Social insurance in the transition to a market economy : theoretical issues with application to Moldova

_ _ wS 1528 POLICY RESEARCH WORKING PAPER 1588 Social Insurance What is the proper mlix of social insurance and social in the Transition assistance in the transition to a M arket Economy economies of Eastern ELirope and the forrner Soviet Union? Theoretical Issues with Application to Moldova Deborah Mabbett The World Bank Europe and Central Asia Country Department IV Country Operations Division 2 April 1996  POLICY RESEARCH WORKING PAPER 1588 Summary findings Social security systems in Eastern Europe and the former family allowances, could also form part of the system. In Soviet Union devote most of their resources to earnings- principle, benefits in a unified system could he means- related pensions and neglect "targeted" interventions to tested, but means-testing is not integral to the system, aid losers from the transition to a market economy. As whereas it is the defining feature of social assistance in social insurance systems, they have the characteristic the two-tier model. weaknesses of the insurance model, paying benefits on Some analysts have argued for a unified system in the the basis of past contributions or work history rather short term (because of fiscal constraints), moving to a than need. Yet contributions and benefits are not closely two-tier system later in the transition. Both types of enough related to enable the system to operate as social system exist in capitalist economies. Indeed, systems in insurance in a market economy. many countries are an untidy mix of insurance and More attention must be paid to the potential impact of assistance. high payroll taxes on employment in a market economy. Mabbert examines some of the issues that arise from Some analysts have argued that contributors to the the relationship between insurance and assistance in the system (future pensioners) should perceive their transition and beyond. She derives three criteria for contributions as "savings," in addition to which determining whether the two-tier model is viable, then unemployment benefits and a new social assistance applies the criteria to the small republic of Moldova. She system could be developed. Such a two-tier system would concludes that Moldova probably does not fulfill the distinguish between social insurance and social three criteria because it has experienced a severe assistance. economic shock and adjustment has been a long drawn- Mabbett discusses an alternative prescription - a out process. "unified system," in which the earnings-related system The two-tier prescription may be more problematic for would be curtailed, making it as "flat" as possible, countries of the former Soviet Union than for the Eastern thereby establishing a universal categorical age benefit. European states. Other benefits, including unemployment benefits and This paper - a product of the Country Operations Division 2, Europe and Central Asia, Country Department IV - is part of a larger effort in the region to assist in reorienting social expenditure to help the poor. Copies of the paper are available free from theWorld Bank, 1818 H StreetNW, Washington, DC 20433. Pleasecontact Louis Biely, room H5-145, telephone 202-473-6280, fax 202-477-3378, Internet address Ibiely@worldbank.org. April 1996. (35 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be used and cited accordingly. The findings, interpretations, and conclusions are the authors' own and should not be attributed to the World Bank, its Executive Board of Directors, or any of its member countries. Produced by the Policy Research Dissemination Center  Social Insurance in the Transition to a Market Economy Theoretical Issues with Application to Moldova* Deborah Mabbett Department of Government Brunel University Uxbridge, Mdx UB8 3PH England Telephone: +1895-274000 Fax: +1895-812595 E-mail: deborah.mabbett@brunel.ac.uk *This paper was prepared with the support of the World Bank and arises out of work done with Helen Sutch on the social protection component of the program supported by the first structural adjustment loan to Moldova. The author would like to acknowledge the help of Philip Goldman, Tom Hoopengardner, Monika Huppi, and Helen Sutch.   Social Insurance in the Transition to a Market Economy Theoretical Issues with Application to Moldova I. Introduction The process of transition to a market economy in Eastern Europe and the former Soviet Union (EEFSU) has seen Western economic advisors engaged in formulating prescriptive analyses in which the central institutions of a market economy are identified and described. Social security has found a place in these analyses, albeit usually in the concluding paragraphs. In particular, the prospective importance of unemployment benefit in facilitating restructuring of overmanned enterprises has been highlighted (IMF et al 1991; Holtzman 1991). Commentators have also emphasized the importance of "cashing out" the consumer subsidies which constituted an important part of the social guarantee provided by EEFSU states to their populations (Ahmad and Chu 1993; Chu and Gupta 1993). The existing social security systems of EEFSU countries have proved to be poorly designed for the functions envisaged in these prescriptions (Ahmad 1992). They devote the bulk of their resources to earnings-related pensions and neglect "targeted" interventions to aid the losers from the transition. 1  The systems have the characteristic weaknesses of the insurance model, paying benefits on the basis of past contributions or work history rather than on the basis of need. At the same time, the relationship between contributions and benefits is not close enough to enable the system to operate as social insurance in a market economy (Kopits 1992). The transition to a market economy means that more attention has to be paid to the potential impact of high payroll taxes on employment than under central planning. This has led commentators to argue that countries should structure their contributory systems so that contributors (future pensioners) perceive their insurance contributions as "saving" (Barr 1994). This entails strengthening the insurance characteristics of the earnings-related system. In addition, in order to meet the other objectives of social security in the transition, an unemployment benefit has to be introduced and a new social assistance system developed. Social assistance is necessary to complement social insurance; otherwise insurance benefits will be appropriated to meet assistance objectives (Kopits 1992, p. 305). There is an alternative to this prescription, which is to curtail the earnings-related system, making it as "flat" as possible, thereby establishing a universal categorical age benefit. Other categorical benefits, such as unemployment benefits and family allowances, could also form part of the system. In the following discussion, the former option, with its strong distinction between social insurance and social assistance, will be referred to as the "two tier" model, while the second alternative is to create a "unified system." In principle, benefits in a unified system could be means-tested, but 2  means-testing is not integral to the unified system, whereas it is the defining feature of social assistance in the two-tier model. Some commentators have argued, in effect, for a unified system in the short term because of fiscal constraints, moving to a two-tier system later in the transition. Both