Unemployment and other non-employment benefits.
In 1994 there were almost 35m people unemployed in the OECD area (a rate of 8.5 per cent), an increase of 10m over the 1990 level (OECD, 1994a). Behind this statistic lie countless tales of personal hardship and deprivation as well as, in the more aseptic language of economics, a massive misallocation and underutilization of resources.
The hardship caused by sudden income loss following unemployment is mitigated to a greater or lesser extent by unemployment and related social benefits. But among the costs of making life less uncomfortable for those looking for employment is that it becomes easier to adapt to life without a (legal) job, potentially leading to an increase in the numbers who rely on benefits as a permanent source of income. Helping the unemployed, so it is argued, begets more unemployed.
In popular discussion, `unemployment' is often treated as synonymous with `being out of work and claiming unemployment benefits'. Technically, however, this is far from true. Unemployment benefits are only paid to some of those of working age without work. Non-employment consists of unemployment plus those not in the labour force. Often considered as a marginal part of the welfare system, there are in fact more than 20m people in the OECD who are in receipt not of unemployment benefit, but of the two other main types of non-employment benefits - early retirement on a public pension and invalidity. Table 1 - which does not include information on all early retirement schemes and excludes sector-specific invalidity benefit insurance schemes - shows that even so, in around half of OECD countries, there are more people in receipt of invalidity benefits (IVB) than are registered as unemployed, a proportion which is even higher when both early retirement and invalidity benefits are considered. Furthermore, absences from work because of illness (usually giving rise to sickness benefit entitlement) can exceed over 5 per cent of those in employment.
[TABULAR DATA OMITTED]
The labour-market effect of these non-employment benefits is important. Labour-force participation rates for men, notably older workers and the less-educated, have declined sharply, and it would seem that many persons capable of work have left the market, relying on such non-employment benefits and/or work in the shadow economy to provide income.
This article explores the impact of public benefits on unemployment and non-employment. It first documents the evolution of the various aspects of unemployment benefit systems in OECD member countries and shows that `generosity' has increased markedly since the early 1960s. The following section presents the salient features of early-retirement, invalidity, and sickness benefit systems. The focus is on replacement rates in the various systems and whether they have influenced the differing use made of the systems across countries. Finally, pooled cross-country time-series equations are employed to examine the relationship between unemployment and non-employment on the one hand and unemployment and other non-employment benefits on the other.
II. UNEMPLOYMENT BENEFITS
(i) Impact on the Labour Market: The Principles
It is widely accepted that the equilibrium unemployment rate is affected by the generosity of unemployment benefits (see Layard et al., 1991; Bean, 1994, for a more extended discussion of the theories). A variety of theories predict such an outcome. The theories fall into two broad groups: those which suggest that benefits alter the behaviour of people actually receiving them; and those which suggest that the existence of unemployment benefits also alters the behaviour of individuals and employers even when not personally experiencing unemployment.
An obvious example of the former type of effect is that the availability of generous financial support for the unemployed makes joblessness less painful, thus allowing a longer job search. The cushioning of income loss weakens the adjustment capacity of the labour market, underpins unrealistic wage expectations (`reservation wages') and postpones needed occupational and geographical change to find gainful employment. These effects can be particularly strong for workers displaced from sectors or enterprises which offer wage premia owing to rent sharing.
Unemployment benefits can also influence wage-setting at large. This effect is emphasized in the wage-bargaining literature. In these theories unions are assumed to maximize utility for their members by trading off wages and employment, knowing that higher wages will have adverse effects on unemployment. When unions know that their members, who are at risk of losing their jobs if wages are bid up, can maintain living standards by drawing on unemployment benefits, they are likely to opt for higher wages and pursue less `responsible' wage demands. Furthermore, in the absence of `experience rating' of employers' social security contributions, employers' behaviour towards lay-offs will be altered - the unstable employment patterns of firms are in effect being subsidized.
However, there are offsetting potential beneficial effects from high unemployment benefits. Longer search periods may result in a better eventual match, which would tend to reduce labour turnover and thus frictional unemployment. This possible beneficial effect must, of course, be set against the risk of longer unemployment duration reducing employability, as the unemployed may lose basic skills and employers use duration of unemployment as a screening device in hiring decisions.
Unemployment benefits may not only influence the level and duration of unemployment and work motivation, but also the level of labour-force participation. The availability of such benefits is likely to encourage people to join the labour force, the reason being that the expected financial reward from participation is increased. The people most strongly influenced by such considerations are likely to be those who face the greatest risk of becoming unemployed, including persons in bad health and persons with no strong desire to have stable jobs. To the extent that such groups influence wage determination, the increase in labour supply will put downward pressure on wages and increase employment. However, if the increases in the supply of labour of such `marginal' workers has little influence on wage-setting, benefit-induced increases in participation rates will be reflected in higher unemployment.
(ii) Unemployment Benefits in Practice:
Evolution over Time
The effects of unemployment benefits cannot be ascertained by reference to the obvious summary statistic of the amount of money paid to the unemployed in a given time period. Unemployment benefit systems are multi-dimensional and impossible to characterize with a single indicator. The systems differ in the degree of coverage, the ease with which entitlement to benefits is attained, the generosity of financial support, the length of benefit periods, and the enforcement of work availability requirements. As discussed in Atkinson and Micklewright (1991), all the different aspects of the benefit systems are important in assessing their impact on the labour market. OECD (1988, 1991) contained a detailed description of the various dimensions of unemployment support systems in the late 1980s in most OECD countries. In the context of the work on the OECD Jobs Study (1994b), data on benefit systems have been collected going back to the early 1960s.(2) These data are the principal source for the discussion below.
Coverage of unemployment benefit systems
By the early 1960s unemployment insurance (UI) coverage had already expanded well beyond full-time employees in industry, the group which was often the first to be entitled to such benefits. Changes in legislation, notably in the late 1960s and the early 1970s, increased the coverage still further in several countries. This often involved extending the coverage to low-risk groups, such as civil servants and high-income earners, or to relatively small groups, such as domestic servants and casual labour. Other changes included making workers in the primary sector eligible for insurance benefits (Canada in 1971 and Norway in 1977), extending the coverage to part-time workers (Germany in 1973), and removing discrimination between primary breadwinners and working spouses in the insurance system (France in 1969, Ireland in 1973,(3) and The Netherlands in 1985). The switch to obligatory unemployment insurance in Japan in 1974 and in Switzerland in 1977 increased sharply the share of employees eligible for benefits, and coverage has also increased in the Nordic countries (Sweden, Denmark, and Finland) where insurance is still voluntary.(4)
Most countries have also made financial support available to job-seekers not entitled to unemployment insurance benefits, either because of insufficient work history or insurance-benefit exhaustion. Several countries provide unemployment assistance benefits for insurance benefit exhaustees, the duration of which may or may not be limited. As the final safety net, in most countries social welfare benefits are usually available to prevent hardship resulting from insufficient resources due to joblessness. However, to make the social safety net more comprehensive and transparent, several countries have established guaranteed minimum income (GI) schemes for all families. Thus, GI schemes were set up in The Netherlands in 1963, in Finland in 1971, in Belgium in 1975, and in France in 1989. All the different non-insurance benefits are, however, means-tested based on family income and wealth, so that unemployed persons with a working spouse and/or with savings will not normally be eligible for such support.
Because of means-testing, many long-term unemployed and first-time job-seekers should not be entitled to any support. However, although different data give a somewhat different picture (OECD 1994b, Tables 8.4 and 8.5), benefit coverage rates seem to be very high in northern Europe, suggesting that means-testing is not disqualifying large numbers of the unemployed, which in turn partly reflects long effective maximum UI benefit periods (see below). The coverage rate is lower in southern Europe because those who are not entitled to unemployment benefits have limited access to other support, Greece and Italy have no schemes for the non-insured. The absence of such schemes in the United States (except for some families with children) and Japan (together with short UI benefit duration) also explains the low coverage rate in these countries.
