Wage Subsidies and Employment: An Analysis of the French Experience1

Until recently, industrial nations have followed the Keynesian prescription of applying general expansionary fiscal and monetary policies as the principal means of fighting unemployment. During the 1970s, however, faced simultaneously with soaring unemployment and double-digit inflation, governments became disenchanted with traditional demand management tools and began resorting increasingly to an arsenal of selective measures related to factor use.2 The underlying rationale is that the latter are likely to provide an immediate stimulus to employment, with less risk of overheating the economy. Prominent among such measures are various fiscal incentives to the private sector, including wage subsidies.3


Until recently, industrial nations have followed the Keynesian prescription of applying general expansionary fiscal and monetary policies as the principal means of fighting unemployment. During the 1970s, however, faced simultaneously with soaring unemployment and double-digit inflation, governments became disenchanted with traditional demand management tools and began resorting increasingly to an arsenal of selective measures related to factor use.2 The underlying rationale is that the latter are likely to provide an immediate stimulus to employment, with less risk of overheating the economy. Prominent among such measures are various fiscal incentives to the private sector, including wage subsidies.3

Until recently, industrial nations have followed the Keynesian prescription of applying general expansionary fiscal and monetary policies as the principal means of fighting unemployment. During the 1970s, however, faced simultaneously with soaring unemployment and double-digit inflation, governments became disenchanted with traditional demand management tools and began resorting increasingly to an arsenal of selective measures related to factor use.2 The underlying rationale is that the latter are likely to provide an immediate stimulus to employment, with less risk of overheating the economy. Prominent among such measures are various fiscal incentives to the private sector, including wage subsidies.3

A marginal wage subsidy is ordinarily provided to employers to offset a portion of wage payments to workers who otherwise would be unemployed; in contrast, a general wage subsidy is paid to employers to reduce the wage bill for their entire work force.4 The subsidy is either categorical—if it is limited to employers or employees with particular industrial, regional, demographic, or other characteristics—or universal, if applied regardless of such characteristics. Although in the past various types of categorical subsidies were deployed to alleviate regional or structural unemployment (or to provide income support to disadvantaged groups),5 lately they have been instituted for the most part in marginal form for a limited time period to combat cyclical unemployment.

After a brief review of temporary wage subsidies implemented in major industrial countries and a discussion of wage subsidies in general, the paper focuses on France’s prime d’incitation à la création d’emploi (Incentive Bonus for Job Creation). A labor input model is specified and estimated using quarterly data (disaggregated by major industrial branch) for that country’s nonfinancial nonfarm sector. The parameter estimates are used to forecast levels of employment for comparison with actual levels beyond the estimation period, in order to assess the subsidy’s effectiveness. The paper concludes with a discussion of the lessons derived from the overall analysis of wage subsidies and the specific evaluation of the French measure.

I. Temporary Wage Subsidies

Over the last four years, the industrialized world has experienced rates of unemployment that are unprecedented in postwar history. Besides offering broad therapy through demand management and symptomatic relief through stepped-up unemployment compensation, several of the hardest-hit countries (including some less developed ones) sought to provide a more effective palliative with temporary wage subsidies and other selective short-term measures. The Federal Republic of Germany, France, Ireland, Jamaica, Japan, the Netherlands, New Zealand, Spain, Sweden, the United Kingdom, and the United States6 have implemented temporary marginal wage subsidies, often in combination with categorical features and, in some cases, with on-the-job training requirements. In addition, Italy and the Netherlands introduced temporary general wage subsidies by way of social security tax reductions.7 Most marginal wage subsidies take the form of cash grants to eligible employers, except the recent American and Spanish income tax credits and the French social security tax exemption. Also, unlike other countries’ marginal subsidies, which are applied to workers hired by an employer or to an employer’s wage bill above a prescribed base level, in Japan, Sweden, and the United Kingdom the subsidies are disbursed with respect to workers declared redundant by the employer yet retained in his work force.

A detailed outline of temporary marginal wage subsidies (excluding those that contain training requirements) enacted between 1974 and 1977 in the Federal Republic of Germany, France, the United Kingdom, and the United States is given in Table 1. Their principal characteristics relate to the nature of the disbursement (method, amount, and duration) and the target population (eligible employers and employees).

Table 1.

Temporary Wage Subsidies in Major Industrial Countries1

article image
article image
article image

Excluding training-related subsidies.

Décret No. 75-436 instituant à titre exceptionnel une prime d’incitation à la création d’emploi, of June 4, 1975, amended September 19, 1975, February 3, 1976, March 31, 1976, and January 26, 1977.

Employees hired under Employment and Training Contract subsidies are not eligible for the Incentive Bonus.

Décret No. 77-713 and loi No. 77-704 portant diverses mesures en faveur de l’emploi des jeunes et completant la loi No. 75-574 tendant à la generalisation de la securité sociale of July 5, 1977.

Richtlinien zur Gewährung von besonderen arbeitsmarktpolitischen Beschäftigungshilfen of December 16, 1974.

A Mobility Subsidy, provided to unemployed persons (eligible for the Wage Cost Subsidy) who move to another municipality for employment, is available in addition to the Wage Cost Subsidy.

Under authority of Employment and Training Act (c. 50, sec. 5 (1)) of July 25, 1973, and Employment Protection Act (c. 71, schedule 14) of November 12, 1975, subsidy was introduced on August 18, 1975; and amended on September 24, 1975; December 17, 1975; February 12, 1976; April 6, 1976; and March 30, 1977.

Between August 18 and September 24, 1975, eligibility was restricted to assisted areas.

Tax Reduction and Simplification Act (P.L. 95-30) of May 23, 1977.

Concerning the nature of the payment, these subsidies are subject to income tax either directly (in the case of cash grants) or indirectly (Both the American income tax credit and the French social security tax exemption reduce the employer’s wage deduction pari passu.), thereby cutting substantially the net value of the disbursement in the hands of the employer.8 Among the various methods, the payroll tax reduction or exemption is the most desirable; it is superior to the income tax credit because it does not disqualify employers who incur losses from receiving the subsidy, and it is both more visible to potential users and easier to administer (through the existing tax enforcement mechanism, obviating the establishment of a separate administrative body) than a system of cash grants.

Whereas France’s Incentive Bonus and Britain’s Temporary Employment Subsidy are specified as a flat amount per full-time worker over a given period of employment, the other subsidies are defined as a percentage of wage payments. Thus, the former are biased in favor of low-wage occupational categories, while the latter are neutral across occupations (although the U.S. New Jobs Tax Credit is biased in favor of low-wage skills and small enterprises as a result of the ceilings on the amount of the credit per employer and per worker). However, the actual choice of a base to calculate the subsidy (in terms of worker, time worked, or percentage of wages) depends not only on the desired patterns of income distribution and resource allocation but also on administrative considerations (such as the availability of information on the employer’s work force in compliance with reporting requirements—for example, for social security records).