types of systems exist in capitalist economies; indeed many countries have systems which are an untidy mixture of insurance and assistance principles. The purpose of this paper is to examine some of the issues which arise in the relationship between insurance and assistance in the transition and beyond. Section II identifies some theoretical issues which arise in the two-tier model. From this discussion, three criteria are derived for determining whether the two-tier model is viable. The paper then goes on to discuss whether the three criteria are fulfilled in one small FSU republic, Moldova. It is concluded that the criteria are probably not fulfilled where a country has experienced a very severe economic shock and where the adjustment process is long-drawn-out. These conditions may apply to several FSU republics. The two-tier prescription may be more problematic for the FSU than for the Eastern European states. II. Theoretical Issues Economic theory provides two main models for conceptualising the operation of an old age pension system. One model is that of a savings contract, whereby people give up current income during their working lives in order to smooth their consumption into retirement. The savings model is most 3  closely complied with where contributors pay into individual accounts in a funded pension scheme. But it is possible to emulate the savings model in a pay-as-you-go (PAYG) system. Benefits can still be related to contributions (individual accounts can be created to make this apparent). The rate of return to participating in the PAYG scheme will be determined by different factors to the funded scheme (the "biological," rather than the financial, rate of return) and different factors also affect the credibility of the scheme. In the PAYG system, political commitment is critical, whereas in the funded system the fiduciary characteristics of the financial system are central. An alternative model is that of deferred compensation. Here, the driving motive for the scheme is employers' desire to structure the compensation package to promote long job tenures. In a market economy, deferred compensation finds expression in employers' willingness to contribute to occupational pensions, but the deferred compensation motive cannot sustain a general pension system, because there is an incentive for employers to free-ride. While all employers benefit from the work commitment engendered by a pension which depends on years of service, a general pension does not encourage long tenure in a particular job. Furthermore, failure on the part of one employer to make contributions to a general scheme will not affect the work motivation of his or her workforce. Free riding was not a problem under central planning, and it is fruitful to see Soviet pensions as a formn of deferred compensation. A close relationship between pensions and work was built into the formulae by which pensions were calculated and the financing structure established to fund them. 4  This relationship was designed around labor allocation objectives. Planners could use pension rights to vary the attractiveness of jobs, discriminating against agricultural workers for many years, for example. Low and differential pension ages could facilitate changes in the allocation of labor across occupations. Significant cross-subsidies between sectors of the economy arose as a result of variations in pension rights (Fox 1994, p. 10). With the development of a labor market based on profit-maximizing employers acting atomistically and self-interestedly, it would seem that the pension system will have to be reformed along the lines of the savings model, if it is to be maintained. An alternative would be to break it up into an occupational pension system, in which employers do have a deferred compensation incentive to contribute. However, this is not an attractive option given the large-scale industrial restructuring that economies in transition must undertake. As is discussed further in section VI, lack of labor mobility and close ties to the enterprise are an important problem in the transition in FSU countries. Those who argue for a two-tier strategy have based their arguments on the savings model. The key idea is that people have an incentive to pay insurance contributions (or to ensure that employers pay contributions on their behalf) in so far as they value the expected future benefit - the insurance cover - which they are purchasing. The incentive to pay contributions is the key element in minimising the distortionary effects of the insurance levy or payroll tax. If contributions are seen as bringing corresponding future benefits, these benefits then constitute part of the remuneration 5  package from working. As a result, labor supply does not fall (or the reservation wage rise) as it would in response to the imposition of a tax (Barr 1993. pp. 230-31). The insurance principle provides the rationale for the payment of payroll taxes into a separate fund dedicated to the financing of earnings-related pensions. If the hypothecated revenues are credibly protected from other claims on govenmment funds, then contributors can make an informed estimate of the expected rate of return on their investment, even though the underlying financing of the system is PAYG. We can therefore conclude that secure hypothecation is the first criterion for making the savings model work. The savings model further requires that the expected value of future benefits matches current contributions. This will depend on the rate of return to investment in the hypothecated fund compared with investors' rate of time preference. Here there is a major problem for an economy in transition. Living standards have dropped sharply, while a "brave new world" of prosperity is in prospect after market reforms are implemented. This means that current income is low relative to expectations of permanent income, which implies a low level of voluntary saving. Even if contributions are expected to bring future benefits, these will be heavily discounted if workers' preferences are strongly in favor of current consumption. This problem is compounded by the liberalization of the labor market in the transition. Under central planning, high levels of forced saving could be achieved by the suppression of informal 6  economic activity and the requirement to work for the state. The development of a liberalized labor market means that labor supply will become more elastic. Furthermore, low desired savings can be seen as one reason for the severity of the problem of diversion of labor into the informal sector. While the informal sector may encompass a wide range of activities (particularly where tax administration is weak), many informal activities are not monetized and provide no opportunities for the accumulation of financial assets. Where this is the case, the desire to accumulate financial assets provides an incentive to participate in the formal sector. If the propensity to save is low - such as in the situation where workers' preoccupation is their current subsistence - incentives to supply labor to the formal sector are weakened. It is difficult to formulate indicators of factors affecting the propensity to save. However, at the very least we can say that workers' current fornal incomes should be sufficiently above the poverty line to enable them to purchase insurance. This constitutes a second criterion for viability of the two-tier approach. This second criterion is clearly