This Selected Issues paper addresses the question of what policy changes in France are needed under European Monetary Union (EMU), as regards the role of fiscal policy in stabilizing the economy. The fiscal strategy over the past two and a half decades is reviewed, and, against this background, an assessment is offered concerning the role and scope of fiscal stabilizers in France under EMU. The main conclusions is that over the past two and a half decades, fiscal policy operated in a clear countercyclical way in France, but this reflected essentially the functioning of automatic stabilizers.


This Selected Issues paper addresses the question of what policy changes in France are needed under European Monetary Union (EMU), as regards the role of fiscal policy in stabilizing the economy. The fiscal strategy over the past two and a half decades is reviewed, and, against this background, an assessment is offered concerning the role and scope of fiscal stabilizers in France under EMU. The main conclusions is that over the past two and a half decades, fiscal policy operated in a clear countercyclical way in France, but this reflected essentially the functioning of automatic stabilizers.

ii. the 35-hour workweek initiative16

41. This chapter gives an overview of the French government’s initiative to reduce the legal workweek from 39 to 35 hours. It is organized around three questions:

  • How significant a departure from current practices in France and elsewhere is such a reduction in legal work time?

  • How do the authorities intend to implement the shift to the new regime?

  • What is the likely micro- and macro-economic impact of the adoption of a shortened workweek?

These questions are taken up in the next three sections, one at a time. Section A gives the historical perspective and some international comparisons. The main conclusion is that the scheduled reduction in the legal workweek is a significant development that makes France an outlier within the group of industrial countries. Section B outlines the current legislation on this matter and the official strategy to promote an early adoption of the 35-hour week. This section is descriptive rather than normative, and provides a comprehensive presentation of the legal framework. The economics of a shortened workweek is finally discussed in Section C. The topic is complex and a formal treatment is beyond the scope of this paper; nonetheless, the discussion goes some way in highlighting from first principles the main sources of concern. The critical issue of wage compensation at the level of the hourly minimum wage receives, however, less attention than it deserves, since an in-depth analysis can be found in the appendix to the staff report on the 1998 Article IV Consultation with France (SM/98/229). Section D concludes.

A. The 35-Hour Workweek in Perspective

42. Over the last 150 years, reduction in work time has been the rule rather than the exception in all industrial countries.17 In this respect, the French experience is typical: since 1831 the average annual hours of work per person have fallen by about 50 percent (Figure II.1). Several factors have driven the secular process, including: the growing importance of the service sector—with shorter work times than in industry and agriculture18; the evolution of union density and collective bargaining19; the increasing popularity of part-time work—linked to both trends in female participation to the labor force and the employers’s search for increased flexibility in production20; the lengthening of paid vacations; the progressive diffusion of work times below the norms of collective agreements; and the overall reduction in these norms, of which the 35-hour week initiative is but the latest example.



Citation: IMF Staff Country Reports 1998, 132; 10.5089/9781451813494.002.A002

Source: Ministère de l’Emploi et de la Solidarité, 1998.

43. The global nature of the process since the 1970’s is illustrated by Table II.1. The broad pattern is one of a steady decline in hours worked over the last three cycles with a significant slow down since the 1980’s. The exceptions are Japan—where the government has been actively promoting a reduction in work time—and, to a certain extent, Germany. In the United States, the United Kingdom, and Sweden the trend appears to have reversed due to country-specific factors—increased reliance on overtime in the US, growing share of the self-employed in total employment in the UK, and in Sweden a rising proportion of women part-timers working longer hours.21 It is noteworthy that in this sample the slow-down has been most pronounced in France.

Table II.1.

France: Trends in Average Annual Hours Worked per Employed Person

(Changes in hours)

article image
Source: OECD, Employment Outlook, Table 5.1

Cyclical periods taken from trough-to-trough according to OECD dating.

44. A cross-country comparison of the number of hours worked—rather than their changes through time—is complicated by differences in the coverage of labor surveys, and the lack of an harmonized definition of workweek. Table II.2 shows the length of the legal workweek as determined by the law or industry-level agreement, the duration of the normal workweek (which differ from the legal norm by the extent that overtime is used or industry agreements take precedence), and the number of paid vacation days and official holidays per year (which further influence how long people actually work).