Microeconomic studies have highlighted the importance of the maximum length of benefit periods in influencing the duration of unemployment spells. Outflow rates are found to increase considerably at around the time that benefits are exhausted in the United States,(5) Canada,(6) and Japan.(7) Although there are some exceptions to such findings in European studies,(8) clear spikes at the time of benefit exhaustion have been detected in several countries, including France,(9) Spain,(10) and Sweden.(11) The possibility of maximum benefit periods influencing the duration of unemployment spells has attracted considerable attention, given that the difference in unemployment across countries reflects primarily differences in the average duration of spells rather than differences in inflows.
There has long been a large dispersion in the maximum length of support periods across OECD countries, but benefit periods have tended to become longer (Figure 1). The UI benefit period has remained unlimited in Belgium for decades, while France changed from having an indefinite benefit period to having a limit in late 1970s. Most of the countries in Europe, which have traditionally limited benefit periods, have extended the maximum length since the early 1960s, whereas the maximum length was similar in the early 1960s and the early 1990s in the big non-European OECD countries (United States, Japan, and Canada). As a result, benefit periods have become much longer in Europe than in the non-European countries. In addition, maximum unemployment benefit periods can be significantly longer than those stipulated for UI. Thus, while both Germany and Austria have relatively short UI benefit periods, UI benefit exhaustees have a long-standing right to open-ended assistance benefits. Other countries have introduced unemployment assistance schemes over the 1980s: France established a scheme of unlimited duration in 1987, and Greece, Portugal, and Spain set up time-limited programmes (1.5 years in Spain, a year in the other two countries .
The standard UI benefit period also understates the effective duration of benefits in some European countries because of the possibility to renew entitlement rights through participation in so-called `active' labour-market programmes. This has been extensively used in Denmark where, as from 1979, the unemployed had the right to get a job offer before the end of the standard benefit period, and the minimum duration of the offer was set with a view to renewing rights to UI benefits. Although the system was tightened subsequently, benefit periods were virtually unlimited until the system was changed in 1994. Extending benefit periods in this way has also been possible in Sweden, the unemployed having the right since 1987 to a public job offer of a length which would result in requalification.
The evolution of standard UI benefit periods often masks the fact that benefit periods have become substantially longer for older workers (Figure 2). In some cases such extension has defacto turned UI benefits into early-retirement benefits, as search and work-availability requirements were often weakened as well. Thus, in the late 1970s The Netherlands and. Denmark extended benefit periods for those aged 58 and 57 or more, respectively, until either early retirement conditions are satisfied or the legal retirement age is reached, and the normally time-limited assistance benefits in Spain were made available until retirement for those over 52. Other countries have increased benefit periods for older workers without turning them into early-retirement benefits. In Germany workers 45 and over became eligible for extended benefit periods in 1987, in Greece benefit periods were extended for the over-55s in the mid-1980s, and in France special rules were applied to older workers when benefit periods became de jure limited in 1979.
One dimension of the generosity of UI (and GI where there are no UI systems) is the replacement rate. Initial replacement rates (i.e. those in the first 3 months of unemployment) have tended to increase over the 1961-91 period. Thus, the simple average over the countries in Table 2 (excluding Portugal and Spain) rose from 41 to 48 for workers with the characteristics described in the footnote to the table, and from 51 to 56 for workers with two-thirds of average income. This general tendency masks important differences across countries. A number of countries, including the four biggest countries in the OECD, have kept replacement rates broadly stable since the early 1960s. The generosity has increased notably in Norway, Denmark, and Sweden, the latter two providing the highest replacement rates in the OECD to lower-wage workers.(12) A few countries have lower replacement rates at present than in the early 1960s. New Zealand, which had the highest replacement rate in industrialized countries in the 1960s, has lowered it to well below the OECD average, and some trimming has also taken place in The Netherlands. Several high unemployment countries, including Denmark, Spain, The Netherlands, and Ireland, reduced replacement rates in the 1980s (but in the first two countries only for high-wage earners who are at lower risk of becoming unemployed).
[TABULAR DATA OMITTED]
In countries where unemployment insurance is supplemented by other support schemes, the benefits in the latter are typically means-tested fixed amounts rather than earnings-related, as is often the case with UI. This implies that the generosity varies more strongly in these alternative schemes than in UI (the earnings-related unemployment assistance schemes in Germany and Austria being an exception). In several countries, listed in Table 3, the amount of unemployment assistance (UA), i.e. financial support for UI benefit exhaustees, or guaranteed income (GI) comes close to the level of UI benefits for low-income workers. This tendency for levelling the different benefits, which has become more marked since the mid-1970s, has arguably reduced the importance of UI benefit exhaustion for low-income workers. UI benefit exhaustion would nevertheless still have important consequences if means-testing were to disqualify the unemployed from UA and GI benefits. Unfortunately, little is known about the evolution of generosity in social welfare systems run by local authorities, but available evidence suggests that the generosity is relatively high in the Nordic countries.
[TABULAR DATA OMITTED]
The benefit indicators in Tables 2 and 3 refer to standard benefits available to the unemployed, and exclude many non-standard benefits. For example, the unemployed may become entitled to housing benefits which could lift the replacement ratio considerably and the standard benefit can be supplemented in some countries (e.g. France and Spain) with free health-care coverage for beneficiaries.
Differences in tax treatment of benefits across countries and across time within countries imply that the pattern of net replacement rates can differ from that of the gross replacement rates discussed above. For example, where UI benefits are not taxed like earned income (e.g. Germany, Austria, and Spain), the gross replacement rate will understate the generosity of the system relative to countries where such benefits are taxed. Also, generosity net of tax may remain unchanged despite increases in the gross replacement ratio if benefits are made subject to income tax at the same time (e.g. the Nordic countries in the 1970s). Unfortunately, detailed data are not available on net replacement rates across countries and across time. First attempts to develop such data, reported in OECD (1994b), suggest that the evolution of net replacement rates follows the broad trends in the gross replacement rate over time,(13) whereas the ranking of countries is somewhat sensitive to whether gross or net replacement rates are used.
Availability and willingness-to-work requirements
Recent studies have stressed the importance of enforcing availability and willingness-to-work requirements in order to put pressure on unemployed job-seekers and to reduce moral-hazard problems. Applying pressure is seen to be necessary to keep the unemployed in contact with the labour market and to counter tendencies for the unemployed to adjust their lifestyles to unemployment and cash benefits. Moral-hazard problems arise because the insured workers themselves can influence the probability of entering joblessness and remain out of work. Such problems, which are present in all types of insurance, are likely to be particularly acute in unemployment insurance, as it is inherently difficult to monitor whether unemployed individuals are actively searching for a job or not. This is the prime reason why private insurance for unemployment is generally not available (Barr, 1992).
The main method of checking availability has traditionally been registration with the Public Employment Service. In the 1960s most countries required the unemployed to sign on in person at employment offices every week or even several times a week (OECD, 1991). However, increased workload in employment offices in the wake of higher unemployment has significantly reduced the frequency of personal contacts between benefit officers and the jobless, several countries replacing such signing-on requirements altogether with an infrequent postal procedure. Rediscovery of the moral hazard problem has led to intensified interviews with the unemployed, notably the long-term unemployed, being increasingly used to check whether requirements are fulfilled, but such interviews are typically conducted after several months, or even years, of joblessness. Such in-depth interviews resulted in a number of persons losing their rights to benefits in the United Kingdom (OECD, 1993b) and in France (OECD, 1993a) because of insufficient search activity.
