Abstract
This paper presents an overview of the different methodologies that can be used to measure the amount of available slack in an economy. A comparison of these alternatives reveals that the evolution of the Nairu is crucial for understanding recent cyclical developments in the French economy. A comprehensive reform of the personal income tax has been assessed. This paper analyzes the expansion of high-tech activities, and reviews a number of policy issues connected with the development of an information and communication technology-based new economy.
I. Measures of Slack in the French Economy1
A. Introduction
5. This Chapter presents an overview of the different methodologies that can be used to measure the amount of available slack in an economy. Three different approaches are discussed: statistical identification, economic identification, and survey-based measures of slack. A comparison of these alternatives reveals that the evolution of the Nairu is crucial for understanding recent cyclical developments in the French economy.
6. Following the cyclical trough in 1993, activity recovered timidly in 1994–96 but then accelerated. Since 1997, GDP growth has grown above potential, more than a million new jobs have been created, and unemployment has declined by almost 3 percentage points to 9½ percent, a level not seen since the peak of the previous upswing. This performance has been achieved in a setting of price stability—inflation has averaged less than one percent and wage growth has been contained—and accompanied by various structural reforms, notably an ongoing privatization program, the reduction in taxes and social security contributions at the low-end of the income distribution and, more recently, the implementation of the 35-hour workweek.
7. At this cyclical juncture, it becomes both more difficult and more important to gauge the amount of slack remaining in the economy. Labor market and tax reforms have arguably reduced the Nairu and raised potential, as evidenced by continued low wage and price inflation in the face of tightening labor markets. But it is difficult to estimate the extent of this reduction. At the same time, lags in the formulation and effect of macroeconomic policy imply a need to forecast future inflationary pressures, rather than react only after they have appeared. And policy can err on both sides, reacting too late and allowing pressures to build, or too early and cutting off the expansion prematurely.
8. The variable most commonly used for this analysis is the output gap, which measures the distance between current and potential GDP. A related measure is the unemployment gap, which measures the distance between the unemployment rate and the Nairu (the unemployment rate consistent with stable inflation). Both gaps can be described in terms of cycles that evolve around a long term trend (that is, potential output and the Nairu respectively). As neither the trend nor the cycle can be directly observed, the crucial aspect of identification is the allocation of movements in output to the trend and the cyclical component. There are two econometric procedures to identify these components, based on statistical and economic properties, respectively. As these methodologies rely on different assumptions for identification, they are bound to deliver divergent results and can be considered as “different windows through which economists can examine their models and data” (Canova (1998)).
B. Statistical Identification
9. Statistical identification amounts to defining the main features (such as order and type of integration of the trend and length or periodicity of the cycle) and the relationship between the trend and cycle. So-called band-pass filters, such as the Hodrick-Prescott (HP) and Baxter and King (BK) filters, are widely used in applied macroeconomics because of the their simplicity and their ability to replicate NBER turning points in U.S. GDP (see Canova (1998)). Identification is achieved by assigning the business cycle a periodicity of 2-8 years, and defining a smooth but variable stochastic trend. Figure I.1 shows the results of estimating the French output gap with the Hodrick-Prescott and Baxter and King filters.2 Because of the traditional criticisms of the crude application of the HP filter,3 a modified version is used here, the HP-Arima (Maravall and Kaiser (2000)), which forecasts and backcasts the series with an Arima model to minimize the end-of-sample problems.
Output Gaps: Statistical Identification
Citation: IMF Staff Country Reports 2000, 148; 10.5089/9781451813517.002.A001
Source: INSEE and Fund staff estimates.10. Two results stand out. The 1996 slowdown did not widen the gap below the level of the cyclical trough in 1993; and output exceeded potential beginning in mid–1997, with the gap rising to around -1 percent at end–1999. The annual rate of growth of potential output implicit in this calculation is 2-2¼ percent over the sample.
