Abstract

High unemployment in France has persisted despite the fact that in recent years France’s macroeconomic performance has compared favorably with the other major industrial or EU countries in other respects. Since 1987, growth has been in line with EU and major industrial-country averages while inflation has been below both. Low inflation has contributed to improved competitiveness and, recently, to trade surpluses. However, employment creation was disappointing even in times of strong growth and the recent recession has led to a sharp fall in employment.

High unemployment in France has persisted despite the fact that in recent years France’s macroeconomic performance has compared favorably with the other major industrial or EU countries in other respects. Since 1987, growth has been in line with EU and major industrial-country averages while inflation has been below both. Low inflation has contributed to improved competitiveness and, recently, to trade surpluses. However, employment creation was disappointing even in times of strong growth and the recent recession has led to a sharp fall in employment.

Unemployment not only puts pressure on government finances and inflicts a heavy social cost; it also has more wide-ranging economic consequences. The higher the structural unemployment in an economy, the lower its potential output, and hence the average standard of living, will be. Furthermore, high unemployment can endanger the credibility of monetary policy. For example, a rise in interest rates—for instance, to protect the level of the exchange rate—may be viewed as unsustainable by the financial markets and have a perverse impact. The crises in the European ERM during 1992 and 1993 are testimony to this.1 There are, therefore, many reasons to seek a better understanding of the problem of high unemployment.

This chapter investigates the causes of unemployment in France and assesses the effectiveness of potential policy responses aimed at reducing it. The chapter is organized as follows: Section I examines the nature of unemployment in France; Section II identifies some of the potential factors contributing to unemployment by comparing France with other industrial nations and by using cross-section regressions; Section III provides some time-series evidence on the long-run and short-run causes of unemployment; Sections IV and V examine recent labor market measures; and Section VI assesses these measures and considers other potential policy responses to structural unemployment.

I. History and Composition of French Unemployment

The Rise and Rise of Unemployment

Following the first oil shock and the onset of recession in 1974–75, unemployment in France rose continuously from just under 3 percent in 1973 to a peak of 10.7 percent in 1987. It then declined gradually to 8.8 percent in 1990 before resuming its ascent (Chart 1).

Chart 1.
Chart 1.

France: Unemployment Growth, and Inflation

Sources: Institut National de la Statistique et des Etudes Economiques (INSEE), National Accounts; and IMF World Economic Outlook.1To April 1993.

Both labor supply and demand factors appear to have contributed to the rise in unemployment. Between 1970 and 1992, the labor force grew by about 16 percent, equivalent to 3.3 million people (Chart 2). This growth came about in spite of a fall in the aggregate participation rate2 during the 1980s—the increase in female participation was more than offset by a decline in male participation (Chart 2).3 The growth in the labor force was thus concurrent with an even larger increase in the population of working age, conventionally defined as those aged between 15 and 64.

Chart 2.
Chart 2.

France: Labor Force, Population, and Participation

Source: Organization for Economic Cooperation and Development (OECD), Historical Statistics.

During the same period, employment growth has been disappointing: there was no net increase in employment between the mid-1970s and the late 1980s. Male employment dropped almost continuously between 1973 and 1987 as employment in the traditionally male-dominated industrial sectors such as mining, chemicals, and metals declined. Growth in the service sector and opportunities for part-time work boosted female employment, which has grown continuously since the early 1970s, albeit not as fast as the increase in the female labor force (Chart 3).

Chart 3.
Chart 3.

France: Labor Force, Employment, and Unemployment

(In millions of persons)

Source: OECD, Historical Statistics.

Why has the high growth of the labor force not been accompanied by an equivalent increase in employment? An increase in the labor supply should, in theory, lead to higher output through lower real wages, increased competitiveness, and increases in production capacity unless the labor market is inflexible. Besides, there have been similar demographic developments in other industrial countries: in both Canada and the United States the average rate of growth of the labor force since 1970 has been more than twice that in France, yet the average rate of growth of employment in the United States and Canada has been over four times that in France.

Over the last two decades the average rates of growth of the population of working age and of the labor force in France have been below the OECD average, though more or less in line with the EU average (Table 1). It is the rate of growth of employment in France that has been disappointing, particularly when compared with the OECD average in the 1980s.

Table 1.

Key Labor Market Indicators

(Average annual percentage growth rates)

article image
Source: OECD, Historical Statistics.

Composition of Unemployment

Although the trend rise in unemployment can be observed in many other industrial countries, two factors characterize the French experience: unemployment peaked somewhat later than in the other major industrial countries, and the subsequent decline was not as sharp (Chart 1). This persistence suggests structural imbalances in the labor market that are also reflected in the composition of unemployment.

Youth Unemployment

The youth unemployment rate has been consistently higher than the total unemployment rate and more sensitive to economic downturn (Chart 4). In general, one would expect the youth unemployment rate to be above the total unemployment rate as the young have lower human capital. Labor market practices such as “last in, first out” also reinforce youth unemployment and explain its sensitivity to economic conditions. Furthermore, there is some evidence that wage bargainers are primarily concerned with the interests of those who are employed, that is, the “insiders,” rather than those who seek employment, the “outsiders.” For example, the insiders may take advantage of high turnover costs or the uncertainties associated with employing an outsider to demand higher wages than are justified by productivity or inflation. This theory also helps to elucidate the persistence in youth unemployment. However, labor market features such as “last in, first out” or “insider-outsider” are not unique to France, yet youth unemployment in France is high compared with other industrial nations (Chart 5). This suggests that other more specific factors are also at work. The minimum wage and high employer payroll taxes could have a more pronounced impact on the employment of the young (see Section II below). Furthermore, many young people are poorly qualified or lack the skills sought by employers. In 1991, the rate of unemployment among 15–24-year-olds with only the certificat d’etudes primaires (CEP) was twice as high as for those with the baccalaureat-level diploma.

Chart 4.
Chart 4.

France: Unemployment—Youth, Female, and Male

(In percent)

Source: INSEE.1To April 1993.
Chart 5.
Chart 5.

France: Relative Unemployment Rates, 1992

(In percent)

Source: OECD.

Female Unemployment

Unemployment among women has been consistently above the average unemployment rate (Chart 4). There are a number of potential causes. Part-time working, especially among women, remains less developed in France than in the other major industrial countries (Table 2).

Table 2.

Size and Composition of Part-Time Employment

(In percent)

article image
Source: OECD, Employment Outlook, 1993.

Although jobs in the service sector—which provides substantial female employment—have grown faster than the average rate of increase in employment, this rate of growth has been lower than that achieved in the United States, the United Kingdom, Italy, and Japan.4 There is also some evidence that, for women in particular, the probability of finding a job rises as the maximum duration of benefits draws near (see Section II below). In France, benefits are available for a longer duration than in most other industrial nations and the ratio of long-term to short-term replacement rates for women was high until January 1993, when the Government introduced a new system whereby unemployment benefits decrease more rapidly over time (see Sections II and III below).

Long-Term Unemployment

About 40 percent of those who are unemployed have been out of a job for at least one year. Although this is not exceptionally high when compared with other industrial economies (Chart 6), it is high in absolute terms. There is considerable evidence that the long-term unemployed exert no pressure on wage inflation (Layard, Nickell, and Jackman (1991)). Therefore, long-term unemployment could reduce the potential output of the economy through a higher natural rate of unemployment. Pissarides (1982) illustrates that human capital depreciates as the duration of unemployment increases, leading to a lower probability of finding a job.

Chart 6.
Chart 6.

France: Relative Unemployment Rates, 1991

Source: OECD, Employment Outlook, 1992.

With regard to the demand-side causes of long-term unemployment, employers faced with continued uncertainty about future orders may refrain from recruitment for long periods of time, causing the persistence of both unemployment and long-term unemployment, particularly if wages are not flexible. Labor market rigidities, such as the relative generosity of long-term benefits, employment protection legislation, the minimum wage, and high employer costs are other potential contributory factors. These factors are investigated further below. The adoption of a number of job creation and training programs makes the study of long-term unemployment even more complex. Cases and Lollivier (1993) characterize the French labor market as being divided into three segments: unemployment, regular employment, and marginal employment. Many job schemes fall into the last category, and there are flows between all three segments. Transition from long-term unemployment to labor market schemes and vice versa could lead to recurrent spells of unemployment rather than to long-term unemployment.

Nonemployment Rate

Another useful concept in comparing the labor market experience of different countries is the nonemployment rate,5 which combines the effects of high unemployment and low participation in the labor force. In addition to the high rate of unemployment, France has a low participation rate, particularly among men (Chart 7). The participation rate in France drops dramatically for those aged 56 and above when compared with the average of EU countries (Table 3), as a consequence of generous early retirement schemes.6

Chart 7.
Chart 7.

