DEPARTMENT OF ECONOMICS Uppsala University
Bachelor Thesis
Authors: Patrik Cras and Christer Rosén Supervisor: Ranjula Bali Swain
Spring 2006
Can income security enhance growth in developing countries?
A study of the effects on economic growth of income support programs for the unemployed and elderly in developing countries.
Abstract
This paper addresses the question if income security can enhance economic growth in developing countries? It takes its starting point in the income security problems of a developing country and summarizes evidence from published empirical research on formal income security mechanisms. We conclude that the findings on incomes security efficiency effects are ambiguous. A limited econometric study based on data from Chile is carried out with a regression showing that social securities total effect on economic growth is negative but more econometric research on total effect on growth are needed to give a definite answer.
Key words: economic growth, income security, social security, developing countries.
Contents
Abstract ... 2
Contents... 3
1. Introduction ... 4
2. Theory of economic growth and social security ... 5
2.1 The Solow growth model ... 6
2.2 The new endogenous growth model... 7
3. Description of income security systems... 11
3.1 Formal Income support for the unemployed ... 12
3.2 Pension ... 14
3.3 Informal insurance mechanisms... 15
4. Effects of income security on Economic behaviour... 16
4.1 Effects of income support for unemployed ... 17
4.1.1 Human capital accumulation... 17
4.1.2 Saving behaviour... 17
4.1.3 Structural transformation... 18
4.1.4 Labour force participation... 19
4.1.5 Economic stability... 20
4.2 Effects of pension... 21
4.2.1 Population growth ... 21
4.2.2 Saving behaviour... 21
4.2.3 Labour force participation... 24
4.2.4 Conclusions of our example from Chile. ... 27
6. Conclusions ... 29
7. References ... 30
Appendix A: Growth regression on Chile... 32
A.1 The data set... 32
A.2 Problems with the data set... 32
A.3 The regression ... 33
A.4 Results ... 39
Appendix B: Pension reform in Chile ... 39
B.1 Economic history of Chile... 39
1. Introduction
Poor people in developing countries are exposed to a wide range of risk. They are also less able to save and that restrict them to deal with income loss due to different crisis. Basic security for the poor can, in this setting, be believed to allow poor people to pursue higher- risk higher-return activities that could pull them out of poverty. We will in this paper study formal social security1 solutions to insecurity due to old-age2, invalidity3 and unemployment4 in developing countries to try to find an answer to the question; can income security enhance growth in developing countries5? The objective of this thesis will be to get aware of some of the microeconomic mechanisms one has to consider when thinking about social security system in a developing country and see if there is empirical evidence for the idea that social security can advance economic growth in the framework of the endogenous growth theory.
The World Development Report6 (2002) put forward the idea that social security could have a positive effect on macro economic variables that affects output and growth. Risk that poor people face includes illness and injury, old age, violence, harvest failure, unemployment and food price fluctuations. According to the report basic security for the poor would probably lay the foundations for investment by poor people that could pull them out of poverty. Poor people are highly risk averse and reluctant to engage in high-risk high-return activities. One downturn could send them deeper in to poverty and they will therefore specialize in low risk, low return activities.
Formal income security programs are today not available to any greater extent in the developing world but in a study by Pal et. al. (2005) they, by using different scenarios on seven sub-Saharan countries, calculate if developing countries could afford introducing basic
1 Social security is accordning to the definition by the International Labour Office (2005, p. 1) “a set of institutions, measures, rights, obligations and transfers whose primary goal is: (a) to gurantee access to health and social services; and (b) to provide income security to help cope with important risk of life (loss of income due to invalidity, old age or unemployment) and prevent or alleviate poverty. “ We choose only to examine the second set of insurance mechanisms to be able to do a more in-dept study.
2 The formal solution to insecurity due to old-age and invalidity is a pension system and “pension” is the term we will use in the continuing of this paper.
3 See previous note.
4 The formal solutions to insecurity due to unemployment are different programs introduced in chapter four such as unemployment insurance, severance pay and public work. The term we use to cover all these programs is
“income support for the unemployed”.
5 We relay on Todaro and Smiths (2004, p. 792) definition that Developing countries are ”characterized by low levels of living, high rates of population growth, low income per capita, and general economiv and technological dependence of developed economies”.
6 The World Bank's annual World Development Report (WDR) is a guide to the economic, social and environmental state of the world today. Each year the WDR provides in depth analysis of a specific aspect of development.
social security (education, health, pensions) and concludes that a basic level of social protection “could be affordable within a reasonable timeframe in the selected countries”7.
Most of the literature on social security examines the effects on economic behaviour that different types of programs generate and presents several effects on people’s incentives.
These effects could have a positive and negative impact on growth.
This paper focuses on the effects of formal income security on the overall effects on economic growth and does not cover the effects on poverty eradication, distribution effects and other if it isn’t important to explain growth effects.
Our main empirical strategy will be to look for effects of the above described possible effects of social insurance on economic growth, in the published empirical findings. The selection of empirical studies included in this paper is made with the ambition to cover different aspects of social security mechanism and should not be seen as a complete summary of the research up to date. As already discussed we will use the new endogenous growth theory as a basic framework to think about economic growth. The basic theory is therefore first presented including its links to social security. Then we present the basic setup of income security followed by the main discussion of its microeconomic implications. This section is mainly based on a review of empirical studies that has been done on the subject. We use Chile as an illustrative example and make a time-series regression on social security effects on growth in Chile. One could question Chile as a suitable example of a developing country but it is the only country which comes close to have enough data available. We also take advantage of the fact that Chile made a major pension reform in 19818 to illustrate graphically the economic effects of one reform in income security. At the end the empirical findings are discussed to see which general conclusions one can make.
2. Theory of economic growth and social security
We first give a short overview of the prevailing neoclassical theory of Economic growth as a starting point for the presentation of the new endogenous growth theory that we use as the theoretical framework in the continuing discussion on social security and economic growth in this paper.
