Turkey: Economic Reforms, Living Standards and Social Welfare Study

Source: SIS published results, LFS surveys.

Why is Turkey not succeeding at generating sufficient jobs for its growing workforce? There are two potential explanations, not necessarily competing, but with different implications for policy. One is that the economy is simply not growing sufficiently to generate jobs for a fast-expanding population. The second is that the economy is growing enough, but somehow this growth is not sufficiently labor -intensiveor in other words, this growth does not generate enough jobs. The former explanation would highlight barriers to growth as the main policy problem. The second would suggest looking at constraints to labor demand, and biases in the pattern of growth.

Meeting this challenge requires faster GDP growth

While respectable, GDP growth has not been sufficient to fully employ Turkey's fast-growing working age population. During 1981-97, total employment grew by only 1.5% per yearwhile the working-age population (the pool of all potential workers) grew by over 3% per annum. A simple decomposition shows that output per person of working-age increased, on average, by only 1.5% per year. This, in turn, broke down into annual productivity growth of about 3%, and a decline in the employment rate of some 1.5% per year. Had Turkey achieved higher overall GDP growth rates, it would have been able to sustain higher increases in output per worker, and/or employ a higher proportion of its potential workforce.

While Turkey sustained very fast growth rates during the 1960s and early 1970s (of over 6% per annum), these rates have since slowed significantly and have become more volatile. During 1981-97, GDP growth averaged 4.5% per yearstill an impressive achievement when compared to the poor performance of other middle-income countries during this period, for example in Latin America. But arguably not enough to sustain Turkey's goal of converging to the income levels of the rest of the OECD.

If we compare Turkey's recent growth performance to that of two "successful" case studies of catching up to the OECD, Spain and Korea, we see that in order to emulate their performance Turkey needs to grow faster than it has during the past two decades. Annual GDP growth in Spain during its "take-off" period, when its income level was more comparable to Turkey's today, averaged 6.4% per year, while that in Korea during its peak growth phase was close to 10% per annum. In both Spain and Korea, high GDP growth provided the basis for large increases in non-agricultural employment and wages, and a sustained rise in living standards. As a result, both countries are now solidly entrenched among the world's richest group of nations. Turkey is not there yet. With an income per capita equal one-third that of Korea, and one-fifth that of Spain, it needs to continue to grow at 7-10% per year in order to deliver to its citizens a similar leap in living standards and incomes, in a relatively short period of time.

Table 2. Turkey's Growth Performance, 1981-97: Comparisons with Spain and Korea During their Take-Off Periods

In Turkey, the first source of productivity growth (sectoral shifts) has operated strongly during the last two decades, accounting for over three-quarters of all productivity growth (Table 3). However, productivity growth within-sectors has been very low, especially outside of industry. Low within-sector productivity growth is at the root of why overall labor demand has not grown faster. Of particular concern is the weak performance of agriculture in the poorer regions of Turkey. It is this sluggish performance, in combination with the importance of agriculture in terms of shares of output and especially employment, that is slowing down overall growth and employment creation. Thus, providing a foundation for improved productivity performance in agriculture, especially in the poorer regions, can have a large payoff in terms of overall growth.

Source: Own calculations from 1994 HICES. a Consumption per equivalent adult; economies of scale.

The comparison of the 1987 and 1994 Household Income and Consumption Expenditure Surveys results suggests that during this period there was a reduction of about 2.3 percentage points in the overall incidence of economic vulnerability (from 38.5 to 36.2 percent of the population). However, the relatively rapid growth of the population meant that despite the drop in incidence, there was an actual increase in the number of economically vulnerable persons, which grew by more than one million. Progress in reducing absolute poverty was more pronounced and actually lead to a reduction in the total number of the poor in Turkey. Although the direction of change is unmistakable, it is also important to note that the magnitude of decline in poverty is not dramatic. Most households that left poverty between 1987 and 1994, would still be categorized as economically vulnerable in 1994. [7]

What contributed to this reduction in poverty? The main factor was the large population shift between urban and rural areas. As the population in the relatively less poor urban areas has expanded with migration flows, poverty in rural areas has fallen dramatically and hence total poverty has also declined. Demographic changes (lower fertility rates among the poor) also contributed to reducing the number of poor in the country as a whole. Almost a quarter of the overall reduction of poverty was due to these "structural" factors. The other big contributor to the drop in poverty was the increase in literacy rates among heads of households: by itself, this accounted for almost one-half of the measured reduction in economic vulnerability and poverty between 1987 and 1994.