In relation to the gross hourly compensation per eligible employee, the subsidies (spread over a full year) confer benefits ranging from 7 per cent of the wage cost to French employers (recipients of the Incentive Bonus) to 32 per cent of the wage cost to British employers, as shown in Table 2.9 The maximum duration of the payments to eligible firms varies between two and eight quarters. It is difficult to judge a priori the adequacy of the magnitude and duration of the subsidy flow. More generally, however, subsidies are to be set on the basis of existing labor market conditions in a particular country or location. In other words, the appropriate size and length of payments depend on the unemployment rate at the going wage rate (determined in part by minimum wage legislation and collective bargaining), as well as on the firing and hiring costs and the resulting inertia of employers in laying off or hiring workers (that is, their propensities to hoard and dishoard labor) in response to cyclical variations in output demand.

Table 2.

Relation of Subsidy to Hourly Compensation in Manufacturing

article image
Sources: U. S. Department of Labor, Bureau of Labor Statistics, and Table 1.

Estimated 1976 average compensation per hour worked. Includes supplementary benefits (family allowances, payments in kind, severance pay, and other bonuses), payroll taxes, and recruiting and training costs.

Estimated maximum subsidy per hour worked, on an annual basis, assuming full-time 40-hour week employment.

Amount is twice as large if computed over the actual duration of the subsidy—that is, on a semiannual basis.

Subsidy can be extended up to two additional quarters at one half of the normal rate.

A critical element in the success of a wage subsidy, expressed in simplest terms by the government’s net revenue loss per job created, is the definition of the target population. Ideally, to maximize the wage subsidy’s beneficial impact thus measured, the subsidy must be applied solely to the employment of (1) workers who would not be hired in the absence of the subsidy, and (2) workers who would be fired in the absence thereof—each criterion being necessary in a different phase of the business cycle. Obversely, under criterion (1), employers should be prevented from receiving the subsidy for the following purposes: (A) to substitute subsidized workers for their existing work force, (B) to hire workers who would be hired anyway, (C) to hire workers who are already employed elsewhere in the economy; and, under criterion (2), they should be prevented from receiving the subsidy (D) to retain workers whom they would retain in their work force anyway.

To the extent that a subsidy satisfies only criterion (1) or (2), it is not likely to promote incremental employment during a decline or a recovery in economic activity, respectively. By the same token, a subsidy devoid of safeguards against its use for purposes (A) through (D) results in a windfall to employers and a loss to the rest of society. Interestingly, a general subsidy can fulfill both criteria, although without including any safeguards, whereas a marginal subsidy usually can be aimed at one criterion alone and yet contain reasonable safeguards. In sum, it does not seem feasible to design a perfect subsidy, consistent with both criteria and embodying all safeguards.

The inherent asymmetry of marginal wage subsidies, and the partial remedy that they may provide for unemployment, are illustrated by the measures under scrutiny. On the one hand, schemes introduced in the Federal Republic of Germany, France, and the United States (based on the optimistic presumption that the recession has come to an end) are intended to encourage additions to employment, in accordance with criterion (1)—although during a mild downturn in economic activity such subsidies may conceivably meet criterion (2) as well.10 These subsidies seek to avoid windfalls from (A) by establishing a base level of employment (set at the level prevailing immediately prior to the promulgation of the subsidy)11 that the employer must exceed in order to receive the subsidy. In addition, both French schemes and the American tax credit prescribe a variable base level (variable either automatically on an annual basis or, in the case of the Incentive Bonus, on an ad hoc basis) that adjusts upward in the event of a recovery, to prevent use of the subsidy for (B). An added advantage of the variable base level is that, through a downward adjustment, it can trigger the subsidy even before reaching the end of a recession. Further, as a safeguard against (C), both the Incentive Bonus, in its initial form, and the Wage Cost Subsidy require a worker to have been continuously unemployed for a specified period (three and six months, respectively) prior to being eligible for the subsidy.12

On the other hand, the British Temporary Employment Subsidy (based on the pessimistic presumption of a deepening recession) attempts to discourage diminutions in employment, in line with criterion (2). Yet, unlike the other subsidy programs patterned in conformity with criterion (1) that apply safeguards through objective tests of eligibility, the Temporary Employment Subsidy leaves the screening of recipients almost entirely to the discretion of interested employers, labor unions, and program administrators, in order to avoid (D). This discretionary element raises serious questions as to the effectiveness of the British scheme; more generally, the lack of a viable alternative13 casts doubt on any attempt to curtail with a wage subsidy an upsurge in unemployment caused by a sharp decline in aggregate demand.

Therefore, whereas it seems necessary for practical reasons to reject temporary marginal wage subsidies directed at discouraging a reduction in employment, it is clearly desirable that subsidies aimed at raising employment incorporate certain basic features—namely, a variable base level of employment and a minimum waiting period—in their definition of the target population. However, just as with the amount and duration of the subsidy, it is difficult to determine a priori the appropriate variation of the base level and the optimal length of the waiting period (or additional categorical requirements), as these are to be calibrated to prevailing labor market conditions, and thus may require considerable fine-tuning by policymakers.

Of the wage subsidies examined, France’s Incentive Bonus for Job Creation emerges as the logical candidate for ex post empirical evaluation because of its near-universal application to the unemployed (from June through November 1975) and the availability of quantitative data for the relevant period. None of the other measures are amenable to a comparable investigation. The coverage of these subsidies is either heavily constrained by special categorical tests or subject to discretionary interpretation (as the British grant), which is almost impossible to take into account adequately in a quantitative measurement of a subsidy’s effect. Thus, the remainder to the paper is devoted primarily to an analysis of the impact of the Incentive Bonus on the level of French employment. But as a prelude to that analysis, it is necessary to discuss in general terms the effects of a wage subsidy.

II. Framework for Analysis

1. economic effects of a wage subsidy

The economic consequences of a wage subsidy can be grouped under four types of effects. Two of them, the factor substitution and scale effects, are first-order microeconomic effects experienced directly by the subsidized firm. The factor substitution effect indicates the rise in employment and the decline in the use of other factor inputs in response to a reduction in wage cost by the amount of the subsidy, which is contingent on the wage elasticity of demand for labor (as determined by the elasticity of substitution between labor and other factors) and on the wage elasticity of supply of labor. The higher the absolute value of the elasticities, the larger will be the increase in employment (and the larger will be the decline in the use of other inputs); the lower it is, the larger will be the rise in the wage rate.

