World Bank Employment Policy Primer April 2004 I No. 5 UNEMPLOYMENT INSURANCE Efficiency Effects and Lessons for Developing Countries * Unemployment insurance (UI) is the most common the choice between entering regular versus informal public income support program for the unemployed in jobs. The program may also interact with adverse developed countries.1 In these countries, it typically shocks, contributing to the persistence of unemploy- offers good protection: it covers the majority of ment. And by interfering with the allocation decisions of employed persons, irrespective of occupation or indus- economic agents, UI benefits may have a separate effect try, and provides adequate smoothening of consump- on economic growth ­ by influencing enterprise restruc- tion patterns. For example, studies on the U.S. find that turing and layoff decisions, for example. the welfare of benefit recipient households is on average This note reviews these efficiency effects of UI only 3-8 percent lower than the welfare of otherwise (other important effects of UI ­ on consumption identical households (Hamermesh and Sleznick,1995), smoothing, income redistribution, and poverty reduc- and that in the absence of unemployment insurance, tion ­ are beyond its scope; for a recent review, see average consumption expenditures would fall by about Vodopivec, 2004). Because our knowledge of these 20 percent (Gruber,1997). In the last decade, UI pro- effects is based predominantly on the experience of grams have been introduced in transition countries, and developed countries, the note also discusses issues that their use in developing countries is on the rise as well. arise in improving or introducing the system in devel- By providing protection against unemployment oping and transition countries. The functioning of UI risk, UI increases the sense of security among employed system depends heavily on the design features of the workers and provides financial compensation to work- program, which are in turn determined by country-spe- ers who lose their jobs. This protection obviously brings cific conditions. The note ends with a summary and welfare gains, but questions arise: are these gains pro- some policy implications. duced at a cost, and, if so, what are the costs and how large are they? How Does Unemployment Insurance The introduction of a UI program changes the Affect Efficiency? opportunity cost of leisure and that, together with a variety of other channels, may affect employment deci- Effects on unemployment sions, labor-force participation, and unemployment. UI Benefits affect unemployment through two main may influence job-search intensity, post-unemployment channels. First, they influence job-search effort and the wages, the labor supply of other family members, and reservation wage of recipients. This may either prolong *This note has been prepared by Milan Vodopivec and edited by Tim Whitehead. It draws heavily from Vodopivec and Raju (2002), one of the background papers in the Employment Policy Primer Series. The World Bank Employment Policy Primer aims to provide a comprehensive, up-to-date resource on labor market policy issues. The series includes two products: short notes, such as this one, with concise summaries of best practice on various topics and longer papers with new research results or assessments of the literature and recent experience. Primer papers and notes are available on the labor markets website at or by contacting the Social Protection Advisory Service at (202) 458-5267 or by email at . I m p a c t E v a l u a t i o n unemployment spells by making leisure more attractive, or shorten them by providing additional resources and BOX 1: HOW LARGE ARE EMPLOYMENT DISINCENTIVE EFFECTS OF UI? thus enabling more effective job search. Second, unem- ployment benefits improve the bargaining position of The effects are measured by the benefit elasticity (the workers. This leads to higher wages and hence a higher percentage increase in the duration of unemployment equilibrium unemployment (Blanchard and Wolfers, resulting from a given percentage increase in the ben- efit replacement rate), and duration elasticity (which 1999). similarly relates the duration of unemployment to the These ambiguous predictions from theoretical potential duration of benefits). models make empirical studies particularly relevant. According to Layard et al (1991), the benefit elasticity And, by and large, they show that unemployment bene- ranges from 0.2 to 0.9, depending on the state of the fits increase unemployment. (For a recent summary, see labor market and the country concerned. (For example, Calmfors and Holmlund, 2000). Many studies of indi- a 0.6 elasticity means that, in response to a 10 per cent increase in the replacement rate, the unemployment vidual countries (using microdata) find that both a duration increases by six per cent, or one week at the higher level and a longer duration of benefits increase average duration of 17 weeks). According to Katz and unemployment, suggesting the presence of the moral Meyer (1990), the duration elasticity in the U.S. is in the range of 0.4 to 0.5: that is, a one-week increase in hazard problem in the job search of benefit recipients. the potential entitlement duration of unemployment (Direct evidence on the intensity of job search by bene- benefits is associated with one to one-and-a-half-day fits claimants is scarce and inconclusive.) increase in the average unemployment spell of recipi- Because equilibrium unemployment is determined ents). by the average duration of unemployment and the Katz and Meyer (1990) estimate that ­ keeping benefit inflow into unemployment, it can be assumed that the expenditures unchanged ­ increases in potential bene- fit duration have much larger adverse incentive effects increase of unemployment duration also increases equi- on unemployment than do changes in benefit levels. librium unemployment. Typically, cross-country studies Moreover, they suggest that the longer duration of ben- ­ directly investigating the relationship between equilib- efits explains about 10 to 30 per cent of the difference rium unemployment and the generosity of the benefits in mean unemployment spell duration between the U.S. and U.K. ­ corroborate such findings. (See Box 1 on the size of these effects.) True, there are also studies (particularly in Europe) which do not find significant effects of UI on unemployment, but most observers agree that the evi- Such employment disincentives might matter less ­ dence on positive effects is more compelling. and UI's overall efficiency effects could even be reversed The evidence based on microstudies is particularly ­ if it also contributed to a better match between the credible. First, a large number of studies, both of devel- worker and his new employer. Indeed, some studies in oped and transition economies, find a positive elasticity the 1970s (mostly on the U.S.) found that unemploy- of unemployment with respect to the level and duration ment benefits increased post-unemployment wages and of benefits. (For the evidence on developed economies, thus facilitated job matches. More recent studies, how- see Devine and Keifer, 1991, and on transition ever, have discerned a weak or negligible effect. economies, Vodopivec et al., 2003.) Second, disincen- Similarly, the evidence on the decision to enter reg- tives created by unemployment benefits show up clearly ular rather than informal jobs is scant and inconclusive. in a pronounced spike in the probability of exit from Some studies find that the availability of unemployment unemployment just before benefit exhaustion. And benefits significantly reduced the entry into precarious third, strong evidence on moral hazard is provided by jobs, while others find no such effects. Interestingly, evi- the U.S. unemployment-insurance experiments: those dence shows that an increase in the generosity of UI in unemployed who were offered a bonus for fast re- Brazil led to increased participation in the self-employ- employment significantly reduced their unemployment ment sector, suggesting that unemployment benefits spells, without affecting their re-employment earnings helped the unemployed to overcome liquidity barriers (Meyer, 1995). which had previously prevented them from becoming self-employed (Cunningham, 2000). 2 I m p a c t E v a l u a t i o n Apart from increasing the equilibrium level of BOX 2: EXPLAINING THE RISE OF EUROPEAN unemployment, unemployment benefits also make UNEMPLOYMENT BY THE INTERACTION OF unemployment more persistent. That is, economies with INSTITUTIONS AND SHOCKS unemployment benefits experience larger and more prolonged unemployment following a transient shock. A stumbling block for proving that institutions (and UI Theoretic models argue that the 'non-UI' economy in particular) affect aggregate unemployment has been the lack of empirical support for such a link in explain- recovers more rapidly as reservation wages adjust more ing the rise of European unemployment over the last quickly and job-search intensity is higher than in the several decades. The same institutions existed when `UI' economy. The interaction of institutions with unemployment was much lower, and their changes adverse shocks also offers an explanation for the rise of alone cannot explain the persistent rise in the average unemployment rate in European economies. European unemployment in the last several decades. (See Box 2). Recently, Blanchard and Wolfers (1999) offered an explanation based on the interaction between shocks and institutions. They show that the impact of a shock Effects on labor force participation on the persistence of unemployment can be amplified The effects of unemployment benefits on labor force by a more generous UI system and higher employment protection (including more generous severance pay). participation are not well researched. Some studies find More generous UI and employment protection change that the existence of UI in OECD countries attracts spe- the nature of unemployment: they increase the average cific groups ­ for example, women and older