France: Relative Participation Rates, 1991

Source: OECD, Employment Outlook, 1992.
Table 3.

Participation Rates by Age Group, 1991

(In percentages)

article image
Source: Eurostat, Labour Force Survey Results, 1991.

The high nonemployment rate in France is more comparable with the southern European and the Benelux countries than with the other industrial countries (Chart 8). This underutilization of labor imposes a heavy burden on government finances, through both lost tax revenue and increased social security expenditure. This burden will become increasingly serious as the proportion of the elderly in the population rises, given the need to finance pension expenditures, which operate on a pay-asyougo basis in France. Not surprisingly, cross-country evidence suggests that the higher the nonemployment rate, the higher the budgetary expenditure on the labor market (Chart 9).

Chart 8.
Chart 8.

France: Relative Nonemployment Rates, 1991

Source: OECD, Employment Outlook, 1992.
Chart 9.
Chart 9.

Labor Market Expenditure and Social Security Contributions

Source: OECD, Employment Outlook, 1992

II. Causes of Unemployment

Employers’ Taxes

Social security contributions in France, particularly for employers, are very high. Total employee and employer contributions as a ratio to income are higher than in all other industrial countries except Italy (Chart 9, lower panel). Personal income taxes, on the other hand, are lower than in most other countries. Table 4 provides a more detailed comparison of labor costs in France and other major European countries for an unmarried worker. Gross labor costs in France, although below those in Italy, are substantially higher as a proportion of net earnings than in Germany and the United Kingdom. In fact, the gap with the latter two countries widened in the 1980s. The main contributor to this differential is employers’ social security contributions, which are not only substantially higher than in Germany and the United Kingdom, but also increased significantly during the 1980s.

Table 4.

Comparison of Earnings, Direct Taxes, and Social Security Contributions

(As a percentage of net earnings for a single worker))

article image
Source: Commissariat général du Plan (1993).

Table 4 also shows that the contribution rates of employees are noticeably higher in France than in Italy and the United Kingdom, but marginally below those in Germany. Furthermore, contributions as a ratio to net earnings increased by some 50 percent during the 1980s. Income taxes in France, on the other hand, are about one third of those in the other three countries as a proportion of net earnings, and they declined marginally during the 1980s.

From a theoretical perspective, the invariance of incidence proposition (IIP) implies that the replacement of an employer tax by an equal employee tax has no effect on the real economy; that is, the product wage, the consumption wage, and the level of employment will be unaffected (Layard, Nickell, and Jackman (1991); Newell and Symons (1986); and OECD (1990)). However, this result may not apply—at least in the short run—if there are market imperfections; for instance, if wages are above market-clearing values and adjust slowly, or wage negotiators care only about the “insiders.” Even if the IIP holds, differences in tax liability could alter the allocation of resources. For example, employer taxes apply only to the wage whereas personal or corporate income taxes also apply to income from capital. A switch from income taxes to employer taxes implies an increase in the overall rate of taxation on employment and a decrease in capital taxation, which may lead to a substitution of capital for labor. Similarly, personal income taxes are usually progressive whereas social security taxes are not. A switch from one to the other could lead to changes in the total tax bill for many individuals and firms and affect labor supply or demand decisions.

Empirical studies (Newell and Symons (1986); and OECD (1990)) suggest that, even when using a model where the IIP holds in the long run, a cut in employer taxes and an equivalent rise in employee taxes could reduce unemployment in the short run. The short-run effects may last for several years because of lags in the adjustment process (OECD (1988)). Cotis and Loufir (1990) find support for the IIP in France (see below). Other authors have found that the tax wedge has a long-run effect on unemployment in France (Bean, Layard, and Nickell (1987)). We investigate this issue further in Section III using time-series data for France.Table 5 provides some cross-section evidence on the impact of employer taxes on the composition of unemployment in the industrial countries.

Table 5.

Employer Taxes, Youth and Long-Term Unemployment

article image

Organization for Economic Cooperation and Development members Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Japan, the Netherlands, Norway, Portugal, Spain, Sweden, the United Kingdom, and the United States.

The first equation gives the result of regressing the ratio of youth to total unemployment on the rate of employers’ social security contributions using cross-section data for 15 OECD countries. The results indicate a significant positive correlation between the two variables. The coefficient on employers’ contributions indicates that if they are 1 percentage point higher in one country than in another, then the ratio of youth to total unemployment is likely to be 1 percentage point higher. In France, employer contributions are about 20 percentage points higher than the sample average, implying a similar gap for the percentage of youth unemployment. The second regression presents results for long-term unemployment that are very similar; they indicate that higher employer contributions are associated with higher long-term unemployment. When the same regressions were performed for men and women separately, the coefficients were higher in the case of women.

Admittedly, these results do not capture any theoretical relationship. However, they do suggest that high employer taxes are associated with a high incidence of youth and long-term unemployment, which are symptoms of structural unemployment. The results imply that increases in employers’ contributions have not been passed on to employees through wage reductions.

The Minimum Wage

France has had an institutional minimum wage since 1950; the current SMIC (salaire minimum interprofessionnel de croissance) was introduced in 1970. There are three mechanisms for revising the SMIC: (1) a rise of 2 percent or more in the consumer price index (CPI) automatically triggers an equivalent rise in the SMIC; (2) on July 1 every year the SMIC is revised by at least half of the increase in the real hourly wages in industry; and (3) the Government can raise the SMIC at its discretion.7 These mechanisms have led to a steady rise in the real value of the SMIC.

The latest report on the SMIC by the Ministere du Travail shows that the number of people being paid the SMIC has risen over the last five years.8 In 1992, 8.6 percent of workers or just under 2 million people received the minimum wage, with the percentage among women being much higher (Table 6). The proportion of individuals earning the SMIC is particularly high for those who are under 26 years old: 35.5 percent of wage earners under 26 were paid the SMIC in 1992. The fact that a large number of young workers, and a growing number of people in general, are paid the SMIC suggests that it is a significant labor market factor that may have an impact on employment.

Table 6.

Proportion of Employees Covered by the SMIC1

(In percent)

article image
Sources: Bazen and Martin (1991); and Ministere du Travail, de I’Emploi, et de la Formation Professionnelle.

The data refer to July of each year and cover wage and salary earners in businesses employing more than ten workers in industry, commerce, and services. They cover all workers whose hourly wage is less than the new hourly SMIC rate that applies from July 1 of each year.

New series.

Empirical studies on the effect of the SMIC on employment have not been conclusive. Bazen and Martin (1991) find that “increases in the real value of the SMIC have exerted significant upward pressure on real youth earnings.” However, they are unable to show conclusively that increases in real labor costs have had a negative impact on youth employment, although they believe this to be the case. Their results do suggest that moderating the relative rate of increase in the SMIC would have a favorable impact on youth employment. Chart 10 also suggests a correlation between the relative minimum wage and youth unemployment. In response to high youth unemployment, the Government has introduced a number of labor market measures (Chart 10, lower panel). Many labor market schemes are targeted at the young (see Section IV below). Given the large increase in the number of places on employment programs between 1984 and 1987, it is difficult to assess whether the fall in youth unemployment after 1984 was due to the fall in the relative value of the SMIC or an increase in labor market programs. It is likely that both played a role.

Chart 10.
Chart 10.

France: Youth Unemployment Relative Minimum Wage, and Labor Market Programs

Sources: INSEE; OECD; and IMF staff estimates.

By adding to wage rigidity—in particular, the rigidity of relative wages by limiting wage differentials—the SMIC has several implications.9 To the extent that the productivity of the less skilled and less qualified is below that justified by the SMIC, they are likely to suffer unemployment. This no doubt helps explain France’s high rate of youth unemployment, and also the persistence of unemployment. The SMIC may also be an obstacle to regional labor mobility, as it is uniform across the country and thus does not reflect regional differences in the cost of living that imply different real values for the SMIC.

The experience of other industrial countries with a statutory minimum wage is also relevant. Empirical work in the United States and Canada has concluded that the minimum wage has had a negative impact on youth employment.10 It is therefore instructive to compare the level of the minimum wage in France with that in other industrial countries.

The cost of employing someone at the minimum wage in France is about twice that in the United States. This is due to two factors. First, the minimum wage in France is about 1.5 times that in the United States, and, second, employers’ social security contribution rates are 38 percent in France but only 7.7 percent in the United States11 (Table 7 on previous page).

Table 7.

Minimum Wage in France and the United States, July 1992

(In francs)

article image
Sources: Bulletin Mensuel de Statistique; and U.S. Employment and Earnings.

On the basis of working 169 hours a month.

Using an exchange rate of $1 = F 5.5.

