Abstract
This paper analyzes a broad range of price and nonprice indicators to assess developments in the international competitiveness of the French economy during the 1980s and early 1990s. The paper provides a brief review of conceptual issues concerning the competitiveness indicators used in this study. Developments in conventional price- and cost-based indicators, both at the aggregate and bilateral levels, are reported. The paper discusses additional price- and quantity-based measures of competitiveness, and also examines the labor market dynamics and economic policy of France.
III. Potential Output and the Output Gap in France and Western Germany 1/
1. Introduction and summary
This note compares estimates of potential output and the output gap in France and western Germany. Estimates are based on a production function relating labor, capital, and capacity utilization to output levels. According to these estimates, potential output growth in France and Germany is now about 2 1/4 percent, reflecting a drop in investment rates in both countries in recent years (Charts 1 and 2). 1/ It was also found that the output gap was about 1 percentage point wider in France than in Germany in 1994. However, it is estimated that this difference largely disappeared in 1995, with the output gap in both countries amounting to some 2 1/4 percent.
FRANCE: Economic Indicators
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: INSEE, Quarterly Accounts; and IMF, International Financial Statistics.1/ In manufacturing.FRANCE: Economic Indicators
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: INSEE, Quarterly Accounts; and IMF, International Financial Statistics.1/ In manufacturing.FRANCE: Economic Indicators
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: INSEE, Quarterly Accounts; and IMF, International Financial Statistics.1/ In manufacturing.
WESTERN GERMANY: Economic Indicators
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Deutsche Bundesbank, Monthly Report.1/ In manufacturing and mining.WESTERN GERMANY: Economic Indicators
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Deutsche Bundesbank, Monthly Report.1/ In manufacturing and mining.WESTERN GERMANY: Economic Indicators
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Deutsche Bundesbank, Monthly Report.1/ In manufacturing and mining.As explained below, estimates of the output gap are subject to relatively large statistical errors, arising in part from the difficulty of estimating the potential, or non-accelerating-inflation, rate of unemployment (NAIRU). Nevertheless, estimates of the output gap can be useful indicators of the scope for non-inflationary economic expansion. Indeed, the output gap in France and Germany has been positively correlated with variations in the inflation rate, even though the correlation has not been especially close.
This paper is organized as follows. Section 2 presents the production function and describes the method used to estimate trend total factor productivity. Sections 3 and 4 discuss the evolution of inputs (capital, employment and working hours) in France and Germany and how their potential or trend values are estimated. They also show how these inputs have contributed to potential output growth in the respective countries. Section 5 presents calculations of the output gap in France and Germany in recent years, with a breakdown of the contribution of employment, working hours and changes in capacity utilization to cyclical variations in GDP. The relationship between the output gap and inflation is discussed in Section 6, while Section 7 discusses potential output growth and inflation in 1995-2000. Appendices provide supplementary information on (i) the NAIRU, (ii) the evolution of the share of labor in output in France, and (iii) the Hodrick-Prescott filter.
2. Production function and total factor productivity
The following Cobb-Douglas production function is used for estimating potential output:
with
L = (Pwa * PR) * (1 - U) * H
where Y is output, c is a constant, a (0<a<1) is the elasticity of output with respect to labor, set to be equal to labor’s share in value added (1-a is the elasticity of output with respect to capital), L is labor input, Pwa is the working age population, PR is the labor force participation rate, U is the unemployment rate, H is average working hours, K is the capital stock, util is the deviation of the capacity utilization rate in industry from its historical average, tfpz is the (flexible) trend of total factor productivity, and e is a random shock. 1/ The capacity utilization variable is a proxy for changes in the intensity of use of capital and employed labor over the economic cycle. 2/ Potential and actual capital stocks are equivalent.
Total factor productivity and the parameter “b” (the elasticity of output with respect to capacity utilization) are estimated jointly following an iterative procedure. First, the contribution of labor and capital to the log of output implied by a given value of the parameter “a” is computed according to equation (1) and subtracted from the log of observed GDP, the residual being denoted resd0. Then, a first estimate of the elasticity “b” is obtained by regressing that residual on a constant, a time trend and the capacity utilization variable:
That part of resd0 which is not explained by variations in capacity utilization provides the first approximation of the contribution of total factor productivity:
This approximation is smoothed using a Hodrick-Prescott (HP) filter and substituted for r + v*t in equation (2). By re-estimating equation (2), a new estimate of b is obtained, which is inserted in equation (3), yielding a new estimate of tfp, which is in its turn smoothed. The process quickly converges to a final estimate of b, which is again inserted in equation (3), to find a final value of tfp. The smoothed series of tfp obtained from the updated equation (3) gives the flexible trend tfp (tfpz).
3. Data and estimates of potential inputs and output for France
a. Capital stock
The capital stock data used is a quarterly OECD series which comprises buildings and machinery in the business sector. This series is based on estimates published by INSEE (e.g., INSEE, 1993) with adjustments to include estimates of the stock of business buildings (Keese et. al., 1991). INSEE and OECD series are constructed using the perpetual inventory method, with the service lives of capital goods described by a bell-shaped statistical distribution. 1/ Because the INSEE series are annual data, the OECD interpolates end-of-year values to obtain quarterly observations. The interpolation is made using data on business investment, which are available quarterly from the National Accounts.
b. Population growth and labor market participation
The major factor affecting the participation rate in France in the last two decades has been the reduction of the minimum retirement age from 65 to 60 years in 1982. The reduction of the minimum age was to some extent a response to the economic slowdown of 1980-81 and to the arrival of the last wave of post-World-War-II baby boomers on the labor market at the end of the 1970s (Chart 3, top panel), which led to a surge in the working age population (i.e., the share of the population between ages 15 and 65). 2/ The lowering of the minimum retirement age, coupled with incentives for early retirement led to a sharp fall in the participation rate (Chart 3, middle panel). In all, the participation rate decreased by 2 percentage points from 68.5 in 1979 to 66.5 percent in 1984. Since 1984 there has been a modest increase in the participation rate.