Countries impose different standards with respect to the right of beneficiaries to reject job offers. In general, the unemployed have, early on in a spell, a large element of discretion. However, it has become more common to reduce the right to this discretion once an unemployment spell has gone beyond a certain limit, with beneficiaries expected to accept `suitable' job offers. But the definition of a suitable job is generally imprecise, and will often vary from one benefit office to another. Even if non-availability is detected, the sanctions in some countries may be minimal and insufficient to act as an efficient deterrent. For example, an unfounded refusal leads only to a temporary reduction in unemployment benefits (e.g. 20 or 40 per cent for 1-6 months) in The netherlands. In contrast, most countries stop or postpone any payments when non-availability is detected.
The limited evidence available suggests that enforcement of these requirements became weaker until the mid-1980s, with some corrective measures taken since then in some countries. Thus, benefit disqualification was quite common in Norway until the early 1980s, but fell sharply until reforms in the early 1990s; the law in the United Kingdom was arguably applied steadily less effectively after the late 1960s, but was tightened in the mid- 1980s; and there was a marked tightening of enforcement in The Netherlands towards the end of the 1980s.(14) Enforcement problems must be seen against the background of the increase in unemployment, the application of work tests becoming problematic when there is a shortage of vacancies. Countries can also use offers to participate in active labour-market programmes to test whether an unemployed person is available for work, and Sweden's favourable unemployment record until the early 1990s has been partly credited to this practice.
A composite indicator
Table 4 attempts to combine information about some of the individual attributes of unemployment benefit systems discussed above in a summary index of overall benefit generosity. The broadbrush picture displayed in Table 4 is that of a significant increase in benefit generosity in the OECD area since the early 1960s. However, there are exceptions to this general pattern, with the generosity index remaining stable or even falling in the United States, Japan, Germany, the United Kingdom, and New Zealand. In the Nordic countries and Switzerland the effect of increases in replacement ratios on the overall generosity index has been amplified by longer maximum benefit periods and expanded coverage. In other countries, the increase in the summary measure has been brought about by a combination of benefit increases and a greater length of permissible benefit periods.
Summary Measure of Benefit Entitlements, 1961-91
1961 1971 1981 1991
Australia 17 14 22 27 Austria 20 23 29 31 Belgium 42 41 45 43 Canada 22 21 25 28 Denmark 20 34 54 52 Finland 5 8 23 39 France 25 24 31 37 Germany 30 29 29 28 Greece 7 7 7 17 Ireland 17 17 28 29 Italy 4 2 1 3 Japan 12 13 9 9 Netherlands 13 48 48 51 New Zealand 42 27 27 26 Norway 4 5 29 39 Portugal 0 0 9 34 Spain 9 12 28 34 Sweden 4 6 25 29 Switzerland 2 1 13 22 UK 24 25 24 18 US 17 11 15 11
Definition: The index is a simple average of replacement rates for three durations (first year, second and third year, and fourth and fifth year, crossed by three family situations (single, dependent spouse, and with spouse at work) and two earnings levels (average earnings and two-thirds average earnings). The worker is 40 years old and has worked without interruptions since age 18. After any UI entitlement is exhausted, it is assumed that the person is entitled to the appropriate rate of benefit in UA and GI programmes. For countries where UI is voluntary, the replacement rate is a weighted average for members and non-members of UI funds, the weights depending on actual coverage rates. Further details of the construction of the index are given in OECD (1994b), which also discusses the limitations of the index.
Source: OECD Database on Unemployment Benefit Entitlements and Replacement Rates.
III. ALTERNATIVE SUPPORT TO NON-EMPLOYMENT
Despite the focus on unemployment benefits in public discussion, people without jobs are often supported by other types of benefits, which typically involve withdrawal not just from employment but from the labour market. Thus, in addition to those who are not capable of working, some people who might be able and prepared to work, were they to find suitable job offers, are to be found in early retirement, invalidity, work-injury, and even sickness benefit systems. Indeed, the objective of schemes enabling early retirement for labour-market reasons is explicitly to be a substitute for unemployment benefits. Even the other benefit systems, although introduced to deal with risks other than those relating to employment, may serve as a substitute for unemployment benefits.
Table 1 indicated that the numbers of those who leave the labour force to be supported by these alternative non-employment benefits is often significantly higher than the numbers of measured unemployed. Figure 3 indicates that, expressed as invalidity rates, there is a variation across countries from 1 per cent to more than 10 per cent of the working-age population. This is unlikely to be explicable by reference to medical factors alone."
A further motivation for considering the labour-market effects of other non-employment benefits is that not only are the numbers in receipt of such benefits very high, they have been increasing. Figure 4 suggests that growth in the number of invalidity-benefit recipients during the 1980s has been spectacular, at least in some countries (especially the UK). If such alternative non-employment benefits can substitute for unemployment benefits in some circumstances, the labour-market consequences need not be negligible, and are growing.
(i) Substitution between Benefit Systems: The
The purpose of different benefits in social-security systems is to insure people against different risks, or compensate them for different needs, when circumstances require. The level and duration of benefits, and the requirements imposed on recipients, are varied to ensure that they cover different needs and leave different incentives, for example to search for jobs. The social objectives being pursued by the giving of invalidity benefits (the support of those medically unable to work) are generally different from those pursued by the award of unemployment benefits (the support of those temporarily unable to find work). But in practice, the distinction between the purposes of schemes may not be clear-cut. A use of early-retirement and invalidity benefits to act as support for those without jobs serves objectives of employers and politicians, as well as the individual benefit recipients. Governments can be tempted to push the unemployed into these benefit systems in an attempt to reduce the politically sensitive `headline' unemployment rate. Employment offices may also send difficult-to-place persons on such non-unemployment support schemes to reduce the workload of benefit officers. The tendency to reclassify the unemployed as being retired, invalid, or sick is likely to be amplified when this can be done at a low cost to the exchequer. Employers may use the existence of invalidity or early-retirement programmes as away to circumvent limitations on their ability to fire workers during times of economic stress (even being prepared to `top up' the public early-retirement schemes to facilitate employment reduction). They may also be under pressure to fire older workers first because they can get non-employment benefits.
The potentially unemployed may also have incentives to use the alternative schemes.(16) The unemployed, or people losing their jobs, can be expected to maximize welfare by selecting the scheme which offers the greatest income(17) and lowest social stigma. For example, if unemployment benefits are lower than invalidity, early retirement, or sickness benefits, there would be a strong incentive for an unemployed individual to migrate from the unemployment benefit system to the alternative systems. Similarly, unemployment benefits will be avoided if not working due to invalidity or retirement is seen as more socially acceptable than being unemployed. Indeed, studies indicate that social stigma attached to the status of unemployment can weigh heavily on the unemployed, whereas other types of supported non-employment, such as early retirement, are accepted as having more honourable social status (Argyle, 1989).
Support of those medically unable to support themselves is an important objective of social security systems. But if, either explicitly or implicitly, the schemes become an alternative means of support to those who might otherwise be unemployed, their impact on the labour market should be considered using the same criteria as would be used when looking at unemployment benefits. As Figure 3 shows, there is no (statistically significant) relationship between the invalidity ratio and the ratio of unemployment to the working age population' Several countries with low unemployment to working-age population ratios have high invalidity to population ratios, and vice versa, but there are outliers where both ratios are low (Iceland, Japan, Switzerland) and both are high (Italy).
