.656
Log foreign debt(-1) 1.472**
(5.873)
0.000
Log foreign debt(-2) -0.214**
(-2.556)
0.016
Note: * Significant at 5 per cent and ** indicate significant at 1 per cent. t-statistic in parenthesis
30
-20
-10
0
10
20
1968 1979 1990 2001 2010
The straight lines represent critical bounds at 5% significance level
Plot of Cumulative Sum of Recursive Residuals
Figure A5: Cumulative Sum of Recursive Residuals
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1968 1979 1990 2001 2010
The straight lines represent critical bounds at 5% significance level
Plot of Cumulative Sum of Squares of Recursive Residuals
Figure A6: Cumulative Sum of Squares of Recursive Residuals
31
-20
-10
0
10
20
1969 1980 1991 2002 2010
The straight lines represent critical bounds at 5% significance level
Plot of Cumulative Sum of Recursive Residuals
Figure A7: Cumulative Sum of Squares of Recursive Residuals
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1969 1980 1991 2002 2010
The straight lines represent critical bounds at 5% significance level
Plot of Cumulative Sum of Squares of Recursive Residuals
Figure A8: Cumulative Sum of Squares of Recursive Residuals
32
Recent Publications in the Series
nº Year Author(s) Title
225 2015 Andinet Woldemichael and Abebe
Shimeles
Measuring the Impact of Micro-Health Insurance on
Healthcare Utilization: A Bayesian Potential Outcomes
Approach
224 2015 Gibert Galibaka
Sophistication of Fruits, Vegetables and Derivatives Exports in the WAEMU space
223 2015 Zorobabel Bicaba, Zuzana Brixiová
and Mthuli Ncube
Extreme Poverty in Africa: Trends, Policies and the Role
of International Organizations
222 2015 Anthony M. Simpasa, Abebe
Shimeles and Adeleke O. Salami
Employment Effects of Multilateral Development Bank
Suport: The Case of the African Development Bank
221 2015 Ousman Gajigo and Mouna Ben
Dhaou Economies of Scale in Gold Mining
220 2015 Zerihun Gudeta Alemu Developing a Food (in) Security Map for South Africa
219 2015 El-hadj Bah Impact of the Business Environment on Output and
Productivity in Africa
218 2015 Nadege Yameogo Household Energy Demand and the Impact of Energy
Prices: Evidence from Senegal
217 2015 Zorobabel Bicaba, Zuzana Brixiová,
and Mthuli Ncube
Capital Account Policies, IMF Programs and Growth in
Developing Regions
216 2015 Wolassa L Kumo Inflation Targeting Monetary Policy, Inflation Volatility
and Economic Growth in South Africa