Table II.2.

France: Duration of Workweek and Paid Vacation: Selected Countries

article image
Source: Revue de l’OFCE, July 1998, p.73.

45. Only Denmark, the U.K., and—since 1991—New Zealand, do not have a legal limit on work time. The duration of the workweek in these countries is determined through collective agreements or individual contracts: as the table shows, the resulting practices are nonetheless in line with those of other industrial countries. The table also shows that Japan is a clear outlier as regards the duration of the normal workweek. By contrast, France already has the lowest legal workweek among industrial countries, although the normal workweek (which is perhaps a better indicator of the organization of work) broadly conforms to the European average.

46. This observation is confirmed by additional factors that are likely to affect the duration of the normal workweek besides the legislative limits and the structure of holidays. These appear in Table II.3 (below). The normal average workweek will tend to be shorter, the shorter the average hours worked by dependent foil-time wage-earners, the larger the shares in total employment of part-time workers and women, and the smaller the share of self-employed workers. On all these accounts, France ranks in the middle of the reference group of major industrial countries.

Table II.3.

France: Selected Countries: Labor Market Indicators

article image
Source: Eurostat (1997), and Revue de l’OFCE, July 1998, p.74.

For Canada and Switzerland, includes part-time workers.

47. The impression that France does not stand out as an exceptional case must be modified, however, when the focus shifts to a measure of the effective average duration of work on an annual basis. This measures corrects the reported normal workweek for the time lost due to absence from work, pauses while on the job, “down-times,” holidays, interruption of production, strikes, etc. Three facts then emerge; (i) in France the average annual effective duration of work per dependent worker is already below the average for other OECD countries; (ii) because of the low participation ratio and the reliance on early retirement, the average effective duration of work per person in the active population is one of the lowest in the industrial world; and (iii) the spell in employment over the life-cycle is also remarkably short, especially for men. Table II.4 presents the data supporting these claims.

Table II.4.

France: Effective Duration of Work for Selected Countries

article image
Source: OECD, Employment Outlook, 1998, p. 207, and Commission d’ Enquête du Senat sur les 35 Heures, 1998.

Column 1 reproduces Table F in OECD, 1998. It gives total hours worked in the year divided by the average number of employed people in 1997, including part-time workers. There are dissimilarities in coverage across countries: data for Italy refers to 1994, for Switzerland to 1995, and for Canada and New Zealand to 1996. Also, in the cases of New Zealand, Switzerland and the United Kingdom, self-employment is included.

For the United States and Japan, under column 3, the average refers to total employed.

48. As said, considerable measurement and methodological problems complicate the international comparisons: the empirical evidence needs to be interpreted with caution. However, even if the current organization of work time in France may not be atypical by international standards, the 35-hour week initiative will make France’s legal workweek more than 2 standard deviations shorter than the European norm. Will this be a substantive change or merely a formal one? While the duration of the legal workweek does not limit an enterprise in its choice of the work organization nor determines the maximum work time permissible under the law, it does establishes the threshold after which overtime is calculated.22 As such, legislative provisions on the workweek can potentially affect unit labor costs directly. Whether this will be the case depends on the response of firms and employee to the new constraint. Tables II.5 and II.6 provide some evidence that the shift to a shorter legal workweek will have pervasive effects: 75 percent of employees currently work more than 35-hours per week (Table II.5), and the sectoral break-down (Table II.6) points to the fact that the mode of the distribution is at about the duration of the legal workweek in both industry and service sector.

Table II.5.

France: Frequency Distribution of Work Time by Year23

article image
Source: Conseil Supérieur de l’Empioi, des Revenues, et des Coûts, 1998, p. 54.
Table II.6.

France: Frequency Distribution of Work Time by Sector (1995)

article image
Source: DARES, Travail et Emploi, No. 74, 1998, p. 97

Not covered by the 35-hour law.