A number of competing influences might explain the absence of any clear correlation between invalidity and unemployment rates. If administrations attempt to move those without jobs away from unemployment benefits, then in the short run there might be a negative relationship between the two rates. But the two rates might well be positively correlated with one another. For example, the permissive stance towards award of invalidity benefits may have a counterpart in lax administration of unemployment benefits, suggesting that the numbers in receipt of both types of benefit will be high. Furthermore, a reduction of the labour force is unlikely to reduce unemployment in the long run. In the short term, pushing the unemployed into invalidity receipt would imply an inverse relationship. But in the longer term, the reduced job-search activity because of the reduced numbers of the openly. unemployed (with consequent job-search requirements), would potentially push open unemployment up once more. Similarly, increased average benefit entitlements arising from allowing the unemployed to enter schemes with higher replacement rates and longer durations will increase the incentives for those with jobs to bargain for higher wages, reducing the aggregate employment level. Whatever the short-term political expediency of allowing or encouraging the use of these alternative non-employment benefits to support those without jobs, this implies that the longer-term effects on the labour market are unfortunate. The burden on individuals (except where such benefits are substantially higher than unemployment benefits) and society is not lessened by defining groups out of the, labour force; poverty, high taxes, and reduced incentives for those in work and a failure to realize the full productive potential of the economy are inevitable consequences. Furthermore, the policy may provide only a temporary reduction in the open unemployment rate. Such considerations suggest a need to ensure not simply that incapacitated workers have sufficient means of support, but that only incapacitated workers receive such support.
(ii) Early Retirement
Table 5 gives replacement rates for some of the early-retirement schemes in operation during the 1980s. The replacement rates refer to the value of schemes available to those with average earnings and who are 5 years younger than the official retirement age.(19) Where there is more than one early retirement scheme, the one which is the largest in terms of numbers of beneficiaries is included (or, where there is no information on beneficiaries, the most generous of the available schemes). In several countries the replacement rates are very high - over 70 per cent in Austria, Belgium, Greece, Italy, and The Netherlands.(20) In virtually all cases they are well above replacement rates in unemployment benefit schemes. There is an incentive, therefore, for both the employed and the unemployed to withdraw from the labour force. This has been explicitly the principal objective of early retirement schemes introduced since the late 1970s. There has also been a move towards increasing flexibility in retirement decisions for the individual without offering financial incentives. Table 6 summarizes the large number of such schemes. In fact, early retirement provisions can be divided into three types, depending on their effects on incentives.
[TABULAR DATA OMITTED]
The first type involves the provision of actuarially-reduced public pensions to those who retire early. For example, in Finland and Sweden pension rights are reduced by 0.5 per cent for every month early that the pension is taken. Canada has the same reduction rate, and also increases the pension if it is taken after the normal retirement age. In Spain and the United States, the reduction rate is somewhat higher; in Germany, rather lower. Such provisions are consistent with the view of public pensions as a form of enforced savings; indeed, many private pension schemes in those countries where they are significant have similar clauses. The motivation for such schemes need not therefore be a desire to reduce the size of the labour force but to take into account personal preferences on work and retirement. Nevertheless, a reduction in the labour force is the normal result of such schemes; over 20 per cent of those aged 60-64 in Canada and the United States were in receipt of actuarially reduced pensions at the beginning of the 1990s (OECD, 1992).
A second type of early retirement benefit is that which allows early retirement with no loss of pension entitlement. For example, the `long-service' pensions given in Austria, Belgium, Germany, Greece, and Italy enable retirement after 35 or 40 years of pension contributions. The motivation for such pensions is that beyond the minimum contribution period required for a full pension, extra years of work have a very low actuarial rate of return in public pension schemes, so the long-service pensions limit the extent of any such unfairness. As with the actuarial reductions in pensions, the motivation for long-service pensions need not be to reduce the size of the labour force, although this is likely to be one outcome (in Austria, for example, over 12 per cent of those aged 55-64 are in receipt of long-service pensions).
The final type of early-retirement benefit consists of a benefit which is set below the normal pension. Such benefits are often given to those elderly workers who have exhausted unemployment insurance entitlement, providing a bridge to the retirement age and pension receipt. Starting with Belgium and the United Kingdom in the mid-1970s, several countries introduced benefits for elderly workers who retired if their employers replaced them with a previously unemployed worker. Such schemes existed in France as `Contrats de solidarite' between 1982 and 1988, covering all those aged between 55 and 59. The age group covered in the German scheme existing between 1984 and 1988 was even larger, ranging from 58 to 65, with the costs of the early-retirement benefit split between employers and government. Other schemes, such as those in Finland and Sweden, did not specify replacement conditions, but instead were supposed to take pressure off the labour market by reducing supply, as well as ensuring a reasonable standard of living for the elderly unemployed who were unlikely to be able to find a new job.
(iii) The Invalidity and Sickness Benefit Systems
The nature of invalidity and sickness benefit schemes All OECD countries have provisions for those who are afflicted by disability. Such invalidity benefits or pensions (hereafter, IVB) are paid to those who are medically unable to be fully effective in the labour market. Sickness benefits (SB) insure against temporary income loss and exist in most OECD countries.(21)
[TABULAR DATA OMITTED]
In many countries, employers are legally obliged to support their employees during the first days or weeks of sickness, but in most countries longer-term sickness benefits are paid by official sickness insurance finds. Collective agreements often stipulate supplements to the legislatory minimum. Sickness benefit recipients are not defined out of the labour force; in labour-force surveys, they fall into the category of `having a job but not at work during the reference period'. However, sickness benefits may substitute for unemployment benefits where people who would otherwise be fired might go on to sickness benefit, and in some countries unemployment benefit recipients may claim sickness benefits, potentially receiving higher benefits and extending the duration of benefit receipt. In general, IVB is paid to those who lose some `working capacity' how much working capacity must be lost before there is some entitlement to benefit varies from 25 per cent in The Netherlands (formerly it was just 15 per cent) to 85 per cent in Australia (Ireland, Luxemburg, the UK, and the US do not specify a degree of incapacity). This loss of working capacity is generally assessed with reference to the `customary occupation' of the individual. (In other countries, including inter alia Denmark, Finland (for workers aged under 55), France, Japan, and the United States, reference is to inability to work in any occupation at all.) In principle, therefore, people who are able to work can be defined out of the labour force because of loss of earnings in their principal occupation, even were they able to work in other occupations.
SB and IVB replacement ratios
The generosity of SB and IVB depends on both the maximum duration and the replacement ratio. The maximum duration for IVB is nearly always until recovery or retirement, hence it is available for longer periods than (most) unemployment benefits; for SB the duration is typically shorter than that for unemployment-insurance benefits.(22) Table 7 shows that the SB replacement ratio declines with the length of the benefit period in many countries, but the initial ratio is often quite high. In Germany, Austria, and Luxemburg sickness benefits are equal to wages in the first month. Judging from the simple average of 16 cases (two income levels, two family levels, and four stages of duration), the overall generosity in 1993 was particularly high in Sweden, Norway, Germany, and Luxemburg, and notably low in the United Kingdom and Canada. (However, in these last two countries, after one year of sickness it is likely that there would be an entitlement to IVB.)
[TABULAR DATA OMITTED]
Replacement rates in IVB often depend on the degree of incapacity for work, as shown in Table 8. Table 8 also illustrates a phenomenon which applies equally to sickness benefits; namely, that replacement rates often are higher for those on lower incomes (reflecting the operation of benefit ceilings and flat-rate elements). The summary IVB generosity figure given in the final column (a simple average of the eight cases presented) suggests that benefit levels are particular high in the Nordic countries, The Netherlands, and in some southern European countries (Greece, Italy, and Spain). The lowest entitlements are found in Australia and Japan, but these results must be interpreted with caution, as they reflect the fact that at incapacity of 85 per cent or less, there is no entitlement to benefit. Definitions of incapacity are likely to vary across countries, and some countries (e.g. Luxemburg, Ireland, the UK) do not specify incapacity as a percentage but as a discrete variable (you either are disabled or are not). The figures in Table 8 assume in these cases that invalidity benefit is available to those with a two-thirds incapacity; the alternative assumption would of course reduce the replacement ratio averages for these countries substantially.