7 The projections are done on Burkina Faso, Cameroon, Ethiopia, Guinea, Kenya, Senegal and Tanzania.
8 In 1981 Chile completely privatised their pension system.
2.1 The Solow growth model
The Solow growth model introduced in 1956 represents the prevailing neoclassical theory of Economic growth (see Dornbush 2000, ch. 18). The Solow growth model identifies which factors that determines growth in output over time and also sheds light on some of the reasons why real GDP per person vary so widely between countries. Solow presented the aggregate production function that links the quantity in aggregate output that can be produced to a given quantity of factor inputs:
Y= A(t) F (K, N) (2.1)
where technological progress (A) is the factor that explaining long term growth and its level is assumed to be determined exogenously and not explained by the model. Real output will increase if there is an increase in the quantity of factor inputs of capital (K) and labour (N).
Figure 2.1 Exogenous technical change
Source: Otani and Villenueva (1990, p. 770)
In a situation when output per worker and capital input per worker are no longer changing a long run equilibrium, called steady-state, occurs. If there is no change in the state of technology, output (Y), capital input (K) and labour input (N) all grow at the same rate. At this state capital input per worker is just the required to offset the labour force growth and depreciation of capital.
Figure 2.1 illustrate such a model. The vertical and horizontal axes measure, respectively, the growth rates of capital and labour, and the capital-labour ratio. The relationship between
Ratio of capital to labour Growth of
capital and labour
K
A B
N K´
C
the rate of capital accumulation and the capital-labour ratio (the K curve) is downward sloping because of diminishing marginal productivity of capital. Labour growth the (the N curve) is horizontal because by assumption it is independent of the capital-labour ratio.
The Solow growth model predicts that if the nation’s saving rate increase the country will have higher steady-state levels of capital input per worker and therefore higher levels of output per worker. However, the model predicts that, while an increase in the saving rate will lead to a temporary period of faster growth (from A to B in figure 2.1), it will not affect the long-run equilibrium or steady-state growth rate (adjusts to C along K´ in figure 2.1). Growth cannot be sustained over time because the labour input ultimately becomes a bottleneck in the production process. Once the new steady-state has been reached the rate of capital formation and growth rate in output will have returned to their initial levels equal to the rate of growth of the labour force.
In the Solow model, sustained growth of output per worker can only be explained by technological progress. The model predicts that in the steady-state both capital input per worker and output per worker grow at the rate of technological progress.
The problem of the Solow growth model is that technological change is exogenous and the model provides no insight into how government policy (for example social security programs) could raise the long-run equilibrium growth rate of output. This model has therefore been contested.
2.2 The new endogenous growth model
The new endogenous growth theory first developed by Romer in 1986 and Lucas in 1988 provided a theoretical framework that links a countries institutional setting to long-run economic growth (See Todaro and Smith 2003, p. 146-150).
The new endogenous growth model have extended the Solow growth model by making the rate of technological change endogenous, the model explains the factors behind the changes in long-run growth rates. In addition, they have abandoned the assumption that all nations have the same access to technological opportunities. In the endogenous growth model human capital is the key engine that leads to improved technology by its effect on research and development. Thus, the rate of growth is partly explained by the stock of human capital.
In these models the level of output per worker depends on both the amount of physical capital input per worker and the human capital input per worker. The endogenous growth model has helped to explain differences in living standards among nations and given insights
into the long- term growth rates that can be enhanced by, for example, education, training, capital formation and research & development.
Figure 2.2 Endogenous technical change
Source: Otani and Villenueva (1990, p. 770)
Otani and Villenueva (1990, p. 770) illustrate the workings of such models, consider Figure 2.2. The relationship between the rate of capital accumulation and the capital- labour ratio (the K curve) is downward sloping because of diminishing marginal productivity of capital. Labour growth the (N curve) is drawn as an increasing function of the capital-labour ratio (in the Solow growth model this line is horizontal), reflecting the fact that society devotes a positive fraction of its per capita income to improving the quality of labour. This improvement in human capital is related to education.
Now suppose that the domestic savings rate rises for some reason. The higher savings rate shifts the K curve upward to K´. The growth rate of capital stock temporary shifts to point B.
Labour input becomes a bottleneck of production and the investment rate will be slowed until the growth of the capital stock is lowered to the level of growth of the labour force at point C. But unlike the Solow growth model where growth is assumed to be independent of the capital-labour ratio this new equilibrium leads to a higher growth rate of output. This is because the increase in per capita output is partly used to development of human resources and the growth rate of the effective labour force increases. The improvement in labour productivity also provides an incentive to raise the share of resources spent on human capital.
Barro (1998, p. 6) concludes that long-term growth rates then depend on institutional settings in this theoretical framework. Empirical research based on the endogenous growth
Ratio of capital to labour Growth of
capital and labour
K
K´ N
C A
B
theory that has been done by cross country comparison supports the model and give insight in which variables that are important explaining growth. The basic set-up of the regressions are:
Growth = β0 + β1X1+β2X2 + …. + βkXk (2.2)
where X1 to Xk are the possible determinants of growth. Barro (1996, p. 13) extended the first work by Mankiw. Mankiw had showed that initial GDP per capita and school enrolment is important to explain growth. In addition to Mankiws finding Barro showed that government expenditure, political stability and inflation affects growth. These results seem stable across several studies. Another study by Lee and Gordon (2004) shows that in addition corporate tax rates has an affect on growth. Studies specifically on developing countries are also available.
In a study on Latin America Loayza, Fahnzyblber and Calderón (2005, p. 51-57) show that economic growth increases with structural policies improvements in education, financial depth, trade openness, and public infrastructure. It decreases when government apply excessive burden on the private sector. Economic growth also decreases when the government does not carry out policies to achieve macroeconomic stability. Otani and Villanueva (1990, p. 771-778) focus their study on developing countries in general and conclude that export performance, expenditure on human capital and the real interest rate on external debt explain the growth performance of these countries well.
According to Todaro and Smith (2003, p. 150) the new growth theory still has problems that can make it inappropriate for developing countries. They claim that the impact of various factors on short- and medium- term growth has been ignored because the new growth theory overemphasis the determinants of long- term growth rates. For example they argue that poor incentive structures may be as responsible for slow GNP growth as low rates of savings and human capital accumulation. But even though this model isn’t perfect explaining economic growth we believe it is clear enough to give a good framework for a discussion on the impact of social security on economic growth in developing countries.