49. The expectation of significant changes in the organization of work at the firm-level has motivated the authorities to draft the law to implement the 35-hour initiative in a way that encourages a flexible approach to implementation.

B. The Legal Framework

50. The Lot d’ orientation et incitation à la reduction du temps de travail of June 13, 1998, and the Decret d’application of June 22, 1998, lay out the broad blueprint for the adoption of a legal workweek of 35-hours. This blueprint reflects the government’s commitment to “create jobs and reduce unemployment by all available means” and—implicitly—the official perception that a positive (social) externality justifies government’s intervention in these matters.24

51. The law mandates that enterprises with more than 20 employees adopt the shorter legal workweek by January 1, 2000. Firms with a smaller workforce must do the same by January 1, 2002. These deadlines apply to firms in the non-agricultural market sector which employed in 1996 about 13.5 million workers or 60 percent of total employment.25

52. Employers and trade unions are called upon to start negotiations in advance of those deadlines to find the modalities most suited to local conditions (i.e., at the level of the plant, firm or the sector) for reducing the workweek. These negotiations will address outstanding issues related to an early adoption of the 35-hour week; the possible re-organization of production processes and work time; the treatment of part-timers and senior workers (cadres); the number of new hires (or preserved jobs); and the internal procedures to monitor and enforce any agreement reached by the social partners.26 The outcomes of these negotiations are expected to influence a law due in the Fall of 1999 concerning as yet unspecified implementation issues (see paragraph 12 below).

53. The law also establishes fiscal incentives for firms that negotiate a reduction in work time before the shift to the mandatory 35-hour workweek and create jobs (or preserve jobs at risk). These incentives are in the form of a rebate of the employer’s social charges (abattement de cotisations sociales). Eligibility for the aid requires three pre-conditions: (i) a reduction of hours worked of at least 10 percent to attain a workweek of 35-hour or less; (ii) new hires or preserved jobs of at least 6 percent of the workforce in the enterprise; and (iii) the signature of a formal agreement among the social partners on the modalities of reducing the work time.27

54. The amount of the rebate and its duration are conditional on the firms’s employment plan, the implementation schedule and other firm-specific factors. It will last five years for firms committed to net job creation and three years (with a possible extension of other two) in the cases where jobs at risk are preserved, provided that the workforce is not reduced for at least two-year.28 The rebate is greatest with an early implementation schedule. In any event, the employer will receive the rebate for each worker whose work time is reduced to 35 hours or less, and for each new hire (or each preserved job, as the case may be.) As Table II.7 shows, the rebate is expected to become permanent after 5 years, although no official decision has been taken on this matter yet.29 These rebates can be increased under special circumstances (for example, if the enterprise concerned has a sufficiently large number of low-wage workers) and are to be pro-rated in the case of part-time workers.30 To put the figures in context, 5,000 francs correspond to about 3 percentage points of social charges at the level of the average wage. The 1998 budget allocates three billion francs for the initial implementation of this scheme.31

Table II.7.

France: Structure of Fiscal Incentives

(per employee, per year)

article image
Source: Journal Officiel de la République Française, June 23, 1998; and Ministère de l’Emploi et de la Solidarité, 1998, p.17.

55. The law does not address operational details beyond the transitional arrangements outlined in the previous paragraphs. It simply defers to another law due in late 1999 the resolution of outstanding issues such as: (i) the regulation of overtime and compensatory leave; (ii) a new statute for of part-time work; (iii) the legal workweek of senior workers and managers; (iv) the scope for defining the legal duration of work with reference to a period other than a week; and (v) how to ensure that the purchasing power of earners of the hourly minimum wage (SMIC) will not be negatively affected by the shift to a shorter workweek.

56. The incompleteness of the current legal framework is, however, by design. In the authorities’ view, lack of specificity is necessary to assure a broad margin for maneuver at the implementation stage, without the inflexibility that would ensue from a centralized, “top-down” approach.