workers ­ duration of unemployment and thus increase the num- into the labor force. With the increase of the fringe ben- ber of the long-term unemployed. Moreover, Blanchard and Wolfers argue that the long-term unemployed who efits, these groups find labor force participation more are not searching for a job do not matter for wage for- valuable - the so-called `entitlement effect.' But there is mation ­ they do not exert enough pressure on wage conflicting evidence as to whether the entitlement effect moderation ­ and so do not contribute to a wage increases employment. Some studies find that the adjustment after an adverse shock. According to the authors, the effect works through two channels: the increase in labor-force participation and the increase of skills of long-term unemployed are rusty and they look unemployment cancel each other, with no net effect on for a job less intensely (duration dependence effect), employment. For the U.S., however, there is evidence and the employers view the long-term unemployed as more risky and are therefore reluctant to hire them in that benefits increase both employment and unemploy- the presence of firing costs (marginalization effect). ment rates. Particularly in the context of developing countries, it is also important to realize that uninsured transient shocks may reduce individual consumption below a threshold needed to retain productivity and Effects on output and growth labor force attachment, and may thus give rise to The effects of UI on output and growth have not "dynamic poverty traps" and lead to chronic poverty been well researched, let alone quantified. The predic- (Ravallion, 2003). tions of theoretical models about the effects on output UI also affects the labor supply provided by family conflict. Some argue that, by mitigating their greater members of benefit recipients. In theory, more generous unemployment risk, UI programs can support the cre- benefit rates suppress the labor supply of other family ation of high-quality, high-wage jobs. Thus, UI may help members because they reduce the family's need for the economy achieve higher output. More generous ben- replacement income, the so-called `income effect.' efits are also expected to help workers acquire more spe- Empirical evidence confirms such predictions. For cialized skills, which may enhance efficiency. Other example, research in the U.S. shows that the labor sup- researchers argue that government-mandated programs ply of wives of unemployed workers is responsive to may break down the social fabric that maintains private unemployment benefits received by their husbands: a $1 transfers and reduce private transfers by more than the increase in the unemployment benefits of a husband amount of the publicly provided insurance benefits, reduces the earnings of his wife by 36 cents. thereby hurting the efficiency of the economy. Similarly, the effects of benefits on growth are not well documented, but empirical studies suggest they are 3 I m p a c t E v a l u a t i o n insignificant (Nickell and Layard, 1999). One way that BOX 3: DOES EXPERIENCE RATING REDUCE UI might influence growth is by encouraging labor real- THE RISK OF UNEMPLOYMENT? location and, in particular, enterprise restructuring. Par- tial-equilibrium results suggest this to be the case. For In the U.S., employer contributions depend on the employer's layoff experience: employers who lay off the U.S., there is considerable empirical evidence that workers more frequently and thus impose a heavier the availability of benefits strongly increases the proba- financial burden on the system are assigned a higher bility of temporary layoffs, but not of quits and perma- rate (this is called experience rating). Does this prac- tice reduce the incentives on the part of employers to nent layoffs. Apparently, when considering temporary layoff workers, and, in general, reduce the risk of layoffs, employers do take into account the availability of unemployment? UI. By imposing additional costs on employers, some These partial-equilibrium results do not necessarily researchers show that experience rating curbed layoffs carry over to the general-equilibrium framework. Theo- and thus increased employment. But more general and retical modeling does not support the argument that complex models show that the effect on employment is ambiguous ­ namely, by increasing labor costs, experi- restructuring could be facilitated by more generous ence rating reduces the number of workers the firm is unemployment benefits: while such benefits might add willing to hire. In a similar vein, models which incor- to the attractiveness of restructuring, they also hinder porate job creation and job destruction show that the job creation because they make it more costly (Blan- transition of the current U.S. system to one of full- experience rating would discourage layoffs, but only by chard, 1997). Similarly, in the context of a job cre- a relatively small amount. Because job creation would ation/job destruction model, an increase in the UI pro- also be