The important theoretical framework we base this thesis on is the new endogenous growth theory presented above, and we mainly treat social security as a possible institutional framework among others in the new growth theory. The theoretical predictions of social security on growth are conflicting. The International Labour Office (2005), the international organization dealing with social security issues, summarizes some of the possible mechanism in which growth can be effected by social security and we present this summary in table 2.1.
Table 2.1 Possible growth effects of social security
Possible positive effects Possible negative effects
If social security is financed out of general revenues, this will affect the government budget balance, increase interest rates and, therefore, by crowding out effects, reduces public and private investment.
Savings and Investment
National pension’s savings can become a major supply in financial markets and may thus, play an important role in economic growth policies.
If administration costs are high, opportunity costs occur. This economic waste could be used for investments in capital.
Labour productivity may be positively affected if the social security system is financed through taxing labour.
This argument is linked to the theory that when people have unemployment insurance they can wait for the more suitable job, where they will be more productive.
Institutions providing income protection in case of unemployment or in the form of early retirement schemes trigger withdrawals of potentially productive employment from the labour force.
When social security is present it functions as an overall reduction of risk due to loss of income. This leads to less incentive for raising the present income level in order to able to save money if you are unemployed tomorrow.
This can effect employment and inflation.
If social security is financed through taxing labour, labour markets are negatively effected. This is due to the theory that work incentives are decreased and job creation decline when taxes are imposed.
Labour force participation
Social security themselves generate substantial employment.
The social security system are links through which positive income effects of globalization can be allocated to the vulnerable persons in societies; through this mechanism they contribute to acceptance of globalization and reinforce global pro-growth policies.
Structural change
Unemployment insurance may help to facilitate labour force adjustments to structural changes of economies.
Social security reduces existential insecurity;
counterbalance the possible need to turn to illegal means to generate income. This can increase long-term investment because of the social stability it creates.
Economic Stability Social security that provide income to unemployed, the
disabled, the elderly and other social groups stabilizes consumption during recessions and therefore support firms sales.
Source: Authors´ own table construction based on the International Labour Office (2005) summary of the possible mechanism in which growth can be effected by social security.
All the arguments in table 2.1 are used in the debate about social security and its impact on economic growth. The objective of this paper as already stated will be to explore the empirical evidence of how social security effect economic growth in developing countries.
This is researched more in depth in section four. Now we turn to describe the relevant insurance systems.
3. Description of income security systems
Income security can be achieved by both formal and informal mechanisms. In developing countries the main strategy to cope with income risk are different informal solutions, for example relying on relatives for support, described in section 3.3. This paper focus on formal insurance which we describe in sections 3.1 and 3.2. Today these types of programs are not generally available. The International Labour Office (2005, p. 1) definition of Social security is “a set of institutions, measures, rights, obligations and transfers whose primary goal is: (a) to guarantee access to health and social services; and (b) to provide income security to help cope with important risk of life (loss of income due to invalidity, old age or unemployment) and prevent or alleviate poverty. “ We choose only to examine the second set of insurance mechanisms to be able to do a more in-dept study.
The most widespread formal income support program is government work programs available in 80% of the low income countries9 but the participation rate is only 0,8 percent.
(see Vodopivec 2004, p. 59). The lowest coverage of 5-10 percent of the labour force exists in Sub-Saharan Africa and South Asia in contrast to Latin America where the coverage ranges from 10 to 80 percent and in Southeast and East Asia with coverage from 10 to 100 percent.
One out of six low-income countries offers no public support programs at all. Formal pension is also limit in scope in the developing world covering only 16 percent of the labour force. In the poorest countries of Sub-Saharan Africa the coverage is less than 10 percent of the labour force. This lack of coverage is because to be able to setup such programs the country must be capable to allocate resources to finance the programs. In an interesting study by Pal at. al.
(2005) they discard the notion that social protection is unaffordable for developing countries.
They look at a wide basic package of social security including universal old-age and
9 The World Bank uses a classification system with low income, middle income and high income countries. This paper is based on the term “developing countries” but to be able to present statistics accurately we use the World Bank classification, were needed, and regard low income countries and middle income countries as developing countries although it may be questionable that some middle income countries should be regarded as developing countries.
invalidity pension under different scenarios and conclude that they “seem within reasonable and affordable limits”. Difficulties in achieving funding of the social security system explain the lack of programs in developing countries but the above study tells us that social security is possible in these countries as well. Understanding the possible different setups of income security systems will be fundamental to discuss general growth effects of the systems and we now describe these.
3.1 Formal Income support for the unemployed
In comparison with developed countries the concept of unemployment is very different in developing countries (see Vodopivec, p. 7). Unemployment is a social construct that emerges only in industrial societies. This is because the development of labour markets is essential for unemployment to emerge. It is in an urban industrial society that workers either work or don’t work; one could say that employment become discretionary. In this setting a worker cannot resort to self- or home-production when they become unemployed.
The supply of workers far exceeds the demand in many developing countries with extensive migration into the cities leading extremely high rates of unemployment as for example South-Africa with an urban open unemployment rate of 33% 1993-1998 (see Todaro 2003, p. 333).
The objective of income support is from the perspective of an individual to compensate for income loss due to unemployment (see Vodopivec, p. 7). The reason why the public has to be involved in these programs and private initiative isn’t enough is that informal insurance mechanism may often be ineffective because of the loss of employment is too large a shock and to frequent and thereby informal insurance tends to be least effective when most needed.
There are also strong information asymmetry that prevents private formal insurance solutions (ibid, p. 17-19). Unemployment insurance reduces self-protection (moral hazard) and information problems make it hard charging high risk individuals higher premiums (adverse selection). Vodopivec (2004, ch. 3) gives example of five different kinds of formal programs;
unemployment insurance, unemployment assistance, unemployment insurance savings accounts, severance pay and public work. Developing countries primary use severance pay and work programs but some developing countries have also introduced a larger set of programs.
Unemployment insurance: The benefits in an unemployment insurance program require that employees and/or the employer make contributions. It is different from ordinary
insurance principals because the premium doesn’t reflect individual risk. In many developed countries the system is mandatory across all formal sectors of the economy but the existing systems in developing countries it mostly covers the industrial- and commerce sector. Usually it is required that the worker is capable of, available for and looking for work. Benefits are most often calculated as a percentage part of average earnings previous to unemployment (with a ceiling) but in some transitional countries they use a flat rate. The benefits are limited in duration and decline over time.