C. The Economic Effects of a Reduction in Work Time

57. While generality in the legal framework may facilitate the transition to the new regime, it complicates an ex ante assessment of the effectiveness of the 35-hour workweek in promoting employment and work-sharing. For example, the likely employment effect depends critically on the evolution of unit labor costs in the new regime, but these cannot be estimated without knowing the impact on factor productivity of a shorter work time, and the induced changes in the wage structure especially in the neighborhood of the minimum wage. These factors are clearly influenced by the legislative provisions defining the dimensions of choice of the decision makers.

58. A noteworthy research effort has attempted to clarify these issues, leading to a broad consensus along the following lines: (i) employment creation will be favored by gains in the productivity of labor and capital; (ii) these are more likely to materialize if a flexible approach to implementation allows firms to re-organize production processes and work schedules; (iii) moderate wage compensation for a reduction in hours worked is nonetheless necessary if adverse developments in labor costs are not to undermine the benefits from greater productivity; (iv) and—in transition—fiscal resources may be used to re-align at the margin the private and social costs (and benefits) of the reduced workweek.32

59. If the 35-hour week leads to an increase in unit labor costs, the outlook for employment will suffer for at least three reasons. First, competitive firms faced with an increase in the cost of labor will reduce production and factor demands, other things equal. Second, an increase in the relative price of labor will induces substitution towards more capital- intensive techniques of production.33 Thirdly, in an open economy external competitiveness may be lost as firms increase prices to preserve profitability, further reducing labor demand. So, it is critical that unit labor costs be safeguarded. This will be so if labor productivity is enhanced, capital utilization does not fall, a wage moderation prevails. We consider these conditions in turn.

60. The argument that workers perform better if work time is reduced rests on the presumption that the relationship between effort and the duration of work in non-monotonic: beyond a certain threshold, a reduction in work time lessens workers’ “fatigue” and thus increases productivity. In addition, a shorter work time may make possible a reduction in the number of incompressible down-times and pauses that punctuate the normal work schedule, adding to the productivity gains. More importantly, there is the expectation that, in the context of negotiating the switch to the 35-hour week, firms will be able to over-haul the organization of production processes and work schedules, thus boosting efficiency. For example, the negotiations may lead to the adoption of flexible schedules and a better synchronization between production and sales, or workers may agree to shorter breaks and “bell-to-bell” work. A popular (but unproved) claim is that the gains in labor productivity due to a reduced workweek may be of the order of 50 percent of that reduction, i.e., about 5 percent with the passage from 39 to 35 hours.34

61. If hours worked are reduced in line with the legal workweek, the utilization of capital—and its productivity—will fall pari passu, since the user cost of capital goes up once depreciation is to be amortized over a shorter working life of the equipment. To avoid this, the production process must be re-organized, for example through greater reliance on shift-working and multiple crews, or by scheduling working hours with greater flexibility.35 The expectation of such work re-organization is a recurrent argument offered by the proponents of the 35-hour initiative. It is based on the perception that pervasive rigidities, including in the bargaining process, have prevented firms from exploiting in the past these open profit opportunities. According to this view, the introduction of the shortened workweek will provide the catalyst for revamping industrial relations, inducing bargaining over dimensions of work not before discussed by the social partners, and thus re-optimizing the production process.

62. The question of how the reduced workweek will affect labor costs is more complex. Two aspects are considered here: (i) the effect of greater reliance on overtime; and (ii) the role of income maintenance at all level of wages, including the minimum wage, and the contribution of the fiscal incentives in the law to a containment of labor costs.