C. Economic Identification
11. Economic theory can be used to identify trend and cycle in two different ways. The first relies on economic theory to statistically identify the trend. For example, King et al. (1991) develop a general equilibrium model where the main endogenous variables share a common stochastic trend. Blanchard and Quah (1989) embed economic assumptions about the nature of supply and demand shocks in a structural VAR to identify potential output. A second approach is to derive reduced-form equations from economic models. For example, the Nairu may be estimated using a Phillips curve (see, among others, Staiger, Stock, Watson (1996)) or using structural models of the labor market (see Layard, Nickell, and Jackman (1991)). 4
12. Recently, Apel and Jansson (1999) have proposed a procedure for the joint estimation of potential output and the Nairu, based on reduced-form equations. This approach has the advantage of exploiting the intrinsic economic relationship between the two: they both refer to an equilibrium state of the economy where the labor market clears, the economy is operating at potential, and inflation is stable. This methodology allows explicit incorporation of the covariation restrictions on cyclical output and cyclical unemployment, while taking into account the available information on inflation. Formally, the model contains the following equations:
where πt is the log difference of the CPI, ut the unemployment rate,
13. Equation (1) is a representation of the Gordon “triangle model”(Gordon 1997), in which inflation is a function of inertia, demand, and supply. It includes an expectations- augmented Phillips curve that controls for supply shocks. Equation (2) is a version of Okun’s Law, linking the cyclical components of output and unemployment5. Assumptions about the stochastic characteristics of the unobserved variables close the model: equations (3) and (4) assume that both the NAIRU and potential output contain a stochastic trend, and equation (5) specifies the evolution of cyclical employment as an authoregressive distributed lag.
14. A recursive Kalman filter algorithm (see Hamilton (1994)) is applied to find a sequence of optimal predictions of the observable variables for a given set of coefficients and a sequence of the unobservables. These predictions are compared to the actual values of the observables, and the forecast errors are then plugged into a maximum-likelihood routine to compute the optimal set of parameters and the corresponding estimates of potential output and the Nairu. Using data over the period 1975-1999 yields the estimates for the output gap shown in Figure I.2. The vector of exogenous supply variables, zt, includes relative import prices, the relative oil price (both deflated by the CPI), the exchange rate, and productivity. In addition, a dummy variable was included to account for the apparent change in regime in inflation volatility after 1985, probably tied to the franc fort policy.
Output Gap: Economic Identification
Citation: IMF Staff Country Reports 2000, 148; 10.5089/9781451813517.002.A001
15. The results are broadly similar to those obtained with the band-pass filters in terms of dating of turning points, but differ in terms of the amplitude of the cycle. The Apel-Jansson methodology suggests a more volatile cycle, with a output exceeding potential by almost 4 percentage points in 1989, and falling below potential by 2½ percentage points in 1993. The results differ from the band-pass estimation in two other aspects: the slowdown in 1996 generated a larger gap than in 1993, and the output would still be below potential by end-1999. The implicit annual rate of growth of potential output is 2–2½ over the sample, but it rose to 2¼–2½ percent after 1994.
16. The main difference between these results and the band-pass filter results turns out to be whether or not the unemployment rate is used for the calculation of the output gap. For example, survey-based measures of economic slack are an alternative to econometric estimation of the output gap is. Figure I.3 shows capacity utilization in industry and the index of business confidence, normalized by their respective means and standard deviations. They show a pattern remarkably similar to the one displayed by the band-pass definition of output gap, and both would suggest that output has exceeded potential since 1997. Estimating the Apel-Jansson model with either capacity utilization or business confidence, rather than unemployment, as a measure of demand yields results (not reported here) more in line with those obtained with the band-pass filters.