Unemployment assistance: Unemployment assistance is a means-tested income support for working-age individuals. The beneficiaries have to be unemployed and without the means to maintain a minimum standard of living. In most countries the assistance program work in tandem with unemployment insurance and sets in when the insurance program expirers.
Benefits can be in-kind but are usually in cash. The program is usually funded by the government’s general tax revenues.
Unemployment insurance savings accounts (UISA): In a UISA system an individual savings account is set up for each employee and the employer makes deposits regularly.
When workers become unemployed they can make withdrawals from their individual account as they see fit. Some systems have a rule that the accounts only can be accessed upon involuntary unemployment but most does not have such a requirement. At retirement the account balance are added to the old-age pension.
Severance Pay: Severance pay is a lump-sum payment made to discharged workers. It can be made voluntary by the employers or be restricted by the government. It is typically provided to redundant workers. Severance pay is usually financed by employers.
Governments can also provide financial assistance, especially when a large-scale restructuring operation is initiated.
Public Works: Public work provides low-wage employment opportunities. It has generally been used to encounter economic and natural shocks. The concept is basically that people must work to obtain benefits. The programs can for example be construction and maintenance of roads and irrigation. These programs are design to be more labour intensive than commercial projects. Program wages are set to attract only the poor and are kept around the minimum wage. People should thereby be encouraged to seek employment outside the program. Also these programs are financed trough general taxation. Social funds that finance small-scale community projects have also been introduced as an alternative.
3.2 Pension
Rosen (2005, p.191-198) gives the basic components of how pensions system to cope with old-age security works. In the basic pension system members of the system makes contributions via tax on payrolls during their working lives. On retirement members are eligible for benefits.
When individuals during their working lives deposit some portion of their salaries into a fund and on retirement receive the principal and accrued interest, the system of old age social security system is called fully funded. Workers' and employers' compulsory contributions are earmarked for each individual's account. The workers than invest the funds in various financial assets, particularly mutual funds. At the end of their working lives, individuals finance their retirements out of the accumulations in their accounts.
A different structure is what is called pay-as-you-go. This mean that benefits paid to current retirees come from payment made by those who are currently working. Each generation of retirees is supported by payments made by the current generation of workers, not by drawing down an accumulated fund.
Many countries first introduced a fully funded system but soon thereafter transformed it to a pay-as-you-go system. The reason for this switch can be as in the case of the United States that the savings of many elderly had been wiped out by a depression, and they needed higher levels of support than what just a few years of contributions could give. Another more general reason for a switch has been the fear that some politicians would manage the fund inefficiently.
An individual's pension benefits depend on the earnings history, age and other personal circumstances. In general the first stage in calculating the levels of benefits received is to calculate the average indexed monthly earnings (AIME). This figure represents the individual's average wages in covered employments over the length of his or her working lives. Only annual wages up to a given ceiling are included in the calculation. The next step is to substitute the AIME into a benefit formula to find the individual's primary insurance amount (PIA). This figure is the basic benefit payable to a worker who retires at the normal retirement age or who becomes disabled. The benefits formula is often constructed in a way so that the PIA increases with the AIME but at a slower rate. The actual benefit paid out depends also on two other factors.
The age at which an individual qualifies for a full pension benefits is usually called the normal retirement age. In general workers who don't start collecting benefits until some time after their normal retirement age receive a permanent increase in their benefits.
When the worker is fully insured, single and retires at the normal retirement age the actual monthly benefit is simply equal to the primary insurance amount. A worker with a dependent family can receive more.
A pay-as-you-go system is generally financed through the payroll tax as a flat percentage of an employee's annual gross wages up to a certain amount. Usually the employee and the employee shift the tax burden so whether the true economic cost is actually split in half is a complicated question of tax shifting.
Many countries do now have different kinds of multipillar-systems that mix the setup of both pay-as-you-go and a fully funded system. The process of turning a pay-as-you-go system into a fully funded is called privatization and has taken place across the world during the last decades. To be able to manage a multipillar-system well considerable administrative capacity is needed (see Gill, Packard and Yermo 2003).
3.3 Informal insurance mechanisms
In contrast to the formal social security systems that dominate in the developed world the developing world today mostly relay on informal insurance mechanism. To discuss the implementation of a formal income security system it is important to know what the prevailing alternative is. The World Development Report (2002) gives an overview of risk managing mechanisms which we base this section on. Poor people are highly risk averse because of the large consequences when a crisis strikes. Risks can occur on different levels Micro shock affects the individual or a household, meso shocks affects a group or a whole community and macro shocks are on the national or international level. In light of this one could conclude that people are not insured well with village based insurance when a shock hits a whole village and there are no informal solutions to target macroeconomic shocks.
The mechanisms in use are: First, mechanisms for reducing risk such as migration, getting more secure income, collective action by the community such as building ditches and terraces and common property management. Second, mechanism mitigating risks by income source diversification, marriage and extended family, collecting buffer stocks and at group level
participating in occupational associations and investing is social capital10. Third, mechanisms for coping with shocks such as sale of assets, loans from money lenders, child labour, reduced consumption, seasonal or temporary migration and transfers from support network.
Problems with the informal solutions can be that income options for farmers typically move together during crises. With limited effective diversification possibilities poor people will specialize in low risk, low return activities.
4. Effects of income security on Economic behaviour
In this section we look empirically at the growth effects of income security. There is not enough data11 available to study overall social security effect on economic growth in the cross-country empirical research tradition of Mankiw and Barro presented in the end of the theory chapter.
To say something about income security in this growth framework we will look separately at each channel through which income security may affect growth and see what empirical finding there is on such partial effects. We study how economic behaviour is changed because of different social security systems. One must remember to consider links of different components of social security to draw any conclusions of the overall effects. The complexity of social security systems and laws can make this rather difficult and we can at this point only be speculative about this wherefore we leave this for the final conclusions.