63. The shift to the shortened legal workweek has the immediate effect of increasing the average hourly wage in a firm that will keep the effective workweek unchanged at or above 39 hours. On the basis of current legislation, an extra 188 hours per year (or (39-35)* 47) will have to be counted as overtime. This will, first, require special dispensation from existing provisions (since there is a legal limit on annual overtime per employee of 130 hours,) and, secondly, special remunerations to the worker affected. Existing regulations set the overtime premium over the straight-time rate at 25 percent for the first 8 hours in a week, and 50 percent thereafter. In addition, a “compensatory rest” (repos compensateur) of half an hour for each hour of overtime has to be awarded beyond a threshold of 42 hours of overtime, bringing the cost of overtime to 50 percent of the straight-time rate.36 The award of compensatory rest is further increased when the total of overtime hours passes the 130 hour limit, with a marginal cost of overtime then equal to 100 percent of straight-time rate. As an illustration, if a firm keeps the effective workweek at 42 hours, its cost per employees will increase by the equivalent of 2 hour of straight-time (6 hours of overtime at a 25 percent premium and one hour at a 50 percent premium) or about 5 percent. Although legislative dispositions on these matters are deferred to the second law on the 35-hour initiative, it seems clear that a significant burden will be placed on firms that will not—or cannot—reduce the effective workweek.

64. Greater reliance on overtime in the new regime seems likely since the costs of adjusting employment are higher at the straight-time margin due to the presence of fixed costs inter alia related to search, training, and job protection legislation. This observation is relevant even for firms that are considering adopting the shortened workweek and expand employment, if the straight-time hourly wage increases as workers seek compensation for a reduction in purchasing power. In this case, in fact, a profit maximizing firm will reduce overall employment in the short run and change the mix of straight-time to overtime hours since adjustment at the latter margin is cheaper.

65. The degree of wage compensation or income maintenance is a second channel through which labor costs will be directly affected by the 35-hour initiative, since a reduced workweek implies a higher hourly wage if the weekly wage is not reduced proportionally. The more reluctant workers are to “buy” greater leisure, i.e. to accept a reduction in their income in exchange for less work, the greater the increase in labor cost and the less favorable the employment outlook: wage moderation is a sine qua non for job creation. For workers earning more than the minimum wage, the degree of income maintenance is to be decided in the context of the negotiations envisaged by the law between the social partners. Regardless of the (re-)distribution of rents in the short run, over time the market-determined wage will (tend to) return to the level compatible with the equilibrium distribution of value added between labor ad capital. To the extent that a greater proportion of labor income will be linked to overtime work, the straight-time hourly wage will eventually be reduced below its level with a 39-hour workweek.

66. The situation is different for earners of the hourly minimum wage (SMIC) since their compensation is set administratively and is not influenced by market forces. The authorities are committed on social grounds to safeguard fully the purchasing power of SMIC-earners (and workers earning close to the SMIC), which implies a 11.4 percent (or 39/35-1) increase in the cost of employing them, ceteris paribus, under the 35-hour legal workweek. Possible problems arising from this approach are discussed in the appendix to the staff report on the 1998 Article IV Consultation with France (SM/98/229). Here, it suffice to say that the system of fiscal incentives described in Section II is intended to cushion the impact of this wage shock. However, the prospective (permanent) rebate of F 5,000 would compensate only about 40 percent of the increase in labor costs. Moreover, the system of subsidies is not targeted to support the employment of low-wage earners, and it would be ineffectual in preventing a shift in labor demand away from the low-skilled. Thus, an extremely significant productivity gains would have to occur to contain unit labor costs to their earlier level, and/or wage moderation must prevail for some time, lest employment prospects of the low skilled be negatively affected in the long run.37

67. Some information on the likely wage and productivity developments can be gathered from the experience under the 1993 Loi Robien. This law was introduced to encourage on a voluntary basis the adoption of a shorter workweek in exchange for a reduction in social charges of the employers.38 The historical record shows that income maintenance has in general being full in the instances of net job creation (the so-called “offensive” cases covered by the Loi Robien), and partial in the “defensive” cases where job preservation was at stake—a situation more common in the industrial sector. Even when full income maintenance has been the rule, the social partners have agreed to a follow-up period of wage moderation. As regards productivity gains associated with a reduction in work time under the Loi Robien, the evidence is largely circumstantial: about 50 percent of the firms involved reached agreements to define legal work time on an annual basis—suggesting that the opportunity to reorganize the work schedule was often seized.39

68. To summarize, the switch to a 35-hour week is less likely to have an adverse effect on employment in the short and long run if it leads to gains in total factor productivity, and wage compensation is moderate, other things being equal. Existing regulations on overtime also would need to be modified to contain unit labor costs, and fiscal resources may have to be mobilized initially to cushion the wage shock from the reduction in the legal workweek.