FRANCE: Labor Inputs
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: OECD, Analytical Database; INSEE, National Accounts; and staff calculations.FRANCE: Labor Inputs
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: OECD, Analytical Database; INSEE, National Accounts; and staff calculations.FRANCE: Labor Inputs
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: OECD, Analytical Database; INSEE, National Accounts; and staff calculations.The participation rate also responds to changes in employment, although its sensitivity to economic cycles is lower in France than in Germany. The lag of the response in France also appears to have varied over time. Despite the strong growth in employment observed in the late 1980s, participation did not increase significantly until 1991, after the cycle had peaked. 1/ By contrast, quarterly data for 1994 suggest that in the current upturn, the participation rate may have responded more quickly to the improvement in economic conditions. A quicker response in the recent period could be explained by the increase in jobs in the service sector (including part-time jobs). Because these jobs often require fewer skills and are less subject to seniority rules, they may seem more accessible to those previously outside the labor force, thus increasing the sensitivity of the participation rate to employment. A firm judgement about the most recent developments is, however, difficult, because high frequency data permitting an estimation of the labor force in the recent period are under revision owing to changes in the collection methodology. Annual data from surveys carried out by INSEE suggest only a minor increase in the labor force between March of 1994 and March of 1995. 2/ 3/ The potential participation rate is computed by smoothing the actual participation rate series. A 10th degree polynomial was used for this purpose.
c. Working hours
Potential average weekly working hours are computed using data from INSEE and the Ministry of Labor. Working hours decreased by more than 10 percent in the 1970s. Following the reduction of the working week in 1981 (of one hour), working hours declined further before stabilizing during the boom years in 1987-1990. They declined again in 1992-93 (Chart 3, bottom panel). A part of this decline can be attributed to cyclical reductions in overtime and short-term layoffs due to the recession, but a large part (between one-third and one-half) reflects an increase in part-time jobs (Dares, 1995). Potential working hours were computed by filtering actual hours of work. As expected, there is a substantial gap in hours in recent years. Nevertheless, trend hours continue to decrease. This is consistent with the increasing expansion of part-time work, reflecting more favorable regulations introduced in recent years and voluntary agreements on reduced working hours allowed by the 1993 Employment Law.
d. The NAIRU
The NAIRU is estimated using an econometric model based on an expectations-augmented Phillips curve. 1/ Estimates of this or similar measures (e.g., the non-accelerating-wage rate of unemployment (NAWRU)) for France have widely diverged. Recent estimates for the early 1990s range from 5.3 percent (Confais and Muet, 1994) to 9.6 percent (OECD, 1994). 2/ Based on the model presented in Appendix I, the NAIRU was estimated at 9.3 percent of the labor force in 1994. This estimate is subject to a standard error of 0.6 percentage point.
The estimated NAIRU has increased from about 6 percent in the mid 1970s to around 8.5 percent in the 1980s, rising to above 9 percent in the 1990s. The actual unemployment rate has increased from less than 3 percent in the early 1970s to 12.3 percent in 1994 (Chart 4, top panel).
FRANCE: Unemployment Rates and the Beveridge Curve
(In Percent of the Labor Force)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: INSEE, National Accounts and staff calculations.FRANCE: Unemployment Rates and the Beveridge Curve
(In Percent of the Labor Force)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: INSEE, National Accounts and staff calculations.FRANCE: Unemployment Rates and the Beveridge Curve
(In Percent of the Labor Force)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: INSEE, National Accounts and staff calculations.The Beveridge curve, which is an alternative approach to assess the “natural” unemployment rate suggests results compatible with the above econometric estimate of the NAIRU. 3/ In the case of France, the Beveridge curve shows several small “vertical” regions (Chart 4, bottom panel). Although leaving room for different interpretations, these peaks may reflect a continuous shift of the natural rate of unemployment: it rose from less than 2 percent in the 1960s to 3.5 percent before the first oil shock, to around 5 percent immediately after it, and to 8.0 percent following the second oil shock. Based on observations from the last episode of labor market tightness (1990), the natural rate would be around 9 percent at present.
e. Capacity utilization and total factor productivity
The elasticity of output with respect to capacity utilization was estimated using the methodology described in section 2 under the assumption that the parameter “a” in the production function (equation 1) for France is equal to 0.625. The value of the parameter “a” was chosen based on the average share of labor compensation in value added by firms excluding the Large National Enterprises, GENs (Chart 5, top panel), as computed by INSEE. Appendix II discusses this choice in the light of the decline in the share of labor in output registered since the mid-1980s. Using the series of capacity utilization in manufacturing computed by INSEE (Chart 5, bottom panel), the elasticity of output with respect to capacity utilization (the parameter “b” in equation 1) was estimated to be 0.44. 1/
FRANCE: Labor Share and Capacity Utilization
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: INSEE, National Accounts; and staff calculations.FRANCE: Labor Share and Capacity Utilization
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: INSEE, National Accounts; and staff calculations.FRANCE: Labor Share and Capacity Utilization
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: INSEE, National Accounts; and staff calculations.After accounting for the contribution of trend tfp and capacity utilization, the production function presented in section 2 explains two thirds of the year-to-year variations of quarterly GDP. The remaining variation is due to the pro-cyclical nature of factor productivity. 2/
f. Potential output growth and individual input contributions
The path of potential output is obtained by inserting inputs at “potential” levels into the production function. The growth of potential output declined from around 3 percent in the 1970s to below 2 percent in the early 1980s; it recovered to about 2 3/4 percent in the late 1980s, but fell again to about 2 1/4 percent in the 1990s (Chart 6, top panel).