To get a clue of the overall effects in one specific country where data are available we took Chile as an example and made an econometric study on total social security benefits (not exclusively income security because the specific data was not accessible) effects on growth based on previous econometric studies as presented in chapter 2. The study presented in more detail in appendix A finds that social security spending has a statistically significant negative impact on growth the following year. The regression shows that an increase in the resources allocated to social security with 1% of GDP lowers GDP growth by 0,77 percentage points the following year. The study however have several problems (see appendix A) and should not be taken as solid evidence.
The existing empirical research is mostly done on developed countries which are most experienced with social security programs. But it is important to look at country specific
10 A person can invest in social capital by for example participating in networks, associations, rituals and mutual gift giving. The person involved later relay on social connections for support in harsher times.
11 ILO gathers statistics on social security expenditure in the “Cost of Social Security” database since 1949 but the datasets are not complete, different county statistics are missing each year.
settings when trying to say something about the validity of the empirical results for developing countries and one should be aware of this when reading this section.
We will take one reform in one country as an example. This is done to see the effects of one single reform in one part of the income security system. Chile has had a long history of social security and we use Chile’s 1981 reform when they completely privatized their pension system as a graphical example. We present and discuss the data from Chile along with the empirical presentation. In appendix B we give a brief presentation of the economic history of Chile that we relay on as background information in this chapter. These examples are merely examples and should be regarded as such.
4.1 Effects of income support for unemployed
We have identified five different channels through which growth is affected by income support for the unemployed. This is through effects on human capital accumulation, saving, structural transformation, labour force participation and economic stability.
4.1.1 Human capital accumulation
In the endogenous growth theory human capital is the most important determinant of long-run growth. Vodopivec (2004, p. 81-104) predicts that income support for unemployed can have a positive effect on human capital accumulation in developing countries because children do not have to drop out of school and start working when parents become unemployed. Income shocks can reduce consumption below a level needed to keep up productivity; such dynamic poverty traps can be prevented by income security. Vodopivec suggests that the program can target these vulnerable family members by for example providing school subsidies as part of the unemployment benefits.
Esguerra el. al. (2002) find empirical evidence in a household income and expenditure surveys in the Philippines that the poor absorb a great portion of income shock through reduced consumption. They reduced food intake and took children out of school.
4.1.2 Saving behaviour
Savings generate growth through investment in the new endogenous growth theory. Engen and Gruber (cited in Vodopivec 2004, ch. 4) has studied how much unemployment insurance generosity affects saving of the employed. They find that unemployment insurance crowd out savings by 2,8% for each 10 percent increase in unemployment insurance generosity. For a
more comprehensive discussion on income security affects on savings se the section on pension in this paper.
But the question is what kind of precautionary saving is done in developing countries before the introduction of a formal insurance system. Rosenzweigh and Wolpin (1993) uses longitudinal household data on farm profits in India and studies investment decisions by Indian farm households. They examine the investments in an important production factor of the area, the bullock12. Sales of bullocks increase significantly when the weather becomes bad and if the weather becomes better the purchase of bullocks increase. They conclude that the farmers are quit averse to risk and the insurance mechanism isn’t optimal. Increased opportunity to income support in bad conditions would have a substantial effect on production efficiency. Farmers could then choose to invest with a focus on increasing production instead of using the investment primary as insurance and not be forced to sell the productive assets when a crisis strikes.
4.1.3 Structural transformation
In the growth theory institutional setting affecting structural transformation is important to explain growth. Vodopivec (2004, ch. 4) concludes that more productive jobs and industries are often riskier and insurance can stimulate a transfer to such sectors. Costly production choices such as using outdated but less risky production technologies can also be reduced by insurance. Unemployment insurance can thus encourage labour reallocation and in particular restructuring of enterprises. One theory is that social security may help facilitate a downsizing of enterprises when needed.
Acemogulo and Shimer (1999) show in a study on the US that states with higher replacement ratio13 shows a relative increase in the number of high-wage occupations. They conclude that in an economy with risk-averse workers a positive level of unemployment insurance can lead to higher output. This is in contrast to the fictive economy with risk neutral workers that will have maximal output without unemployment insurance
Studies of the US also show a reverse effect that unemployment insurance is most widely used in the sectors with seasonal jobs so unemployment insurance helps workers remain in sectors that aren’t productive all year round (see Vodopivec 2004, p. 91).
12 The Bullock is used as draught animal.
13 The replacement rate is how large benefits the unemployed receives as a percentage of his salary previous to unemployment.
4.1.4 Labour force participation
Because of the many incentives that social security has on labour force participation it’s an important aspect to regard. The security system must be financed and if there is a public system this is achieved by taxation. Economic theory says that total labour taxation has a negative effect on growth. A study of this aspect is done by Nickel and Layard on the US (cited in Vodopivec 2004, p. 92) and they conclude that a 10 percentage point rise in the tax wedge will reduce overall labour force participation by around 2 per cent of the population of working age. The tax wedge is the sum of the payroll, income and consumption tax rates.
Thus, only a small fraction of differences across countries can be explained by taxes. High effective marginal tax can also leads to decision not to take on employment offers.
Vodopivec (2004, p. 100) put forward analysis of work disincentives of social assistance programs. Those studies finds that high effective marginal tax rates, that can be the result of phasing out the benefits and lead to poverty traps that discourage recipients from working.
Studies also claim that higher labour taxes lead to higher real wages. This would lead to slower growth when firms choose to substitute labour for more capital.
Mortensson (cited in Vodopivec 2004 p. 92) showed in a study from 1994 that higher replacement rate of the insurance reduces job creation and through that channel aggregate output. The spouse’s decision to work can also be effected by marginal effects for the household in total.
UISA are not yet used to any wider extent but Colombia replaced severance payments system with a system of severance payments accounts14 in 1990. Kugler (2004) has studied this transformation. Severance pay does not create a moral hazard problem by lowering job search efforts, but it does affect incentives to enter unemployment and hence creates another moral hazard problem Severance payments increase costs for the employer and that can distort hiring decisions. The transformation of this system can shift cost of severance pay from employers to workers and thus the moral hazard problem is eliminated. The empirical results from Colombia shows that the introduction of the new system lowered wages by between 60% and 80% of total severance payments contributions by the worker. This shift should according to Kugler have reduced costs employers as well as reduced distortions in hiring and firing decision by the companies.