A comparison with Belgium and the Netherlands is interesting because, unlike France, these countries have a lower minimum wage for the young than for other workers (Table 8). In France the minimum wage, at least in theory, applies to anyone aged 18 or above.12 In the Netherlands the full minimum wage applies to those aged 23 and above. For those under 23, the minimum wage is reduced sharply.13 In Belgium the full minimum wage applies to those aged 21 and above and there are lower rates for those under 21, although the reduction is not as sharp as in the Netherlands.14

Table 8.

Minimum Wage for an 18-Year-Old, July 1992

(In francs)

article image
Sources: Bulletin Mensuel de Statistique; U.S. Employment and Earnings; and Van den Heuvel (1992).

Using $1 = F 5.5, and July 1992 exchange rates for other currencies.

In the United States there is a reduced minimum wage for 16-19-year-olds that is 85 percent of the full minimum wage.

The existence of the youth minimum wage means that, in absolute terms, it is cheaper for an employer to employ an 18-year-old in the Netherlands or Belgium than in France, although the full minimum wage is higher in the former countries. Interestingly, the ratio of youth to total unemployment in the Netherlands is among the lowest in the industrial countries (Charts 5 and 11).

Chart 11.
Chart 11.

Ratios of Youth to Total Unemployment 1992

Source: OECD.

There is, therefore, considerable evidence that the minimum wage in France is high by international standards, especially for the young. There is also some evidence that the minimum wage affects youth, if not total, unemployment. We will pursue this issue further by doing some time-series analysis of data for France in Section III, after we have briefly considered the other potential causes of unemployment.

Generosity of Benefits

Most models of wage determination imply that benefits have a significant effect on wages (Layard and Jackman (1991)). The impact of the replacement rate (i.e., the ratio of unemployment benefits to wages) on the decision to seek employment very much depends on the individual’s circumstances and is difficult to measure at an aggregate level. Besides, some countries with a high replacement rate, such as Sweden and Norway, have low unemployment and a high participation rate. The important issue is not so much the absolute level of benefits, or even the replacement ratio, but the incentive structure of the benefit system and the characteristics of the individual (see, for example, Schmitt and Wadsworth (1993)).

One aspect of this is the duration structure of benefits. Table 9 gives the result of regressing the proportion of long-term unemployed on the ratio of the long-term to short-term replacement rate using cross-section data for 15 OECD countries. Separate regressions were performed for men and women.

Table 9.

Long-Term Unemployment and Relative Generosity of Benefits

article image

Organization for Economic Cooperation and Development members Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Japan, the Netherlands, Norway, Portugal, Spain, Sweden, the United Kingdom, and the United States.

Table 9 shows that there is a significant positive relationship between long-term unemployment and the generosity of long-term relative to short-term unemployment benefits for both men and women. The equation for men suggests that if the ratio of long-term to short-term unemployment benefits is 1 percentage point higher, then long-term unemployment will be 0.46 percentage points higher. Interestingly, the corresponding figure for women is 0.60 percentage points. In the data set, which corresponds to 1991, the ratio of short-term to long-term replacement rates for France is 17 percentage points higher than the sample average for men, and 22 percentage points higher for women. The introduction of the new unemployment benefits system will reduce these ratios (see Section IV).

There are other features of the benefits system that may also have an impact on unemployment. According to the EU Commission, unemployment benefits were paid to 43 percent of those without work in France in 1989, which is substantially above the EU average of 30 percent. One cause of this phenomenon could be the duration unemployment benefits are available in France. Table 10 provides data on the maximum duration of benefits for the OECD countries. Benefit duration in France is among the longest in the OECD countries. Only in Belgium, Denmark, and the Netherlands are benefits available for a longer period.

Table 10.

Maximum Duration of Unemployment Benefits and Replacement Rates

(In weeks)

article image
Source: OECD, Employment Outlook, 1993.

Institutional Factors

Structural unemployment is also affected by the benefits administration system, hiring and firing regulations, and other institutional factors that could discourage employers from taking on new workers or the unemployed from searching for jobs. A few of these factors are examined in this section.

Unemployment Benefit Administration

France does not seem out of line with regard to the minimum waiting period before the receipt of benefits; as in many other industrial countries, there is no minimum period. A number of countries have a minimum waiting period of a few days. However, if a person has left his job voluntarily, he would be completely disqualified from receiving benefits in France, whereas in many countries there is a waiting period of several weeks, but not a complete disqualification (Table 11).

Table 11.

Periods for Which No Insurance Benefits Are Paid at the Start of an Unemployment Spell

article image
Source: OECD, Employment Outlook, 1991.

Regulations concerning signing on in France are rather liberal: confirmation of unemployment status is done on a monthly basis by post (Table 12). However, France does have a system of intensified interviews by the placement and benefit administration services (Table 13).

Table 12.

Signing On and Other Regular Reporting by Benefit Claimants

article image
Source: OECD, Employment Outlook, 1991.
Table 13.

Scheduling of Intensified Interviews by Unemployment Duration

article image
Source: OECD, Employment Outlook, 1991.

Hiring and Firing Legislation

The 1986 work dismissal legislation in France requires that those being dismissed are given one to two months’ written notice of dismissal and informed in writing of the reasons for their dismissal. At the same time this legislation removed the need to seek the permission of the Labor Ministry for collective dismissals. Other legislation in 1989 obliged companies to prove the cause of dismissal. The law in France also requires that workers be paid up to a maximum of one and a half months’ salary as severance pay. Though the legal restrictions on dismissal in France are less strict than those in the southern European countries, they are more strict than those outside Europe, particularly in the United States.

Barriers to nontraditional forms of work such as part-time or temporary work are often viewed as a form of rigidity in the labor market. We noted above that part-time working is less developed in France than in many other industrial countries (Table 2). In order to assist part-time working, the Government passed legislation in December 1992 exempting employers from 50 percent of social security contributions if they took on part-time workers.

The proportion of people working on temporary contracts in France has risen continuously over the last decade from 3.3 percent of employment in 1983 to 10.2 percent in 1991 (Table 14). However, the regulations governing fixed-term contracts remain more strict in France than in most other industrial countries (Table 15). For example, in some circumstances the use of temporary contracts is restricted; there is a legal maximum duration for fixed-term contracts; and employees are entitled to termination benefits. The barriers to temporary work have to be considered in conjunction with the constraints on permanent employment such as employer taxes or hiring and firing regulations. If the regulations governing permanent employment were not rigid then employers would make use of temporary contracts only to the extent that they help in adjusting to changes in demand. However, the existence of barriers to permanent employment would encourage greater use of temporary contracts. Therefore, given the existence of barriers to permanent employment in France and the fact that the regulations governing temporary work are more strict than in most other countries, it is difficult to argue that the rise in the proportion of temporary workers in France is a sign of enhanced labor market flexibility.

Table 14.

Temporary Workers

(As a percentage of total employment)

article image
Source: OECD, Employment Outlook, 1993.
Table 15.

Temporary Work: Fixed-Term Contract Regulations and Requirements, 1990

article image
Source: OECD, Employment Outlook, 1993.

Maximum duration in months.

Possibility of renewal and the number of times a contract may be renewed.

Another potential constraint on hiring nontemporary staff in France is that the employment agency, ANPE (Agence Nationale pour l’Emploi), currently has a monopoly in placing nontemporary staff. This was originally enacted to prevent the exploitation of the unemployed, but it is no longer enforced. Firms regularly employ outside the ANPE. Such legal restrictions also exist in a number of other European countries (Table 16). Even though this legislation is not currently enforced, it could act as a disincentive to private firms considering entering the placement market if they fear that the legislation may be invoked in the future.

Table 16.

Legal Constraints on Methods of Filling Vacancies

article image
Source: OECD, Employment Outlook, 1991.

Mismatch

Evidence on the mismatch between the skills held by workers and those demanded by employers is rather difficult to assemble. For example, indicators based on unemployment by occupation consider only the supply side of the market, and, since skills are defined on the basis of the last job held, they exclude all the unemployed without previous experience.

Data on the characteristics of the unemployed suggest some degree of mismatch. Youth unemployment is highest among those with the least qualifications (OECD (1992b)). The long-term unemployed tend to have lower educational attainment than average and are concentrated in the least-qualified occupations (Caracosta, Fleurbaey, and Leroy (1991)). As for regional mismatch, both the uniformity of the SMIC across the country and the high taxation of property transfers are hindrances to regional mobility.

Using time-series measures of mismatch based on the distribution of vacancies and unemployment by industry and occupation, Jackman and Roper (1987) and Jackman, Layard, and Savouri (1991) argue that mismatch in France, and indeed in most of Europe, did not increase markedly during the 1980s. However, Entorf (1993) illustrates that time-series measures of mismatch may lead to the wrong conclusions since these measures are biased downward when unemployment is rising. Entorf also presents some micro data that suggest that regional and occupational mobility declined in Europe during the 1980s.