FRANCE: Contributions to Potential Output
(Annualized Quarterly Growth Rates)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.FRANCE: Contributions to Potential Output
(Annualized Quarterly Growth Rates)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.FRANCE: Contributions to Potential Output
(Annualized Quarterly Growth Rates)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.The contribution of labor to potential output in France was negative during 1977-86, but became slightly positive thereafter. 3/ The fall in labor’s contribution observed in 1982 resulted from the reduction in both the participation rate and working hours, reflecting the reductions in the minimum retirement age and the work week. In addition, the contribution of labor was depressed by the increase in the NAIRU in 1981-84. In more recent years, the small negative impact of further increases in the NAIRU was offset by trend increases in the participation rate and the working age population (Chart 6, bottom panel). The contribution of capital declined in 1975-87. It increased in the late 1980s, owing to a surge in investment, but has fallen again in the wake of three years of decreasing investment. Indeed, low investment has been the main factor behind the decline in potential growth in recent years. Finally, the contribution of trend tfp fell from about 2 percent in the late 1970s to around 1 1/4 percent in 1994.
4. Data and estimates of potential inputs and output for Germany
a. Capital stock
Data on capital stock for Germany are annual and cover 1970-94. They are computed by the Federal Statistical Office by applying the perpetual inventory method and correspond to the gross capital stock (excluding dwelling). For the purpose of estimation, the year-average capital stock is constructed as the average of two year-end values.
b. Population growth and labor market participation
Working age population in western Germany was boosted by unification. It had previously shown a pattern not entirely different from that in France, where the early 1980s were marked by the entrance of the baby-boom generation into the labor force (Chart 7, top panel). The participation rate in Germany showed a decline from 1960 to the mid-1980s, followed by an increase during the wave of immigration in the late 1980s-early 1990s (Chart 7, middle panel). 1/ Trend labor force participation rates were estimated by using a HP filter on data for 1960-94. As expected, the actual participation rate in 1993-94 was found to be below the trend participation rate.
WESTERN GERMANY: Labor Inputs
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: Deutsche Bundesbank, Montly Report; and staff calculations.WESTERN GERMANY: Labor Inputs
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: Deutsche Bundesbank, Montly Report; and staff calculations.WESTERN GERMANY: Labor Inputs
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: Deutsche Bundesbank, Montly Report; and staff calculations.c. Working hours
As elsewhere in Europe, working hours have significantly decreased in Germany since 1970 (Chart 7, bottom panel). The trend decline in working hours observed over the last two decades will probably continue in the 1990s, but not at the same pace. In order to incorporate this judgment into the mechanics of the (endpoint-sensitive) HP filter used to obtain potential levels, working hours were extrapolated over the next decade assuming a slower decline (0.2 percent a year). The HP filter was applied to the resulting series for 1969-2005. The results show the expected slowdown of the trend decline in hours in the early 1990s, and working hours in 1994 are, as expected, below their trend level. 2/
d. The NAIRU
The NAIRU is estimated to have risen gradually from 1970 to the late 1980s, based on evidence summarized in the 1994 report on Recent Economic Developments (SM/94/213, chapter V). This pattern is consistent with studies which put the NAIRU at 6-9 percent in the late 1980s, the fact that all measures of inflation began to rise between 1987 and 1989 when unemployment in Germany was still above 7 percent (Chart 8, top panel), and the shift in the Beveridge curve since 1983 (Chart 8, bottom panel). In particular, the NAIRU is put at 6 3/4 percent in 1987-92, following the sharp rise in unemployment in the mid-1980s, and is estimated at about 7 percent in 1993-94.
WESTERN GERMANY: Unemployment Rates and the Beveridge Curve
(In Percent of Labor Force)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: Deutsche Bundesbank, Monthly Report; Bundesanstalt fur Arbeit; and Statistisches Bundesamt.WESTERN GERMANY: Unemployment Rates and the Beveridge Curve
(In Percent of Labor Force)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: Deutsche Bundesbank, Monthly Report; Bundesanstalt fur Arbeit; and Statistisches Bundesamt.WESTERN GERMANY: Unemployment Rates and the Beveridge Curve
(In Percent of Labor Force)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Sources: Deutsche Bundesbank, Monthly Report; Bundesanstalt fur Arbeit; and Statistisches Bundesamt.e. Capacity utilization and total factor productivity
The parameter “a” in the production function is set equal to 0.7, i.e., approximately the average labor share in GDP at factor cost (Chart 9, top panel). In the absence of official estimates for the imputed wages of the self-employed, total labor income is calculated as gross income from dependent employment per employee, multiplied by total employment (all using the workplace rather than the residence concept). 1/ The elasticity of output with respect to capacity utilization in manufacturing (Chart 9, bottom panel) was estimated at 0.12, well below the value estimated for France, reflecting the higher variability of the German series in the past. 2/ Total factor productivity, which had fallen from more than 2 percent in the early 1970s to close to 1 percent in the early 1980s, increased to 1 3/4 percent in the years following the reunification. This increase is largely explained by the massive restructuring pursued by German firms in response to higher wages and the appreciation of the deutsche mark in the 1990s.
WESTERN GERMANY: Labor Share and Capacity Utilization
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.1/ Including the self-employed.2/ Break in series in 1978-79. Old and new series overlap for a period.WESTERN GERMANY: Labor Share and Capacity Utilization
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.1/ Including the self-employed.2/ Break in series in 1978-79. Old and new series overlap for a period.WESTERN GERMANY: Labor Share and Capacity Utilization
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.1/ Including the self-employed.2/ Break in series in 1978-79. Old and new series overlap for a period.f. Potential output growth and individual input contributions
Potential growth was on a broadly declining trend from 3 3/4 percent in the early 1970s to 1 3/4 percent in the mid-1980s, reflecting slowdowns in tfp growth and, especially, in investment (Chart 10). Labor’s contribution to potential growth was negative almost throughout this period, because not only were working hours falling, but also the NAIRU was rising and the trend participation rate declining. Only around 1981 did the contribution from labor turn briefly positive, as the coming of age of the baby boom generation briefly raised the growth of the working age population to close to 1 1/2 percent a year (compared with an average of only 1/2 percent in the period 1970-85).