The scale effect consists of the rise in employment (and in the use of other factors) that accompanies the expansion in output brought about by the reduced wage cost. To the extent that the firm shifts the subsidy to the consumer in the form of lower product prices, there will be a growth in output demand; this, in turn, will engender increased derived demand for all inputs, including labor. Thus, the scale effect depends on the price elasticity of demand for the product, and the elasticity of demand for inputs with respect to output (i.e., the inverse of the degree of homogeneity of the production function), constrained by the supply elasticities of factor inputs. Again, the higher the absolute value of the elasticities, the larger will be the increment in factor use, including employment; the lower it is, the larger will be the pressure on factor prices.

In addition to these direct effects, indirect repercussions are felt throughout the economy—interindustry effects and multiplier effects. Interindustry effects measure changes in relative prices and in quantities of products and factor inputs in different industries (including feedbacks on those of the subsidized firm or industry) resulting from the initial substitution and scale effects. The magnitude and direction of these repercussions depend on the interindustry relationships (i.e., the degree of complementarity and substitutability among products and among factors, and their relative proportions in consumption and production) within, and the openness of, the economy. Thus, in contrast to the direct substitution and scale effects, the interindustry effects on output and factor inputs and their relative prices cannot be predicted unambiguously without knowledge of the technological relationships in production and of taste patterns in consumption. With respect to labor, however, it can be said that such effects will tend to reinforce the increase in employment and wages (depending on the unemployment rate and the labor participation response) mentioned earlier.

While the above effects originate on the supply side, the multiplier effects originate on the demand side. The latter effects comprise the macroeconomic consequences of the subsidy through successive rounds of expansion in aggregate demand on the strength of a rise in: the disposable income of households (generated by the increment in employment and the wage rise owing to the subsidy, net of the foregone unemployment compensation received prior to the subsidy), the after-tax earnings of firms (to the extent that the subsidy is not shifted to workers in the form of higher wages and/or to consumers in the form of lower prices), and the government’s deficit (given by the subsidy payments less the automatic increase in tax receipts and reduction in unemployment compensation payments). The resulting multiplier effects on the level of output, employment, investment, and prices depend not only on the technological relationships, supply constraints, and consumer preferences outlined above but also on the marginal propensities to spend of the various sectors, as well as on the built-in elasticities of the tax and unemployment compensation systems. Although the effects on output, employment, investment, and factor prices should be positive with any set of realistic parameter values, the general price level may or may not decline, depending on the slack in capacity utilization and the means of financing the budgetary deficit.

Obviously, the cumulative impact of a wage subsidy is contingent on the characteristics discussed in the previous section. As against a general wage subsidy, which would be fully reflected in all the enumerated effects, a temporary marginal wage subsidy would be felt appreciably through only some of these effects. Specifically, assuming that the marginal subsidy is well designed, in the sense that it meets at least criterion (1) and contains reasonable safeguards against (A) through (C), it will lead (in the absence of a sharp drop in effective demand), through the substitution effect, to a rise in employment in the short run and (upon approaching full employment) to higher wages in the long run,14 with minor scale, interindustry, and multiplier effects.15 Otherwise, without appropriate safeguards, the subsidy will result chiefly in a windfall for subsidized firms and in a budgetary deficit with expansionary multiplier effects.

2. summary of the literature

In his seminal contribution on wage subsidies, Kaldor (1936) focused on their substitution and scale effects on employment and on their budgetary cost. In comparing marginal and general wage subsidies, he concluded that the former is a more appropriate tool to fight cyclical unemployment, though it might entail a larger net revenue cost to the government than the latter; he indicated that the best manner of implementing either subsidy would be through a reduction in payroll taxes. Since the early 1960s, a number of authors have discussed the use of wage subsidies as an alternative to other policy tools (such as tariff protection, output or capital subsidies) and as a cure for regional or structural unemployment in developed and developing countries.16 However, Weiner and others (1969) provided a rigorous theoretical treatment of the substitution and scale effects of a wage subsidy in a simple one-sector model, whereas Ahluwalia (1973) and Mieszkowski (1974) also investigated interindustry effects in two-sector models. The multiplier effects of wage subsidies were traced in the framework of full macroeconomic models17 of a closed economy by Fethke and Williamson (1976) and of an open economy under flexible exchange rates by Argy and Salop (1977).

On an empirical level, following the first quantitative estimate of a hypothetical cut in payroll taxes by Frisch (1949) in the context of an early model of the Norwegian economy, most simulations of the impact of wage subsidies have been ex ante measurements of their substitution effect on employment, income, and government expenditure, performed on U.S. data. Barth (1972) estimated the impact of a suggested universal subsidy paid to workers, assuming a wide range of wage elasticities of demand for, and supply of, labor. Biederman (1972) quantified the consequences of an employment-training income tax credit under a unitary elasticity of demand and an infinite elasticity of supply. Somewhat more realistically, Greenston and MacRae (1975 b) simulated the employment impact of the U.S. Work Incentive Tax Credit (a categorical subsidy applied to welfare recipients) on the basis of a full labor market model developed by Crandall and others (1975).18

More recently, Kesselman and others (1977) estimated elasticities of substitution among three factors of production from a translog cost function that, along with assumed infinitely elastic factor supplies, were used to calculate the rise in employment attributable to general and marginal wage subsidies proposed in the form of income tax credits. Further, Hamermesh (1976 and 1977) calculated both the substitution and scale effects of several general wage subsidy options on employment, using, first, typical parameter estimates based on a number of econometric studies of factor demand and production functions and, second, his own estimates of a quarterly labor demand function. Also incorporating parameter values from other studies in their economy-wide model, Fethke and Williamson (1976 and 1977) measured the multiplier effects of a hypothetical tax credit with a variable base level.

Among these works, only the most recent ones contain results that may be directly relevant to our inquiry. Specifically, Kesselman and others (1977) found that a subsidy that lowered the marginal cost of labor by 1 per cent would lead to an increase of about 0.2 per cent in the demand for blue-collar labor, a decline of nearly 0.1 per cent in the demand for white-collar labor, and a drop of 0.5 per cent in the demand for capital—assuming that the level of output was kept constant. This outcome is explained by the complementarity between white-collar labor and capital, and by the substitutability between blue-collar labor and any other factor input. For his policy simulations, Hamermesh (1976 and 1977) adopted a four-quarter wage elasticity of employment (reflecting substitution and scale effects) ranging between -0.1 and -0.6. According to the revised Fethke and Williamson (1977) model,19 a 1 per cent marginal wage subsidy would produce a 0.3 per cent rise in employment, a 0.2 per cent drop in the price level, and a 0.2 per cent increase in real wages (upon full adjustment to the subsidy). Interestingly, the impact of a 1 per cent general wage subsidy would depend on the method of financing it: there would be a 0.2 per cent increase in employment, a 0.6 per cent reduction in prices, and a 0.4 per cent boost in real wages, if the subsidy were accompanied by an equivalent reduction in other government expenditures; or, alternatively, there would be a 0.4 per cent rise in employment, a 0.8 per cent increase in prices, and a 0.2 per cent decline in real wages, if there were no offsetting discretionary changes in taxation or expenditures.