a necessary, but not a sufficient, condition to enable social insurance to develop. Workers must be not only able but willing to contribute to the insurance system. Their willingness depends on the expected rate of return from their investment. While this is shrouded in uncertainty, one concrete issue can be identified. If the country has a social assistance system as well as a social insurance system, it is only worth contributing to the insurance system if the expected future income for the contributor exceeds the assistance level. If people use the present as 7  a guide to the future, they may ask whether those who have experienced the contingency for which they are buying insurance (that is, the current elderly) are major recipients of social assistance. If they are, then, unless the situation is thought to be only transitional, the incentive to insure is reduced. The exact effect depends on the design of social assistance. Clearly if assistance takes the form of flat-rate supplements for pensioners, the effect on the incentive to insure is less than if assistance is inversely related to insurance (for example, if it is means-tested). It follows from this that a third criterion for successful implementation of the two-tier model is that insurance recipients should only exceptionally be in receipt of social assistance. If this criterion is not met, it is necessary to convince contributors that this is only a transitional situation. Clearly an issue of timing affects both the second and third criteria. Current conditions may be against the development of a thriving insurance tier, but if economic recovery is rapid, the criteria may be fulfilled in the near future. The problem is that if the criteria are not currently fulfilled, then attempting to maintain the insurance principle may actually jeopardize recovery. In the transitional period, the expected value of future benefits is less than current contributions, which means that the legally-mandated contributions are actually taxes, and will have all the distortionary effects of taxes. Furthermore, if criteria (2) and (3) are not met, then it may become inefficient to comply with criterion (1). If the insurance levy is a tax, hypothecation does not accord with efficient operation of the tax system. The government should aim to raise revenue for all purposes through the 8  combination of taxes which minimizes the resulting economic distortions. If the government is not to be constrained in choosing an optimal tax mix, then it may not be possible to retain the hypothecation of finance from one particular tax to one particular area of expenditure. II. Social Security in Transition in Moldova: Introduction With the breakup of the Soviet Union, Moldova experienced one of the worst terms of trade shocks of any of the FSU republics. Moldova is totally dependent on imports for its energy. As Russia pushed oil and gas prices up to world levels, the value-added of major Moldovan production activities in both agriculture and industry collapsed (World Bank 1994, p. 6). The terms of trade decline directly affected the government's budgetary position. In the social sector, housing, health, and education costs increased via the direct impact of higher energy costs on schools, hospitals and the provision of communal services. Subsidy expenditure also increased, because subsidy arrangements often committed the budget to meeting the gap between variable producer prices and a fixed consumer price. As the effect of higher prices for imported inputs fed through the economy, producer prices soared, and subsidy commitments increased accordingly. The dismantling of the universal welfare system constituted by consumer subsidies was one of the main measures taken by the government to reduce real incomes in the face of the deterioration in Moldova's terms of trade. It is important to note that reduction in subsidies was undertaken for macroeconomic as well as microeconomic reasons, as this affected the form which compensation 9  could take. Average incomes had to fall if the process of subsidy reduction was to lead to macroeconomic equilibrium (bearing in mind that the burgeoning cost of subsidies reflected the deterioration in Moldova's terms of trade). The macroeconomic imperative was that only the "most needy" households could be compensated. The specific implication of this in the Moldovan case was that compensation could not be implemented by increasing the minimum wage. This is because the minimum wage functions as the "numeraire" for all wages. This feature derives from the Soviet system of centrally-determined wage tariffs. In Moldova this system was renewed and revamped in the Law on Salaries passed in February 1993. The Law specifies an indexation system for tariff wages based on the minimum wage. For budget sector employees, a detailed 29-point tariff, in the form of coefficients of the minimum wage, was established. For other enterprises, a looser framework of 11 categories with 9 coefficients was set out in the law; this was later amended to become an 8-point scale. The tariffs constituted a minimum guarantee of "payment for labor" and were obligatory on all enterprises, although the Law also stipulates that enterprises have discretion to pay amounts over the tariff "to stimulate productivity." Recognizing that increases in the minimum wage would basically increase all labor incomes proportionately, the government refrained from full indexation of the minimum wage. However, incomplete indexation had a rather perverse effect. Pressure on real wages led to "drift" of wages 10  above tariff levels. While the average wage remains strongly linked to the minimum wage, it has increased more than the minimum wage, as shown in table 1. Table 1 The Minimum Wage at Uprating Dates Minimum Average wage/ Date wage Minimum wage (Lei) March 1992 0.4 3.97 April 1992 0.85 2.00 October 1992 0.85 4.55 November 1992 1.7 3.34 February 1993 1.7 4.28 March 1993 3 3.29 June 1993 3 5.70 July 1993 7.5 3.66 October 1993 7.5 5.44 November 1993 10 5.19 December 1993 10 9.80* January 1994 13.5 5.01 May 1994 13.5 6.25 June 1994 18 5.93 * This figure is affected by the payment of annual bonuses in December. 11  The increase in the average wage relative to the minimum wage had implications for pensions. The wage tariff provides the mechanism whereby wages are linked to "work" pensions (so-called because amounts payable are based on previous earnings and work history). Before 1993, pensions were periodically uprated according to changes in the average wage. Hyperinflation made this process cumbersome and complex, and in July 1993 the Ministry of Labor and Social Protection introduced a new system for uprating pensions, which clearly followed on the principles established in the Law on Salaries passed in February. The new system involved the calculation of a coefficient based on previous earnings and work history, which was then applied to the minimum wage to arrive at the pension. This meant that pensions were formally indexed in the same way as wages and salaries. Any increase in the minimum wage would see proportionate adjustments in the incomes of both workers and pensioners. However, pensions were more strictly tied to the minimum wage than wages, because enterprises could pay wages above the tariff. The outcome of this system was that real pensions fell more than real wages, because the real minimum wage fell more than the real average wage. In this sense, pensioners bore the brunt of hyperinflation. (For a discussion of a similar pattern in Russia, see Liu 1993, pp. 67-9.) This was inevitable, in so far as the main losers from the breaking-up of the state-guaranteed income structure would always be those most reliant on the formal system of guarantees. Incomplete indexation of the minimum wage meant that the wage determination system became more flexible: the tariff minima became less binding on enterprises. 