14 Severance payments account has the same financing setup as UISA but instead of recurring benefit payouts there is a lump-sum payment made to the discharged worker. See section 3.1 for description of severance pay and UISA.
Fallon and Lucas (cited in Vodopivec 2004, p. 98) show in a study on India and Zimbabwe that a stronger job security regulation lowers employment drastically. Severance pay also, according to empirical findings, increase part-time employment and self- employment and reduces labour market flows.
Administrative capacity is important. Effective monitoring and use of sanctions increase the transition rate to employment. To achieve this, the administrative body must be able to process information effective enough. Vodopivec (2004) suggests that the modern technology make this information processing affordable even for low income countries. Even low-income countries have pension system that requires this type of information processing. The problem is that the program has to check if the unemployed are willing to accept employment and searching active for employment. Monitoring this kind of eligibility is problematic even for developed countries. The Moral hazard problem is thus worse in developing countries and indicates that disincentives to participate in the labour force could be even worse in developing countries. Poor people in developing countries are almost always underemployed instead of unemployed because they cannot afford being unemployed. These people may choose unemployment instead if unemployment benefits were available and therefore lower labour force participation. Avoiding that benefits go to the non poor may require large resources. One example is drawn from a study on the Philippines by Euguerra et. al. (2002) where they found that the screening cost for an applicant to tuition fee subsides for the university is 10 dollars, equivalent to total per capita social spending. Less administrative capacity is needed to manage a public work program than other types of income support programs.
4.1.5 Economic stability
The growth theory has shown that economic stability can enhance growth. Cyclical pattern of growth can be soften by the benefits as stabilizer of the economy). In a recession with high unemployment the unemployment-benefits smoothens consumption and helps slow down the down turn, in an expansion payroll taxes dampens the economy. Empirical research has found that income support reduces GDP losses during a downturn in the economy by 10-15 percent (Vodopivec 2004, p. 91).
4.2 Effects of pension
An economic argument for introducing Pension systems in developing countries is that it will influence people childbearing decisions. We have also identified that Pension systems can distort people's behaviour mainly through the channels of saving behaviour and labour force participations decisions. This section presents empirical findings on this topic.
4.2.1 Population growth
Zangh and Zangh (1995) have studied pension’s systems effect on growth rates.
Developing countries struggles with over-population and have been trying to reduce population growth. With a to high population growth parents are unable to invest enough in each children’s education and health and it is not possible to keep capital accumulation to increase in the same speed and therefore capital per person ratio decline and thus growth.
According to Zangh and Zangh introduction of a pension system reduces fertility. Pension is seen as an alternate investment instead of children to secure income during old age. They imply therefore that the introduction of a pension system leads to higher output growth. But their model ignores several many real world elements so the conclusions should only be seen as suggestive and not final.
4.2.2 Saving behaviour
The level of saving is important because it affects the level of investments in the economy and thereby generates growth. Rosen (2005, p. 203) put up the theoretical framework to think about saving behaviour. The life-cycle theories15 of savings say that individuals' consumption and saving decisions are based on lifetime considerations. According to this theory individuals save some portion of their income while working in order to accumulate wealth from which they finance consumption during retirement. Such funds are invested until they are needed, therefore increasing society's capital stock. The introduction of a social security system can very much change the level of lifetime savings. This can happen through three channels. First, Wealth substitution effect is when workers realize that their contributions to the social security system will pay back as guaranteed retirement later on, they view the social security taxes as a mean of “saving” for these future benefits and they will save less of their own. With a pay as you go system, the contributions are not actually saved and increase the
15 For a deeper presentation of the life cycle hypothesis see for example Rosen (2005). To understand the reasoning of this paper it’s enough to know that it tells us that individual’s consumption and saving decisions are based on lifetime considerations.
society capital stock; instead the contributions are paid out to beneficiaries directly. In summery this leads to a reduction in the total amount of capital accumulation. In a fully funded system the same distortions on over-all savings of the country do not occur. Second, retirement effect is that pension may induce people to retire earlier than they otherwise would have, but to receive benefits they have to leave the labour force. When more years of retirement has to be financed with fewer years of work the savings rate must increase. The effect is an increase in the saving rate. Third, bequest effect occurs if the system give insensitive to people to want to leave inheritances for their children. Parents may save more too offset the distributional effect that a pension system can have distributing income from the working generation to the older generation.
In order to analyze the workings of any pension system one must recognize the fundamental fact that in any year in the future, the consumption of both retirees and workers must come out of that year's production. A structural reform, as for example privatization of the pension system, can help finance future retirees' consumption only to the extent that it allows future output to increase. And the only way it can do this is by increasing savings, because a larger capital stock probably increases the productivity of future workers.
The best- known privatized pension system is Chile's. The main attraction of a private pension system is that over the long run, stocks have earned much higher returns than the implicit rate of return that the pension pays on individual' contributions into the system. The most important argument in favour of the private system goes; if social security taxes were invested in the private market, there higher returns would allow retirees to enjoy large benefits without imposing huge taxes on the current workforce. This argument does not hold, consumption of both retirees and workers must come out of that year's production.
Privatization can help finance future retirees' consumption only to the extent that it allows future output to increase by increasing savings.
Because the three effects work in different directions theory alone can not tell how social security's pensions insurance effect servings and indirectly investments.
In a study made by Feldstain (1996) using annual US data from 1930 to 1992, he estimated using regression analysis that the social security wealth variable was 0.028 and statistically significant. Social security wealth is the expected present value of the benefits to which an individual is entitled. This positive sign suggests that increases in social security wealth increase consumption and, therefore, decrease savings. In other words the wealth substitution effect dominates the retirement and bequest effect. To assess the quantitative importance of the coefficient, the value of social security wealth in 2002 in the USA was
$16,000 billion. A coefficient of 0.028 implies that social security reduced personal savings in 2002 by $448 billion (= 0.028 times $16,000). In comparison, during 2002 personal savings was about $ 296 billion. The $ 448 billion is then 60 % of the potential personal savings of $ 744 billion (the sum of $448 billion and $ 296 billion) If all these calculation and estimates are correct, the cost of the pay-as-you-go system in the USA is major and it has had a huge negative impact on capital accumulation in the United States.