III. The French Experience

1. determinants of employment behavior

In order to ascertain the employment effect of a wage subsidy, our initial task is to identify the overall determination of employment behavior. Much of the literature on the determinants of employment is predicated on the relationship between the demand for labor services by profit-maximizing firms and the supply of such services by utility-maximizing households.20 Notwithstanding the merits of a fully specified labor market (and of allowing for interactions with the goods market and the money market), especially for capturing long-run behavior, the model formulated here is a partial-equilibrium short-run one with emphasis on labor demand.21 This approach is justified by the particular policy orientation of our investigation, on the one hand, and is dictated by data limitations, on the other.

For the aggregate of firms in a given industry, facing perfectly competitive markets, profit maximization leads to the well-known marginal productivity condition for labor input:22


where Q denotes net output originating in the industry, L denotes labor services, and w denotes the wage rate (all in real terms). Further, the marginal product of labor is contingent on the industry’s production function. Assuming constant returns to scale, constant elasticity of substitution (between capital and labor), and embodied technological change, the level of output is determined as follows:


where, besides the variables already defined, K denotes capital services, δ is a distribution parameter, γ and ϕ denote the respective rates of technological change embodied in labor and capital, σ denotes the elasticity of substitution, and t denotes time.

Differentiation of equation (2) with respect to labor, and substitution of the partial derivative into (1), yields


in equilibrium. By solving for L, equation (3) can be converted into the stochastic relationship


which indicates that the employment of labor input in period t depends on the expected levels (denoted by asterisks) of output and wages and of the error term η. There are several reasons why employment may not respond instantaneously and precisely to the current level of exogenous variables, the main ones being costs of hiring and firing employees and imperfect managerial knowledge. Hence, we may postulate a mechanism of expectations, in which the expected level of each variable is a function of the distributed lag of its preceding actual levels:


where αi ≥ 0 and βi ≥ 0 are weights subject to the condition that


Given equations (5) and (6), equation (4) can be expressed in logarithmic notation


Of major interest from the policy perspective is the definition of the real cost of labor services, as perceived by the firm. Ordinarily, the average real wage rate for the employer is defined as


where v denotes the money wage paid to the employee per unit of time, y the rate of payroll taxes (contributions to social security and other funds), and p the price of output. However, the net wage cost under a wage subsidy is reduced to 23


which contains s, the subsidy per unit of time—subject to the firm’s income tax. This formula can accommodate alternative types of wage subsidies as well. For example, in contrast to equation (9), which applies under the French Incentive Bonus, a subsidy computed as a percentage of the basic wage and exempted from income tax (or without reducing the wage deduction from the tax base) would result in


where u denotes the company income tax rate.

2. estimation

For empirical estimation, terms are rearranged in equation (7) to obtain each industry’s labor input function over quarterly observations t = 1, 2, …, n,


where it is hypothesized that the employment of labor input (normalized by the current level of output) adjusts within two quarters to changes in output,24 and within one year to changes in the wage rate.25 The actual effect of exogenous variables is measured by the λ coefficients, from which the structural parameters can be derived in the following manner:


The meaning of λ4, λ5, and λ6 depends on the value of j (i.e., j = 0, 1, 2, or 3) that fulfills the conditions underlying equations (5) and (6). For example, if j = 0 because of prompt adjustment, then


With slower adjustment, if j = 1, then


If j = 2, then


or if j = 3, then λ4, λ5, and λ6 vanish.26

The data used to estimate equation (11), compiled by the Institut National de la Statistique et des Etudes Economiques (INSEE),27 consist of seasonally adjusted quarterly series corresponding to France’s eight major nonfinancial nonfarm industrial branches from the first quarter of 1959 through the second quarter of 1975—that is, the period preceding the enactment of the Incentive Bonus. The equation was fitted for each industry on two alternative dependent variables: number of man-hours and number of employees. While the former is the appropriate definition of labor as a factor of production, the latter is the relevant measure for evaluating the effectiveness of the subsidy (the objective of which is to generate employment). Both sets of results are reported, but no attempt is made at a rigorous reconciliation between the two, with a separate explanation for hours worked per worker28—a task that lies beyond the scope of our analysis, given the lack of data on certain key variables.

Ordinary least-squares regression revealed the presence of higher-order autocorrelation that was subsequently corrected with the Cochrane-Orcutt approach.29 The results are presented in “unscrambled” form, so that they refer to the original equation, except for the Durbin-Watson statistic, which was computed on the residuals of the transformed equation. Further, each regression has been run stepwise; in successive rounds, terms were discarded that violated the aforesaid constraints on α and β or whose coefficients were smaller than their respective standard errors. As a final methodological point, two dummy variables that account for extraordinary historical events were included in the estimated equation.30 Since they are devoid of analytical meaning, they are not listed along with the other regression coefficients.

Tables 3 and 4 show excellent fits and satisfactory correction for autocorrelation for all industrial activities. The independent variables display the expected signs, except the time trend, which in one regression has a positive coefficient. However, Tables 5 and 6 are more informative about the role of each variable in explaining employment. On the whole, especially as summarized in the bottom line (namely, the industry-weighted average of each parameter at the end point of the sample) of each table, the structural parameter estimates provide ample support for the theoretical model.

Table 3.

Man-hours Equation: Regression Coefficients1

article image

The figures in parentheses are the standard errors of estimate of the relevant coefficients. The double asterisk (**) denotes statistical significance at the 1 per cent level. The single asterisk (*) denotes statistical significance at the 5 per cent level.

SEE¯ and R¯2 are, respectively, the standard error of estimate and the coefficient of determination, corrected for loss in degrees of freedom, and D-W is the Durbin-Watson statistic.

Table 4.

Employees Equation: Regression Coefficients1

article image

The figures in parentheses are the standard errors of estimate of the relevant coefficients. The double asterisk (**) denotes statistical significance at the 1 per cent level. The single asterisk (*) denotes statistical significance at the 5 per cent level.

SEE¯ and R¯2 are, respectively, the standard error of estimate and the coefficient of determination, corrected for loss in degrees of freedom, and D-W is the Durbin-Watson statistic.

Table 5.

Man-hours Equation: Structural Coefficients

article image

Long-run output elasticity is unity for all industries. Short-run elasticities α0 and α1 refer, respectively, to current and lagged changes in output.

Short-run elasticities σβ0, σβ1, σβ2, and σβ3 refer, respectively, to current and lagged (one. two, and three quarters) changes in wages.

Table 6.

Employees Equation: Structural Coefficients

article image

Long-run output elasticity is unity for all industries. Short-run elasticities α0 and α1 refer, respectively, to current and lagged changes in output.

Short-run elasticities σβ0, σβ1, σβ2, and σβ3 refer, respectively, to current and lagged (one, two, and three quarters) changes in wages.