[TABULAR DATA OMITTED]
Tables 9 and 10 give a limited time-series of summary replacement rates described in the previous two tables. The present systems of sickness insurance and IVB insurance date back to the 1960s or early 1970s. During this period several continental European countries increased the overall generosity of these systems by increasing replacement ratios and (in the case of sickness benefits) expanding the maximum length of benefit periods. The recession in the 1970s arrested the expansion of the sickness benefit system. During the 1980s, several countries in Europe started to cut back on payments of both benefits. The United Kingdom and Ireland abolished the earnings-related component of sickness and invalidity benefits, the replacement ratio was lowered significantly in The Netherlands, and Denmark imposed limits on the sickness benefit period. The overall generosity of sickness benefits increased in a few countries (Australia, Finland, Greece, and Portugal) during the 1980s, as did the typical value of invalidity benefits in Italy and, to a lesser extent, Canada, Ireland, and Sweden. In summary, for both benefits there is also some limited evidence that generosity increased during the 1970s, was fairly constant during the 1980s, but may have declined somewhat in some countries since the late 1980s.
Table 9 Average IVB Replacement Rates over Time
1974 1981 1987 1993
Australia 22 21 16 Austria 70 53 53 Belgium 43 51 52 47 Canada 21 31 31 Denmark 47 49 47 44 Finland 60 60 53 France 36 36 36 34 Germany 46 46 46 46 Greece 71 58 54 Ireland 34 34 38 38 Italy 39 42 60 60 Japan 23 24 17 Luxemburg 46 57 54 59 Netherlands 79 85 68 63 New Zealand 47 51 50 Norway 66 65 65 Portugal 51 53 47 Spain 75 60 56 Sweden 66 64 74 Switzerland 28 27 32 United Kingdom 42 53 48 38 United States 26 28 27 30 Definition: See explanation of the average figure in the note to Table 7. Note: Unlike the data on unemployment benefits in Tables 2-4, these data have not been reviewed for accuracy by national administrations, and should not be treated as official OECD statistics. Estimates can be highly sensitive to other earning-profile assumptions, and do not take into account differences in administrative determination of incapacity, so should be treated as indicative. Source: Authors' calculations, based on sources and assumptions similar to those indicated in the footnote to Table 7.
Table 10 Generosity Index in Sickness Insurance: Evolution over Time
1961 1974 1981 1993
Australia 40 51 Austria 42 42 Belgium 35 50 52 52 Canada 14 15 Denmark 79 79 58 Finland 22 50 France 36 38 32 37 Germany 51 65 65 65 Greece 29 65 Ireland 50 56 37 Italy 25 34 33 33 Japan 45 45 Luxemburg 35 75 75 75 Netherlands 60 60 60 53 New Zealand 42 41 Norway 75 75 Portugal 60 70 Spain 56 56 Sweden 90 78 Switzerland 0 0 United Kingdom 38 15 17 United States 0 0 Definition: The index is the average of the cases set out in Table 7. Assumptions and sources are as given in Table 7. Source: Authors' calculations.
Generosity and numbers of recipients
If it were true that there is an element of choice over benefit receipt, then the relative generosity of a benefit should influence the numbers in receipt of the benefit. This is indeed the case for IVB. Figure 5 illustrates that there is a correlation between the proportion of those aged over 55 in receipt of IVB and the average replacement ratio, as described above.(23) Furthermore, in several of those countries where take-up of IVB seems low relative to the replacement rate (Belgium, Italy, Spain, The Netherlands), there are early retirement provisions which give high replacement rates (see the discussion above).(24)
There is, however, little correlation between the generosity of public sickness benefits and the sickness-absentee rate for the countries where such data exist (Figure 6).(25) The lack of any close correlation between the generosity of sickness benefits and sickness-absenteeism could reflect that high generosity is accompanied by tighter control. To prevent abuse of the system most countries have a waiting period before sickness-insurance benefits can be claimed,(26) typically 3-7 days, and/or make the employer responsible for financial support during the initial stage of sickness. Some of the countries with high sickness-absenteeism, notably some of the Nordic countries, have traditionally applied very short, if any, waiting or employer periods and the high generosity in Germany has gone hand in hand with unusually long employer periods (6 weeks). But the effectiveness of control of long-term sickness benefit periods will crucially depend on the prevailing practice in the medical profession concerning the granting of medical certificates and the degree to which controls are exercised on this practice. The same arguments apply even more strongly with relation to IVB. If doctors yield easily to demands for certificates, high benefit generosity is likely to lead to intensive use of the systems, whereas strict application can prevent an abuse of the system.
Labour market conditions and the award of non-employment
The extent to which IVB and SB substitute for unemployment benefits will depend on how entitlement to these benefits is assessed. The greater the weight given to medical considerations, the less the possibility of those without work ending up in receipt of these non-employment benefits. It is widely argued that, for various reasons, more people find their way into such non-employment benefit systems when unemployment is high than when it is low.
A relationship between IVB and SB receipt and unemployment need not in itself indicate abuse of the benefits. Because health is related to economic well-being, it is to be expected that the numbers in receipt of IVB will rise as unemployment increases, reflecting reduced incomes, greater inequality, and the increased mental pressures on workers and those without jobs. As discussed below, there are, however, good grounds for believing that IVB and SB do substitute for unemployment benefits, at least in some countries, in the short term.
In a number of countries, medical diagnosis of IVB entitlement is explicitly supplemented by an assessment of the state of the labour market (Table 11). Instead of being limited to individuals unable to work at all, individuals who are incapable of working in those sectors or occupations where jobs are available are assessed as being entitled to IVB.
[TABULAR DATA OMITTED]
In Sweden, the changes in the early 1970s were clearly in response to increasing unemployment among older workers. Disability compensation legislation was revised to require adjudicators `to concern themselves mainly with the applicant's ability and possibility of obtaining a continued income through such work as has previously been done or through other suitable and available work' (see OECD, forthcoming; Casey and Bruche, 1983). Furthermore, Bjorklund and Holmlund (1991) report informal evidence that disability requirements are enforced less strictly when workers are at risk of unemployment. In fact, Swedish employers frequently `top up' disability pensions until retirement as a way of gaining acceptance for reductions in the size of their work-forces. While officially the number of people who benefit from the labour-market criterion was low (around 22 per cent of total IVB awards made to men in 1985, and slightly under 20 per cent for women in the same year), the change in rules signalled a more liberal attitude generally to the award of IVB. The result can be seen in the time-series for regular IVB recipients assembled by Adema (1993); in the 1960s, the number of recipients rose by 21 per cent. In the 6 years following the relaxation of the requirements, the numbers in receipt of the benefit increased by almost 10 per cent per year. The formal abolition of the labour market criterion failed to change attitudes towards award of IVB, new cases of which, with the onset of the recession, rose by a quarter in 2 years.
Labour-market conditions were a factor in determining eligibility for IVB in The Netherlands until 1987. In particular, education, work experience, and previous earnings are all factors against which any medical incapacity was assessed. Aarts and de Jong (1990) claimed that the labour-market conditions were responsible for much of the very rapid growth in IVB recipients. They argue that assessors were administratively lenient.(27) Over time, it became normal practice to assume that, in effect, failure to find a job was indicative of disability. To prevent this, in 1987 reforms removed the labour market criterion. This should have meant an increase in those on partial IVB pensions who, under the old criteria, would have been upgraded to full IVB. In fact, this did not happen, as assessors had no incentive to dispute claims by claimants. Psychological reasons for incapacity have become increasingly dominant, accounting for 30 per cent of the total of new claimants in 1989 (Ministerie van Sociale Zaken en Werkgelegenheid 1991, as reported in Adema, 1993). The failure of the 1987 reforms led to a further attempt to tighten the criteria in 1990, which seems to have had some success in slowing the rate of new inflows into the scheme.