Survey Data
Citation: IMF Staff Country Reports 2000, 148; 10.5089/9781451813517.002.A001
Source: INSEE and Fund staff estimates.17. This result can be explained by the fact that unemployment remained persistently high during the initial phase of the recovery, and therefore while both the statistical measures and the survey-based measures point to a recovery since 1994, the estimate that includes the unemployment rate does not turn around until 1996. Thus, understanding the evolution of the unemployment rate and the Nairu seems key to ascertaining the extent of slack available in the French economy.
D. The Evolution of the Nairu
18. The Nairu represents the rate of unemployment that is compatible with stable inflation. Its origins are in the wage-price mechanisms advanced by Friedman (1968) and Phelps (1968), and it has recently been modeled as a time-varying concept that responds to macroeconomic and institutional shocks (see Gordon (1997), Staiger, Stock, Watson (1996), and Blanchard and Wolfers (2000)). The estimates that arise from equations (1) to (5) appear in Figure I.4. The Nairu rose significantly during the early-1990s recession, and reached a peak of close to 11 percent in 1995; it then declined to a value of about 9½ percent by end-1999. This compares to estimates of around 10 percent by the OECD and the Banque de France.
France: Unemployment and Nairu
Citation: IMF Staff Country Reports 2000, 148; 10.5089/9781451813517.002.A001
Source: INSEE and Fund staff estimates.19. Although the unemployment rate continued to rise until 1997, the estimate of the Nairu starts to decline in 1994. This decline would have allowed for an increase in the rate of growth of potential output, which could explain why the measure of the output gap obtained with the multivariate model (Figure I.2) remains negative throughout the period. By contrast, the band-pass filters, based solely on the information contained in the GDP series, allocate the fluctuations of output during this period mainly to the cyclical component; hence the rapid reduction in the output gap. As regards the difference with respect to the survey-based measures, it could be argued that individual economic agents are not well-placed to gauge the extent of gradual structural changes. For example, entrepreneurs used to an environment with a high rate of unemployment, where finding additional employees was therefore easy, may find it difficult to hire additional resources in the new lower-unemployment environment if they fail to adapt their search process accordingly. Thus, survey-based measures may overreact to these types of structural changes and, by adding up myopic individual opinions, report capacity constraints that are not present overall.
20. Therefore, in order to discriminate between these two interpretations of the recent cyclical recovery, it becomes important to assess the plausibility of this estimated decline in the Nairu since 1993. In order to do so, a discussion of the main factors that affect its evolution, namely labor market policies, institutional changes, and factors affecting the rate of growth of labor productivity seems necessary.
21. Since the early 1990s, several labor market policies have been implemented in France with the objective of reducing structural unemployment (see Box I.1). Some programs, such as the reduction in the cost of labor and the increase of the employability of the low-skilled aimed to reduce the unemployment level compatible with a given level of wages (see also Chapter 2 for details on recent tax reforms). Other strategies, such as the increase in labor market flexibility, have essentially aimed to raise labor productivity. The effects of the 35-hour workweek on the Nairu seem favorable: the annualization of work time may have lowered the Nairu through an induced increase in productivity; in addition, the increase in the scope of collective bargaining, with the introduction of more flexible work arrangements, and the pervasiveness of wage moderation in the context of rapidly declining unemployment seem to suggest a decrease in insiders’ bargaining power and thus a reduction in the Nairu6.
22. Recent research also points to the importance of the interaction between macroeconomic shocks and labor market institutions for the evolution of the Nairu (Blanchard and Wolfers (2000), Fitoussi et al. (2000)). In particular, institutions would affect both the size and persistence of the effect on the Nairu of a given macroeconomic shock. Under this perspective, the recent rapid decline in the Nairu could be reflecting the coalescence of positive macroeconomic shocks with a string of labor market reforms, creating a sort of positive and self-reinforcing hysteresis effect. Indeed, a plausible hypothesis could be that the accumulated impact of certain labor market measures at the moment of strong economic growth and declining real interest rates has allowed for a significant reduction in long-term and low-skilled unemployment, the main two sources of structural unemployment.