The quarterly rate of labor-augmenting technological progress, quantified by the estimate of γ, is, on the average, 1 per cent with respect to man-hours and 0.6 per cent with respect to employees. The larger estimate in the man-hours equation may reflect government pressure exerted in the past to reduce the workweek, besides improvement in the quality of labor. Across industries, the rate of technological change ranges from a negative 0.4 per cent in trade to about a positive 2 per cent in fuels.

Estimates of the effect of output, measured by elasticities α0 and α1 indicate that over one half of the labor response to a change in output takes place in the same quarter and the remainder in the following quarter.31 There is some interindustry variation between the fastest response in intermediate goods to the slowest response in trade, though with no noteworthy difference in the adjustment speed between the man-hours and employees equations—contrary to the a priori view that firms react to a change in the demand for their product by varying first the number of hours worked, and then the number of employees in the work force.

The average long-run wage elasticity of employment, indicated by σ, is -0.32 for man-hours and -0.36 for employees. Although these values are in broad conformity with the substitution elasticities derived from similar models,32 they are in conflict with the expectation of a higher elasticity for man-hours than for workers. In the short run, over half of the response to a change in the wage rate occurs within two quarters, and it is fully accomplished within one year. For man-hours, the peak of the response is in the second quarter, and for employees in the third quarter. A plausible interpretation of this evidence is that a rise in wages induces firms to reduce hours of work initially and then to cut down on the number of employees; the path to equilibrium is completed by utilizing the remaining workers more intensively. A similar process should occur in reverse in the event of a decline in the wage rate.

To be sure, both the magnitude and the time profile of adjustment to wages in the nonfinancial nonfarm sector as a whole gloss over differences among industries. Labor input in food and kindred products, fuels and lubricants, consumer goods, construction and public works, and nonfinancial services seems to respond faster to a change in wages than does labor input in machinery and primary metals, intermediate goods, and trade, where the adjustment begins one or two quarters after the change and is completed by the third quarter. Besides the influence of particular institutional and technological characteristics of each industry, the dispersion of long-run wage elasticities33 could also have been caused by the high level of aggregation of heterogeneous activities in certain industries.

Another potential source of bias is the assumption of an exogenous wage rate and the absence of the cost of capital services. As pointed out earlier, the ideal approach is to estimate a multi-equation labor market model (including labor demand and supply functions, broken down by industry and by age-sex composition, respectively, and the corresponding wage equations); also, the influence of the rental cost of capital should be accounted for explicitly.34 However, data limitations do not permit estimation of such a comprehensive model for France.

3. simulation

The abundance of empirical research on employment behavior has not been matched by an equally vigorous effort at investigating the impact of wage subsidies on employment. The few works in this area (reviewed in the preceding section) consist of ex ante simulations of the effect of changes in the wage cost brought about by subsidies introduced or proposed in the United States, none of which is directly comparable to France’s Incentive Bonus.

In this section, an ex post simulation of the employment impact is presented for the period when the Bonus was made available with the broadest coverage (without restriction as to the size of the firm). Since the Bonus was aimed at increasing the number of employees, rather than the number of man-hours, the simulation is based on parameter estimates from the employees equation (11). Accordingly, the dependent variable in each industry was forecast for the third and fourth quarters of 1975, using the actual values of exogenous variables (including w, the gross wage cost) in the respective quarter; these we shall call presubsidy forecast values. An additional forecast, which yields postsubsidy forecast values, was performed for the same period, replacing the w variable by w′, which incorporates the subsidy—as defined in equations (8) and (9), respectively. The difference between the postsubsidy and the presubsidy simulations reflects the maximum theoretical impact of the subsidy.35

In order to ascertain the effectiveness of the Bonus in raising the level of employment, it is necessary to compare two measures: actual deviations between realized values and presubsidy forecast values, and theoretical deviations between postsubsidy and presubsidy forecast values of the dependent variable. A quick perusal of Table 7 reveals that in five industries, actual deviations are negative, contrary to the expected rise in employment measured by the corresponding theoretical deviations. Only fuels and lubricants, intermediate goods, and construction and public works show positive actual deviations, which exceed the predicted deviations for the first two industries. To identify the nature of actual deviations, which are in fact forecast errors, the appropriate RMS (root-mean-square of forecast) and F-statistics (relating the variance of the forecast error to the variance of the estimation error) were computed. The F-test confirms that employment in the forecast period is determined by the same structure as in the sample period in almost all industries, with the notable exception of intermediate goods and, to a lesser extent, of food and kindred products.

Table 7.

Deviations from Forecasts of the Number of Employees, Third and Fourth Quarters of 1975

article image

Difference between forecasts with postsubsidy and presubsidy wage rate. Figures marked with an asterisk (*) are three times as large (in absolute magnitude) as the corresponding standard error of estimate.

Difference between the actual value and the forecast with presubsidy wage rate. Figures marked with an asterisk (*) are three times as large (in absolute magnitude) as the corresponding standard error of estimate.

The forecast statistics refer to the actual deviations from forecast. RMS is the root-mean-square forecast error, and F is the ratio of the sum of squared error of forecast to the sum of squared error of estimate (divided by their respective degrees of freedom, shown in parentheses). The symbol # indicates statistical significance at the 1 per cent level.

Parameter estimates from equation (12) are used for fuels and lubricants, machinery and primary metals, and intermediate goods; and estimates from equation (11) are used for all other industries.

However, the question remains as to whether the actual boost in employment above the presubsidy forecast for three industries has indeed been caused by the Incentive Bonus or by the forecasting quality of our model. To test these alternatives, the employees equation was rerun without the wage terms, over the original sample period,


and the dependent variable was forecast with the new set of parameter estimates. For all industries, the least-squares fit of equation (12) is inferior to that of equation (11), yet the forecasting power improves substantially with equation (12) for intermediate goods, and, to a smaller degree, for fuels and lubricants, and machinery and primary metals—as indicated by a sharp decline in the size of the actual deviations and the resulting RMS values listed in Table 7.

This finding does not, of course, mean that wage rates have no influence on the level of employment in these industrial activities. Rather, it indicates that there may have been structural shifts36 in employment behavior that are not accounted for explicitly in our model, so that the wage elasticities in these industries drifted from their estimated values over the sample period, although probably without altering significantly the average elasticity for the nonfinancial nonfarm sector as a whole.