WESTERN GERMANY: Potential Output
(Annual Growth Rates in Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.WESTERN GERMANY: Potential Output
(Annual Growth Rates in Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.WESTERN GERMANY: Potential Output
(Annual Growth Rates in Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.Potential growth rose sharply in the late 1980s reaching a peak of 3 3/4 percent in 1990-91, as a result mainly of an immigration-driven increase in the labor force and a rising number of inward commuters. When these tailed off and investment declined in the recent recession, potential growth fell back to around 2 1/4 percent.
5. Estimation of the output gap in France and Germany
Estimates of the output gap are subject to a number of uncertainties, in particular those relating to the estimation of the NAIRU. Reflecting only the uncertainty concerning the NAIRU, point estimates of the gap have a precision of no more than 1/2 percentage point of GDP in the case of France (this confidence interval assumes that the correct specification for the NAIRU estimation was chosen but it disregards the uncertainties associated with the estimate of the elasticity of output with regard to capacity utilization and the distinction between cyclical and structural reductions in working hours). 1/ Results from other approaches of computing the output gap, such as the Hodrick-Prescott filter, can also vary considerably depending on small changes in the application of the filter (see Appendix III). Nevertheless, output gap estimates can be useful indicators of the scope of non-inflationary economic expansion. Present estimates suggest that the average output gap was larger in France (3 1/4 percent) than in Germany (2 1/4 percent) in 1994, but has been of broadly the same size (about 2 1/4 percent) in both countries in 1995 (Chart 11). While the unemployment rate is higher in France than in Germany, so is the NAIRU, yielding an employment gap that is only moderately larger in France (Table 1). In addition, the participation rate in France in 1995 was at “potential,” while that in Germany was well below trend. This reflects the generally more pronounced cyclical fluctuations in the participation rate in Germany.
FRANCE and WESTERN GERMANY: Contributions to Output Gap
(In Percent of GDP)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.FRANCE and WESTERN GERMANY: Contributions to Output Gap
(In Percent of GDP)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.FRANCE and WESTERN GERMANY: Contributions to Output Gap
(In Percent of GDP)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.Output Gap
(In percent of potential output)
In 1980 prices.
In 1991 prices.
Output Gap
(In percent of potential output)
| 1989 | 1990 | 1991 | 1992 | 1993 | 1994 | 1995 (Proj.) | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| France | |||||||||||
| Output Gap | 2.4 | 2.3 | 0.6 | -0.4 | -4.0 | -3.2 | -2.2 | ||||
| Contribution of | |||||||||||
| Participation Rate | -- | -0.1 | 0.1 | -- | -- | -- | -- | ||||
| Unemployment | -0.2 | 0.2 | -0.1 | -0.7 | -1.7 | -2.0 | -1.5 | ||||
| Working Hours | 0.2 | 0.3 | 0.3 | 0.1 | -0.5 | -0.7 | -0.6 | ||||
| Capacity Utilization | 1.8 | 1.7 | 0.3 | -0.5 | -1.7 | -1.0 | -0.2 | ||||
| Residual | 0.6 | 0.1 | -- | 0.7 | -- | 0.5 | -- | ||||
| Output Growth Rate (in percent per year) 1/ | |||||||||||
| Actual GDP | 4.3 | 2.5 | 0.8 | 1.3 | -1.5 | 2.9 | 2.9 | ||||
| Potential GDP | 2.5 | 2.5 | 2.4 | 2.3 | 2.1 | 2.1 | 2.0 | ||||
| Germany | |||||||||||
| Output Gap | 0.4 | 2.4 | 3.7 | 2.2 | -2.3 | -2.3 | -2.3 | ||||
| Contribution of | |||||||||||
| Participation rate | 0.3 | 0.5 | 0.4 | 0.3 | -0.2 | -0.6 | -0.7 | ||||
| Unemployment | -0.1 | 0.4 | 0.9 | 0.7 | -0.3 | -1.0 | -1.0 | ||||
| Working Hours | -- | -0.1 | -0.3 | -- | -0.7 | -0.2 | -0.2 | ||||
| Capacity Utilization | 0.8 | 0.9 | 0.6 | -- | -0.7 | -0.1 | 0.3 | ||||
| Residual | -0.6 | 0.7 | 2.1 | 1.2 | -0.3 | -0.5 | -0.8 | ||||
| Output Growth Rate (in percent per year) 2/ | |||||||||||
| Actual GDP | 3.6 | 5.7 | 5.0 | 1.8 | -1.8 | 2.4 | 2.2 | ||||
| Potential GDP | 2.7 | 3.7 | 3.7 | 3.3 | 2.7 | 2.4 | 2.2 | ||||
In 1980 prices.
In 1991 prices.
Output Gap
(In percent of potential output)
| 1989 | 1990 | 1991 | 1992 | 1993 | 1994 | 1995 (Proj.) | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| France | |||||||||||
| Output Gap | 2.4 | 2.3 | 0.6 | -0.4 | -4.0 | -3.2 | -2.2 | ||||
| Contribution of | |||||||||||
| Participation Rate | -- | -0.1 | 0.1 | -- | -- | -- | -- | ||||
| Unemployment | -0.2 | 0.2 | -0.1 | -0.7 | -1.7 | -2.0 | -1.5 | ||||
| Working Hours | 0.2 | 0.3 | 0.3 | 0.1 | -0.5 | -0.7 | -0.6 | ||||
| Capacity Utilization | 1.8 | 1.7 | 0.3 | -0.5 | -1.7 | -1.0 | -0.2 | ||||
| Residual | 0.6 | 0.1 | -- | 0.7 | -- | 0.5 | -- | ||||
| Output Growth Rate (in percent per year) 1/ | |||||||||||
| Actual GDP | 4.3 | 2.5 | 0.8 | 1.3 | -1.5 | 2.9 | 2.9 | ||||
| Potential GDP | 2.5 | 2.5 | 2.4 | 2.3 | 2.1 | 2.1 | 2.0 | ||||
| Germany | |||||||||||
| Output Gap | 0.4 | 2.4 | 3.7 | 2.2 | -2.3 | -2.3 | -2.3 | ||||
| Contribution of | |||||||||||
| Participation rate | 0.3 | 0.5 | 0.4 | 0.3 | -0.2 | -0.6 | -0.7 | ||||
| Unemployment | -0.1 | 0.4 | 0.9 | 0.7 | -0.3 | -1.0 | -1.0 | ||||
| Working Hours | -- | -0.1 | -0.3 | -- | -0.7 | -0.2 | -0.2 | ||||
| Capacity Utilization | 0.8 | 0.9 | 0.6 | -- | -0.7 | -0.1 | 0.3 | ||||
| Residual | -0.6 | 0.7 | 2.1 | 1.2 | -0.3 | -0.5 | -0.8 | ||||
| Output Growth Rate (in percent per year) 2/ | |||||||||||
| Actual GDP | 3.6 | 5.7 | 5.0 | 1.8 | -1.8 | 2.4 | 2.2 | ||||
| Potential GDP | 2.7 | 3.7 | 3.7 | 3.3 | 2.7 | 2.4 | 2.2 | ||||
In 1980 prices.