Table 8. Decomposition of Poverty Change into Growth and Redistribution Components

• Social assistance schemes are dispersed and disjointed. The level of benefits is very small, and biased towards certain categories of the population. Some accrue equally to the poor and non-poor (e.g. old age income assistance). The best-covered groups are the elderly and the disabled; while the worse off are the working poor and the unemployed. There is a need to consolidate these disparate efforts, and develop a comprehensive and more uniform social assistance program that adequately targets the poor.
• Despite the wide coverage of the system, social insurance (and especially the pension system) fails to reach the most vulnerable households. The wide coverage of Turkey's social security schemes is clearly a very positive achievement, which places Turkey ahead of other countries with a similar income level. However, to the extent that the system is primarily linked to holding a formal-sector job, it may fail to reach the poorest individuals (Figure 2). While this is not a failure of the social insurance system per se, the lack of systematic anti-poverty interventions that fall outside of the insurance framework raises concerns for the future. If future social protection efforts are channeled exclusively through social insurance mechanisms, they are likely to keep excluding the poorest of the poor. Hence, in parallel to the continued development and improvement of its social insurance system, Turkey also needs to focus on systematic interventions aimed at those who cannot be reached through the system. Moreover, given the lack of redistributive instruments in Turkey, the Government may want to assess the need for introducing a well-distinguished redistributive component within the social insurance system.

• The social insurance system is fiscally unsustainable, and is generating large deficits that need to be covered by the State budget. By absorbing a large and growing fraction of state revenues, these deficits may crowd out any additional resources that would be directed towards the poorest in society. Moreover, the deficits in the general budget are a major contributor to inflation, which acts as a regressive tax. Efforts to place the system on a financially sustainable path are already underway with the new pension reform law approved in August 1999. However, the system will continue to require significant subsidies through the medium term.
• The system of agricultural subsidies represents a significant drain on the budget, and is biased towards richer regions and larger farmers. Agricultural support policies accentuate rather than mitigate the existing regional disparities. The system needs to be revisited, and consolidated into a limited lump-sum transfer, and targeted towards poorer farmers and poor regions.
• The educational system, while comprehensive, does not provide enough access for the poorest. The problem of access is particularly critical for secondary education. There also needs to be a greater push towards ensuring that rural girls have improved entry into schooling. The existing distributional gap between urbanized and rural Turkey, and between men and women, will not be eradicated unless there is truly equal educational opportunity for all Turkish children.

Key elements of a strategy to improve living standards and reduce poverty

• Provide a macroeconomic environment that is conducive to growth and price stability. Address fundamental structural reforms, including the much-needed reform of the public sector, to underpin a sustained reduction in inflation and the basis for continued growth.
• Remove biases against employment creation outside of agriculture. Reduce public sector borrowing requirements, to bring down real interest rates and stimulate investment. Eliminate barriers to competition in product markets.
• Facilitate the outflow of resources from agriculture and provide a basis for productivity growth in the sector. Facilitate rural-urban migration flows. Improve the availability of public infrastructure, especially roads and water supply, in poorer agriculture areas.
• Invest in education, and especially in that of poor children. Facilitate public compliance with the program to extend universal schooling through the eighth grade. Provide incentives for educating girls and flexibility in schooling arrangements. Eliminate barriers that reduce access and attendance among children of poor and rural families. Reduce adult illiteracy through adult education programs.
• Reallocate Government expenditures so that they are better targeted to the economically vulnerable. Continue the reforms of the social insurance system, and assess the need for a well-defined redistributive component. Improve the targeting and coordination of existing social assistance schemes, and gradually increase their size and coverage. Over the medium term, merge existing multiple system into a comprehensive single benefit scheme linked to means or proxy-means testing. Consolidate the system of agricultural subsidies into a limited lump-sum transfer, and target towards poorer farmers and poor regions.

Notes:

1. Concerns over the quality of the data used have been raised at different times during the review process, particularly in light of what is perceived as underreporting of income in household surveys. In view of these concerns, the team would like to stress that the analysis in the Report is solidly based on a variety of data sources (see technical appendix) and on the triangulation of findings. Only in the instances when different data sources tell a consistently similar story does the Report take a definite stand on an issue. Where lack or low quality of data is perceived to be an obstacle, for example in the analysis of use of health services, the report highlights this clearly. Moreover, the team would also like to emphasize that relative to other countries, the quality of data in Turkey is very good, particularly with respect to the information derived from the household surveys carried out by SIS. Information from both the HICES and the LFS are used extensively in this Report.

2. Reported census data for 1997 are from published projections.

3. A similar case is Brazil, where despite huge macroeconomic volatility, the distribution of income has not changed much since 1976.

4. The pension system in Turkey is, in principle, conceived as a social insurance scheme and hence need not have a progressive distributional impact. On the contrary, to the extent that pensions are linked to past wages, they are likely to be unequally distributed. What is more problematic, however, is that the pension system runs a current deficit of about 3% of GDP, which is financed out of direct budget transfers, and hence from general revenues. In this regard, a non-negligible fraction of the state's budget revenues are going to finance a transfer to the top of the distribution, squeezing out expenditures that could be more progressively targeted to the poor.

5. The cost of the minimum food basket in 1994 was about US$36 per month per equivalent adult.

6. Approximately equivalent to twice the level of the minimum food basketor about US$190 per household per month.

7. Since 1994 was a crisis year, the comparison may underestimate the true decrease in poverty that has occurred since 1987. Unfortunately, at this time, the 1994 HICES data are the latest available that document household expenditures and incomes in detail.