2-social protection index technical workshop - review and assessment of estimation issues (terry...

SOCIAL PROTECTION INDEX Review and Assessment

of Estimation Issues

Terry McKinleyDirector, Centre for Development Policy & Research,

School of Oriental and African StudiesInception Workshop on the SPI Update

ADB Headquarters3-4 April 2014, Manila

Estimation Issues: Social Insurance

Social Insurance often dominates the SPI, usually due to expenditures on pensions, sometimes due to universal Health Insurance

1. Pensions: Who are designated as the ‘intended beneficiaries’ of pensions? Answer: the Elderly

a. But how do you define the Elderly? For 2009 we chose the threshold of 60 years of age or older

b. Demographic data are readily available for this threshold (UN Population Division)

2


c. But in some transition economies, people drew pensions when they reached 55 years of age

d. In addition, pension coverage was supposed to cover all the population, worsening the problem

e. Result: the actual beneficiaries exceeded the intended beneficiaries in some cases

f. Even the overall average for Asia and the Pacific as a whole was boosted up unrealistically

g. Partial solution: we capped coverage at 100%h. Option: add in the population share for 55-60

years of age?? But are the data available?3


2. Health Insurance: this is a contributory scheme and is usually based on formal employment (effectively, in the public sector or in sizeable private firms)

a. We chose the ‘Employed Population’ as the intended beneficiaries (based on its potential to contribute to the insurance)

b. It is practically impossible to derive credible estimates of workers who are formally employed

c. Moreover, workers who are self-employed or labour in the informal sector should be eligible

4


d. But what if health insurance is universal (a direction in which some countries have been moving)?

e. If so, we defined the intended beneficiaries as the entire population (otherwise the results would have been distorted)

3. Other Forms of Social Insurance: Provident Funds and Passive Labour Market Programs

a. Provident Funds: Who are their intended beneficiaries?

5


b. Provident Funds (such as in Singapore or Malaysia) are essentially pension schemes

c. But the complication is that they also serve other purposes: participants can withdraw funds early for such activities as financing education, covering health costs and even buying a house

d. So we kept Provident Funds as a separate component of Social Insurance, but without adding to the total number of ‘intended beneficiaries’ (the elderly and the employed population)

6


e. What to do about Passive Labour Market Programs???

f. Should we move them to the third major component of social protection, Labour Market Programs? This third component includes only Active Policies to help put people back to work

g. If so, then who would be the ‘intended beneficiaries’? The unemployed and ‘underemployed’ (not the employed population)?

h. Such passive programs are now hidden within Social Insurance

7


i. Passive Labour Market Programs include several small programs: unemployment benefits, worker injury programs, disability compensation and severance payments. They could also be considered to include Maternity Leave

j. More focussed efforts are needed to gather information on Maternity Benefits: they were small and scattered in the 2009 exercise

k. Also, should we continue to include worker disability pensions in Social Insurance, instead of as part of Social Assistance?

8

Estimation Issues: Social Assistance

Social Assistance has six Sub-Programs:1) Social Transfers (32.5% of SA expenditures)2) Child Welfare (33% of expenditures)3) Disaster Relief (14% of expenditures)4) Assistance to the Elderly (or Social Pensions)

(12% of expenditures)5) Health Assistance (5% of expenditures)6) Disability Programs (3.5% of expenditures)

9


Social Assistance accounts for 36% of all expenditures on social protection but includes 58% of all beneficiaries

1. Social Transfers: cash transfers or in-kind transfers (e.g., food) to poor groups

a. Who are the ‘Intended Beneficiaries’? The income poor, defined by national poverty lines

b. Social Transfers can also include assistance to vulnerable groups, e.g., children or the elderly (but these are included elsewhere)

10


2. Child Welfare: a broad category including allowances, feeding programs, fee waivers or targeted welfare services

a. The broader these programs, the more that they overlap with regular education initiatives: a) Are all children inherently vulnerable? or b) should particularly poor or vulnerable children be targeted with special initiatives?

b. Who are the ‘Intended Beneficiaries’? All children 14 years or younger (this was expanded from 4 years or younger for the 2005 exercise)

11


3. Disaster Relief: an increasingly important form of social assistance in Asia & Pacific

a. Who are the ‘Intended Beneficiaries’? Those affected by ‘Disasters’ (natural or ‘man-made’)

b. But how can we distinguish the actualbeneficiaries from the intended beneficiaries??

c. The 2009 exercise failed to make this distinction (largely because of a lack of data)

d. If we also include pre-emptive measures to prevent disasters, this kind of assistance could become very large

12


e. Given the widespread lack of relevant data, we often had to at least set the ‘intended beneficiaries’ equal to the ‘actual beneficiaries’

f. We should develop some more realistic (i.e., less generous) protocols for dealing with this problem

4. Assistance to the Elderly: Social Pensions (in the form of cash transfers or allowances) accounted for almost all of this sub-program

a. Though rising in importance, such expenditures remain small (4% of all social protection)

13


b. Few social pensions are universal; most still target benefits, such as to the poor (or to the very old)

c. But we did not include a separate category of ‘intended beneficiaries’ for this form of social assistance

d. Was this a mistake? We assumed, in effect, that the great majority of the beneficiaries were poor

e. Should we instead add all of the elderly as the ‘intended beneficiaries’?? Or, instead, estimate the size of the poor elderly?