adversely affected, the net effect could even be gram's income-replacement rate reduces job creation a small increase in the unemployment rate. The effects of experience rating thus show primarily as a reduction and thus aggregate output (Mortensen, 1994). There- of flows to and from unemployment, but not as an fore, the overall potential of income support programs increase in employment. It has to be emphasized that to spur enterprise restructuring is likely to be limited. experience rating is more important when employment It is also interesting to note that experience rating ­ a protection is low, as it is in the U.S.; in the European context, employment protection legislation takes over method of financing of UI which `penalizes' employers the role of experience rating ­ with similar effects on for laying off workers by imposing larger contribution labor market flows and employment. rates on them ­ reduces inflows to and outflows from unemployment. Experience rating, therefore, discour- ages the very restructuring that might make an economy higher equilibrium unemployment (evidence from more productive. At the same time the evidence is that cross-country studies) ­ although the magnitude of experience rating does not increase employment nor such effects is not firmly established. (See an evaluation reduce unemployment. (See Box 3). summary in Table 1). Benefits also contribute to the per- Finally, let us emphasize that UI acts as an automat- sistence of unemployment. Their effects on restructur- ic macroeconomic stabilizer. Unemployment benefits ing and growth are less researched and are probably not soften the impact of adverse shocks on GDP ­ and, by significant, except for the positive effects arising from the same token, they also restrain expansion when the the macroeconomic stabilization function of the bene- economy starts growing again. During downturns, fits. The evidence on some other effects - for example, unemployment benefit payments increase and UI taxes on the impacts of benefits on post-unemployment fall, and the net injection of purchasing power moder- wages ­ is also inconclusive. There seems to be mount- ates the severity of the contraction. During upturns, ing and persuasive evidence of the negative efficiency however, UI taxes increase and UI benefits decrease, effects of UI, much more so than of the positive effects. restraining the expansion. Empirical evidence in North However, the magnitude of these effects has not been America shows that UI reduces GDP losses during precisely determined and is undoubtedly influenced by a downturns by 10-15 percent (Chimerine et al., 1999). host of country-specific features. (See Box 4.) To summarize, unemployment benefits increase the duration of unemployment spells of recipients (evi- dence from single-country studies) and contribute to 4 I m p a c t E v a l u a t i o n TABLE 1: EFFICIENCY EFFECTS OF UNEMPLOYMENT INSURANCE INDICATOR EFFECTS A. UNEMPLOYMENT Job-search effort Significant disincentives for leaving unemployment (moral hazard problem) Equilibrium unemployment Benefits increase the equilibrium unemployment rate Post-unemployment wages Inconclusive evidence on the improvement of job matching (via post-unemployment wages) Encouragement of taking regular vs. · Inconclusive evidence on entry into precarious jobs informal jobs · In Brazil, benefit payments increase probability to enter self-employment Persistence of unemployment Benefits slow down adjustment to shocks ­ make unemployment more persistent (European unemployment) B. LABOR FORCE PARTICIPATION Employment, unemployment For some groups benefits increase their labor force participation, and inactivity but reductions in inactivity primarily show up as increases in unemployment Labor supply of other family members Reduces labor supply of the spouses of unemployed workers C. OUTPUT AND GROWTH Overall effects · Theoretical predictions about the effects on output inconclusive · The effects on growth insignificant Enhancing restructuring of enterprises · Attractiveness of restructuring increases; in the U.S., and overall adjustment strong evidence on the increase of temporary layoffs · Because job creation is hindered, overall adjustment not assisted Macroeconomic stabilization By acting as automatic macroeconomic stabilizer, UI reduces GDP losses during downturns by 10-15 percent extent monitoring of continuing eligibility requirements How to Account for Specific is ignored and leakages are allowed, it may add signifi- Conditions of Developing Countries? cantly to the overall expenditures of the system, and thus The functioning of UI programs is heavily colored also indirectly affect unemployment. by country-specific conditions. Specific features prevail- Administrative capacities clearly vary by developing ing in developing or transition countries may affect the country. In the Philippines - a country which does not performance of UI programs. have a UI program but is contemplating its introduction - none of the offices of the public employment service Weak administrative