23. In summary, the results of this paper imply that the output and labor market gaps have been narrowing rapidly in the past few years, notwithstanding a reduction in the Nairu. Indeed, the gaps can be expected to close in the course of next year. This conclusion, which flows from the Apel-Jansson methodology, is at odds with other, more traditional methods and indicators, although these suggest that markets are tighter still. Nevertheless, the differences in the estimates of macroeconomic slack illustrate the broader uncertainties regarding estimation of potential output and the Nairu. These are particularly acute because various reforms—key ones of which are discussed in the next Chapter—can be expected to have lowered the Nairu by an amount that is difficult to determine with great precision. In addition, as discussed in Chapter III, the new economy may be boosting productivity growth and therefore potential output, as appears to have been the case in the United States.
Labor Market Policies in the 1990s: Reducing Structural Unemployment
Beginning in the early 1990s, the French authorities implemented policies to reduce structural unemployment through lowering the cost of low-skilled and inexperienced labor, increasing the employability of the long-term unemployed, enhancing labor market flexibility, and enhancing work incentives. The main building blocks of this strategy are listed below.
Reduction of the cost of labor. The 1993 Five year law on labor, employment, and training set the stage for a durable reduction of the cost of low-skilled labor, including exemptions of employer contributions to the family branch of the general social security regime targeted to employees earning between 1 and 1.2 times the minimum wage (SMIC). In 1995, these exemptions were expanded and consolidated in the so-called ristourne dégressive, a system of tapered rebates on other categories of employers’ social security contributions (notably health and basic pensions) up to 1.3 times the SMIC. This ristourne was revised in 1998 and in 2000, in the context of the implementation of the 35-hour workweek; it is currently applied to workers earning up to 1.8 times the SMIC.
Increase the employability of the unemployed. Over the last decade, active labor market programs have proliferated, focused on enhancing the opportunities of the young, the long-term unemployed, and those with low earning potential. Broadly, these measures encompass three types of programs: (i) programs targeted to the market sector involving fiscal transfers (usually in the form of reductions in social charges plus monthly subsidies), such as the Contrat-Initiative-Emploi; (ii) programs involving minimum wage concessions and training, such as the contrat d’apprentissage; and (iii) subsidized programs targeted to the non-market sector, such as the Nouveaux Services, Emploi-Jeunes.
Reform of working time. With the intention of increasing the flexibility of work arrangements through the annualization of working hours, the 1993 Five year law on labor, employment, and training introduced an experimental program that allowed the number of hours worked to be varied at the enterprise level, up to a maximum of 48 hours per week, so long as the annual number of hours worked was reduced. This system was later formalized in the 1996 Loi Robien, which provided a subsidy for firms reducing working time by 10 to 15 percent and providing an equivalent increase in employment. Finally, in 1998, the 35-hour workweek initiative was launched, culminating in 2000 (2002 for small firms) of a reduction in the workweek to 35 hours and the widespread adoption of annualization.
Increase labor market flexibility. Although the legal framework had existed since the early 1980s, part-time work increased rapidly in the second half of the 1990s, following several initiatives that provided partial exemptions from employer’s social security contributions and increased the number of hours that could be considered part time. Similar measures were implemented to promote temporary and fixed-term work, such as the increase in the duration and maximum number of renewals of fixed-term contracts. Temporary and fixed-term contracts account for about one-third of the new jobs created since 1997;similarly, one-fifth of the new jobs created over this period are on a part-time basis.
Strengthen the incentives to return to work and avoid poverty traps. Unemployment insurance was reformed in 1993, with the reduction in the duration of benefits, a tightening of the eligibility requirements, and the introduction of tapered benefits. The mechanism of interessement was later introduced for both unemployment benefits and the basic income support scheme (revenue minimum d’insertion (RMI). This allows claimants to maintain, for a certain period, part of their entitlement rights even as they earn income, thus reducing the disincentive to work. More recently, other measures (including changes to the taxe d’habitation and a proposed graduated rebate on the contribution sociale généralisée, CSG) can also be expected to reduce threshold effects and strengthen work incentives.