In a system, such as that of The Netherlands, where labour markets are tightly regulated, the giving of generous IVB is a way of gaining acceptance for voluntary redundancy. Workers get unlimited duration benefits until retirement, and employers can adjust the size of their work-force without any great social stigma. Unsurprisingly, therefore, the rate of entry into disability pension has been highest in those industries where employment contraction has been greatest (Casey and Bruche, 1983). In 1992, reforms attempted to address this incentive directly, by penalizing financially those employers who laid off workers into the invalidity system.
The system in the UK is very different. Alterations to the size of the labour force are not regulated and labour-market conditions have never explicitly been brought into the determination of IVB entitlement, which is supposed to be determined entirely on medical criteria. Yet even here there is evidence that employment conditions influence benefit award. Holmes and Lynch (1990) suggested that IVB was being used as a form of early retirement; and Molho (1991) and Disney and Webb (1991) suggested that the level of benefits and rate of unemployment were correlated with IVB take-up. However, when unemployment starting falling in the UK (around 1987-90), there was no reduction in the growth of IVB recipients, as these earlier studies would have suggested (Adema, 1993). A qualitative survey of recipients (see Lonsdale et al., 1993) suggests a link between labour-market conditions, the relative value of benefits, and the growth in recipients. One of the mechanisms through which this link existed was that doctors (whose certification is required for entitlement) were influenced by their assessment of the probability of patients finding a job, rather than strictly determining whether they could work (Ritchie et al., 1993).
In three rather different countries, then, a good case can be made that invalidity benefit is being used to support those who in the past would have been working (the case of The Netherlands) or unemployed. Have sickness cash benefits been used as a substitute for unemployment benefits? Studies show that the same individuals tend to be heavy users of both sickness and unemployment benefit systems. In Denmark in 1989 around 25 per cent of those who had unemployment benefits as their main income support also received sickness benefits in the course of the year (Finansministeriet, 1991). Studies in Sweden also show that over a 13-year period (1978-90) the individuals most dependent on unemployment support received the bulk of sickness benefits as well (Finansdepartementet, 1994), and the prime-age unemployed had 4-5 times the risk of being certified sick for long periods compared with their employed counterparts (Arbetsmarknadsdepartementet, 1994). Evidence from Australia also suggests that a large proportion of sickness beneficiaries had transferred from unemployment benefits. The fact that the same individuals show up as beneficiaries of sickness and unemployment benefits is not entirely surprising. People with a record of bad health are likely to run a higher risk of becoming unemployed (see e.g. Pedersen and Westergard-Nielsen, 1993), and unemployment has been shown to have adverse effects on health (see e.g. Mayer and Roy, 1991).
However, transition between the two systems could possibly reflect financial incentives to use one system rather than the other. High sickness benefits relative to unemployment benefits will encourage the unemployed to seek the former, and vice versa. To prevent such distorting incentives, several countries have aligned benefit levels across the two systems. However, an unemployed person would benefit financially from receiving sickness rather than unemployment benefits in a number of European countries. Thus, at least over the short term (up to 6 months), sickness benefits are significantly higher than unemployment benefits in Germany, Austria, Belgium, Italy, Finland, and Greece. In many countries sickness benefits are substantially higher than unemployment benefits for the better off, since ceilings on benefits are not applied or are higher in the sickness benefit system. But even if the benefits in the two systems are equal, the fact that UI benefit periods can be prolonged by taking sick leave may tempt the unemployed to use the sickness benefit system extensively during their unemployment spell.
IV. EMPIRICAL EVIDENCE
(i) Unemployment Benefits: Effects on
Unemployment and Participation
It has proved difficult to explain variation in the unemployment rate across countries and across time by differences in unemployment benefit systems alone. As shown in OECD (1991) there was no cross-country correlation between the average UI (or GI) replacement ratios and unemployment rates in the late 1980s: high replacement ratios were found in both high-unemployment countries (e.g. Denmark) and low-unemployment countries (e.g. Sweden), and low replacement ratios were found in high-unemployment countries (e.g. Italy). However, initial replacement ratios interacting with maximum benefit duration do appear to be correlated with the variation in unemployment rates across countries in the 1980s, as shown in Burda (1988).
To assess the importance of unemployment benefit variables in explaining the rate of joblessness across countries and over time, the data discussed in section II were employed in pooled cross-section time-series regressions in OECD (1994b). That study suggested that cycle-to-cycle changes in unemployment rates were correlated with lagged changes in the summary measure of benefit entitlement. A variety of other relationships were also explored, which suggested correlations between the duration of benefit entitlement and duration of unemployment and between benefit entitlements and unemployment rates by sex. The two sets of regressions given in Table 12 attempt to supplement that work by using a standard lag structure, rather than comparing inter-cyclical changes, and by including an explicit treatment of the maximum duration of unemployment insurance. The dependent variable in both is the unemployment rate, and the data cover every odd year in the 1967-91 period. In the regressions reported in the upper panel, the explanatory variables are the initial replacement ratio and the maximum duration of UI benefit periods. Countries with no unemployment insurance or with unlimited benefit periods were excluded from these regressions. The independent variable in the regressions reported in the lower panel is the summary measure of benefit entitlement discussed earlier, and these regressions include all 21 countries in the dataset. As the hypothesis of common intercepts across countries was rejected in F-tests, only regression results from a fixed-effects model are reported.
In the parsimonious forms the coefficients on the benefit variables take on expected signs (both the replacement rate and duration variables having positive coefficients) and are all significant, which would support the conjecture that benefit systems play a role in explaining the unemployment rate across time and countries. However, the proportion of the variance of the unemployment rate explained by country intercepts and unemployment benefit variables is low, indicating the importance of omitted variables. A notable feature of the regression results is that they point to relatively long lags between changes in the benefit variables and changes in the unemployment rate. Thus, the goodness-of-fit of the equations improves with longer lags, and a lag of 6 years seems to be favoured by the data. The long lags could indicate that the benefit variables interact with the economic cycle in generating unemployment. For example, as argued in OECD (1994b), an increase in benefit generosity may only be felt when the economy turns down and displacement takes place.
The results reported in the upper panel of Table 13 suggest that generous unemployment benefits also tend to attract people into the labour force. The dependent variable in these regressions is the labour-force participation rate, while the independent variable is the summary measure of benefit entitlements. The parameters on the lagged generosity variables are statistically significant, and together with country intercepts they explain from a fifth to a quarter of the variation in participation rates over time and countries. This implies that the impact on the employment rate, i.e. total employment relative to the population of working age, is ambiguous. The employment rate equations reported in Table 13 tend to confirm this: the generosity variable is generally not statistically significant.
[TABULAR DATA OMITTED]
(ii) Other Non-employment Benefits: Impact on
the Labour Market
The impact of invalidity and sickness benefits on unemployment can be assessed by augmenting the regression equations above with summary indicators of benefit entitlements in these alternative systems. If the parameters on the augmented variables take on statistically significant and negative values, this would indicate that the alternative systems serve partly as a substitute for unemployment benefit systems and that participation in the systems is partly a matter of choice for the beneficiaries. If the parameters are not statistically significant or take on wrong signs, participation in the alternative schemes may still influence the unemployment rate through other non-financial features of the non employment benefit systems, such as administrative measures, duration of entitlement, the reduced stigma attached to non-employment as opposed to unemployment, etc.
In order to run the augmented regressions it is necessary to shorten the estimation period, given that data on entitlements in invalidity and sickness benefit systems are not available prior to 1981 for a sufficient number of countries. The estimation period in the pooled cross-section time-series regressions reported in Table 14 is from 1983 to 1991. The indicators of entitlements, which are identical to those discussed in section Ill, were calculated for 1981, 1987, and 1993. For the missing years data were either interpolated or based on information on changes in legislation.