capacity come close to the capacity necessary for checking a UI requires strong administrative capacity to moni- claimant's labor-force status and job-search efforts. Even tor its sophisticated initial and, particularly, its continu- in more developed regions, the majority of these offices ing eligibility requirements. Other things equal, the have only one worker, and the offices are active only in stricter the monitoring of the behavior of the recipients, peak periods (Esguerra et al, 2001). While the adminis- the smaller the employment disincentives created by UI. trative capacity in upper-middle-income countries is Because developing countries have a weaker monitoring more sophisticated, it still lags significantly behind the capacity, a UI program is likely to produce more disin- capacity of developed countries. In Argentina, for exam- centives than it does in developed economies ­ or, to the ple, the capacity for screening the initial eligibility for UI 5 I m p a c t E v a l u a t i o n BOX 4: WHY A CONSENSUS HAS NOT BEEN REACHED Large share of underemployed workers ON THE EFFECTS OF UI? In developing countries, the existence of benefits may attract the underemployed ­ workers who cannot Why, despite a wealth of studies, do disagreements afford to stay out of work and thus take low-wage, low about the (size of) efficiency effects of UI persist? Here is a partial list of explanations: productivity jobs, often with short working hours ­ into insured unemployment. It is likely that, in choosing First, theoretical and empirical results are often of par- tial equilibrium nature, and such results may or may between relying on unemployment benefits or taking a not be validated in a general equilibrium framework, temporary, low-paid job, many such workers would pre- (i.e., when more complex interactions are accounted fer unemployment benefits, which means that insurance for). There are many institutional and program features would reduce their incentives to work ­ and thus to self- with rich possibilities for interaction which could ­ and often should ­ be incorporated in a general equilibri- protect. This is consistent with the notion that the tran- um model. Leaving them out may be responsible for sition from underemployment to open unemployment the different results of various models. can be partly explained as an income effect: as house- Second, empirical results are often country-specific hold incomes rise, it becomes possible for some workers and so generalizations may be difficult. For example, to go through periods without work while waiting for a while most studies find a positive relationship between job to open (Cox Edwards and Manning, 2001). Unless benefits and the duration of recipients' unemployment spells, estimates of the magnitudes of these effects longer unemployment spells produce larger post-unem- vary. Quite likely, some of these differences are pro- ployment wages or otherwise increase the quality of a duced by differences in the effectiveness of monitoring job match, the efficiency effects of the postponed exit and enforcement of job search across countries. If job search requirements and work tests are strictly from unemployment are negative. enforced and benefits withdrawn when job offers are rejected, the generosity of benefits is less important High political risk and the moral hazard problems less pronounced. One UI is susceptible to political risk: cost pressures can reasonably expect that the strictness of monitor- ing as well as the enforcement severity of the work test often lead to the reduction of the benefit level, and gov- vary among countries, but these features of benefit ernments are tempted to increase the benefit level to systems, being hard to measure, are inadequately con- appease specific groups (for example, older workers). trolled for in empirical research. The typically high exposure of developing countries to Third, interpretations of different authors vary. While such risk therefore renders the program less effective some may consider certain effects as being small and and exacerbates its negative efficiency effects. UI is insignificant, other may interpret the same results as important. found to be susceptible to political risk because (i) redis- tribution of program funds is highly non-transparent; (ii) being a social insurance, the program may be per- ceived as a `social contract' and may exhibit `natural per- is adequate (the existing capacity of other social protec- sistence', and (iii) it is publicly managed and its liabilities tion programs has been used) but ten years after UI's are largely unfunded. See Vodopivec, 2004. introduction the country has still not acquired effective capacity to monitor continuing eligibility. (See Box 5.) High incidence of private transfers When introducing a public income-support pro- Large informal sector gram, its effects on existing private transfers should be The more informal an economy, the more abundant considered. Simple analyses that do not account for and diverse are the opportunities for