14


5. Health Assistance: this is a complement to Health Insurance, helping poor and vulnerable groups unable to contribute to insurance (such as offering exemptions from co-payments)

a. This sub-program should be integrated with some form of universal health coverage

b. It remains quite small, accounting for only 5% of all expenditures on social assistance but it does reach 9% of all SA beneficiaries

c. But there is no separate category of ‘intended beneficiaries’ for this form of social assistance

15


6. Disability Programs: These programs are quite small throughout most of Asia & the Pacific, reaching only about 15% of all intended beneficiaries

a. In two-thirds of the countries in Asia & Pacific, such programs do not exist or account for less than 1% of all social protection expenditures

b. Who are the ‘Intended Beneficiaries’? Estimates during 2009 were quite variable. We spent a lot of time reviewing and adjusting them

16


c. Our main reference was the 2011 World Report on Disability (by the WHO and the World Bank)

d. We defined the ‘intended beneficiaries’ as the severely disabled. Across Asia and the Pacific this group averaged about 3% of the total population

e. A broader group that includes the ‘moderately disabled’ would have comprised about 15% of the total population

f. But even the smaller group of ‘severely disabled’ were not being reached by social protection

17


g. Should disability pensions (which are part of social insurance) be incorporated with social assistance for the disabled?

h. Together, these two forms of social protection only reach, in any case, 1.3% of all beneficiaries of social protection

i. The difference is whether the disabled have been part of a contributory scheme (such as contributing to social insurance by means of previous employment)

18

Estimation Issues: Labour Market Programs

Labour Market Programs are the third major program of social protection for the SPI

A. Such programs are small, accounting for only about 5% of all expenditures on social protection and 5% all beneficiaries

B. Such programs include two major forms of activepolicies: 1) skill development and training and 2) special work programs (e.g., India’s employment guarantee scheme)

C. Each accounts for about half of all expenditures and beneficiaries of labour market programs

19


I. Skill Development and Training: this form of social protection provides benefits to the unemployed or underemployed in order to help them gain new employment

So note that not all forms of skill development and training are necessarily part of social protection

II. Special Work Programs (food- or cash-for-work schemes): these can be quite large, such as in India and Bangladesh, and usually provide temporary employment to those out of work

20


Who are the ‘intended beneficiaries’ of such active policies? Answer: the unemployed and ‘underemployed’

But the ‘underemployed’ are difficult to define (do they lack hours or lack adequate income?)

In 2009 consultants had great difficulty in identifying this group. For those that made a credible effort, the average proportion was about 25% of the employed population. This included the unemployed, which are easier to identify by statistics (but also probably less poor)

21


We need to improve our efforts on the 2009 data but what statistic can we use?

One possible option: the employed who earn less than $2 per day (PPP dollars)

This grouping includes the extremely poor employed (earning less than $1.25 per day) and the moderately poor employed (less than $2 per day) but this is still likely a significant proportionof the employed

Data are available for a significant number of countries in Asia & the Pacific from the ILO

22


Should passive labour market programs (which are part of social insurance) be combined with active labour market programs?

Probably, especially since passive programs have traditionally been part of social protection

The beneficiaries of passive policies are usually out of work (though they could have contributed to their benefits through previous employment)

Unemployment benefits are particularly important, accounting for 18% of all expenditures on all labour market programs

23

Summing Up Suggested Estimation Changes

1) Strive to incorporate ‘early retirees’ into the ‘intended beneficiaries’ for pensions

2) Move Passive Labour Market Programs from Social Insurance to Labour Market Programs

3) Make more concerted efforts to identify and estimate maternity leave

4) Make more credible assumptions about the relative size of the ‘intended beneficiaries’ of disaster relief

5) Utilize a more sensible definition of the ‘underemployed’ (such as <$2 per day income????)

6) Include the ‘poor elderly’ as intended beneficiaries for social pensions

24

Summing Up Earlier Suggested Changes for Distributional Issues

The first presentation of the workshop outlined some issues on the Distributional disaggregation of the SPI (poor/nonpoor; women/men)

a. Poor vs. Nonpoor: we had originally intended to present a poverty index in 2009, e.g., the ratio of the SPI for the poor to the SPI for the nonpoor but this ratio was often over 1

b. We have to review the realism of our estimates for poverty targeting for each form of social protection (drawing perhaps on recent surveys)

25


c. The 2005 SPI exercise incorporated an indicator on ‘Social Protection Impact’: per capita benefits received by the poor as a ratio to the national poverty line

d. We might want to review the usefulness of such an approach: we already generate estimates for both the numerator and the denominator of this indicator

e. One drawback: different measures of poverty might populate our estimates: national poverty lines, a regional poverty line and PPP poverty estimates

26


c. Women vs. Men: we also need to review the realism of our estimates of the benefits received by women and men. Program-level data shouldbe disaggregated by gender but they are not

d. We should, at least, incorporate the actual shares of women and men in the total population—instead of lazily assuming that each share is 50%

e. And we should also carry over these shares into assumptions about the employed (SI) and the unemployed plus underemployed (LMP)

27

2-social protection index technical workshop - review and assessment of estimation issues (terry...

Documents