69. This prescription focuses on first-round responses, and as such it is not a full characterization of the conditions for a favorable impact of the 35-hour week. A general equilibrium assessment is, however, a complex task, in part because the relevant channel of transmissions are not usually spelled out in conventional macro-economic models. In the textbook paradigm of an open economy with a fixed exchange rate and perfect capital mobility, the dynamic response to the reduced workweek is determined by the initial change in unit labor costs. The most unfavorable scenario looks at a reduction in work time as a negative productivity shock. Starting for ease of reference from an equilibrium at the natural rate of unemployment, a shortened workweek raises costs to firms which increase prices in the short run to preserve their profit margins (i.e., the aggregate supply curve shifts up). The price level increase leads to a real appreciation and net exports fall accordingly, reducing equilibrium output. The initial short run equilibrium is then characterized by a higher price level, lower output, a worsening current account, a deteriorating fiscal balance, and unemployment in excess of the natural rate. The unemployment gap will in time depress wages and reduce unit costs. As this dynamics unfolds, output and employment recover towards their natural rates and the real appreciation unwinds.40

70. This description of the dynamics glosses over two important points. First, in a medium-run perspective, the transition period will be longer—and costlier—the more pervasive are nominal wage rigidities. This observation underscores the importance of promoting from the onset wage moderation, including at the level of the minimum wage, to speed up convergence. Second, the original long-run equilibrium may not be restored if the transitional increase in unemployment has permanent consequences. This would be the case, for example, if real wage rigidity at the level of the minimum wage prevent an undoing of the initial shock in a segmented labor market, or if insider-outsider effects come into play. In addition, it could be argued that in the new equilibrium, the total supply of hours may in any case be somewhat lower if workers opt for leisure and find it easier to reduce their supply of labor by resisting overtime work. Moreover, some employers, faced with an increase in the hourly wage after 35 hours of work rather than 39, may similarly decide to reduce total hours demanded and use more capital instead. On both accounts, total labor input is adversely affected even if long-run employment is unchanged, and equilibrium output may suffer.41

71. Under more favorable circumstances, unit labor costs are preserved through wage moderation and productivity gains. Then, output and employment will expand only if work-sharing involves a redistribution of disposable income in favor of a segment of the population (the formerly unemployed) with a higher propensity to consume than the workers currently forced to work shorter hours, or if confidence effects play out in the household and in the business sectors. This hardly seems a foregone conclusion.

72. The case for the 35-hour week rests on even more benign assumptions. These have been incorporated in three macro-models to estimate probable employment gains: the models of the Banque de France (BDF) and of the Observatoire français des conjunctures économiques (OFCE) and the model of the Diréction de la Prevision (DP). In the BDF simulations it is assumed that the reduction in work time affects 9 million workers and amounts to about 7 percent of the effective workweek; productivity increases by about 3 percent; the rate of capital utilization is kept constant; and reductions in social charges and productivity gains leave unit labor costs unaffected. Under these conditions, it is projected that in three years employment would increase by about 700,000 jobs.42 If, on the other hand, there is no re-organization of work processes and monthly income is maintained fully, employment increase only by 115,000 as a result of favorable aggregate demand effects. By contrast, the OFCE model assumes only 70 percent of wage earners see their legal workweek reduced;43 productivity increase by half the reduction in work time; 35 hours are paid on average like 38 hours (i.e,. a 70 percent income maintenance); and the rebate in social charges is as described in the law. The projected increase in employment is then 150,000 each year for two year, on average, but the figure is sensitive to modifications in the assumptions. For example, the employment effect is a gain of only 130,000 jobs if workers demand full income maintenance, and could be a loss of 100,000 jobs if firms offset the higher cost of overtime by reducing proportionally the number of hours worked. The model of the DP assumes productivity gains on the order of 3.5 percent for 8 million worker and wage income maintained at about two-third of its level with a 39-hour week. Under this scenario, employment could increase by about 300,000 jobs by the year 2000. However, if firms effectively boycott the reduction in work time, there could be an employment loss of about 20,000 jobs.