undeclared paid these transfers exaggerate the effectiveness of public work, and thus the higher are the costs of monitoring insurance programs. The expansion or introduction of eligibility requirements. Given the large size of the infor- public income-support programs could lead to cutbacks mal sector in developing countries, the costs of effective in private transfers. This risk is particularly true for pro- monitoring of the continuing eligibility of benefit recip- grams where the main beneficiaries are the non-poor, ients - particularly their availability for work ­ are likely whose transfers are more responsive to income shocks. It to be large. has been shown that, if the introduction of public insur- 6 I m p a c t E v a l u a t i o n ance breaks down the habit of self-help among local BOX 5: ADMINISTRATIVE HURDLES AND THE communities, replacing private transfers by social insur- INTRODUCTION OF UI IN ARGENTINA ance is welfare-reducing because it reduces overall insur- ance possibilities (Attanasio and Rios-Rull, 2000). Argentina introduced its UI system in 1992, following a macroeconomic crisis that raised the fear of large- It should also be remembered that the nature of scale, open unemployment. The number of recipients unemployment has an important bearing about the is relatively small ­ on average, about 100,000 to desirability of a public insurance program. As argued by 125,000 workers receive benefits, out of two million officially unemployed workers. Administration of the Gill and Ilahi (2000), if unemployment spells are more program (the processing of claims and payment of ben- frequent and shorter, self-insurance measures may be efits) was handed over to the social security system more appropriate; on the other hand, less frequent and (ANSES ­ Administración Nacional de la Seguridad longer unemployment spells speak in favor of public Nacional) which operates a national network of offices and reports to the Ministry of Labor and Social Securi- insurance programs. ty. Workers register at and receive payments from one of 150 local ANSES offices; there are no job placement Summary and Policy Options or other re-employment services provided. Thanks to its wide risk-pooling, UI enables a high While the administration of benefits has seemingly degree of consumption smoothing for all categories of proceeded smoothly - workers are informed of their eli- workers, thus offering a high degree of protection. But gibility and receive payments on a timely basis - only recently could the system detect recipients who had the above review shows that UI typically brings various found new jobs in the formal sector. And it still applies negative effects on efficiency that most likely outweigh only a few measures to prevent the leakage of benefits its positive efficiency effects. By affecting job-search to those who have found jobs in the informal sector. Through a newly introduced system of common per- intensity and wage bargaining, benefits are found to sonal identification numbers, the government has been prolong unemployment spells and increase equilibrium able to cross-check whether UI recipients are also on unemployment. In addition, by interacting with shocks, social security payment rolls. (Personal identification benefits contribute to the persistence of unemployment. numbers were introduced in 1994, and it took several years to develop this cross-checking capability.) This Some other hypothesized effects ­ for example, on way, significant numbers of benefit recipients actually employment, labor-force participation, and output and working in the formal sector have been purged from growth ­ are less clear and are likely insignificant. On the benefit-receipt lists. However, a far greater num- the positive side, UI acts as an automatic macroeconom- ber of recipients is likely to be working in the informal sector, with no measures being taken to detect them ic stabilizer. and to take their benefits away. The performance of UI depends on the design of the Source: Mazza (1999). program ­ and also on country-specific conditions. Because its smooth and successful performance relies on a strong administrative capacity to monitor program eli- gibility, an informal sector of modest size, a low inci- problems inherent to UI, programs such as UI Savings dence of underemployment, and low political risk ­ Accounts or its variant, the Integrated UI System, may conditions that are typically lacking in developing and be promising options, particularly for countries where transition countries - the case for UI's introduction in initial conditions seem to be especially suitable (see these countries is less compelling than it is in developed Vodopivec, 2004, for the description of these two pro- countries. Particularly in developing countries, its exis- grams). East Asia and Latin America, where the existence tence may also reduce incentives for self-protection and of severance pay programs may ease the transition, may break down the habit of self-help within local commu- also include candidates. And the