12  III. Social Assistance The Soviet welfare system can be thought of as comprising not only (or even mainly) transfer payments, but also the provision of various goods and services free or at subsidized prices. These included health services, education, transport, housing (including heating costs) and food. While governments in many market economies also provide or subsidize some goods and services. subsidization in the FSU encompassed some expenditures which could not really be defended on public good. merit good or externality grounds. It was recognized that sharp rises in the prices of key consumer goods would exacerbate the problem of low living standards and poverty, but many commentators argued that it was efficient to counteract this through social assistance. Available studies (and a priori reasoning) suggested that the bulk of subsidy expenditure went on consumption by the nonpoor (see, for example, McAuley 1991, p. 100-1; IMF 1992, p. 30). Overall social expenditure could be reduced without an increase in poverty if increased cash transfers were targeted to the poor. This rationale for an enhanced "social safety net" became a standard part of economic reform proposals put forward by the IMF in FSU republics (IMF IV.6; IMF 1993a, p. 7; IMF 1993b, pp. 5-6). While some commentators have advocated the replacement of subsidies with noncategorical, means-tested social assistance, this proved to be beyond the institutional capacity of the social 13  security system. The institutional problem was to create a social assistance system which had clear and fair rules for entitlement but at the same time was sufficiently calibrated and controlled to ensure that entitlements could be met out of the available budget allocation. One difficulty with a means-tested system is to predict how many people will qualiI, in the absence of good information on household incomes. Furthermore, lack of information presents not only a budgeting problem, but also an administrative problem. It is tempting to empower local administrations to use local knowledge to ascertain informal income. but this then means that the central financing body must find a way to prevent overspending due to liberal decisions by local administrations. A block grant system can get around the problems which arise from local adjudication with central financing (Sipos 1994, p. 246). However, this control is achieved at the cost of clear rules for entitlement. In Moldova, there is a structure of local Social Support Funds (SSFs) which control small budgets and devise their own ways of accomplishing the best distribution of the money they can manage. However, to channel large amounts of subsidy compensation money through SSFs would mean that a great deal of attention would focus on the composition of the SSF committees and the justice and fairness of their decisions. The allocation of central government money between localities would be contentious and potentially politically divisive (for a review of the rather arbitrary allocation of social protection funds between regions in Russia, see Liu 1993, p. 74). An alternative approach to implementing means-tested social assistance is to specify detailed qualification conditions which can be monitored and audited by central government. Central 14  government thereby maintains control over the system. Provided it calibrates the qualification conditions correctly, it can avoid excessive expenditure. In Moldova, this approach was attempted in September 1993. Under the September measures, pensioners were eligible to receive different amounts of compensation according to three income bands, while families with children would receive compensation only if the family's per capita income was less than the minimum wage. The Ministry of Finance formulated very tight controls, requiring extensive documentation pertaining to family composition, all incomes of family members, and private plots of land owned by the family. For children over 13, the benefit was also conditional on their school attendance. These control mechanisms proved to be too rigorous to implement. Very little money was paid out for children under the September compensation arrangements. Pensioners also found the procedures difficult to negotiate. In the end, the Ministry of Finance and the Ministry of Labor and Social Protection agreed to simplify the arrangements for pensioners, abandoning the comprehensive income test and basing compensation payments on the amount of work pension alone. This approach was continued in May 1994. when bread and milk subsidies were finally removed. Three rates of compensation to pensioners were provided, inversely related to the work pension. The cumulative effect of successive compensation measures was that, by mid-I 994, pensioners who qualified only for the minimum "work" pension were receiving two-thirds of their total income from various "social assistance" measures implemented to compensate them for the 15  removal of subsidies, while the average pensioner received one-third of his or her income from these sources. Basically this happened because the social insurance system was unable to keep many of its beneficiaries out of poverty. The government therefore had to top up the social insurance system with budget-financed compensations. At the same time, the administrative difficulties of operating social assistance meant that most of the compensation expenditure went to social insurance recipients, that is, to the existing "clients" of the social security system. The result of this process is that the third criterion outlined in section I is not fulfilled. In effect, what has happened is that the economic crisis, specifically reflected in the removal of subsidies, lowered the level of real insurance benefits below the poverty line, placing most beneficiaries in need of "assistance." It was necessary to increase the share of noninsurance payments in pensioner income very substantially, which indirectly undermined the insurance principle. The compensation process, accompanied as it was by targeting towards the main insurance recipients, altered the basic dynamics of the relationship between pensioners' and workers' incomes. V. The Financial Situation of the Social Fund Some insight into the first criterion outlined in section II can be gained by tracing through the evolution of the financial situation of the Social Fund. Protection of the hypothecated revenues of the Social Fund fell-foul of the adverse financial situation of the general government budget in 1992-93. During this period, the budget was very short of money while the Social Fund was awash 16  with funds, and the government could not resist drawing on the Social Fund's resources for noninsurance purposes. The Social Fund is the revenue-collecting organization for three constituent funds: the Pension Fund, the Employment Fund, and the Social Insurance Fund (SIF). The SIF pays sickness and maternity benefits and is administered by the trade unions at the place of work. Its financing is quite autonomous: enterprises pay their SIF contributions into a separate account. By contrast, the Employment Fund notionally receives an earmarked tax of 1% of payroll, but in practice its financing is integrated with the Pension Fund. Employment Fund and Pension Fund revenue goes into the same account, and the Employment Fund is managed by the Employment Department, which is within the Ministry of Labor and Social Protection. When government decisions refer to finance coming from the Social Fund, this means the combined resources of the Pension Fund and the Employment Fund. The first hypothecation issue which is presented by this structure concerns the allocation of payroll taxes between the constituent funds. One potential problem is that the Employment Fund could gradually require more funds as unemployment increases, and, if the government tries to control the overall payroll tax rate, this would reduce finance for pensions. In this scenario, rising unemployment could undermine incentives to contribute to the Pension Fund, if contributors believe that future pensions will be reduced because of unemployment. In practice, the opposite situation appears to have occurred: unemployment has not emerged (for reasons discussed further 17  below) and the Employment Funld has spent less than its notional allocation of 1% of payroll. While it is possible that uLnemllploymlllent is affecting, the Pension Fund through an increase in early retirement, of'ficial data suggest that thiere is little earlv retirement in Moldova, by contrast with other FSUJ couLntries. The budget balance conditioni for the Pensionl Fund can be written as follows: t.( I -n).AW.E AP.N .... (I) or t = AP/(I-n).AW . N/E where "t" is the payroll tax or contribution rate, "n" the nonpayment rate, "AW" average wage, "AP" average pension, iN" number of beneficiaries. and "E" is the number of workers employed. It can be seen that continuing Pension Fund balance can be ensured with a stable payroll tax rate provided: - the replacement ratio (AP/AW); - the dependency ratio (N/E); and - the payment rate (I -n) are stable. 18  The maintenance of a stable replacement ratio is a basic element in the "insurance"' rationale of the Pension Fund and a key commitment to the pensioner population. This meant not only maintaining a general linkage between pensions and wages. but also liiking the pension structure to the wage structure, so that workers wlho occupied a high position in the earnings league could expect this to be reflected in their pension. This rationale was consistent with the funding structure for pensions. Provided the nonpayment rate and the dependency ratio did not rise, the Pension Fund could finance a stable replacement ratio with a stable tax rate. With the pension uprating system established in July 1993 (described in section III above), the linchpin of the replacement rate was the minimum wage (MW). The ratio of average pensions (AP) to average wages (AW) would be stable provided that the ratios AP/MW and MW/AW were stable. The first ratio was affected only by the earnings and work history coefficients of new pensioners; the second would be stable if average wage developments were governed by the Law on Salaries. However, as was shown in table 1, the ratio of average wages to minimum wages was not stable. The average wage increased by more than the minimum wage as the government refrained from full indexation of the minimum wage in its attempts to restrain inflation. So long as the nonpayment rate remained stable (which it did until early in 1994), the rise in AW/MW meant a financial windfall for the Pension Fund. Behind this windfall lay the bleak fact that the income position of pensioners relative to workers had deteriorated, as explained in section III. 19  With the government budget seriously in deficit, the Ministry of Finance was very interested in policies which would enlarge the areas of social expenditure covered by the Pension Fund. The first area of expenditure to be affected by this pressure was family allowances. Family allowances have been administered by the Pension Fund since their inception, but in Moldova the budget was initially assigned responsibility for financing. Enterprises made the payments to their eligible workers, and then netted the amounts paid out of their contributions to the Pension Fund, which was then reimbursed by the Budget. In 1991, the Pension Fund received this reimbursement. In 1992, the reimbursement was budgeted for, but not paid in full, and in 1993, no provision was made in the Budget for reimbursement of family allowance expenditure. The procedure for payment and reimbursement of family allowances made it easy for the government to shift their financing, by failing to make reimbursements. This shift in the financing of family allowances obviously undermines the insurance principle, because contributions to the Pension Fund are not hypothecated to pension payments only. It also illustrates a general point about the relationship between administration and financing. It is administratively efficient to use the network of offices managed by the Ministry of Labor and Social Protection to administer all social security receipts and payments, both those relating to "insurance" and other benefits. In principle, joint administration should not preclude the operation- of separate financing arrangements. However, in practice, administrative arrangements can affect financing in a situation where budgetary control and financial management is weak. 20  The main issue in maintaining an insurance rationale for pensions concerns the financing of the minimum pension. To the extent that a minimum pension is provided out of insurance funds for those with inadequate work histories or earnings records, the insurance principle is weakened. In Moldova, there are two minima in operation: a minimum "work" pension, which equals the minimum wage, for those with enough years of work to qualify for a pension but insufficient earnings; and a "social" pension, for those with insufficient years of work, which has been variously set between two-thirds and three-quarters of the minimum wage. Both these pensions are financed by the Pension Fund. The amounts involved for the social pension are small, but about 25% of pensioners receive only the minimum "work" pension. As the real value of the minimurn wage fell in 1992 and 1993, the government became increasingly concerned about the position of those whose income was set equal to the minimum wage. The Presidential Decree which increased the minimum wage at the beginning of 1994 provided for higher increases for workers at the bottom of the wage tariff, and for special supplements for the worst-off pensioners. These supplements are financed by the Pension Fund. The same Decree contained measures to try to deal with the rise in the price of gas. Since household energy subsidies were mainly off-budget (being partly charged to the profits of the energy utilities, and partly cross-subsidized by industry), the rise in energy prices did not release resources to finance compensations. The government therefore looked to whatever sources of finance were 21  available. Since compensations were largely directed to pensioners (given the institutional barriers to any alternative arrangements discussed in the previous section) the Pension Fund was a natural target to bear part of the cost of financing. Fifty percent of the cost of energy compensation payments introduced in winter 