All in all, the forecasts suggest that the Bonus as implemented has had, at most, a mild influence in raising or maintaining the level of employment. This is illustrated in Table 7 by the actual deviations in the last quarter in 1975. Of a gross increase of 58,000 jobs (measured in terms of all actual deviations above presubsidy forecasts) in fuels and lubricants, intermediate goods, and construction and public works, only 27,000 jobs in construction and public works (which match the theoretical deviations) may have been induced by the subsidy—even though this rise has been swamped by declining employment (below presubsidy forecasts), producing a net negative actual deviation of more than 80,000 jobs for all activities. These figures fall substantially short of the potential impact of the subsidy, reflected in theoretical deviations totaling about 182,000 jobs, which may be slightly overstated because of possible bias in the underlying wage elasticities.37

Information on actual disbursements and number of subsidized employees, presented in Table 8, indicates that by the end of 1975 a total of almost 38,000 employees (one third in construction and public works, and the rest dispersed more or less evenly across other activities) had been hired under the Bonus—that is, one third less than the sum of positive actual deviations.38 Interestingly, however, the number of employees thus hired increased by nearly 15,000 in the first quarter of 1976 (as against less than 10,000 in each of the following two quarters), despite the statutory restriction of eligibility to small businesses after November 1975, indicating a possible administrative lag in processing applications for the Bonus.39

Table 8.

Incentive Bonus for Job Creation: Cumulative Payments and Employment, by End of Quarter, Third Quarter, 1975-First Quarter, 1977

article image
Source: France, Ministry of Labor.

As anticipated in the previous section, the simulation understates the employment impact of the Bonus in its original form (effective between June and November 1975), in that it measures only the direct partial-equilibrium substitution effect, while neglecting all other repercussions—which are expected to be of minor importance given the marginal nature of the subsidy. In sum, although the subsidy’s actual impact amounted to about one seventh of the potential employment effect, it can be viewed as a success in relation to its net financial cost, which appears to have been small.40

In any event, several explanations can be advanced for the relatively modest rise in employment (in comparison with the potential increment). First, apparently employers were not informed in time about the enactment of the subsidy.41 Second, despite the availability of the Bonus and the mild recovery in late 1975, employers were reluctant to hire workers because they had been forced to hoard labor under a legal requirement to seek government approval to lay off workers.42 Third, the relatively small size of the Bonus was probably insufficient to cover the hiring costs incurred by employers in most industries. Fourth, the six-month minimum waiting period as a registered unemployed person may have been too long to enable firms to qualify for the grant, since the bulk of the cyclically unemployed are likely to be idle for shorter periods. Finally, a lesser deterrent was that employees hired under Employment and Training Contracts were not eligible for the Incentive Bonus.43

IV. Lessons for Policy Formulation

The foregoing study of temporary marginal wage subsidies in general, and of France’s Incentive Bonus for Job Creation (as implemented during the second half of 1975) in particular, leads to a number of tentative policy conclusions. Some of them are based on administrative considerations, while others follow from economic analysis. Although wage subsidies must be tailored to prevailing conditions in the local labor market, a number of general guidelines can be formulated.

In practice, a marginal wage subsidy, aimed principally at reducing cyclical unemployment, should be provided to firms for making additions to their work forces rather than for retaining redundant employees. The subsidy, preferably in the form of a reduction in (or exemption from) existing payroll taxes and administered by the government agency entrusted with the collection of these taxes, should be applied to workers hired above the firm’s base level of employment, set at the inception of the program and variable thereafter, at intervals spaced either regularly or according to fluctuations in output demand over the business cycle.44 Eligible workers should be subject to a minimum unemployment period, probably not longer than three months, prior to being hired under the subsidy. The amount and the length of the subsidy must be sufficient to cover hiring costs and to permit the firm to employ the marginal worker until the recovery in economic activity is under way—that is, probably ranging between one and two years in duration.

Special categorical features may be incorporated in the definition of the target population, in an attempt to direct the subsidy to the hardest-hit groups of unemployed workers, identified by age, region, or skill. Thus, the subsidy may be made available without a waiting period to unemployed youth (up to a specified age limit) and to regions or occupations in which the unemployment rate exceeds (by a certain percentage) the national average. Also, it may be defined as an absolute amount per worker, rather than as a percentage of his wages, so as to favor low-income occupations.45 However, other categorical requirements, such as those that favor firms in particular size categories or industrial activities, may hinder the effectiveness of the subsidy as an employment generating device.

In view of these refinements and the underlying need of policymakers for reliable and prompt information about labor market conditions, a temporary marginal wage subsidy is an instrument to be used primarily in industrial countries. Even there it must be recognized, however, that such a subsidy is subject to certain limitations. If implemented adequately (that is, following the above guidelines), it can retard layoffs toward the end of a decline in economic activity and can accelerate hiring during a recovery, but it will have practically no impact during a deepening recession. In the latter situation, the subsidy would have to be introduced in conjunction with demand management measures.

A preferable instrument during a recession would be a temporary general wage subsidy, applied to the wages of marginal as well as inframarginal workers.46 Such a subsidy renders unnecessary the definition of the target population (and solves the attendant administrative difficulties), but, more importantly, it is also likely to be a more effective tool for the simultaneous reduction of unemployment and containment of inflation than traditional demand management policies. The danger of overheating the economy could arise, however, if the subsidy—ratified by the monetary authority—were too large or if it were kept in effect for too long. In either case, as the economy approached capacity, the subsidy would be shifted to labor in the form of higher wages, boosting aggregate demand and generating an upward pressure on the price level.


The labor input function (11) was estimated in the following operational forms:

ln(MH/VA) = λ0 + λ1TIME + λ2 Δln VA + λ3 ln(GCMH/PD)−j + λ4 Δln(GCMH/PD)−j + λ5 Δln(GCMH/PD)−j−1 + λ6 Δln(GCMH/PD)−j−2 + λ7D63I + λ8D68II


ln(E/VA) = λ0 + λ1TIME + λ2 Δln VA + λ3 ln(GCMH/PD)−j + λ4 Δln(GCMH/PD)−j + λ5 Δln(GCMH/PD)−j−1 + λ6 Δln(GCMH/PD)−j−2 + λ7D63I + λ8D68II

with j = 0, 1, 2, or 3 (See Section III.2.). Further, for estimation and prediction of presubsidy levels of employment, according to equation (8),


and for postsubsidy prediction, according to equation (9), GCMH is replaced by


The variables are defined as follows:

article image

All are measured with seasonally adjusted quarterly data compiled by the Institut National de la Statistique et des Etudes Economiques (with the exception of the trend, the dummy variables, and IBMH), for the period between the first quarter of 1959 and the fourth quarter of 1975; of these observations, those through the second quarter of 1975 were used for estimation, while the remaining two quarters were set aside for simulation. The underlying data base consists of annual totals spliced with quarterly sample information obtained from various published and unpublished sources. The data are disaggregated by major nonfinancial nonfarm industrial branch. There are eight branches (The number preceding each is the INSEE industry code for France’s quarterly national accounts):

article image
article image

In contrast to the other variables, PTR, IBMH, TIME, D63I, and D68II are uniform across all branches. For a description of the splicing technique and data sources used, and a preliminary tabulation of the INSEE series (through the end of 1974), see Laroque and others (1975).