Several studies have come up with rather different result and point out that Feldstain’s analysis does not include all relevant variables, measurement errors, or inappropriate definitions of social security wealth. Some also finds little support for effects of the size that Feldstain reported in his 1974 study.
Laimer and Lesnoy (1982) found evidence that pension might even have increased saving.
Replacing a pay-as-you-go pension system with a fully funded scheme could eliminate the incentives to in formalize production and employment
Figure 5.1 Gross Domestic Savings as percentage of GDP in Chile
0 5 10 15 20 25 30 35
1960 1962
1964 1966
1968 1970
1972 1974
1976 1978
1980 1982
1984 1986
1988 1990
1992 1994
1996 1998
2000 2002 Year
Procent of GDP
Year of Pension reform
Source: Graphical construction by authors based on data from World Development Indicators.
In figure 5.1 we see that the total savings rate continue to drop two years after the reform in Chile but then there is a heavy increase from 1983 to the 90s where it seems to stabilize at this higher level.
Theory tells us that there is a substitution effect that makes people lower their personal savings when a pension system is introduced. With a pay as you go system this makes total savings of the economy decline. When going to a fully funded system also the tax
contributions of the employee is really saved and total savings of the economy should increase. Our graphical analyse above support that this effect dominate. However, this graphical analysis does not control for other changes that can have an effect on savings. For example we know that in 1984 Chile, in response to the decline in savings, made a tax reform aimed at increase insensitive for savings in order to boost output growth and this reform should have worked in tandem with the effects of the pension reform to increase savings.
The lag for two years before we se an increase can be expected because it takes time for any structural change to show up in economic statistics. A change in law is not the same as an immediate change in economic behaviour. These facts is consistent with theory and one is temped to conclude that the pension reform has been the reason for the positive change in savings.
In an attempt to quantify the effect of structural pension reform on growth in Chile, Corbo and Schmidt-Hebbel (2003) controlling for other reforms, used time-series regressions for the period 1981-2001 to estimate separately the impact of pension reform on the capital stock, labour supply, and the total factor productivity. They then substituted these estimates in a Cobb-Douglas production function to estimate the overall impact of pension reform on economic growth. They found that the reform increased saving by 0.7 % to 4.6% of GDP per year. Consequently, the investment rate increased by the equivalent of 1.2% of GDP during the same period.
4.2.3 Labour force participation
For people older than the normal retirement age the pension system provides incentives for partial or complete retirement. Many scholars believe according that social security has played a key role in the change in retirement pattern (Rosen 2005, ch. 9). The retirement incentives associated with social security is linked to the social security wealth. In addition this depends on several considerations, but the most important is the adjustment made to future benefits to compensate the individual for the fact that that person hasn't drawn upon the system that extra year. For studies made in the US it turns out, however, that once a worker
reaches the normal retirement age (64 years old), the adjustment is always worse than actuarially fair, which means that social security wealth falls if the individual works another year. According Diamond and Gruber’s (1999, p. 456) calculations the ratio of the fall in the value of social security wealth to net earnings can be quite high, about 23 percent for a 66 year old.
Several econometric studies have been done in order measure the impact of pension on retirement decisions. Many are consistent with the hypothesis that the system increases the likelihood of retirement and reduces the amount of labour supplied by those who continue to work. In a study of the social security systems in eleven industrialized countries, Gruber and Wise (1999) found that the age at which benefits are first available has an important effect on the likelihood of retirement. However, there is considerable uncertainty about the magnitude of the response. One important reason for the uncertainty is the fact that many of the variables influencing labour supply decisions of the aged are difficult to measure and sometimes unavailable altogether. These include health status, local labour market conditions, and the amount of wealth accumulated in private pensions.
Figure 5.2 Labour force participation in Chile
0,52 0,54 0,56 0,58 0,6 0,62 0,64 0,66
1960 1962
1964 1966
1968 197
0 1972
1974 1976
197 8
1980 1982
1984 1986
1988 1990
199 2
1994 1996
1998 200
0 2002
2004 Year
Labour force / Population (15-64 years old)
Year of Pension reform
Source: Authors calculation based on data from the Human Development Indicators16
16 We have calculated Labour force participation as follows:
Y = (Labour force (%) * Total population) / Population 15-64 years old
Figure 5.2 presents a clear upswing in labour force participation from 1981 when the pension reform was introduced in Chile. The increase over a 10 year period is 4 percentage points. Because only formal sector (taxed and regulated) employment contributes to a person’s retirement fund the incentive to seek formal sector employment increase with the fully-funded system. The direct link between contributions and future benefits is stronger under a fully funded system than under a pay-as-you-go system were benefits also depend on age and other personal circumstances for example family status. The graphical analyse could indicate such a formalization of the labour markets and an increase in incentive to want to participate in the labour force has been taking place. As with the analysis of savings we can be far from certain that this increase isn’t due to other parts of Chilean economic reforms and the general positive development of the Chilean economy.
A study already discussed, Corbo and Schmidt-Hebbel (2003) concluded that the reform significantly reduced the pure tax on employment, accounted for by the difference between social security contributions and expected future benefits, which was very large in the old pay-as-you-go system. This reduction in the implicit labour tax caused total employment in the Chilean economy to grow between 1.3% and 3.7%. Formal employment increased even more, between 3.2% and 7.6%, while informal employment contracted. This change in the composition of employment reflected in a rise in the average labour productivity in Chile.
4.2.4 Conclusions of our example from Chile.
To summarize our Chilean example we now look at the growth development in Chile.
Graph 5.3 GDP Growth (%) in Chile
-15 -10 -5 0 5 10 15
1961 1963
1965 1967
1969 1971
1973 1975
1977 1979
1981 1983
1985 1987
1989 1991
1993 1995
1997 1999
2001 2003
Year
GDP Growth (%)
Year of Pension reform
Source: Graphical construction by authors based on data from World Development Indicators.
According to the graphical presentation above the GDP growth dropped by 15% after the pension reform was introduced. When knowing from the history presentation (see Appendix B) that Chile went through its worst economic crisis since the 1930s in 1982-83 it is not reasonable to derive this dramatic drop in growth rates from the reform. This major depression was partly a result of several external shocks and several domestic policy mishandlings. Moreover, as already discussed, because of the lag factor in most system changes the pension reform is expected to take some time to influence macroeconomic variables such as GDP growth. One can therefore also argue that the increase around 1883 may be consistent with this line of argument. Per capita GDP growth has risen significantly since the mid 1980s, exceeding 5% per year. This high growth rate can surely partly be explained by higher factor productivity. Chile’s radical pension reform may be contributing to less distorted factor markets and therefore, to higher growth.