In 1991 prices.
Historically, the output gap in France was positive in the second half of the 1970s, but fell sharply following the second oil shock, eventually turning negative in the first half of the 1980s. 2/ It became positive again as the boom of the late 1980s unfolded, reaching a peak of 3 percent in the first quarter of 1990. Thereafter, it narrowed and eventually turned negative in 1992, widening to -4 percent at the trough of the recession in 1993, before narrowing to -2 1/2 percent in late 1994.
In Germany, the output gap was positive in the 1970s, with the exception of 1975. The negative gap in the first half of the 1980s was larger than in France due to a lower NAIRU and a negative contribution of cyclical productivity. The output gap turned sharply positive in the late 1980s, as the boost provided by unification resulted in a positive employment gap in 1990-92. Subsequently, the sharp decline in activity in 1993 turned the output gap from +2 1/4 percent in 1992 to -2 1/4 percent in the following year.
6. Output gap and inflation in France and Germany
Chart 12 suggests a positive relationship between the output gap and changes in inflation in France and Germany (e.g., inflation decelerates when the gap is negative). However, this correlation has weakened in recent years, especially in the case of France, as inflation fell to very low levels (in fact, inflation in France was zero for several months in 1994). 1/ The coefficient obtained from regressing year-to-year changes in inflation on the output gap in France is close to 0.4 and significant at 5 percent level in 1974-94. However, this coefficient falls to less than 20 percent in 1984-94.
FRANCE and WESTERN GERMANY: Output Gap and Inflation
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.FRANCE and WESTERN GERMANY: Output Gap and Inflation
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.FRANCE and WESTERN GERMANY: Output Gap and Inflation
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.7. Outlook for potential growth and inflation in France and Germany
As noted, potential output growth in France has fallen in recent years, mainly owing to low levels of investment. As investment recovers in 1995 and beyond, potential growth is forecast to increase to 2 1/2 percent by year 2000. This increase also incorporates a reduction in the NAIRU in coming years as a result of labor market measures, a continuous moderation in the increase in the working age population, and only a modest increase in the participation rate. However, with an output gap currently estimated at 2 1/4 percent of potential GDP and an employment gap of 2 percent, inflationary pressures are not expected to arise in France.
Using projections for fixed investment from the World Economic Outlook and labor force projections from the Federal Statistical Office, keeping the NAIRU unchanged, and setting TFP growth equal to its recent value, potential growth in Germany is projected to stay around 2 1/4 percent until the end of the decade. Although investment is expected to recover from its depressed level in the recent recession, the growth of the working-age population, which is currently small but positive, is expected to turn negative in the latter part of the 1990s. A possible offset might come from a decline in the NAIRU; this offset is, however, not currently incorporated in the above figures.
APPENDIX I: Estimates of the NAIRU in France
A crucial aspect of the calculation of the output gap is the estimation of the NAIRU (the non-accelerating-inflation rate of unemployment). Estimates of the NAIRU in France have diverged widely, ranging from 5.3 percent (Confais and Muet, 1994) to 9.6 percent (OECD, 1994), which is not surprising because economic theory does not provide a unique way to estimate the NAIRU, and different specifications can yield quite disparate results (Setterfield, et al., 1992). In view of these difficulties, three specifications are estimated here, including one in which the usual linearity assumption about wages and price increases is relaxed. 1/ All these specifications yield estimates of the NAIRU in the 1990s within 1.5 standard deviations from each other, i.e., ranging from 9 to 10 percent. They diverge more substantially, however, with regard to the late 1970s and early 1980s (Chart 13).
FRANCE: Estimates of the NAIRU
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.FRANCE: Estimates of the NAIRU
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.FRANCE: Estimates of the NAIRU
(In Percent)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.All specifications are derived from a system of equations based on an expectations-augmented Phillips curve. Where the linearity assumption about wages and prices is relaxed, the log of (annual) inflation and wage increases is used. In all specifications, the set of variables is confined to changes in wages, prices and total factor productivity, and to the unemployment and capacity utilization rates and their first difference. 2/ Models in the literature on the Phillips curve often include institutional factors aiming to explain why the downward pressure exerted by unemployment on wages may be limited. The most commonly used variables for this purpose are measures of the generosity of unemployment benefits and the strength of unionization. However, in the case of France, the series on unemployment benefits show breaks and capture only part of the actual level of social and economic support provided to the unemployed. 1/ The unionization rate, which has fallen from 30 percent to 10 percent of the working force since 1970, is also of limited use, because it does not capture the effect of collective bargaining agreements in core sectors on wage increases in the economy as a whole. 2/ Hence, the experience that these variables frequently produce disappointing results (Ford and Rose, 1989) was confirmed in the current study, and these variables were, therefore, not incorporated in the model.