A simple measure of mismatch based on the variance of unemployment15 indicates that mismatch increased in France during the 1980s. For example, regional mismatch increased from 1.27 percent in 1982 to 1.88 percent in 1990. However, regional mismatch fell to 1.48 percent in 1992 because the rate of increase in unemployment was higher in regions with lower unemployment.16 The same picture emerges when considering unemployment by professional status; because unemployment has risen sharply among white-collar workers (who had a relatively low unemployment rate), mismatch falls from 12.5 percent in 1990 to 11.9 percent in 1992.

Chart 12 depicts a more formal measure of mismatch, the unemploymentvacancy (UV) ratio (or Beveridge curve), for the period 1974–92. An outward shift of the curve seems to have occurred: between 1978 and 1983 the same ratio of vacancies to the labor force was coupled with a higher unemployment rate than before. During the period 1984–87, unemployment rose while vacancies were on the rise, albeit modestly. This behavior is indicative of a small increase in mismatch.17

Chart 12.
Chart 12.

France: Vacancies and Labor Market Expenditure

Source: OECD, Employment Outlook, 1992.

France spends 2.7 percent of its GDP on the labor market; the average for the OECD is 2 percent, and for the major industrial countries excluding France, 1.1 percent. However, France spends much less than most other industrial countries on active labor market measures such as training, and more on passive measures such as unemployment compensation (Chart 12, lower panel). Only the former are likely to alleviate labor market mismatch.

III. Causes of Unemployment—Time-Series Evidence

This section presents a time-series econometric model of the determinants of the NAIRU (non-accelerating-inflation rate of unemployment) or the natural rate of unemployment. The primary interest here is not to estimate a numerical value for the NAIRU but to identify a set of policy instruments that could be used to reduce it.

The Model

To understand the causes of unemployment in Britain, Layard and Nickell (1986), in an influential study, constructed an empirical model consisting of equations for prices, wages, employment, and the trade balance. Here we estimate a simplified version of their model that consists of two equations: a price equation and a wage equation. In this model, firms are assumed to operate in an imperfectly competitive environment, setting their prices on the basis of cost and demand. A key feature of the model is that wages are influenced by pressure variables such as those reviewed in the last section, for example, mismatch, the minimum wage, generosity of benefits, or import prices. If increases in the wage pressure variables lead to higher real wages, then unemployment would have to rise for price inflation to be kept in check. More formally, the model assumes that stable inflation requires consistency between two concepts:

(i) the target real wage—firms and workers bargain about nominal wages as a markup on expected value-added prices

w* = target(w – p);18 and

(ii) the feasible real wage—in an imperfectly competitive market the firms are thought of as setting value-added prices as a markup on wages

p* = feasible(p – w).

The economic variable that brings about this consistency in the long run is unemployment (Layard, Nickell, and Jackman (1991)).

The model is given below in equations (1)–(4). Since we are interested in the NAIRU, which is the long-run equilibrium concept for the unemployment rate, we have separated the long-run relationships from the dynamic ones. This is also desirable from an estimation point of view (Wren-Lewis (1990)). Equations (1) and (2) give the long-run wage and price equations; equations (3) and (4) give the dynamic framework for wages and prices.

w*=pr+α1U+α2Zw+α3(pmp),(1)p*=β1(ydy¯)+β2Zpr,(2)w=w*+p+θ1(L)Δwθ2(L)Δp+θ3(L)Dw(3)p=p*+w+φ1(L)Δpφ2(L)Δw+φ3(L)Dp(4)

where w* = target real wage (i.e., target (w - p)), p* = feasible real wage (i.e., feasible (p - w)), w = nominal hourly earnings, pr = productivity or unit labor costs, p = output prices, U = unemployment rate, Pm = import prices, (yd - ӯ) = actual demand relative to potential output, Zw = other wage pressure variables, Zp = other variables affecting firms’ margin, Dw, Dp = other factors affecting the dynamics of wages and prices, and Δ is the difference operator. All lower-case variables are logged. θ and Φ are polynomials in the lag operator L and have both positive and negative powers of L, with forward terms appearing as expectations. It is also assumed that polynomials are such that the target real wage is independent of the rate of inflation, that is, dynamic homogeneity holds.

Combining equations (3) and (4) gives

[θ2(L)φ1(L)]Δp=w*+p*+[θ1(L)φ2(L)]Δw+[θ3(L)Dw+φ3(L)Dp].(5)

The left-hand side of equation (5) is a polynomial in changes in prices. The right-hand side of equation (5) can be regarded as a measure of inflationary pressure, which is made up of three elements: (i) w* + p*, the difference between the target real wage and the feasible real wage; (ii) a polynomial in the dynamics of wage contracts; and (iii) other short-run influences on wages and prices. The first element, w* + p*, can be derived from combining equations (1) and (2):

W*+p*=α1U+β1(ydy¯)+α2Zw+β2Zp+α3(pmp)(6)

In a steady state where inflation is constant, equation (5) implies that w* + p* = 0. Therefore equation (6) becomes a long-run relationship between the NAIRU, the level of actual demand relative to potential, (yd - ӯ), Zw, Zp, and import prices. For a given level of demand, equation (6) would determine the long-run influences on the NAIRU.

The Results

Because we are interested in estimating long-run relationships, an obvious way forward is to employ cointegration techniques.19 The basic idea of cointegration is that two or more variables may be regarded as defining a long-run equilibrium relationship if they move closely together in the long run, even though they may drift apart in the short run. This long-run relationship is referred to as a cointegrating vector. Because there is a long-run relationship between the variables, a regression containing all the variables of a cointegrating vector will have a stationary error term, even if none of the variables taken alone is stationary.

We have directly estimated the long-run relationships in equations (1) and (2) above using the Johansen procedure and quarterly data for the period 1971:1-1992:IV.20 The test statistics and the estimated cointegrating vectors from the Johansen procedure are reported in Appendix I. The cointegration tests reveal the following unique wage and price equations:

wp=pr+0.59smic0.07U+0.15SK+0.12(pmp),and(7)
pw=pr+0.02CU+0.26ρ+0.43t1.(8)

The wage equation contains U and pm - p as suggested by equation (1) as well as the log of the real value of the SMIC (denoted as smic) and an index of skill shortages, SK, from the EU Commission’s quarterly survey of employers.21 The latter two represent wage pressure variables Zw in equation (1). The price equation contains capacity utilization, CU, the cost of capital, ρ, and the employers’ tax wedge, t 1. The capacity utilization variable is a proxy for demand relative to potential output, (yd - ӯ). This variable is available from business survey data. The cost of capital, ρ, and the employers’ tax wedge, t 1, represent Zp in equation (2) and are discussed further below. A unit coefficient on the productivity term, pr, is imposed in both equations as suggested by equations (1) and (2). A number of other specifications for both the target real wage and the target price markup were also tried (see below).

In the cointegrating vector for the real wage, all of the estimated coefficients have the expected signs and are of plausible magnitudes. Two of the variables, U and SK, enter not as logs but as levels. Unemployment has the expected negative effect on wages. The equation indicates that if the import price wedge rises, wage earners would resist a fall in the real consumption wage (i.e., deflated by consumer prices) by pushing up the real product wage. The other two variables, smic and SK, are of policy interest. The minimum wage variable captures the downward rigidity of wages and the index of skill shortages can be viewed as a proxy for mismatch.22 The wage equation here is similar to that used in a paper by Blanchard and Muet (1993); however, instead of the smic and the SK variables, they used a price inflation variable in the wage equation. We were unable to obtain cointegration using this formulation. The approach here has the advantage of directly identifying wage pressure variables; besides, conceptually it would be unusual to have inflation in a long-run vector.

In the cointegrating vector for prices, the estimated coefficients again have the expected signs and are of reasonable magnitudes. The coefficient of the cost of capital, ρ, and the employers’ tax wedge, t1, imply that, in the long run, employers pass on some of the increase in their costs by reducing earnings. Both of these variables are strongly influenced by economic policy choices. We find that tax wedges have no long-run effect in the wage equation, a result similar to that obtained by Cotis and Loufir (1990). However, in our formulation the employers’ tax wedge has a longrun effect on unemployment through the price equation.

A number of other variables and formulations were also tried but with no success. In the wage equation we tried other forms of mismatch; direct, indirect, and the employers’ tax wedge; unemployment in logs rather than as a rate; and an aggregate measure of benefits. In the price equation we also tried tax wedges and changes in consumer expenditure. However, none produced plausible results.