[TABULAR DATA OMITTED]
The regression results do not support the hypothesis that generosity in invalidity and sickness benefit systems influences the aggregate unemployment rate. In regressions where a common intercept is forced on all countries, the parameter on the summary measure of entitlements to unemployment benefits continues to be significant, but the value of the parameter is significantly lower than reported in Table 12. While the parameter on sickness benefits is negative, that on disability benefits is positive, which would indicate that higher generosity in disability benefit systems is accompanied by higher unemployment. Such results are consistent with the hypothesis that a more accommodating benefit stance should lead to higher wages. However, neither parameter is statistically significant. The results become more difficult to interpret when each country has its own intercept. Such a fixed-effects model, which is favoured by an F-test, gives implausible results.
[TABULAR DATA OMITTED]
There is some evidence that the size of entitlements to invalidity and sickness benefits influence labour-force participation rates. The parameter on the IVB variable in all the regressions reported in Table 14 is negative, suggesting that high benefit generosity encourages people to withdraw from the labour market. Though statistically significant in OLS estimation, the parameter in the fixed-effects model, which continues to be favoured in F-tests, is imprecisely estimated. The parameter on sickness benefits is positive in OLS estimation. This is in line with expectations: people on sick leave are still classified as in the labour force, and the availability of generous benefits is likely to encourage marginal workers to join the labour force. However, the parameter is statistically insignificant and even switches sign in the fixed-effects model.
V. CONCLUDING REMARKS
There has been a significant increase in the generosity of unemployment benefit systems in most OECD countries since the early 1960s: the coverage of potential beneficiaries has been extended, benefit periods have been lengthened, and financial support relative to pre-displacement wages increased. As a result, it is possible that benefit systems have contributed to higher unemployment. Simple pooled cross-section time-series analysis is indeed consistent with this hypothesis. The effects of higher generosity have most likely been compounded by less monitoring of work-availability and willingness-to-work requirements, the reduction in vacancies per unemployed person and greater workload of public employment agencies making such tests more difficult to apply. Nevertheless, the relationship between benefit changes and unemployment should not be exaggerated; it is only a partial explanation of the rise in unemployment.
Those unable to find jobs may also have found their way into other non-employment benefit systems, which were originally designed to support individuals partly or fully incapable of working for reasons of age or bad health. There is indeed a striking difference across countries in the extent to which people of working age are receiving invalidity, early-retirement, or sickness benefits, and there is some tendency for low-unemployment countries to have a relatively large proportion of the population in these alternative programmes. While likely to reflect different policies towards the unemployed, this may also be related to the difference in generosity of benefits: invalidity benefit systems have been used more intensively in countries which provide generous levels of such benefits. In many countries the alternative benefit systems offer higher benefits than are available in unemployment compensation, encouraging the unemployed to migrate to non-unemployment benefit systems, either through their own volition or through employers choosing workers entitled to such benefits as least likely to resist being fired, or through administrators choosing to assign those without jobs away from unemployment benefits to reduce administrative workloads and to provide temporary reductions in the politically sensitive headline unemployment rate.
When considering in detail the development of non-employment benefits in some countries, a case can be made that various benefits not intended to support the jobless can under certain circumstances substitute for unemployment benefits. But there is nothing inevitable about this. Simple pooled cross-section time-series analysis provides no support for the hypothesis that invalidity and sickness benefit entitlements substitute for unemployment benefits.
That no evidence is found that such benefit entitlement affects the unemployment rate in the long run may not be surprising, for several reasons. First, `outliers' may occur where enforcement of benefits is either particularly tight or particularly lax. For example, failure to police entitlement to invalidity benefits might also indicate weak enforcement of work tests in unemployment benefits. Conversely, strict enforcement of job-search requirements is unlikely to be accompanied with a liberal interpretation of incapacity. The extent to which numbers in receipt of invalidity benefits and long-term sickness benefits respond to the level of benefit entitlement critically depends on the attitude of the medical profession in certifying work incapacity, high benefits inevitably leading to high numbers in receipt of the benefit if the `gate-keepers' do not exert tight controls. Second, those in receipt of such benefits are not in the labour force, not searching for work, and so not influencing the market wage rate. On this reading of the evidence, these benefit entitlements influence the size of the labour force, with the level of unemployment benefit entitlement and duration of entitlement influencing - to some extent - the level of unemployment. Third, the alternative non-employment benefits are often more generous than unemployment benefits - both in terms of level and duration. Knowledge of the availability of such benefits could influence wage-setting behaviour, pushing wages higher as mean expected income loss for those without jobs becomes lower. Overall effects on the labour market would be complicated were this shown to be true; higher early-retirement and invalidity benefits would reduce labour-force participation, but those remaining in the labour force would also face altered incentives. All three of these explanations of why no evidence of substitution between unemployment benefits and non-employment benefits is found in multiple-correlation estimates require further empirical testing; the possibility that they might be proved suggests the need to avoid complacency in public policy in this area. (1) The article draws, particularly in section II, on work done for the OECD jobs study: Evidence and explanations, ch. 8. Many thanks are due to the following for helpful comments: Willem Adema, Andrew Burns, Jorgen Elmeskov, David Grubb, John Martin, Peter Scherer, and the editors of this review. All errors remain those of the authors. This article reflects the personal views of the authors, not those of the OECD or its member countries. (2) See Hemmings (1994). The OECD will shortly be issuing software with the basic data and a benefit model to calculate replacement rates for different durations of unemployment, different earning levels in previous job, and different family situations. Work is also under way to extend the range of benefits covered to include child care and housing benefits and to integrate them with tax models developed by the OECD Secretariat. (3) Only female employees in agriculture were excluded prior to 1973. (4) Insurance was made compulsory in Sweden for the second half of 1994, but from start of 1995 was made voluntary once again. (5) Katz and Meyer (1990). (6) Ham and Rea (1987). (7) OECD (1991, pp. 207). (8) Pedersen and Westergard-Nielsen (1993). (9) Florence et al. (1990). (10) Alba-Ramirez and Freeman (1990). (11) Carling et al. (forthcoming). (12) In 1993 the replacement rate was cut to 80 per cent in Sweden. (13) Where there has been no change in the taxability of benefits over time, the net replacement rate has tended to rise, because the average rate of tax on people in work has increased by more than that of those on benefit incomes. (14) See OECD (1993 b,c). (15) The total labour force provides an alternative denominator which would not overstate the size of the potential employment by including (for example) women working at home with no intention of entering paid employment although it is one which is in turn affected by the numbers receiving IVB (defined out of the labour force). With this as the denominator, there is a still a negative relationship between the unemployment and invalidity ratios, but it is smaller and even less significant. The Netherlands stands out as having a particularly high invalidity ratio using this measure. (16) The principles regarding substitutions between benefit systems are discussed in Fortin and Lanoie (I 992). The actual paths through which entrants pass on their way into IVB vary across countries. Adema (1993) suggests that in the UK, for example, a significant number of inflows into IVB come via unemployment, whereas in The Netherlands the flow is mainly from employment via sickness benefits. (17) Apart from the fact that benefit income in non-employment benefits is usually higher than in unemployment benefits, the former is likely to be a lower risk of detection of informal-sector work. (18) Similarly, a relationship between unemployment and invalidity cannot often be detected using time series of national data. Adema (1993) was able to find a clear impact of disability on measured unemployment only for The Netherlands, results for the UK not being robust and no such relationship being found in Sweden. (19) Where there are no schemes which apply to those 5 years from the standard retirement age, no replacement rate is calculated. (20) As reported in OECD (1992),different assumptions to those made here could lead to similarly high replacement rates in all the Nordic countries. (21) There are no statutory provisions for sickness benefits at the federal level in the United States, but seven states have introduced them. The federal government in Switzerland stipulates insignificant cash benefits and no canton has made sickness insurance compulsory. (22) In most countries the maximum duration is 1 year or less, whereas a few countries (Sweden, Australia, and New Zealand) have unlimited sickness benefit periods. (23) An OLS regression yields a coefficient of 0.48 on the IVB replacement rate, with a standard error of 0.2 and an R-squared of 0.3. (24) If d(ivb) is the proportionate change in the absolute numbers of those in receipt of IVB over the 1981-90 period and D(ivb 1981) and D(er1981) the difference between IVB entitlement and early retirement benefits and the insurance or assistance which is received in the fourth year of unemployment in 1981, then:
d(ivb) = 1.09 + 1.03 D(ivb 1981) - 0.61 D(er1981) R2=0.41
which is at least suggestive that - looked at over a very long period - the number in receipt of IVB is sensitive to the incentive to be on this form of support rather than long-term unemployment benefits or early retirement benefits. If the two outliers, Italy and the United Kingdom, are excluded from the analysis, then the same equation yields:
d(ivb) = 2.8 + 0.84 D(ivb1981) - 0.26 D(er1981) R2=0.59.