As explained in Setterfield, et al. (1992), to estimate the NAIRU a system comprising a wage equation and a price equation needs to be estimated simultaneously. In the linear specifications, the wage equation assumes that year-on-year wage increases (dw4) are affected by past wage increases, past year-on-year consumer price increases (dwpc4), the unemployment rate (unprate), and its year-on-year change (dunprate). The price equation assumes that quarterly price increases (dp) are determined by contemporary increases in wages, past increases in prices (including smoothed energy prices, d4oilz), contemporary changes in total factor productivity (dtfp), and the capacity utilization rate (util) and its first difference (dutil). The price equation can be interpreted as a partial mark-up model with flexible prices, i.e., one in which prices respond to increases in individual cost components, but not necessarily one-to-one. The log-linear specification relates the log of wage increases (ldw4) to the log of CPI increases (ldpc4). As shown in Table 2, coefficients are not significantly different in specifications 1 and 2, except for the impact of technical change, which is smaller in the first equation, since part of it seems to be picked up by changes in trend oil prices. 1/ The importance of changes in total factor productivity is smaller in the third specification. For all specifications the explained variance ratio (R2) of wage increases is almost equivalent, while the part of price changes explained by the second equation is higher in the third (non-linear) specification.
Wage and Price Equations for France 1/
(1973-1994)
Figures in parentheses are standard errors.
Coefficient multiplied by 100.
Wage and Price Equations for France 1/
(1973-1994)
| Wages | Prices | ||||||
|---|---|---|---|---|---|---|---|
| spec1 | spec2 | spec3 | spec1 | spec2 | spec3 | ||
| dw4 | dw4 | 1dw4 | dp | dp | 1dpc4 | ||
| constant | 5.44 | 5.39 | 0.35 | constant | 0.29 | 0.73 | 0.07 |
| (0.47) | (0.48) | (0.05) | (0.25) | (0.19) | (0.11) | ||
| dw4(-4) | 0.17 | 0.17 | dw | 1.04 | 1.00 | ||
| (0.07) | (0.07) | (0.09) | (0.09) | ||||
| 1dw4(-4) | 0.47 | d1w4 | 0.46 | ||||
| (0.08) | (0.12) | ||||||
| dpc4(-1) | 0.83 | 0.83 | dp(-1) | -0.04 | 0.00 | ||
| (0.07) | (0.07) | (0.09) | (0.09) | ||||
| 1dpc4 | 0.53 | 1dpc4(-1) | 0.54 | ||||
| (0.08) | (0.12) | ||||||
| unprate | -0.49 | -0.49 | -0.03 | doilrz4 | -1.33 | 0.61 | |
| (0.05) | (0.05) | (0.01) | (0.52) | (0.20) | |||
| dunprate4 | -0.76 | -0.77 | -0.15 | dtfp | -0.46 | -0.75 | -0.15 |
| (0.24) | (0.24) | (0.03) | (0.17) | (0.13) | (0.08) | ||
| util | -0.06 | -0.04 | 1.41 2/ | ||||
| (0.03) | (0.03) | (0.85) | |||||
| dutil | 0.03 | 0.02 | |||||
| (0.03) | (0.03) | ||||||
| R2 | 0.945 | 0.942 | 0.943 | R2 | 0.853 | 0.858 | 0.952 |
Figures in parentheses are standard errors.
Coefficient multiplied by 100.
Wage and Price Equations for France 1/
(1973-1994)
| Wages | Prices | ||||||
|---|---|---|---|---|---|---|---|
| spec1 | spec2 | spec3 | spec1 | spec2 | spec3 | ||
| dw4 | dw4 | 1dw4 | dp | dp | 1dpc4 | ||
| constant | 5.44 | 5.39 | 0.35 | constant | 0.29 | 0.73 | 0.07 |
| (0.47) | (0.48) | (0.05) | (0.25) | (0.19) | (0.11) | ||
| dw4(-4) | 0.17 | 0.17 | dw | 1.04 | 1.00 | ||
| (0.07) | (0.07) | (0.09) | (0.09) | ||||
| 1dw4(-4) | 0.47 | d1w4 | 0.46 | ||||
| (0.08) | (0.12) | ||||||
| dpc4(-1) | 0.83 | 0.83 | dp(-1) | -0.04 | 0.00 | ||
| (0.07) | (0.07) | (0.09) | (0.09) | ||||
| 1dpc4 | 0.53 | 1dpc4(-1) | 0.54 | ||||
| (0.08) | (0.12) | ||||||
| unprate | -0.49 | -0.49 | -0.03 | doilrz4 | -1.33 | 0.61 | |
| (0.05) | (0.05) | (0.01) | (0.52) | (0.20) | |||
| dunprate4 | -0.76 | -0.77 | -0.15 | dtfp | -0.46 | -0.75 | -0.15 |
| (0.24) | (0.24) | (0.03) | (0.17) | (0.13) | (0.08) | ||
| util | -0.06 | -0.04 | 1.41 2/ | ||||
| (0.03) | (0.03) | (0.85) | |||||
| dutil | 0.03 | 0.02 | |||||
| (0.03) | (0.03) | ||||||
| R2 | 0.945 | 0.942 | 0.943 | R2 | 0.853 | 0.858 | 0.952 |
Figures in parentheses are standard errors.
Coefficient multiplied by 100.
The NAIRU is obtained from the above wage and price equations by assuming that, at potential, wages and prices increase at constant rates (i.e., dp=dp(-1)=dpc4/4, dw=dw(-1)=1/4*dw4) and that exogenous variables are at their steady state (See Setterfield et. al, 1992). Under these hypotheses, it is estimated that in 1994, the NAIRU corresponded to 9.3 and 9.8 percent of the labor force in specifications 1 and 2, respectively (in both cases with a standard deviation equal to 0.6 percentage point). The third specification yields a NAIRU of 9.1 percent, also with a standard deviation of 0.6 percentage point. The NAIRU used in the computation of the output gap was that obtained from the first specification, which lies between the other two in recent years.