Since our primary interest is in the long-run determinants of unemployment, we shall not dwell any further on the long-run wage and price equations and shall proceed to obtain a long-run equation of the form given by equation (6) above. This can be done in the steady state by simply adding the vectors in equations (7) and (8) and eliminating wages and prices to obtain

U0.3CU=8.4smic+3.7ρ+6.1t1+2.1SK+1.7(pmp).(9)

Equation (9) is not, strictly speaking, a NAIRU equation because it has a term in capacity utilization, CU.23 This can be illustrated using Chart 13 below. Equation (9) is represented by the lines AA’ and BB’. The right-hand side variables in equation (9) will determine a schedule of values for unemployment and capacity utilization. A cut, for instance, in the real value of the SMIC would shift AA’ to, perhaps, BB’ so that, for a given level of capacity utilization, employment would be higher and unemployment lower.

Chart 13.
Chart 13.

Employment and Capacity Utilization

To obtain an equation for NAIRU it is necessary to eliminate CU from equation (9). This can be done by estimating an upward-sloping schedule, such as CC, between unemployment and capacity utilization that reflects a linkage through aggregate demand: a rise in capacity utilization would increase the demand for labor and reduce unemployment. Here we have estimated a very simple cointegrating relationship between CU and U using the Johansen procedure:

0=CU+0.25U.(10)

Substituting for CU from equation (10) into equation (9) gives

U=7.8smic+3.4ρ+5.7t1+2.0SK+1.6(pmp).(11)

Equation (11) implies that, in the long run, unemployment would be lower when the following variables are lower:

  • (i) the real minimum wage

  • (ii) employers’ tax wedge

  • (iii) skill shortages

  • (iv) the cost of capital, and

  • (v) real import prices.

Using the actual values of the right-hand variables in equation (11) gives a NAIRU of 8.2 percent in 1992.

Appendix II provides estimates of the dynamic equations (3) and (4). The dynamic equations illustrate that reductions in direct and indirect taxes would also have an impact on unemployment through wage and price dynamics, but only in the short run. However, given the long lags on these variables, the impact could last as long as two years. The fact that reductions in direct and indirect taxes have only a temporary impact, whereas a cut in employers’ taxes has a long-run effect, is a violation of the IIP. However, this could be because the cost to the employer of employing someone at a wage of the SMIC plus employers’ taxes is well above the market-clearing rate. In these circumstances, reductions in employer taxes and in the real value of the SMIC could help generate new jobs.

IV. Recent Labor Market Measures

Sections II and III identified some of the causes of unemployment in France; in this and the next section we examine the recent policy responses. The most notable recent labor market initiative is the five-year employment plan that was approved by parliament in November 1993. Prior to that, four principal labor market measures had been implemented: a new unemployment benefit system was put in place in January 1993; an action plan for the long-term unemployed was introduced; a number of labor market programs, particularly those aimed at the young, were expanded; and payroll taxes for the low paid were reduced. These four measures are reviewed below; a more detailed analysis of the five-year employment plan follows in the next section.

The Unemployment Benefit System

The new benefit system, the allocation unique degressive (AUD), came into effect in January 1993. The benefits are financed by social security contributions, as before. The novel feature of the new system is that, after an initial period, benefits are progressively reduced every four months. The previous salary and the period of contribution to the system determine the initial level of the benefit. For someone who has worked for at least six months earning the SMIC, the initial replacement rate would be 75 percent gross or 83.8 percent net of contributions (Table 17); it would then decline over time. For someone earning F 20,000 per month, the initial replacement rate is equal to 57.4 percent gross or 67 percent net (Table 17).

Table 17.

Replacement Ratio

article image
Source: French authorities.

The initial period for which benefits are paid at the maximum level and the subsequent four-monthly reductions in the benefit are determined by the age of the claimant and the period of contribution to the system (Table 18). For example, someone earning F 9,000 per month who is less than 25 years old and has paid 14 months’ contributions in the last two years would receive the maximum level of benefit for nine months. Thereafter, benefits would be reduced by 17 percent every four months, and the person would receive benefits for a maximum of 30 months (Table 18).

Table 18.

Duration of Benefits

article image
Source: French authorities.

For this category, benefits are reduced by 25 percent at commencement.

Action Plan for the Long-Term Unemployed

In February 1992 the Government introduced a program of systematically interviewing the long-term unemployed, action d’insertion et deformation (AIF). The program aimed to identify the training needs and job prospects of the long-term unemployed. Between February and November 1992 about one million long-term unemployed were interviewed. Following these interviews, the rate of outflow of the long-term unemployed increased appreciably in spite of increasing difficulty in the economic situation. Thirty percent of the interviews led to a job or a training scheme, 25 percent of the interviews led to no action, and 16 percent of the interviewees were taken off the unemployment register. The rise in the placement of the long-term unemployed during the interview period was accompanied by a fall in the placement of short-term unemployed, indicating a substitution effect. However, this was in a period of economic slowdown and the exact magnitude of substitution is difficult to ascertain.

Labor Market Programs

Labor market programs have played an increasing role in the effort to stem the rise of unemployment (Chart 14). Numerous programs have been introduced: in July 1993, approximately 40 labor market programs were in existence providing assistance to some 1.9 million individuals.24 These programs provide aided employment in the public and private sector as well as early retirement and training provisions, and have had a significant impact on the labor market (Chart 14). The Observatoire Francais des Conjonctures Economiques (OFGE) estimates that between 1985 and 1987, when observed employment fell by 60,000 and unemployment rose by 170,000, labor market programs provided 370,000 jobs. Therefore, unemployment would have been higher in the absence of labor market programs.

Chart 14.
Chart 14.

France: Employment Unemployment, and Labor Market Programs

(In millions of persons)

Sources: French authorities; and IMF staff estimates.1Including those no longer seeking employment.

Many programs are targeted at those facing the greatest difficulty: the young, the long-term unemployed, and those on the minimum social security benefit revenu minimum d’insertion (RMI). In 1992 about half of the participants on employment and training schemes were under 26 years old. Table 19 provides a brief description of the schemes aimed at the young. Similar programs have been initiated for the long-term unemployed and those on minimum benefits. The most prominent are contrat emploi-solidarite (CES), which creates jobs deemed to be of public value in the noncommercial sector for the long-term unemployed and those on minimum benefits; and contrat deretour a l’emploi (CRE), which subsidizes the hiring of those who have been out of a job for over a year.

Table 19.

Labor Market Programs for the Young

article image
Source: French authorities.

In June 1993, the new Government announced a number of emergency measures to encourage employment. These measures essentially increased the level of the subsidies and the total funding of a number of existing programs, including

  • Increased funding for contrat d’orientation: the state will provide a subsidy of F 2,000 for the first three months and F 3,000 for the subsequent three months to an employer taking on a poorly qualified under-23-yearold for six months. The wage need be no higher than 30–65 percent of the SMIC.

  • Under contrat de qualification the Government will waive all social security contributions and provide up to F 10,000 to enterprises taking on an unqualified young person. The young person will be paid 35–75 percent of the SMIC and benefit from a training program that will lead to professional qualifications.

  • The state will pay F 5,000 for taking on apprentices under contrat d’apprentissage, and social security contributions will be waived. The apprentices would be entitled to 400 hours of training and be paid 25–78 percent of the SMIC.

  • Increased funding for contrat d’adaptation to enable young people with qualifications to gain experience. Social security contributions are reduced and employers will be paid F 2,000 for taking on a qualified under-25-year-old who will be paid less than 80 percent of the SMIC.

  • Finally, the number of places on CES were increased to 650,000 and the state aid to employers for taking on someone on CRE was raised from F 10,000 to F 20,000.

Introducing Payroll Tax Exemptions

In order to reduce the cost of hiring people at the bottom end of the pay scale, the Government exempted employers from paying family allowance contributions for workers earning up to 10 percent above the minimum wage in July 1993. The contribution for those paid between 10 and 20 percent above the SMIC was cut in half. This was financed by an increase in a more broadly based tax, contribution sociale généralisée (CSG). This measure is to be expanded under the five-year employment plan (see below).

V. The Five-Year Employment and Training Law

In August 1993, the Government unveiled a five-year employment and training plan aimed at addressing the structural problems of the labor market. The plan, which contains over 50 measures, was approved by parliament in November 1993. The following are some of the major initiatives contained in the plan:

Reductions in Employers’ Contributions

To build on the initiative introduced in July 1993 (see above) and further reduce the cost of employing low-skilled workers, the plan proposes that by 1998 employers will be exempt from making family allowance contributions for those paid less than 1.5 times the SMIC, and will contribute half the usual amount for those paid between 1.5 and 1.6 times the SMIC. This measure will be phased in gradually over the 1994-98 period, although it is immediately effective (as from January 1, 1994) for newly created firms (Table 20). In addition to this measure, until December 31, 1995, employers taking on up to three new employees will be exempt from paying employers’ contributions for two years. Furthermore, other employers’ social security contribution rates, except those for unemployment insurance and pensions, will be frozen until 1998.