governments should nities. also foster self-protection and self-insurance mecha- Introducing UI may thus be viewed as a possible nisms, particularly by promoting well-functioning longer-term goal for many developing countries. In pro- labor, insurance, and financial markets (de Ferranti et al, tecting workers from the risk of unemployment, coun- 2000). tries also have to consider other programs and policies. For example, by avoiding or reducing the moral hazard 7 I m p a c t E v a l u a t i o n Hamermesh, D. S., and D. T. Slesnick (1995)."Unemployment References Insurance and Household Welfare: Microeconomic Evi- dence 1980-93." NBER Working Paper No. 5315. Attanasio, O., and J. V. Rios-Rull (2000). "Consumption Katz L. F., and B. D. Meyer (1990). "The Impact of the Poten- Smoothing in Island Economies: Can Public Insurance tial Duration of Unemployment Benefits on the Duration Reduce Welfare?" European Economic Review 44(7) of Unemployment." Journal of Public Economics 41:45-72. (June): 1225-1258. Layard, R., S. Nickell, and R. Jackman (1991). Unemployment: Blanchard, O. (1997). The Economics of Post-Communist Tran- Macroeconomic Performance and the Labour Market. New sition. Oxford: Oxford University Press. York: Oxford University Press. Blanchard, O., and J. Wolfers (1999). "The Role of Shocks and Mazza, J. (1999). "Unemployment Insurance: Case Studies Institutions in the Rise of European Unemployment: The and Lessons for Latin America and the Caribbean." Inter- Aggregate Evidence." NBER Working Paper No. 7282. American Development Bank, RE2/SO2, Technical Study. Calmfors, L., and B. Holmlund (2000). "Unemployment and Meyer, B. D. (1995). "Lessons from the U.S. Unemployment Economic Growth: a Partial Survey," Swedish Economic Insurance Experiments." Journal of Economic Literature Policy Review 7(1): 109-153. 33, March: 91-131. Chimerine, T., T. S. Black, and L. Coffey (1999). "Unemploy- Mortensen, D. T. (1994)."Reducing Supply-side Disincentives ment Insurance as an Economic Stabilizer: Evidence of to Job Creation." In Federal Reserve Bank of Kansas City, Effectiveness Over Three Decades." Unemployment Reducing Unemployment: Current Issues and Policy Insurance Occasional Paper 99-8, Washington, D.C.: U.S. Options, Proceedings of a symposium, Jackson Hole, Department of Labor. Wyoming, August 25-27, 1994. Cox Edwards, A. C., and C. Manning (2001). "The Economics Nickell S., and R. Layard (1999). "Labor Market Institutions of Employment Protection and Unemployment Insur- and Economic Performance." In Ashenfelter and Card, ance Schemes: Policy Options for Indonesia, Malaysia, the eds., Handbook of Labor Economics, Vol.3. North Holland, Philippines, and Thailand." In G. Betcherman and R. 1999. Islam (eds.) East Asian Labor Market and the Economic Ravallion, M. (2003). "Targeted Transfers in Poor Countries: Crisis: Impacts, Responses, and Lessons. Washington, D.C.: Revisiting the Trade-offs and Policy Options." World The World Bank. Bank, Policy Research Working Paper 3048. Cunningham, W.V. (2000). "Unemployment Insurance in Vodopivec, M. (2004). Income Support Systems for the Unem- Brazil: Unemployment Duration, Wages, and Sectoral ployed: Issues and Options, The World Bank, forthcoming. Choice." World Bank, processed. Vodopivec, M., A. Woergoetter, and D. Raju (2003). "Unem- De Ferranti, D., G.E. Perry, I. S. Gill, and L. Serven (2000). ployment Benefit Systems in Central and Eastern Europe: Securing Our Future in a Global Economy. World Bank A Review of the 1990s." The World Bank, Social Protec- Latin and Caribbean Studies, Washington, D.C.: World tion Discussion Paper No. 0310. Bank. Devine T. and N. Keifer (1991). Empirical Labor Economics: 1 Unemployment insurance is typically a mandatory, contribution-based program provid- The Search Approach. Oxford University Press. ing compensation for the loss of a job. It often covers the majority of employed persons, Esguerra, J., M. Ogawa, and M. Vodopivec, 2001. "Options of the most notable exception being the self-employed. Benefit levels are earnings-related Public Income Support for the Unemployed in the (usually with predetermined minimum and maximum amount) and the duration of enti- Philippines." Paper prepared for the World Bank-ILO- tlement often depends on the beneficiary's employment history. To qualify for the benefit, JMOL-PDOLE seminar on "Labor Market Policies: Their the unemployed must satisfy the minimum covered employment or contribution require- Implications for East and South East Asia,"Manila, March ment and should not have been fired from the job; quitters are often disqualified or they 2001. have to observe a waiting period. To keep the benefit, recipients must be capable of, Gruber, J. (1997)."Consumption Smoothing Effects of Unem- searching for, and available for work. Non-compliance with other labor market require- ployment Insurance."American Economic Review 87: 192- ments (such as willingness to participate in vocational training or public works) can also 205. result in the permanent or temporary suspension of benefits. 8