1993-94 was charged to the Fund. It is apparent that the Pension Fund is not securely hypothecated, and the first criterion outlined in section II is therefore not fulfilled. To some extent, the same factors which have led to an increase in the importance of "assistance" payments for pensioners (as outlined in section IV) have also undermined hypothecation, as the government has used Pension Fund as well as budget financial resources to combat poverty. However, assistance could have been increased without undermining hypothecation if the financial position of the general government budget had been stronger, and if social security had been better placed to claim budgetary resources. Arguably, the latter factor is the critical problem. To put the issue bluntly, not all government expenditure is equally under the control of the Ministry of Finance, with the result that the government's command over resources is not properly regulated through the budget process. Some ministries are better placed than others to conduct their -affairs on a debt-financed basis, subsequently presenting the Ministry of Finance with obligations as afait accompli (expenditure incurred as a result of default on government-guaranteed loans is a prime example). Social security, by contrast, is entirely dependent on the release of tightly-controlled cash. It must compete with other very pressing claims on cash, notably the payment of public sector wages. 22  Given the weakness of budgetary management and the strength of the "economic" ministries relative to those undertaking social functions, it might be thought desirable for both social insurance and social assistance costs to be met from a separate fund. However, this hypothecation would have to be based on public choice logic, rather than on the insurance principle. flypothecation on public choice grounds evades the real issue, which is that the government has not established control over all its branches and has not been able to set and enforce priorities in determining their command over resources (Allan 1994). VI. The Informal Economy and Insurance In section II it was argued that the extent to which insurance contributions will be seen as savings will depend in part on households' propensity to save. A severe fall in real income accompanied by the expectation of a future increase in income will engender a low propensity to save, and bring transitional difficulties in getting workers or their employers to pay insurance contributions. It was also argued that a low propensity to save could be reflected in a growth in informal, unmonetized economic activity. The poor quality of Moldovan national accounts makes these propositions difficult to document. The available estimates of GDP are based on output data and cannot be reconciled with income and expenditure data. Information on the proportion of the population in poverty is also problematic. 23  According to the Moldovan authorities, most of the population is in poverty. This view is reached by comparing wages with the "minimum consumption basket" (MCB) of goods needed for living. The Moldovan MCB suffers from the deficiencies of other Soviet consumption baskets, as it is based neither on actual consuimption patterns nor on nutritional and other requirements of life (McAuley 1994, p. 32). Other FSU republics and the Russian Federation have developed a "physiological subsistence minimum" poverty line (ibid: Liu 1993, p. 67), but the Moldovan government has not adopted this poverty line. In any case, the physiological subsistence concept has not been developed to provide insights into whether households have sufficient income to be able to save. More information about households' assets and productive resources would be needed to address this question. In the current social insurance structure, enterprises rather than households pay the contributions to the Social Fund. In this context, the impact of a low household propensity to save is that workers have little incentive to work in the formal economy to obtain an insurance record, or to ensure that contributions are paid on their behalf by employers. Since early 1994, enterprises have accumulated substantial arrears on their payroll tax liabilities to the Social Fund, and the reasons for this give some insight into the institutional structure of the labor market and its effect on social insurance in Moldova. Payroll tax arrears began to increase sharply in the first quarter of 1994. This increase was accompanied by an increase in arrears on wages owed to workers. Both problems in their turn 24  reflected the tightening of monetary policy after the establishment of a separate currency in November 1993. One might expect that, as the flow of cheap credit to enterprises was curtailed and some enterprises found that they could not pay their workers, the next step in the adjustment process would be that unemployment would increase. This would worsen the financial situation of the Social Fund by reducing the payroll tax base and expanding unemployment benefit payments, but arrears would stabilize. However, open unemployment has not increased, and Social Fund arrears have continued to grow. The proximate cause of the nonemergence of unemployment is that enterprises have not taken the step of formally dismissing workers, preferring instead to put their workers on "forced vacations." This pattern of enterprise behavior is common to many of the countries of the former Soviet Union, and the motives for it have been extensively analyzed (see, for example, Roxburgh and Shapiro 1994). One striking feature of the Moldovan situation is the extent to which enterprises may retain workers in employment but fail to pay wages. According to the law, workers are meant to be paid two-thirds of their tariff wage when they are involuntarily idle. However, enterprises have evaded this requirement by obtaining workers' "agreement" to the nonpayment or part-payment of wages. While officials at the Ministry of Labor and Social Protection argue that this agreement is often fictional, the close relationship between management and unions in many workplaces has prevented open conflict. 25  The combined effect of forced vacations and the emergence of wage arrears must be that many people in employment are not making a living from their formal work. It is possible that large numbers of workers have already "effectively" quit their jobs, and are engaged in other activities, although without giving notice to their employer. This concealed movement of labor out of the enterprise sector has not been accompanied by growth in new sources of contributions to the Social Fund, as such activity remains informal and unrecorded. It could be argued that the effect of hidden unemployment on Social Fund finances is less severe than the effect of open unemployment. While hidden unemployment reduces the payroll tax base, open unemployment would also increase benefit expenditure. However, it is possible that the medium-term prognosis for the Social Fund under the current pattern of adjustment is worse than if open unemployment had emerged. The reason is basically that hidden unemployment is also accompanied by patterns of industrial relations and enterprise behavior which have an adverse effect on the social security system. The most apparent signal of the emerging problems in the relationship between the Social Fund and the enterprise sector is the development of the relationship between wage arrears and payroll