Both the gradual elimination of the pure tax component of pay-as-you-go and the deepening of financial markets resulting from pension reform could have a significant influence on growth. Corbo and Schmidt-Hebbel found that pension reform’s impact on savings and investments, labour markets, and total factor productivity led to average annual economic growth of 0.49 percent, or almost one-tenth of Chile’s average annualized growth of 4.63 percent over the period 1981-2001. Corbo and Schmidt-Hebbel also argue that the mandatory social security savings channelled to the pension funds, contributed significantly to deepen the financial maturity of the Chilean economy.
6. Conclusions
Theory alone does not support a definite answer to the question can income security increase growth? Our empirical presentation only gives a partial view of the existing empirical findings. Different studies we have looked at give conflicting answers to this papers question. A general comment about the empirical work we have studied is that income security system can have a large effect on people’s behaviour but these effects can both enhance and impair economic efficiency, and the overall effect on growth is not clear. The problems with the empirical work are that they only study changes in different parts on the income security system separately. Another large drawback is that most studies on income security effects are carried out on data from the developed countries. Most of the developing countries that have income security systems only recently put them in place and more time is needed before the effects can be further examined.
We can however, conclude that the environment in a developing country without income security is not optimal for good economic decisions by the poor. But the empirical evidence presented in this paper makes us sceptical that formal income security is the ultimate solution to enhance growth. The negative effect of formal income security seems to be quite large if the program setup is not well balanced. After considering the thesis findings it is tempting to conclude that a very basic income support program, limited in scope to target the poorest with minimal negative insensitive effects may boost growth the most.
The main purpose with our own regression study was to gain insight about the total growth effect. The social security variable in the regression on growth was -0,77. This mean that an increase in the benefits as a percentage of GDP with one unit lower the GDP growth the following year with 0,77 percentage points.
Though the model and data used clearly has some shortcomings we strongly believe that if there where only more data available to do a cross-country study, the econometric strategy we present could lead to a more definite answer to the question of this paper.
7. References
Acemogulo, D. and Shimer, R. (1999) Efficient Unemployment Insurance; The Journal of Political Economy, vol. 107, no. 5, pp. 893-928.
Barro, R. (1998); Determinants of Economic Growth: A cross-country empirical study;
Cambridge; MIT Press.
Diamond, Peter, and Jonathan Gruber (1999); Social Security and Retirement in the United States; Chicago; University of Chicago Press, pp. 437-73.
Dornbush, R. Fischer, S. Kearney, C. (2000): Macroeconomics; mcGraw-Hill
Esguerra, J. Ogawa, M. Vopopivec, M. (2002); Options of Public Income Support for the Unemployed in the Philippines; Washington; The World Bank.
Exportrådet (2000); Marknadsguide Chile; Stockholm; Exportrådet.
Feldstain, M. (1996); Social Security and Saving: New Time Series Evidence; National Tax Journal, vol. 49, pp 151-64.
Gill, I. Packard, T. and Yermo, J. (2004): Keeping the promise of Social Security in Latin America: Washington; The World Bank.
Gruber, J. and Wise, D. (1999); Social Security and Retirement around the world, eds Introduction and Summary; Chicago; University of Chicago Press.
International Labour Office (2005); Social protection as a productive factor; Geneva, International Labour Office.
Karuna, P. Behrendt, C. Léger, F. Cichon, M. and Hagemejer, K. (2005); Can low income countries afford basic social protection? First results of a modelling exercise; Geneva;
International Labour Office.
Kugler, A. (2005); Wage-shifting effects of severance payments savings accounts in Colombia; Journal of Public Economics, vol. 89
Lee, Y. and Gordon, R. (2004); Tax structure and economic growth; Journal of Public Economics, vol. 89, no. 5-6, pp. 1027-1043.
Leimer, D. Lesnoy, R. Selig, D. (1982); Social Security and Private Saving: New Time-Series Evidence; Journal of Political Economy, No. 3, pp 606-29.
Loayza, N. Fajnzylber, P. and Calderón C. (2005); Economic Growth in Latin America and the Caribbean: Stylized facts, explanations, and forecasts.
Otani, I. and Villenueva, D. (1990); Long-Term Growth in Developing Countries and Its Determinants: An Empirical Analysis; World Development, Vol. 18, No 6, pp. 769-783.
Pal, K. Behrendt, C. Léger, F. Cihon, M. and Hagemejer, K. (2005): Can low income countries afford basic social protection? First results of a modelling exercise; Geneva;
International Labour Office.
Rosen, H. (2005); Public Finance – 7th edition; Singapore; McGraw-Hill.
Rosenzwigh, M. and Wolpin, K. (1993); Credit Market Constraints, Consumption Smoothing, and the Accumulation of Durable Production Assets in Low-income Countries: Investments in
Bullocks in India; The Journal of Political Economy, Vol. 101, No. 2, pp. 223-223.
Todaro, M. and Smith, S. (2003); Economic Development; Harlow, Pearson Education Limited.
Vodopivec, M. (2004); Income support for the unemployed: Issues and options; Washington;
The World Bank.
World Development Report (2002); Attacking Poverty: Opportunity, Empowerment and Security, Chapter 8: Helping poor people manage risk; World Bank
Zangh, J. and Zangh J. (1995); The Effects of Social Security on Population Growth;
Southern Economic Journal, vol. 62, No. 2, pp. 440-450.
Data:
World Development Indicator on CD-Rom; Washington; World Bank.
World Education Indicators, UNESCO Institute for Statistics:
http://www.uis.unesco.org/statsen/statistics/indicators/i_pages/indic_2.htm (2006-05-19) Cost of Social security; International Labour Organization – Social Security Department:
http://www.ilo.org/public/english/protection/secsoc/areas/stat/css/index.htm (2006-05-19) For data earlier than 1990 we used the printed book series “The cost of social Security”:
Geneva; International Labour Organization.