Table 20.

Five-Year Employment Plan: Employers’ Family Allowance Contributions

article image
Source: French authorities.

Measures Aimed at the Young

As regards youth training, the plan envisages that the responsibility for training all under-26-year-olds will be devolved to the regions by the end of 1998. The Government also intends to double the number of places on apprenticeship programs and to reduce the qualifying age from 16 to 14. In each employment office a section dedicated to providing information and guidance to the young is to be set up, and a national committee will monitor progress on youth training and apprenticeship and publish a triennial report on the subject.

To promote youth employment, a wage subsidy of F 1,000 per head per month for nine months is to be paid to employers taking on a first-time young worker for at least 18 months. This measure will be in place until 1998, but the subsidy will be doubled for any young workers taken on before October 1994. This subsidy will be available to young workers irrespective of their qualifications, as long as they are not participating in another government program. Also, all private sector employers can take advantageof this subsidy as long as they have not announced any redundancies in the previous six months.25

Work-Time Flexibility and Reductions in Work Time

To enhance work-time flexibility, the 39-hour working week is to be replaced with an annual equivalent for an experimental period of two years.26 Part-time workers will be able to work up to 1,200 hours during an 18-month period and still receive unemployment benefits. Also, to encourage companies facing financial difficulties to switch from full-time to part-time work, rather than make people redundant, the law proposes to subsidize employers’ social security contributions and extend unemployment benefit coverage for workers making the transition from fulltime to part-time work.

The tight regulations on Sunday working are to be eased. Shops in touristareas will be able to open, and the local authorities will have greater freedom to grant local shops permission to open on Sundays.

As regards reductions in the hours of work, for an experimental period of three years any company that achieves a 15 percent reduction in work time, accompanied by at least a 10 percent increase in employment, will be eligible for a 40 percent reduction in employers’ contributions in the first year and 30 percent in the subsequent two years.27 The law also allows changes in working time to be made through direct negotiations between the employers and the employees at enterprise, industry, or sectoral level. In the absence of an agreement, the legal duration of full-time work remains at 39 hours per week.28

Other Initiatives

A number of existing training programs will be expanded. For example, the maximum duration of the CES will be doubled from 12 to 24 months and will be renewable up to three times, with state subsidies available for up to five years. In addition, employers will be exempt from social security contributions for up to five years if they hire someone on a CES.

To encourage people to provide part-time services such as running errands for the elderly, the government is to introduce a chèque service whereby individuals will be paid for these services in vouchers bought at the post office. Those paid in vouchers will be exempt from social security contributions. This is also seen as a way of discouraging black market work.

To promote entrepreneurship, the aid for setting up new businesses will be increased to F 32,000 and will be available not only to those who receive benefits but also those who are registered as seeking employment.

The law also requires the creation of enterprise committees consisting of elected members of the personnel in each company. The management has to submit to the enterprise committee an annual report covering the financial state of the company, training, part-time work, and related issues.

VI. Assessment and Conclusion

The slowdown in economic activity in France and in her trading partners in the early 1990s has undoubtedly contributed to higher unemployment. However, the persistence and the composition of unemployment suggest that it is necessary to address the structural imbalances in the labor market in order to ensure a substantial decline in unemployment when the recovery takes hold.29

The Government has taken a number of welcome measures in this regard. Table 21 attempts to assess the potential impact of the recent measures. The introduction of progressive reductions of unemployment benefits as the spell of unemployment lengthens (AUD) should, by lowering the ratio of the long-term to short-term replacement rate, reduce long-term unemployment (see Section II above). The initial replacement ratio, however, will remain rather high.

Table 21.

Potential Impact of Government Measures on Causes of Structural Unemployment

article image
Source: IMF staff assessment.

Evaluation of job and training prospects by means of individual interviews through AIF should assist the long-term unemployed in finding suitable jobs or training, thereby reducing mismatch as well as excessive reliance on unemployment benefits. The array of reinforced labor market measures for the young (contrat d’orientation, contrat de qualification, contrat d’apprentissage, contrat d’adaptation) has the effect of reducing the cost of hiring young people through reductions in employers’ taxes and the SMIC. The CES and CRE reduce the cost of employment through wage subsidies and contribution exemptions. The latter, in common with most youth programs, also has an element of training and should help reduce mismatch. There remains, however, a multitude of training programs with overlapping aims and complicated eligibility criteria.

The five-year employment plan has provided the opportunity to formulate a medium-term strategy to address the structural issues in the labor market. The plan contains a number of very positive measures. For example, the gradual reduction in employers’ family allowance contributions over the next five years will progressively reduce employment costs for the low paid. The plan also contains many beneficial training initiatives, notably the expansion of the apprenticeship programs. However, following the withdrawal of the CIP, the plan does not really address the minimum wage problem. Furthermore, some elements of the plan are unlikely to enhance labor market flexibility and reduce structural unemployment. For instance, although easing the regulation regarding working on Sunday and annualizing the 39-hour week will enhance work-time flexibility, reductions in hours of work will, at best, have no impact on unemployment. Reducing working time would lower unemployment if output were unaffected. If output were affected then such a policy might either be inflationary or leave unemployment unchanged. The alternative is to have an equivalent reduction in earnings, which is difficult to bring about. Not surprisingly, therefore, empirical evidence casts doubt on the effectiveness of this policy in reducing unemployment.30 Reductions in working time are best left to individual negotiations between employees and employers.

On the whole, most of the initiatives taken by the Government over the last few years are steps in the right direction. However, given the scale of unemployment in France, more ambitious measures are needed to ensure a substantial and rapid decline in structural unemployment when the recovery gathers pace. A few potential measures are given below; some could facilitate reductions in unemployment in the short term while others would only bear fruit in the long run.

Reduce the Duration and Generosity of Benefits

In spite of the recent reform of the benefits system, Table 17 shows that the initial net replacement ratio in France remains high. Tables 10 and 18 indicate that unemployment benefits are available for longer in France than in most other industrial countries: benefits are available for up to two and a half years for those under 45 and for just over four years for those over 45. Florens, Fougère, and Werguin (1990) show that there is a significant correlation between the expiry date of benefits and the probability of finding a job. This suggests that reducing both the duration of benefits and their initial level could help to encourage search effort and to reduce unemployment duration, and hence the total number of unemployed. Distributional objectives can be achieved more efficiently through taxation rather than through the benefits system. Furthermore, any savings from reductions in benefits and from other passive measures such as early retirement could be transferred to reduce labor costs (through a further reduction in contributions) and fund more active measures (such as training programs) to enhance the skills of those searching for jobs.

Reduce Employers’ Taxes Further

The recent government decision to waive employers’ family allowance contributions for those being paid up to 20 percent above the minimum wage and gradually to extend this to those earning up to 1.5 times the SMIC by 1998 is likely to help job creation by reducing employers’ labor costs. However, at most, it will reduce labor costs by 5.4 percent of the net wage for the low paid (Table 20). Total employee and employer contributions as a ratio to income will still be higher than all other industrial countries except Italy and Belgium (Chart 9, lower panel).

Therefore, it is necessary to take further measures in this area in order to boost employment generation. The planned exemption schedule for employers’ family allowance contributions to 1.5 times contributions by 1998 could be extended and accelerated. If exemptions are not extended to all employees, then they may act as a disincentive to employ workers at wages just above the ceiling. They could also discourage training, because those with training could exact wages above the exemption threshold.31 Cuts in benefits and budgetary expenditure are the most efficient means of financing an extension or acceleration of exemptions. However, in the long run, it is desirable completely to delink the financing of family allowances from employers’ taxes; they could be financed through more broadly based income taxes.32

There have been suggestions for levying environmental (e.g., carbon) taxes at a European level and using the income generated to reduce employers’ contributions. This proposal has to be treated with caution as it would not be a neutral measure: it would discriminate against energyintensive sectors and in favor of those that are labor intensive. Besides, previous recessions have been associated with large oil price shocks. Therefore, any increases in environmental taxes should be made in small steps and on an experimental basis.33

Reform the Minimum Wage Legislation

The analysis in this chapter suggests that the minimum wage has a significant adverse impact on unemployment in general and on youth unemployment in particular. Reducing the real value of the SMIC relative to average earnings will help to lower unemployment. However, under the current mechanism for revising the SMIC, a rise of 2 percent or more in the CPI automatically triggers an equivalent rise in the SMIC. Furthermore, the SMIC is raised by at least half the increase in real hourly wages in industry every year. It would be desirable to have a reappraisal of the legal framework for revising the SMIC to ensure that it will not rise by more than the increase in CPI. At a minimum, the SMIC should not be increased by more than the current minimum legal requirement.34

In spite of the withdrawal of the CIP, many existing employment schemes facilitate a reduction in the SMIC for the young.35 However, as an indirect mechanism these remain a second-best solution. A better option is to introduce a reduced minimum wage for the young as this would be transparent, permanent, and easy to apply. The introduction of a youth minimum wage in the Netherlands has had a positive impact on employment, and the analysis in this chapter suggests that it could help reduce youth unemployment in France. Failing this, the existing youth training programs, with remuneration below the minimum wage, could be simplified and the conditions for participating in them eased to ensure that they are easy to understand, targeted at the long-term unemployed and those with inadequate qualifications, and widely available.