tax arrears. If enterprises' failure to pay payroll tax was a reflection of the nonpayment of wages, then payroll tax arrears would be about one-third of wage arrears (given the payroll tax rate of 35%). However, this is not the case. Payroll tax arrears are greater than wage arrears, standing at 316.7 26  million lei at I April 1995 compared with wage arrears of 167.3 million lei. (This latter figure represents about a month's wages over the whole economy.) One explanation of this pattern is that a high proportion of wages are being paid in kind. When wages are paid in kind, payroll tax is assessed, but tax is not paid until the enterprise has money in its bank account. Workers may be paid in the output of the enterprise (especially in agriculture) or in goods obtained through barter transactions. It is possible that some enterprises are finding ways to obtain cash to pay workers without using their bank accounts (the banks automatically remit payroll tax when cash is withdrawn to pay wages), but such transactions are illegal. Both nonpayment of wages (forced vacations) and the use of wage payment in kind reflect the very weak bargaining position of workers in the current environment. The history of wage payment in kind in other economies suggests that the behavior of the labor market can be profoundly affected when the nexus between workers and the employer is based on provision of goods and services instead of cash. Payment in kind is connected with close, paternalistic relationships, in which the employer exercises a high degree of control over workers' lives and the worker is highly dependent on the enterprise. When workers are paid in kind, their opportunities to save are limited, which ties them more closely to the enterprise, because of the difficulty of accumulating enough resources to leave. Under central planning, enterprises used benefits in kind to attract and retain workers in the face of underdeveloped consumer markets and labor shortages. The use of noncash payment in a situation of labor surplus potentially has much more adverse consequences. Workers are tied to 27  failing enterprises and are faced with declining real remuneration and a lower standard of employer- provided goods and services such as health care and housing. Paternalistic management is accompanied by the absence of independent trade unions. In Moldova, the trade unions remain engaged in their old functions of managing the distribution of the enterprise's "Social Fund." Wvhile the national union confederation campaigns for increases in the minimum wage, shopfloor pressure is virtually invisible. Protests about nonipayment of wages appear to be confined to the sectors where workers are paid directly from the budget (health and education, in particular). Such protests are directed against the state, rather than enterprise managements. The absence of bargaining pressure on management may be compounded in the future as more enterprises become worker-owned in the course of privatization. One implication of this institutional structure is that real wages will be kept lower than if there were independent trade unions. While this has a positive side, that less open unemployment is likely than in countries with stronger worker organisations, its implications for social insurance are adverse. It was argued in section II that one necessary condition for developing a social insurance system is that workers' incomes are sufficiently above the poverty line to enable them to purchase insurance (criterion 2). In western capitalist economies, the main threat to meeting this condition comes from unemployment. If part of the working age population is persistently unemployed or experiences frequent interruptions in employment, then people will reach the pension age without adequate 28  insurance cover. In Moldova, the threat to criterion 2 comes from low cash wages, reflecting in turn low enterprise value-added and demagnetization of the economy. In the paternalistic enterprise structure which appears to be entrenched in Moldova, particularly in the countryside, old-age provision is likely to become based directly on the enterprise. Some elements of this tendency are already present. Enterprises often top up the incomes of "their" pensioners. In agricultural districts, pensioners may be recipients of allocations of food, while urban enterprises distribute Christmas bonuses. One sign that such provision is likely to become more important comes from the way that the Ministry of Labor and Social Protection has responded to the shortage of funds caused by payroll tax arrears. The Ministry has ensured that pension payments have been kept up-to-date in the main cities (Chisinau and Balti). The Ministry notes that payroll tax arrears are much lower among urban industrial enterprises than in agriculture. While in theory the resources of all contributors are pooled nationally, in practice there has been a tendency to pay pensions first in the districts where the flow of contributions has been adequate. The implication is that the national pension system is too weak to pool resources across the country, and that enterprises in noncontributing districts will be left to make their own provision for their pensioners. VII. Conclusion Many discussions of social security in Eastern Europe and the former Soviet Union draw a strong distinction between social insurance and social assistance. This paper has sought to show that the 29  interaction between insurance and assistance can be very problematic. It has been argued that incentives to contribute to social insurance in the transition are weak. There are three main reasons for this: 1. Insurance funds are not securely hypothecated. 2. The propensity to save is low, because current income is depressed. 3. Institutional and administrative factors led to a situation where insurance recipients receive a high proportion of their income from "assistance" payments, and, conversely, the bulk of assistance expenditure goes to the insurance categories. In section VI it was argued that the current condition of the enterprise sector is an obstacle to the continuation of social insurance based on the collection of contributions from employers. One possible inference from the discussion is that household asset accumulation should be promoted. This would imply that any "second tier" of social security provision should concentrate on the provision of savings opportunities for households, rather than being organized around employment. The difficulty, as even some of the most prosperous Western economies have found, is to promote a sufficiently high household propensity to save. In the FSU, households are likely to be strongly attracted to real asset accumulation, rather than entrusting their wealth to the financial markets. It was noted in the introduction that a unified social security system could either pay flat-rate benefits to specified categories in the population, or. alternatively, it could provide means-tested benefits. Means-testing would appear to improve the effectiveness of the system in ameliorating 30  poverty, and has therefore been advocated by many commentators. However, the reality is that hope has faded that "targeted" social security provision could significantly counteract poverty and contribute to social harnony and the achievement of widespread political support for the transition process. Instead, the importance of the social security system must increasingly be seen in terms of