  • Ahluwalia, Montek S., “Taxes, Subsidies, and Employment,” Quarterly Journal of Economics, Vol. 87 (August 1973), pp. 393-409.

  • Argy, Victor, and Joanne Salop, “Price and Output Effects of Monetary and Fiscal Policy Under Flexible Exchange Rates” (unpublished, International Monetary Fund, September 7, 1977).

    • Search Google Scholar
    • Export Citation
  • Barth, Michael C., “Universal Wage-Rate Subsidy: Benefits and Effects,” in The Economics of Federal Subsidy Programs: Part 4Higher Education and Manpower Subsidies, U. S. Congress, Joint Economic Committee (Washington, August 28, 1972), pp. 497-540.

    • Search Google Scholar
    • Export Citation
  • Berner, Richard B., “A General Equilibrium Model of International Discrimination” (doctoral dissertation, University of Pennsylvania, 1976).

    • Search Google Scholar
    • Export Citation
  • Biederman, Kenneth R., “Alternative Tax Subsidies for the Training and Employment of the Unemployed,” in The Economics of Federal Subsidy Programs: Part 4—Higher Education and Manpower Subsidies, U. S. Congress, Joint Economic Committee (Washington, August 28, 1972), pp. 541-69.

    • Search Google Scholar
    • Export Citation
  • Black, S. W., and H. H. Kelejian, “A Macro Model of the U. S. Labor Market,” Econometrica, Vol. 38 (September 1970), pp. 712-41.

  • Brittain, John A., The Payroll Tax for Social Security (The Brookings Institution, 1972).

  • Crandall, Robert W., and others, “An Econometric Model of the Low-Skill Labor Market,” Journal of Human Resources, Vol. 10 (Winter 1975), pp. 3-24.

    • Search Google Scholar
    • Export Citation
  • Dhrymes, Phoebus J., “A Model of Short-Run Labor Adjustment,” in The Brookings Model: Some Further Results, ed. by James Duesenberry and others (Chicago, 1969), pp. 110-49.

    • Search Google Scholar
    • Export Citation
  • Dumont, Jean-Pierre, “French Employment Policy During 1974-1976,” in Reexamining European Manpower Policies, National Commission for Manpower Policy, Special Report No. 10 (Washington, August 1976), pp. 101-20.

    • Search Google Scholar
    • Export Citation
  • Feldstein, Martin S., “Tax Incidence in a Growing Economy With Variable Factor Supply,” Quarterly Journal of Economics, Vol. 88 (November 1974), pp. 551-73.

    • Search Google Scholar
    • Export Citation
  • Fethke, Gary C., and Samuel H. Williamson (1976), Employment Tax Credits as a Fiscal Policy Tool, U. S. Congress, Joint Economic Committee (Washington, July 21, 1976).

    • Search Google Scholar
    • Export Citation
  • Fethke, Gary C., and Samuel H. Williamson (1977), “Employment and Price Level Effects of a Variable Base Wage Credit,” University of Iowa Bureau of Business and Economic Research, Working Paper Series No. 76-26A (Iowa City, January 1977).

    • Search Google Scholar
    • Export Citation
  • Frisch, Ragnar A., Price-Wage-Tax-Subsidy Policies as Instruments in Maintaining Optimal Employment, U. N. Economic and Social Council, Economic and Employment Commission (United Nations, April 18, 1949).

    • Search Google Scholar
    • Export Citation
  • Greenston, Peter M., and C. Duncan MacRae (1975 a), “Evaluation of the WIN Tax Credit,” The Urban Institute, Working Paper No. 3609-01 (Washington, June 20, 1975).

    • Search Google Scholar
    • Export Citation
  • Greenston, Peter M., and C. Duncan MacRae (1975 b), “Categorical Wage-Bill Subsidies: Theory and Application,” The Urban Institute (Washington, August 1975).

    • Search Google Scholar
    • Export Citation
  • Hamermesh, Daniel S. (1976), “Econometric Studies of Labor Demand and Their Application to Policy Analysis,” Journal of Human Resources, Vol. 11 (Fall 1976), pp. 507-25.

    • Search Google Scholar
    • Export Citation
  • Hamermesh, Daniel S. (1977), “Subsidized Job Creation in the Private Sector: Problems and Prospects,” prepared for the Brookings Conference on Job Creation (unpublished, The Brookings Institution, May 1977).

    • Search Google Scholar
    • Export Citation
  • International Labor Office, Fiscal Measures for Employment Promotion in Developing Countries (Geneva, 1972).

  • Kaldor, Nicholas, “Wage Subsidies as a Remedy for Unemployment,” Journal of Political Economy, Vol. 44 (December 1936), pp. 721-42.

  • Kesselman, Jonathan B., and others, “Tax Credits for Employment Rather Than Investment,” American Economic Review, Vol. 67 (June 1977), pp. 339-49.

    • Search Google Scholar
    • Export Citation
  • Kopits, George F., International Comparison of Tax Depreciation Practices (Organization for Economic Cooperation and Development, 1975).

    • Search Google Scholar
    • Export Citation
  • Laroque, Guy, and others, Comptes trimestrielles: Méthodes statistiques et series retrospectives, Les collections de l’INSEE, 40C (Paris, December 1975).

    • Search Google Scholar
    • Export Citation
  • Lucas, Robert E., “Labor-Capital Substitution in U. S. Manufacturing,” in The Taxation of Income from Capital, ed. by Arnold C. Harberger and Martin J. Bailey (The Brookings Institution, 1969), pp. 223-74.

    • Search Google Scholar
    • Export Citation
  • Lucas, Robert E., and Leonard A. Rapping, “Real Wages, Employment and Inflation,” Journal of Political Economy, Vol. 77 (September-October 1969), pp. 721-54.

    • Search Google Scholar
    • Export Citation
  • McCracken, Paul W., and others, Towards Full Employment and Price Stability (Organization for Economic Cooperation and Development, 1977).

    • Search Google Scholar
    • Export Citation
  • Mieszkowski, Peter, “The Indirect Market Effects of Wage Subsidy and Public Employment Programs,” in Studies in Public Welfare Paper No. 19: Public Employment and Wage Subsidies, U. S. Congress, Joint Economic Committee (Washington, December 30, 1974), pp. 138-63.

    • Search Google Scholar
    • Export Citation
  • Nadiri, M. Ishaq, “The Effects of Relative Prices and Capacity on the Demand for Labor in the U. S. Manufacturing Sector,” Review of Economic Studies, Vol. 35 (July 1968), pp. 273-88.

    • Search Google Scholar
    • Export Citation
  • Nadiri, M. Ishaq, and Sherwin Rosen, A Disequilibrium Model of Demand for Factors of Production (National Bureau of Economic Research, 1973).