Convert Unemployment Benefits into Employment Subsidies

The employment subsidy that replaced the CIP (see above) should have a positive impact on youth employment because it reduces the cost of employing young people. However, this subsidy is not targeted, as it applies to all under-26-year-olds and is not conditional on any kind of training. It would be more efficient to have any subsidies targeted at those with greatest needs, for example the young unskilled or poorly qualified workers who have difficulty finding a job or training.

An interesting variant of employment subsidies is the idea put forward by Snower (1994a, b) that links subsidies to unemployment duration and training. Under this proposition the unemployed, particularly the longterm unemployed, can use part of their unemployment benefits to provide vouchers to the firms that hire them. The longer someone has been unemployed, the more vouchers they will be given to use. To cash the vouchers, the employer has to provide training that leads to nationally recognized qualifications, and, the more training an employer offers, the higher will be the value of the vouchers when they are cashed. Subsequent to finding a job, the monthly value of the voucher declines as the period of employment lengthens. This scheme is not inflationary, since the long-term unemployed exert little pressure on wage inflation; its cost is minimal because the funds would have been spent on unemployment benefits; and it encourages training for those who are most in need of it.

Improve the Effectiveness of Labor Market Programs

As explained in Section IV, there are a large number of active labor market programs already in existence in France. Many are well targeted and some have been reinvigorated recently. However, if anything, there are too many programs—some 40 in total, and about ten aimed solely at the young. Simplifying the multitude of programs, many of which share objectives as well as means, would make them more transparent and more easily understandable by employers and the unemployed, and thereby improve their effectiveness.

Furthermore, experience so far suggests that measures that help create jobs in the private sector are more likely to lead to permanent jobs; a shift of resources to private sector job subsidies is likely to be beneficial. According to a recent survey, 58 percent of those participating in the CRE and 67 percent of those taking on a contrat de qualification held a job a few months after the program.36 However, only 50 percent of those participating in the CES in the noncommercial sector were in employment subsequent to the program and one fifth of those were on another CES program. The survey also found that in many cases the CES did not have a training element. In this regard it is regrettable that the CES is expected to expand more than the other labor market programs.

Given that signing on for benefits in France is done monthly by mail, there is little contact with the unemployed. However, the intensive interviewing program for the long-term unemployed, AIF, has produced some encouraging results and its expansion would be helpful.

Targeted labor market programs, such as those aimed at the young or the long-term unemployed, will help to create employment in the long run only if they enhance the skills of the individuals who participate in them.37 It is, therefore, necessary to have a well-defined, nationally recognized system of qualifications for any such programs and to encourage employers to participate in the training schemes in order to ensure that the skills gained by the participants correspond to those that employers demand.

Introduce Incentives for Profit Sharing

Firms that practice profit sharing in France appear to enjoy higher profitability and productivity, as well as higher employment growth, than those that do not have any method of profit sharing (Commissariat general du Plan (1993)). Tax incentives for profit sharing would be particularly effective if they were aimed at new firms, because they do not face the problem of replacing a part of existing wages with performance-related pay. Profit sharing may not reduce unemployment if it results in strong productivity gains; however, it would introduce an element of wage flexibility and hence help to smooth the variation of employment over the cycle.

Appendix I: Econometric Results

In Table A1, panel (a) reports the maximal eigenvalue test of the null hypothesis that there are at most r cointegrating vectors against the alternative of r + 1 cointegrating vectors.38 Starting with the null hypothesis that there are no cointegrating vectors (r = 0) against the alternative of one (r = 1), the test statistic (36.7) is greater than the 95 percent critical value (33.3), rejecting the null hypothesis and indicating that there is at least one cointegrating vector. The null hypothesis of r ≤ 1 against r = 2, however, cannot be rejected, suggesting that there is a unique cointegrating vector. Panel (b) reports the trace test of the null hypothesis that there are at most r cointegrating vectors against the alternative that there are more than r. Again, the null of r = 0 against r ≥ 1 is rejected. However, the null of r ≥ 1 against r ≥ 2 cannot be rejected, indicating that there is at most one cointegrating vector. The two tests, therefore, indicate the existence of a unique cointegrating vector. Panel (c) of the table presents the estimated cointegrating vectors. The coefficients in parentheses are normalized on w - p - pr.

Table A1.

Johansen Maximum Likelihood Tests and Parameter Estimates

(Long-run vector for real wages; 1971:1 to 1992:IV (84 observations), maximum lag in VAR = 4. Eigenvalues in descending order: 0.33, 0.18, 0.14, 0.12, 0.05)

article image

r denotes the number of cointegrating vectors.

Turning to Table A2, in panel (a) the maximal eigenvalue test of the null hypothesis that there are no cointegrating vectors (r = 0) against the alternative of one (r = 1) is rejected, indicating that there is at least one cointegrating vector. The null hypothesis of r ≤ 1 against r = 2, however, cannot be rejected, suggesting that there is a unique cointegrating vector. In panel (b), the trace test of the null hypothesis of r = 0 against r ≥ 1 is rejected. However, the null of r ≤ 1 against r ≥ 2 cannot be rejected, indicating that there is at most one cointegrating vector. The two tests together indicate the existence of a unique cointegrating vector. Panel (c) of the table presents the estimated cointegrating vectors. The coefficients in parentheses are normalized on p - w + pr.

Table A2.

Johansen Maximum Likelihood Tests and Parameter Estimates

(Long-run vector for markup of prices over wages; 1971:I to 1992:IV (84 observations), maximum lag in VAR = 4. Eigenvalues in descending order: 0.28, 0.12, 0.10, 0.04)

article image

r denotes the number of cointegrating vectors.

Appendix II: Dynamic Models

Table A3.

Dynamic Model for Real Wages

(Sample period 1971:I–1992:IV)

article image
Table A4.

Dynamic Model for the Markup of Prices Over Wages

(Sample period 1973:I-1992: IV)

article image

Data Sources

All the data have been obtained from Institut National de la Statistique et des Etudes Economiques Quarterly National Accounts, Informations Rapides, and Bulletin Mensuel de Statistique unless indicated otherwise.

article image

References

  • Bazen, Stephen, John Martin, “The Impact of the Minimum Wage on Earnings and Employment in France,” OECD Economic Studies, No. 16 (Spring 1991) pp. 199221.

    • Search Google Scholar
    • Export Citation
  • Bean, Charles R., P. Richard Layard, and Stephen J. Nickell, eds., The Rise in Unemployment (New York: Basil Blackwell, 1987).

  • Bismut C., “La France et la RFA avant et après le premier choc pétrolier,” Economie et Statistique, No. 148 (1982), pp. 5774.

  • Blanchard, Olivier, and Pierre-Alain Muet, “Competitiveness Through Disinflation: an Assessment of the French Macroeconomic Strategy,” Economic Policy, Vol. 16 (April 1993), pp. 1256.

    • Search Google Scholar
    • Export Citation
  • Bos, Eduardo, Rodolfo Bulatao, “Projecting Fertility for All Countries,” Policy, Research, and External Affairs Working Paper: WPS 500 (Washington: World Bank, September 1990).

    • Search Google Scholar
    • Export Citation
  • Brown Charles, “Minimum Wage Laws: Are They Overrated?” Journal of Economic Perspectives, Vol. 2 (Summer 1988), pp. 13345.

  • Cahuc, Pierre, Patrick Sevestre, and Hélène Zajdela, “Négociations salariales et segmentation du marché du travail,” Economie et Prévision, No. 92–93 (1990), pp. 4350.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Caracosta L., M. Fleurbaey and C. Leroy, “Compétitivité, croissance et emploi: la France de l’an 2000 en perspective,” Economie et Statisque, No. 243 (1991), pp. 6987.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cases C., S. Lollivier, “Estimation d’un modèle de sortie de chômage á destinations multiples,” INSEE Série des Documents de Travail No. 9327 (Paris: Institut National de la Statistique et des Etudes Economiques May 1993).