APPENDIX II: The Evolution of the Share of Wages in Value Added in France
Chart 5 presents the share of wages in value added since 1970 for the so-called Large Public Enterprises (Grandes Entreprises Nationales, GEN) and other companies shown separately. 1/ It can be seen that the share of wages in total value added has steadily decreased in the case of the GENs, in contrast with the rest of the economy, where it is not significantly smaller today than it was 25 years ago. It may be surprising that the share of wages decreased less in the competitive than in the monopolistic sector, since labor is particularly well organized in the latter. However, because prices in this sector are regulated, the information they convey has to be analyzed with care, as discussed below.
The main reasons explaining the fall of the share of wages in value added of the GENs are a reduction in subsidies and the large investment program of the late 1970s, 2/ which made the GENs more capital intensive. A policy of lower subsidies and higher energy prices would produce larger operating surpluses (excédents bruts d’exploitation) in the accounts of companies, implying that in the case of the power company--for the same labor costs per kilowatt--it would appear that the “share of capital” has increased. More generally, while previously the government provided subsidies (in particular to cover capital cost), in recent years output prices were adjusted towards fully covering labor and capital costs. Thus, the share of wages was compressed. To a smaller extent, the increase in real interest rates since 1980 has had a similar effect on the share of wages in the value added by the private sector.
APPENDIX III: Estimates of the Output Gap Using the Hodrick-Prescott Filter
The Hodrick-Prescott filter is a procedure to smooth time series. Following the intuition that potential output should not fluctuate very much in the short-run, it has become increasingly popular to estimate the output gap based on the ratio between actual GDP and its smoothed value. In the Hodrick-Prescott filter, the following sum is minimized:
Where τt is the smoothed value (“potential”) and the weight λ is by convention set at 100 when annual data are used, and at 1600 when quarterly data are used. Although these weights have interesting properties for the GDP series in the United States, there is no compelling theoretical reason to set them at these values. In fact, for most countries, the use of these weights for quarterly and annual data leads to quite divergent results for the respective frequencies. Often quarterly data yield a smoothed curve closer to the original curve than annual data (i.e., smaller gaps). In addition, because the filter is a moving average, the last elements of the filtered series are significantly affected by the last observations of the original series. In the case of France, both problems arise: (i) the gap derived from quarterly data is, on average, smaller than that from annual data, and (ii) the measure of the 1993 gap obtained using annual data for 1970-93, or using data including projections up to 2000, 1/ is 1/2 percentage point larger than that obtained using annual data for 1970-95 (Chart 14). Using data for 1970-93 only, the slowdown in 1991-93 has only a limited effect on the smoothed series, and thus the gap between smoothed and actual GDP is large. When data for the period up to 1995 are included, the level effect of low GDP in 1994-95 lowers the smoothed series, and the output gap is correspondingly smaller. By contrast, when projections up to 2000 are used, the smoothed series fluctuates less in the low output years resulting in a large gap in 1993. This elementary exercise shows that caution has to be used when interpreting the apparently “neutral” estimates of the output gap yielded by the Hodrick-Prescott filter.
FRANCE: Output Gaps Derived from Hodrick-Prescott Filter
(In Percent of Potential Output)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.1/ Based on WEO projections of real GDP growth rates for 1996-2000.FRANCE: Output Gaps Derived from Hodrick-Prescott Filter
(In Percent of Potential Output)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.1/ Based on WEO projections of real GDP growth rates for 1996-2000.FRANCE: Output Gaps Derived from Hodrick-Prescott Filter
(In Percent of Potential Output)
Citation: IMF Staff Country Reports 1995, 141; 10.5089/9781451813470.002.A003
Source: Staff calculations.1/ Based on WEO projections of real GDP growth rates for 1996-2000.References
Baraldi, L., and B. Lamotte, “La négociation des salaires en France: une perspective institutionaliste,” Revue Francaise d’Economie, Vol IX, n.4 (1994).
Barrat, O., T. Coutrot, and S. Mabile, “Structures et tendances de la négociation salariale en France,” Dares, Ministere du Travail, mimeo (1995).
Confais, E., and P.-A. Muet, “Les rigidités du marché du travail,” in Le Chômage Persistant en Europe, OFCE, Paris (1994).
Dares, “Duráe du Travail et Conjoncture,” n.82. Ministère du Travail, Paris (1995).
Jackman, R. and C. Leroy, “The Nairu in France: main determinants of its evolution over time,” Banque de France, mimeo (1995).
INSEE, Comptes et Indicateurs Economiques, INSEE Résultats, Paris (1993).
Keese, M., G. Salou, and P. Richardson, “The Measurement of Output and Factors of Production for the Business Sector in OECD Countries,” Working Paper No. 99, OECD, Paris (1991).
Lecoupeur, Y., “La duree d’utilisation des équipements dans l’industrie en 1994,” Bulletin de la Banque de France, n.14 (February 1995).
OECD, “Estimating Potential Output,” mimeo, Paris (1994).
Rose, D. and R. Ford, “Estimates of the NAIRU Using An Extended Okun’s Law,” Bank of Canada, Working Paper, 89-3 (1989).
Setterfield, M., D. Gordon and L. Osberg, “Searching for a Will o’ the Wisp: An Empirical Study of the NAIRU in Canada,” European Economic Review, Vol. 36 (1992).
All references to “Germany” hereafter refer to western Germany.
In view of the importance of nonresident workers in Germany, the labor input for Germany was adjusted for the contribution of commuters:
L = ((Pwa * PR) * (1 - U)) + C) * H
where C stands for the number of inward commuters, and the working age population and the corresponding participation rate refer to the resident labor force.
The capacity utilization variable captures changes in the intensity of the use of equipment as well as changes in the organization of work. For instance, surveys published by the Bank of France (Lecoupeur, 1995) indicate that French managers considered that in 1993 they could have increased production by some 8 percent without hiring more people, or even using machines for longer hours.
The implicit average service lives used by INSEE were 33 years in the 1970s and 32 years in the 1980s for structures, and 12 and 13 years in the 1970s and 1980s, respectively, for plant and equipment (OECD, 1991). Although some French economists have expressed the view that these current service lives appear to be too long, thus leading to an overvaluation of the capital stock, no revisions have been made to the INSEE data. Shorter service lives would increase the sensitivity of potential output to fluctuations in investment.