73. These studies show that the 35-hour week initiative is likely to increase employment only under fairly optimistic assumptions, and to damage employment creation only under fairly pessimistic ones. However, the simulations are mechanic exercises and tend to down play the uncertainty associated with each outcome, inter alia by ignoring the endogenous changes in preference and technology that might accompany the regime switch. An additional limitation is that capital-labor substitutability is ruled out, and labor itself is considered an homogeneous commodity—a potentially misleading simplification when the wage structure is likely to be compressed with the adoption of the 35-hour week as low wages increase relative to the average wage. The models are also linear, which precludes the analysis of non-convexities arising from work-reorganization. On balance, these simulations are suggestive at best, and do not produce a compelling case for a reduction in the legal workweek as a strategy for job creation.

74. This perception is corroborated by the experience to date. Since the passage of the 35-hour law, 258 agreements to reduce the work time of about 30,000 workers have been signed, 170 of which have led to requests for government subsidies. Net job creation (or preservation) has amounted to about 2,500; and typically the agreements have involved full maintenance of weekly wages.44 In the metal-processing sector, the industrial agreement recently reached between the employer association, UIMM, and the trade unions avoids any attempt at job creation and instead centers on how to accommodate overtime: the limit on overtime has been increased from 94 hours per employee to 180 with the option for an extra 25 hours. Companies will also be allowed to annualize the time worked over the full year with a limit before overtime of 1,645 hours.45 So far, the 35-hour initiative has failed to live up to the expectations of its most enthusiastic supporters. Moreover, Figure II.2 hints at the fact that the observed weak impact on employment may be representative of future developments: in a cross-section of industrial country, there does not seem to be a linear relationship between the length of the workweek and the unemployment rate.



Citation: IMF Staff Country Reports 1998, 132; 10.5089/9781451813494.002.A002

Source: Revenue de l’OFCE, 1998.

D. Conclusions

75. On balance, the economic consequences of a switch to the 35-hour workweek remain difficult to predict with great confidence. There is little doubt, however, that a necessary condition for positive outcomes is flexibility at the implementation stage, in the two-fold sense of an unfettering legal framework and an innovative approach by the social partners. If the economic agents are left with ample latitude to respond to the changed environment, the potential for productivity gains—and, hence, for offsetting possible negative developments in labor costs—will be maximal. There will also be a need for wage moderation, including at the level of the SMIC. If wage increases are not curbed and best responses are constrained by institutional or procedural rigidities, there is a risk that the 35-hour initiative would amount to a negative supply shock, and could well harm the employment situation and the long-run output and fiscal prospects.


Prepared by Alessandro Zanello.


In 1870, annual hours worked per person employed averaged 2,900 for a group of countries comprising Austria, Belgium, Denmark Finland, France, Germany, Italy, the Netherlands, Norway, Sweden, Switzerland, the United Kingdom, Australia, Canada, the United States and Japan. That average was down to about 1540 by 1993. See Maddison, A.: Monitoring the World Economy, OECD, 1995, Table J-4, p. 248.


Over the period 1982–95 three-fourth of the economy-wide reduction in work time can be linked to reductions in the service sector, and one-fourth to reductions in other non-agricultural sectors. (Conseil Supérieur de l’Emploi, op. cit. p.49.)


See OECD, Employment Outlook, 1998, p. 166.


Without the increase in part-time, the average annual work time would have been in 1995 the same as in 1982. (Ibid.)


OECD (1998), Ibid. p. 155.


Maximum legal work time is 10 hours a day, although this limit can be by-passed by agreement among the social partners. The maximum permissible workweek is 48 hours except if special derogations are accorded.


The frequency of people working more than 46 hours is 11.2 percent for the European Union as a whole. (See, Conseil Supérieur de VEmploi, 1998, p. 44.)


The quote in Italic is from the Instruction of June 24, 1998, signed by Minister Aubry.