its role in the structure of the state. Social security constitutes a large part of the government's administrative infrastructure and the ability to operate the system is an important indicator of the government's capacity to govern. For all their limitations, existing social security systems provide rules for reallocating resources which are relatively transparent, compared with other areas of continued government intervention in the FSU. Provision of pensions and benefits is a very important part of the government's direct contact with the population, and is the focus for a great deal of political debate and discussion. The importance of maintaining a viable social security system can therefore be seen as part of the task of maintaining a viable state. One implication of this is that social security reform proposals should pay close attention to their institutional and administrative effects. This is particularly important to any attempts to replace categorical benefits with means-tested benefits. When the administrative details are examined, means-testing appears very problematic in its effect on the relationship between the state and the population. Its information requirements are very high, and the incentives for people to conceal their incomes are strong. If local administration is used to surmount the information problem, difficulties in the allocation of funds from central to local government arise. 31  Social security is one part of the institutional legacy of central planning which might be capable of forming part of the institutional basis of a market economy. However, there are many obstacles to reforming existing systems around the dual principles of social insurance and means-tested assistance. Unified categorical systems may be less satisfying to theorists, but more durable in practice. 32  References Ahmnad, E. 1992. "Social Safety Nets." In V. Tanzi, ed.. Fiscal Policies in Economies in Tr-ansition. Washington, DC: International Monetary Fund. Ahmad, E., and K.Y. Chu. 1993. "Russian Federation: Economic Reform and Policy Options for Social Protection." In V. Tanzi, ed., Transition to Market: Studies in Fiscal Reform. Washington, DC: International Monetary Fund. Allan, B. 1994. "Toward a Framework for a Budget Law for Economies in Transition." IMF Working Paper 149. International Monetary Fund, Fiscal Affairs Department, Washington, DC. Barr, N. 1993. The Economics of the Welfare State, second edition. Oxford: Oxford University Press. . 1994. "Income Transfers: Social Insurance." In N. Barr, ed.., Labor Markets and Social Policy in Central and Eastern Europe. World Bank: World Bank and Oxford University. Chu, K.Y., and S. Gupta. 1993. "Protecting the Poor: Social Safety Nets During Transition." Finance and Development, June. Fox, L. 1994. "Old Age Security in Transition Economies." Policy Research Working Paper 1257. World Bank, Policy Research Department, Washington, DC. Holtzman, R. 1991. "Safety Nets in Transition Economies: Concepts, Recent Developments, Recommendations." Chap. 10.2 of The Transition to a Market Economy, vol. 2. Paris: Centre for Cooperation with European Economies in Transition, OECD. 33  International Monetary Fund. 1992. Ukraine: Budgetary Implications of Social Safety Net Option. Fiscal Affairs Department, Washington, DC. . 1993a. Kyrgystan: Social Safety Nets - Reform Options and Sustainable Economic Adjustment. Fiscal Affairs Department, Washington, DC. . 1993b. Moldova. Social Safety Net Options and Budgetary Implications. Fiscal Affairs Department, Washington. DC. International Monetary Fund, World Bank, OECD, and EBRD. 1991. "Labor Market, Social Safety Net, Education and Training." Chap. 4.6 of A Study of the Soviet Economy, vol. 2. Paris: IMF, World Bank, OECD, and EBRD. Kopits, G. 1992. "Social Security." In V. Tanzi. ed., Fiscal Policies in Economies in Transition. Washington, DC: International Monetary Fund. Lui, L. 1993. "Income Security in Transition for the Aged and Children in the Soviet Union and in the Russian Federation." Social Security Bulletin 56(1). McAuley, A. 1991. The Economic Transition in Eastern Europe: Employment, Income Distribution, and the Social Security Net." Oxford Review of Economic Policy 7( 4). . 1994. "Social Welfare in Transition: What Happened in Russia." Policy Research Paper 6. World Bank, Policy Research Department, Washington, DC. Roxburgh, I., and J. Shapiro. 1994. "Employment Incentives of Taxes Based on the Average Wage: Lessons from the Russian Experience." Paper presented at the European Economic Association Meeting, Maastricht, 3-5 September.  Sipos, S. 1994. "Income Transfers: Family Support and Poverty Relief." In N. Barr, ed., Labor Markets and Social Policy in Central and Eastern Europe. Oxford: World Bank and Oxford University Press. World Bank. 1994. Moldova. Moving to a Market Economy. World Bank Country Study, Washington, DC.     Policy Research Working Paper Series Contact Title Author Date for paper WPS1570 Protecting the Old and Promoting Estelle James January 1996 S. Khan Growth: A Defense of Averting the 33651 Old Age Cnsis WPS1571 Export Prospects of Middle Eastern Alexander Yeats February 1996 S. Lipscomb Countries: A Post-Uruguay Round 33718 Analysis WPS1572 Averting the Old-Age Crisis: Robert J. Palacios February 1996 M. Pallares Technical Annex 30435 WPS1573 North-South Customs Unions and Eduardo Fernandez-Arias February 1996 S. King-Watson International Capital Mobility Mark M. Spiegel 31047 WPS1574 Bank Regulation: The Case of the Gerard Caprio, Jr. February 1996 D. Evans Missing Model 38526 WPS1575 Inflation. Growth, and Central Banks: Jose de Gregorio February 1996 K. Labrie Theory and Evidence 31001 WPS1576 Rural Poverty in Ecuador-A Jesko Hentschel February 1996 E. Rodriguez Qualitative Assessment William F. Waters 37873 Anna Kathryn Vandever Webb WPS1577 The Peace Dividend: Military Malcolm Knight February 1996 R. Martin Spending Cuts and Economic Norman Loayza 31320 Growth Delano Villanueva WPS1578 Stock Market and Investment: Cherian Samuel March 1996 C. Samuel The Governance Role of the Market 30802 WPS1579 Different Strategies of Transition Marek Dabrowski March 1996 C. Rollison to a Market Economy: How Do 84768 They Work in Practice? WPS1580 Indonesia's Cocoa Boom: Hands- Takamasa Akiyama March 1996 G. Ilogon Off Policy Encourages Smallholder Akihiko Nishio 33732 Dynamism WPS1581 Where Has All the Education Gone? Lant Pritchett March 1996 S. Fallon 38009 WPS1582 Stock Market Development and Ross Levine March 1996 P. Sintim-Aboagye Long-Run Growth Sara Zervos 38526 WPS 1583 Trade Preferential Agreements in Michael Michaely March 1996 D. Papageorgiou Latin America: An Ex-Ante Assessment 31910 WPS1584 Price Support at Any Price? Costs Witold M. Orlowski March 1996 W. Orlowski and Benefits of Alternative Agricultural 37270 Policies for Poland  Policy Research Working Paper Series Contact Title Author Date for paper WPS1585 Public Finances and Economic Luca Barbone March 1996 C. Pelegrin Transition Hana Polackova 85087 WPS1586 Did External Barriers Cause the Azita Amjadi March 1996 S. Lipscomb Marginalization of Sub-Saharan Ulrich Reinke 33718 Africa in World Trade? Alexander Yeats WPS1587 Payments and Finance Problems Constantine April 1996 M. de la Puente in the Commonwealth of Michalopoulos 31206 Independent States WPS1588 Social Insurance in the Transition Deborah Mabbett April 1996 L. Biely to a Market Economy: Theoretical 36280 Issues with Application to Moldova