References
Apel, Mikael, and Per Jansson, 1999, “System Estimates of Potential Output and the NAIRU,” Empirical Economics, 24, 373 ‑388.
Baxter, Marianne and Robert King, 1999, “Measuring Business Cycles: Approximate Band-Pass Filters for Economic Time Series”, The Review of Economics and Statistics, Vol. 81, pp. 575 ‑593.
Blanchard, Olivier, and Danny Quah, 1989, “The Dynamic Effects of Aggregate Demand and Supply Disturbances”, American Economic Review, 79, 655 ‑73.
Blanchard, Olivier, and Justin Wolfers, 2000, “The Role of Shocks and Institutions in the rise of European Unemployment: the Aggregate Evidence”, The Economic Journal, 123 ‑145.
Canova, Fabio, 1998, “Detrending and Business Cycle Facts”, Journal of Monetary Economics, 41, 475 ‑512.
Cogley, Tomas and James Nason, 1995, “Effects of the Hodrick-Prescott Filter on Trend and Difference Stationary Time Series: Implications for Business Cycle Research”, Journal of Economic Dyamics and Control, 19, 253 ‑278.
Fitoussi, Jean Paul, David Jestaz, Edmund Phelps, and Gylfi Zoega, 2000, “Roots of the Recent Recoveries: Labor Reforms or Private Sector Forces?,” Brooking Papers on Economic Activity, 1, 238 ‑291.
Friedman, Milton, 1968, “The Role of Monetary Policy”, American Economic Review, 58, 1, 1 ‑17.
Gordon, Robert, 1997, “The time varying Nairu and its implications for economic policy”, Journal of Economic Perspectives, 11, 11 ‑32.
Hamilton, James, 1994, Time Series Analysis, Princeton: Princeton University Press.
Harvey, Andrew, and Albert Jaeger, 1993, “Detrending, Stylized Facts and the Business Cycle”, Journal of Applied Econometrics, 8, 231 ‑247.
Hodrick, Robert, and Edward Prescott, 1981 “Post-War U.S. Business Cycles: An Empirical Investigation” Carnegie Mellon Discussion Paper 451.
King, Robert, Charles Plosser, James Stock, and Mark Watson, 1991, “Stochastic Trends and Economic Fluctuations,” American Economic Review, 81, 819 ‑40.
Layard R., S Nickell and R Jackman, 1991, Unemployment, Oxford: Oxford University Press.
Maravall, Agustin and Regina Kaiser, 2000, “Estimation of the Business Cycle: A Modified Hodrick-Prescott Filter”, Spanish Economic Review, 1, 175 ‑206.
Phelps, Edmund, 1968, “Money-Wage Dynamics and Labor Market Equilibrium”, Journal of Political Economy, 76, 4, 678 ‑711.
Staiger, Douglas, James Stock, and Mark Watson, 1996, “How Precise are Estimates of the Natural Rate of Unemployment?,” NBER Working Paper No. 547
Prepared by Angel Ubide-Querol.
Potential GDP may be estimated using a production function approach, which is useful to disentangle the contribution of labor, capital, and productivity to potential output growth. However, this approach depends on assumptions about the functional form of the production function and the calibration of its parameters. In addition, in practice it requires detrending some of the variables in the production function, thus being subject to the standard criticisms about detrending methods. In fact, one could interpret this approach as an application of detrending, where the potential output series is a weighted average of the detrended inputs.
Although this is the typical representation of Okun’s Law, one could argue that, from an empirical point of view, unemployment typically lags GDP, and therefore equation (2) should be reversed. The results of this exercise, not presented here, are broadly similar to the main specification of the model.
However, the overall effects of the 35-hour workweek initiative on potential output are ambiguous, since the associated reduction in labor supply would reduce potential output.