The field of application of the law is more restricted than this since about one employee in five works part-time. Pending legal decisions on the treatment of part time workers, the law is expected to affect about 9 million employees, or 40 percent of total.


Special authority to engage in these negotiation may be vested on individual workers if there is no trade union representative in the firm. Temporarily, these workers will enjoy the same legal protection vis-à-vis the employer as union representatives. Firms with more than 50 must negotiate in order to be eligible for employment subsidies. Firms with less than 50 employees qualify if simply adopt the agreement negotiated at the sectoral level once validated by the Ministère de l’Emploi et de la Solidarité (accord de branche étendu). Set Ministère de l’Emploi etde la Solidarité, 1998, p. 15.


This agreement will form the basis of a contract (convention) between the employer and the state for the disbursement of the subsidy. It will have to vetted by the departmental office of Labor, Employment and Professional Training (Diréction Departementale du Travail, l’Emploi, etde la Formation Professionelle).


See Conseil Supérieur de l’Emploi, des Revenus, et des Coûts, 1998, p. 110.


See Conseil Supérieur de VEmploi, des Revenus, et des Coûts, 1998, p.29, and Ministère de l’Emploi et de la Solidarié, 1998, p. 16.


Under the law, a worker is employed part-time if he works on average no more than 80 percent of the legal workweek.


See Maarek, G.: Le fétichisme des 35 heures, in Problèmes économiques, July 1998, No. 2.575, p.6.


See for example the paper by G. Cette and A. Gubain in La Réduction du Temps de Travail: une Solution pour l’Emploi?, Cahuc P., and P. Granier (Eds.), 1997.


It follows from the convexity of the profit function in wages and prices, and Hotelling’s Lemma, that demand curves for inputs are always negatively sloped. See for example, H. Varian, Microeconomic Analysis, 1984, p.51.


See Cornilleau G., E. Heyer and X. Timbeau: Les 35 heures en douceur? in Lettres de l’OFCE, January 21, 1998. The gain from re-organization would presumably be a one-off affair. To put the claim in perspective, recall that the rate of productivity growth over the last 20 years in France has averaged 2.5 percent per year (see Accardo, J. and M. Jlassi: La productivité globale des facteurs entre 1976 et 1996, INSEE, May 1998.) Note that “excessive” productivity growth may damage employment creation to the extent that less hours worked are then needed to produce the same output.


For some evidence of the scope for such reorganization see Blanchard O.J., and J-P. Fitoussi, Croissance et Chômage, La Documéntation Française, Paris, 1998, p. 27.


The 35-hour law mandates that this threshold will be reduced to 41 hours on January 1, 1999.


Two-third of the 2.3 million SMIC-earners are employed in the service sector; of these, one third in hotels and restaurants, recreational activities, and “personal services,” where the scope for productivity gains may be limited by the use of fixed coefficient technologies.


The impact of the law has been limited to about 200,000 workers. Self-selection bias suggests caution in deriving lessons from this experiment.


See DARES, Premières Informations et Premières Synthèses, No. 03.1, 1998.


The presumption that the natural rate is independent of the length of the legal workweek is of course consistent with the historical record. It can be given a theoretical underpinning by thinking of a reduction in work time as a negative productivity shock that shifts by the same amount the wage-setting and the price-setting relationships in real wage-unemployment space. See Blanchard O.J., Macroeconomics, 1997, p.518.


This, in turn, may have negative fiscal consequences if total revenues become insufficient to finance an unchanged supply of public goods. Prospective tax increases will in turn discourage work.


Using the Blanchard-Fitoussi estimate of a -0.8 elasticity of unemployment with respect to employment, unemployment under this scenario would fall by about 2 percentage points.


This corresponds to all workers in firms with more than 20 employees.


See, Le Figaro, August 31, 1998, p. 9, and the Web-site:


Two trade unions—the CGT and the CFDT—refused to sign the agreement because it failed to generate jobs. Minister Aubry has declined to endorse the agreement and has even suggested that it may be invalid. See The Financial Times, July 30, 1998.

France: Selected Issues
Author: International Monetary Fund