    • Search Google Scholar
    • Export Citation
  • Coe, David, “Structural Determinants of the Natural Rate of Unemployment in Canada,” Staff Papers, International Monetary Fund Vol. 37 (March 1990), pp. 94115.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Commissariat général du Plan, L’economie française en perspective (Paris: Editions La Découverte et La Documentation française, 1993).

    • Search Google Scholar
    • Export Citation
  • Cotis, Jean-Philippe, and Abderrahim Loufir, “Formation des salaires, chômage ‘d’équilibre,’ et incidence des cotisations sur le coût du travail,” Economie et Prévision, No. 92–93 (1990), pp. 97110.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cuthbertson, Keith, Stephen G. Hall, and Mark P. Taylor, Applied Econometric Techniques (Ann Arbor: University of Michigan Press, 1992).

    • Search Google Scholar
    • Export Citation
  • Drazen, Allan, and Paul R. Masson, “Credibility of Policies versus Credibility of Policymakers,” Quarterly Journal of Economics, Vol. 109 (August 1994), pp. 73554.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Drèze, Jacques, and Charles R. Bean, eds., Europe’s Unemployment Problem (Cambridge, Massachussetts: MIT Press, 1990).

  • Entorf, Horst, “Do Aggregate Measures of Mismatch Measure Mismatch? A Time Series Analysis of Existing Concepts,” INSEE Série des Documents de Travail No. 9318 (Paris: Institut National de la Statistique et des Etudes Economiques, February 1993).

    • Search Google Scholar
    • Export Citation
  • Florens, J.P., D. Fougère and P. Werguin “Durée de chômage et transitions sur le marché du travail,” Sociologie du travail, No. 4 (1990).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jackman, Richard, and S. Roper, “Structural Unemployment,” Oxford Bulletin of Economics and Statistics, Vol. 49 (February 1987), pp. 936.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jackman, Richard, P. Richard Layard, and S. Savouri, “Mismatch: A Framework for Thought,” in Mismatch and Labour Mobility, ed. by Fiorella Padoa Schioppa (Cambridge, England: Cambridge University Press 1991), pp. 44101.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Johansen, Soren “Statistical Analysis of Cointegration Vectors,” Journal of Economic Dynamics and Control, Vol. 12 (June–September 1988), pp. 23154.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Johansen, Soren, and Katarina, Juselius, “Maximum Likelihood Estimation and Inference on Cointegration, with Applications to the Demand for Money,” Oxford Bulletin of Economics and Statistics, Vol. 52 (May 1990), pp. 169210.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Joyce, Michael, and Simon Wren-Lewis, “The Role of the Real Exchange Rate and Capacity Utilization in Convergence to the NAIRU,” Economic Journal, Vol. 101 (May 1991), pp. 497507.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Layard, P. Richard, and Stephen J. Nickell, “Unemployment in Britain,” Economica, Vol. 53 Supplement (1986), pp. S12169.

  • Layard, P. Richard, Stephen J. Nickell, and Richard Jackman, Unemployment: Macroeconomic Performance and the Labour Market (Oxford, England: Oxford University Press, 1991).

    • Search Google Scholar
    • Export Citation
  • Newell, A., and J.S. Symons, “Corporatism, the Laissez-Faire and the Rise in Unemployment,” Centre for Labour Economics Discussion Paper No. 260: 1–62 (London: London School of Economics and Political Science, November 1986).

    • Search Google Scholar
    • Export Citation
  • Organization for Economic Cooperation and Development, Measures to Assist the Long-Term Unemployed: Recent Experience in Some OECD Countries (Paris: OECD, 1988).

    • Search Google Scholar
    • Export Citation
  • Organization for Economic Cooperation and Development, (1992a), Employment Outlook (Paris: OECD 1990, 1991, 1992, 1993).

  • Organization for Economic Cooperation and Development, (1992b), Economic Survey France (Paris: OECD, 1992).

  • Pesaran, M., Bahram Pesaran, “Microfit 3.0: An Interactive Econometric Software Package” (Oxford, England: Oxford University Press 3rd ed., 1991).

    • Search Google Scholar
    • Export Citation
  • Pissarides, Christopher, “From School to University: The Demand for Post-Compulsory Education in Britain,” Economic Journal, Vol. 92 (September 1982), pp. 65467.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Plassard, Jean-Michel, and Gabriel Tahar, “Théorie du salaire d’efficience et disparités non compensatrices: évaluation à partir de l’enquete Fqp,” Economie et Prévision, No. 92–93 (1990), pp. 6776.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schmitt, John, and Jonathan Wadsworth, “Unemployment Benefit Levels and Search Activity,” Oxford Bulletin of Economics and Statistics, Vol. 55 (February 1993), pp. 124.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Snower, Dennis, (1994a), “The Simple Economics of Benefit Transfers” (mimeographed, London: Department of Economics, Birkbeck College, London University, 1994).

    • Search Google Scholar
    • Export Citation
  • Snower, Dennis, (1994b), “Converting Unemployment Benefits into Employment Subsidies,” CEPR Discussion Paper No. 930 (London: Centre for Economic Policy Research May 1994).

    • Search Google Scholar
    • Export Citation
  • Taylor, Mark P., “The Link Between Real Interest Rates and French Aggregate Private Investment” (mimeographed, Washington: International Monetary Fund, 1993).

    • Search Google Scholar
    • Export Citation
  • Van den Heuvel, K., Minimumbeloningen in de Europese Gemeenschap (Amsterdam: Ministerie van Sociale Zaken en Werkgelegnheid, 1992).

  • Wren-Lewis, Simon, “Nominal Inertia and Keynesian Effects,” NIESR Discussion Paper No. 174 (London: National Institute of Economic and Social Research, 1990).

    • Search Google Scholar
    • Export Citation
1

Drazen and Masson (1994) find support for the hypothesis that persistence in unemployment can undermine the credibility of monetary policy.

2

Defined as LF/PW = (E + U)/PW, where LF is the labor force, E employment, U unemployment, and PW is the population of working age.

3

The fall in the male participation rate is not independent of the rise in unemployment and is partly due to the early retirement of the old unemployed; see discussion of participation rate by age below.

5

The nonemployment rate is defined as

1PWE

where E is employment and PW is population of working age.

6

Those aged over 56 can obtain unemployment benefits without looking for work. If someone over 56 becomes unemployed as a result of company restructuring, he will be entitled to 65 percent of previous income or a maximum of F 12,000 per month, with the employer paying 10 percent of the cost.

7

Following the election of President Mitterand, the SMIC was raised by 10 percent in May 1981.

8

In fact these numbers are an underestimate since the survey carried out by the Ministere du Travail does not cover firms with less than ten employees, nor those in agriculture, coal, public transport, and utilities.

9

In addition to the SMIC, there is evidence of other types of wage rigidity that could also lead to higher unemployment. For example, there is evidence of “insider power” (Cahuc, Sevestre, and Zajdela (1990); Plassard and Tahar (1990)).

10

See Brown (1988) for a survey of the U.S. evidence and Coe (1990) on evidence for Canada.

11

Based on internationally comparable figures compiled by the OECD (1992a).

12

In recent years the Government has introduced a number of measures that reduced employers’ contributions and the minimum wage for young people participating in employment schemes. These measures were reinforced in June 1993, reducing the effective minimum wage substantially when participating in special schemes (see Section IV below).

13

The minimum wage in the Netherlands is reduced according to the following schedule:

article image

14

The minimum wage in Belgium is reduced according to the following schedule:

article image

15

This is defined as ½ VAR(U,/U).

16

For example, the unemployment rate in Yvelines (Ile de France) rose from 4.9 to 6.2 percent, a rise of 27 percent, whereas it rose from 13.5 percent to 15.5 percent in Herault in Languedoc-Roussillon, a rise of 15 percent.

17

Bismut (1982) illustrates that there was a significant shift of the Beveridge curve after the two oil shocks. There is no evidence of such instability during the 1980s. However, the flatness of the UV curve during the 1974–92 period and its upward slope during 1984–87 point to a small increase in mismatch.

18

w and p are in logs.

19

Cuthbertson, Hall, and Taylor (1992) present a survey of cointegration.

20

Johansen (1988) and Johansen and Juselius (1990) present a cointegration estimation methodology that is based on maximum likelihood estimates of all the cointegrating vectors in a given set of variables, and provide two likelihood ratio tests for the number of cointegrating vectors. Johansen (1988) demonstrates that the likelihood ratio tests have asymptotic distributions that are a function only of the difference between the number of variables and the number of cointegrating vectors. Therefore, in contrast with the DF and ADF tests, the Johansen likelihood ratio tests have well-defined limiting distributions. Johansen and Juselius also provide a methodology for testing hypothe