Data on population and labor force are drawn from INSEE and OECD. In contrast with the convention adopted by the OECD and in this paper, INSEE often expresses the participation rate as the ratio of the labor force to the population aged 15 or more.
In addition to the varying response of the labor force to fluctuations in economic activity, the participation rate is affected by demographic changes. New cohorts affect the participation rate by increasing the denominator by more than the numerator (due to lower participation among the young). This factor explains part of the fluctuation in the early 1990s. For instance, part of the increase in the participation rate in 1991 was due to the lower birth rate in 1975-76.
High frequency data come from the unemployment agency--ANPE. In mid-1994 the agency was directed to change the classification of job-seekers to conform with ILO standards. This change has been implemented since early 1995. The INSEE survey has historically conformed with ILO standards. Employment figures have also been revised recently. According to the latest revisions employment growth in the business sector in 1994 was 1.2 percent instead of the initially estimated rate of 1.5 percent.
A modest increase would be consistent with results from regressing changes in the labor force on various lags of changes in employment.
See Appendix I.
Jackman and Leroy, 1995, state that the NAIRU is around 10 percent.
The Beveridge curve is a graphical presentation of the inverse relationship between unemployment and vacancies. The “natural” rate of unemployment is defined as the locus where the curve becomes nearly vertical, i.e. large increases in vacancies are associated with only small reductions in the unemployment rate. As the curve is constructed from time series of vacancies and unemployment, it can have more than one place where it tends to be vertical. Such multiple peaks are interpreted as shifts in the natural rate of unemployment.
The coefficient has a standard error of 0.05. Lagged capacity utilization did not improve the regression in a significant way.
Short-term swings in “cyclical” productivity can be interpreted either as supply shocks (as in real business cycle models) or as responses to transitory demand changes. In the latter case, changes in demand are absorbed by firms through labor hoarding and other organizational changes. Changes in the capacity utilization variable are only an imperfect proxy for such adjustments.
Contributions are computed by multiplying the growth rate of inputs by the elasticity of output with respect to this input.
Working age population is taken to be the population aged between 15 and 65. It is available only through 1992; estimates for 1993-94 were made by interpolation using the Federal Statistical Office’s population projections of July 1994.
By contrast, trend working hours estimated using a HP filter over the period 1969-94 showed a continued decline at a rate of 0.7 percent a year in recent years, and suggested that working hours in 1994 were above their trend level.
The average labor share over the sample period is 0.72, which was rounded down to 0.7 in order to give somewhat more weight to more recent, lower observations.
This result was found using the historical average of capacity utilization. Averaging capacity utilization over peak-to-peak or trough-to-trough subperiods gave similar results. The standard deviation of the coefficient estimated is 0.05.
This implies a probability of two-thirds that, under the assumptions stated, the true NAIRU is within plus or minus 1/2 percentage point from its point estimate.
The positive gap in 1981-82 resulted from stimulative demand policies in 1981, and the reduction in potential labor input in 1982 following the decrease in the minimum retirement age and working hours.
Overall, the inflation rate was one fifth lower in 1994 than in 1993, although this decline implied little change in absolute terms.
Intuition suggests that when inflation is at, say, 15 percent, a 1 percentage point increase in unemployment may have an impact on the price index that is different than when inflation is more or less stable around 2 percent.
In addition to the traditional Phillips curve method, cointegration analysis was used, producing mixed results. The interest in the latter technique stems from the fact that increases in real wages (net of productivity growth), the unemployment rate, and smoothed changes in oil prices in France seem to be well described by trendless random walks. In this case, a measure of the NAIRU can be obtained from the linear combination of those variables yielding stationary residuals (the so-called cointegration vector). Assuming that real wage increases equal productivity growth, and that oil price remain constant, the unemployment rate balancing the cointegrated vector can be viewed as a measure of the NAIRU. A cointegrated vector using data for the period since the 1983 devaluation of the franc was estimated, yielding:
unprate -3.752*dwhpr4-0.542*d4oilz-0.097 = 0
where unprate is the unemployment rate, dwhpr4 is the year-on-year change in wages (net of productivity growth) deflated by the GDP deflator and d4oilz is the year-on-year change in smoothed oil prices deflated by the GDP deflator. In this case, the NAIRU would be 9.7 percent. Over a longer period, however, coefficients changed signs, making interpretations difficult.
Two “unemployment benefit” variables were tested. The value of the basic benefit paid by UNEDIC and the average value of total unemployment benefits. Both series show benefits decreasing in real value in the 1980s. The first series has a discontinuity in 1979, when the payment of benefits was centralized. Since there is no comprehensive measure of the value of benefits before the 1979 reform, any comparison between protection levels in the 1970s and today has to be viewed cautiously. The second series is distorted by the fact that benefits are often proportional to wages, and in the 1980s people with long careers became unemployed, increasing the average benefit. In addition, the second series does not fully capture the effect of recent reforms that have shifted some of the unemployed to long-term minimum income programs, or the effect on labor supply of comprehensive welfare support for housing, health and other basic needs.
The overall decrease in unionization masks the relative strength of unions in the public sector and state enterprises, which influence wage negotiations in the private sector. In addition, because agreements set by unions and employers are often made mandatory for all workers (unionized or not) in a given sector, workers do not need to join unions in order to benefit from these agreements (see Baraldi and Lamotte, 1994, and Barrat et. al., 1995).
The systems are estimated using three stage least squares. Standard errors are computed from the quadratic form of analytic first derivatives.
The GENs comprise the electricity, telecommunication, gas, railways, metro and coal companies; which are monopoly companies in their corresponding sectors.
For example, the construction of a large number of nuclear power stations, required by the policy of “all-electrical” power adopted by the government following the first oil shock.
Under the assumption of GDP growth of 2 3/4 percent a year for 1996-2000.