(SE=0.25) (SE=0.16). (25) Microeconomic studies show that the degree of sickness is influenced by the generosity and the design of sickness benefits (see e.g. Barmby et al., 1991). (26) The importance of waiting periods has been demonstrated in work-injury insurance systems, see Krueger (1990). (27) OECD (1993a) attributes lax administration to the fragmented management structure which left individual boards with little incentive to control entry into the schemes.
Aarts, L. J. M., and de Jong, P. R. (1990), `Economic Aspects of Disability Behaviour', Doctoral Thesis, Erasmus Universiteit, Rotterdam. Adema, W. (1993), `The Beveridge Curve and Institutional Arrangements', Doctoral Thesis, University of Oxford. Alba-Ramirez, A., and Freeman, R. B. (1990), `Job Finding and Wages when Long-Run Unemployment is Really Long: The Case of Spain', National Bureau of Economic Research Working Paper no. 3409. Arbetsmarknadsdepartementet (1994), Arbetsmarknad & Arbetsmarknadspolitik 1993, Stockholm. Argyle, M. (1989), The Social Psychology of Work, new edn, Penguin. Atkinson, A. B., and Micklewright, J. (1991), `Unemployment Compensation and Labor Market Transitions: A Critical Review', Journal of Economic Literature, 29, 1679-727. Barmby, T. A., Orme, C. D., and Treble, J. G. (1991), `Worker Absenteeism: An Analysis Using Microdata', Economic Journal, 101, 214-29. Barr, N. (1992), `Economic Theory and the Welfare State: A Survey and Interpretation', Journal of Economic Literature, 30, 741-803. Bean, C. (1994), `European Unemployment: A Survey', Journal of Economic Literature, 32. Bjorklund, A., and Holmlund, B.(1991),`The Swedish System', in A. Bjorklund, R. H. Haveman, R. Hollister, and B. Holmlund (eds), Labour Market Policy and Unemployment Insurance, Oxford, Oxford University Press. Burda, M. (1988), `Unemployment', Economic Policy, 7, 851-86 Calamita, M. C., and Ferrante, A. (1991), `La Prevoyance Sociale Italienne. Informations', Note by the Directorate-General for Social Security, Rome. Carling, K., Edin, P.-A., Harkman, A., and Holmlund, B. (forthcoming), `Unemployment Duration, Unemployment Benefits, and Labour Market Programs in Sweden', Journal of public Economics. Casey, B., and Bruche, G. (1983), Work or Retirement: Labour Market and Social Policy for Older Workers in France, Great Britain, the Netherlands, Sweden and the USA, Aldershot, Gower. Disney, R., and Webb, S. (1991), `Why are There so many Long-term Sick in Britain?' Economic Journal, 101. Finansdepartementet (1994), En Social Forsakring, Rapport till espertgruppen for studier i offentlig ekonomi, Stockholm. Finansministeriet (1991), Rapport om arbejdsmarkedets strukturproblemer, Del II. 3 8-40, Copenhagen. Florence, J. P., Fougere, D., and Werquin, P. (1990), `Durees de Chomage et Transitions sur le Marche du Travail', Sociologie du Travail, 32(4). Fortin, B., and Lanoie, P. (1992), `Substitution between Unemployment Insurance and Workers' Compensation: An Analysis Applied to the Risk of Workplace Accidents', Journal of Public Economics, 49,287-312. Ham, J. C., and Rea, S. A. (1987), `Unemployment Insurance and Male Unemployment Duration in Canada', Journal of Labour Economics, 5(3), 325-53. Hemmings, P. (1994), `Benefit Entitlements and Replacement Ratio Data for the OECD Countries', Paris, Organization for Economic Cooperation and Development, (mimeo). Holmes, P. R., and Lynch, M. (1990), `An Analysis of Invalidity Benefit Claim Durations for New Male Claimants in 1977-8 and 1982-3', Journal of Health Economics, 9. Katz, L. F., and Meyer, B. (1990), `The Impact of the Potential Duration of Unemployment Benefits on the Duration of Unemployment', Journal of Public Economics, 41, 45-72. Krueger, A. B. (1990), `Incentive Effects of Workers' Compensation Insurance', Journal of Public Economics, 41, 73-100. Layard, R., Nickell, S. J., and Jackman, R. (1991), Unemployment, Oxford, Oxford University Press. Lonsdale, S., Lessof, C., and Ferris, G. (1993), Invalidity Benefit: A Survey of Recipients, DSS Research Report no. 19, London, HMSO. Mayer, F., and Roy, P.-M. (1991), `La Relation Chomage-Sante: Une Etude Longtitudinale', Canadian Journal of Economics, 24, 551-62. Molho, I. (1991), `Going on to Invalidity Benefit: A Study of Women (1977-8-1983-4)', Applied Economics, 23. OECD (1988), Employment Outlook, Paris, Organisation for Economic Cooperation and Development. - (1989), Employment Outlook, Paris, Organisation for Economic Cooperation and Development. - (1991), Employment Outlook, Paris, Organisation for Economic Cooperation and Development. - (1992), Employment Outlook, Paris, Organisation for Economic Cooperation and Development. - (1993a), Employment Outlook, Paris, Organisation for Economic Cooperation and Development. - (1993b), The Public Employment Service in Japan, Norway, Spain and the United Kingdom, Paris, Organisation for Economic Cooperation and Development Documents. - (1993c), The Labour Market in the Netherlands, Paris, Organisation for Economic Cooperation and Development Documents. - (1994a), Employment Outlook, Paris, Organisation for Economic Cooperation and Development. - (1994b), The OECD Jobs Study: Evidence and Explanations. Part II: The Adjustment Potential of the Labour Market, Paris, Organisation for Economic Cooperation and Development. (forthcoming), The Labour Market Position of Older Workers, Paris, Organisation for Economic Cooperation and Development. Pedersen, P. J., and Westergard-Nielsen, N. (1993), `Unemployment: A Review of the Evidence from Panel Data', OECD Economic Studies, 20, 65-114. Ritchie, J., Ward, K., and Duldig, W. (1993), GPs and IVB, DSS Research Report no. 18, London, HMSO.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||impact of benefits on the labor market|
|Publication:||Oxford Review of Economic Policy|
|Date:||Mar 22, 1995|
|Previous Article:||Evaluating unemployment policies: what do the underlying theories tell us?|
|Next Article:||The assessment: regional policy in the European Union.|