    • Search Google Scholar
    • Export Citation
  • Nordhaus, William D., “Recent Developments in Price Dynamics,” in The Econometrics of Price DeterminationConference, ed. by Otto Eckstein (Washington, 1972), pp. 16-49.

    • Search Google Scholar
    • Export Citation
  • Sims, Christopher A., “Output and Labor Input in Manufacturing,” Brookings Papers on Economic Activity: 3 (1974), pp. 695-728.

  • Solow, Robert M., “Macro-Policy and Full Employment,” in Jobs for Americans, ed. by Eli Ginzberg (Englewood Cliffs, New Jersey, 1976), pp. 37-58.

    • Search Google Scholar
    • Export Citation
  • Ullman, Lloyd, “Manpower Policies and Demand Management,” in Jobs for Americans, ed. by Eli Ginzberg (Englewood Cliffs, New Jersey, 1976), pp. 85-119.

    • Search Google Scholar
    • Export Citation
  • U. S. Congress, Congressional Budget Office, Employment Subsidies and Employment Tax Credits, Background Paper (Washington, April 1977).

    • Search Google Scholar
    • Export Citation
  • Van Rompuy, P., “Simultaneous Estimation of Short-Run Employment Functions,” European Economic Review, Vol. 11 (Summer 1971), pp. 403-35.

    • Search Google Scholar
    • Export Citation
  • Waud, R., “Manhour Behavior in U. S. Manufacturing: A Neoclassical Interpretation,” Journal of Political Economy, Vol. 76 (May-June 1968), pp. 407-27.

    • Search Google Scholar
    • Export Citation
  • Weiner, Neil S., and others, “Report on the Feasibility of Estimating the Effects of a National Wage-Bill Subsidy,” Institute for Defense Analyses, Research Paper P-545 (Arlington, Virginia, September 1969).

    • Search Google Scholar
    • Export Citation

Mr. Kopits, Senior Economist in the Tax Policy Division of the Fiscal Affairs Department, holds degrees from Georgetown University. He is also a professorial lecturer at the Johns Hopkins School of Advanced International Studies. Previously, he was associated with the Brookings Institution and the U. S. Treasury Department.


The cooperation of the Institut National de la Statistique et des Etudes Economiques in making available an updated quarterly series of the French national accounts, and comments by B. Grais of the Institut, are gratefully acknowledged. This paper has benefited also from suggestions by Albert Eckstein and C. Duncan MacRae. Iqbal Zaidi provided computational assistance.


Historically, fiscal and financial incentives to capital formation have been the most widely adopted selective policy tools. For an international comparison of tax-subsidy schemes affecting the cost of capital, see Kopits (1975).


Subsidization of wage costs to reduce unemployment was advocated more than four decades ago by Kaldor (1936). More recently, the desirability of complementing macro-economic instruments with such fiscal incentives has been suggested by Solow (1976), Ullman (1976), and McCracken and others (1977, chapter 7).


In theory, the subsidy has the same economic effect whether it is disbursed to employers or workers. Some authors draw a distinction among wage-bill subsidies, wage-rate subsidies, wage subsidies, and employment subsidies, depending on the subsidy base (wage bill, man-hours, or number of employees) or on the recipients (employers or employees). Given the lack of uniformity in usage (despite the similarity in theoretical treatment), Kaldor’s generic term, wage subsidy, is adopted throughout this paper.


The first known wage subsidy was implemented in Germany for six months in 1932.


For a summary of U. S. legislative proposals, see U. S. Congress (1977, Appendix B-I).


In the United States, the Carter Administration proposed unsuccessfully a temporary general social security tax credit. A similar across-the-board payroll tax reduction has been enacted on a permanent basis in Sweden.


The reduction in the value of the subsidy depends, of course, on the employer’s income tax rate, as shown by the difference between equations (9) and (10), which appear in Section III.


For subsidies disbursed during a period shorter than one year (the Incentive Bonus and the Wage Cost Subsidy), the subsidy ratios are calculated on an annual basis to account for the prescribed minimum period of employment.


A sufficiently large subsidy (e.g., one equivalent to the entire wage of the marginal worker) could induce the employer to hoard inframarginal workers so as to maintain the base level of employment that would make him eligible for the subsidy. An analogous outcome may obtain with implementation of a subsidy with a base level set at a percentage below the employment in the base period, whereby workers hoarded in excess of that base level would also entitle the employer to the subsidy. A variable base level of this kind has been suggested by Fethke and Williamson (1976), though it has not been adopted by any government. Either approach (large payments or reduced base level) is essentially applied to inframarginal employment, and thus could lead to substantial windfalls during a recovery, much like a general wage subsidy does.


In the case of the American subsidy, this safeguard is somewhat weakened by the definition of the base level in terms of the wage bill instead of employment.


The Federal Republic of Germany’s Wage Cost Subsidy is more restrictive in this regard, for it combines the waiting period requirement with a regional unemployment test.


One possible objective approach to calculating redundancies for purposes of deciding eligibility for such a program would be the application of a growth-of-sales test. This method, of course, would be highly arbitrary and unlikely either to eliminate windfalls arising from (D) or to retard layoffs with regard to firms experiencing a sales performance that deviates from the measure of cyclical fluctuations used in the test.


This outcome is symmetrical to the long-run incidence of payroll taxes, investigated by Brittain (1972, chapters 2 and 3) and Feldstein (1974).


Although, according to neoclassical price theory, a general and a marginal wage subsidy of an equivalent amount in relation to the wage rate at the margin should have identical impacts on product prices (and thereby identical scale and interindustry effects), it is doubtful that firms actually follow marginal cost pricing behavior to the point where they would pass a significant portion of a temporary marginal wage subsidy along to consumers by way of a price reduction. For a survey of the theory and the evidence on the pass-through of changes in labor costs to prices, see Nordhaus (1972).


See, for example, the studies in International Labor Office (1972), and the references in Fethke and Williamson (1976, p. 2).


Comprised of a goods market, a money market, and a labor market.


In an extension of their work, Greenston and MacRae (1975 a) also conducted a feasibility study of an ex post evaluation of the tax credit.


This version of the model realistically assumes a wage elasticity of labor demand of -0.36, in contrast to the first version (1976) based on an elasticity of -1.


See the empirical contributions by Lucas and Rapping (1969), Black and Kelejian (1970), and Crandall and others (1975), with regard to the labor market in the United States, and by Berner (1976, chapter 5) with regard to the labor market in the original member countries of the European Economic Community.


Assuming an infinitely elastic supply of labor over the relevant range (realistic under conditions of less than full employment). For a similar treatment, see the models used by Nadiri (1968), Waud (1968), Dhrymes (1969), Nadiri and Rosen (1973), and Hamermesh (1977), to explain employment in the United States, and by Van Rompuy (1971) to explain employment in Belgium.


Other factor inputs (capital in our model) are held constant.


If the subsidy is a general one, equation (9) represents the average effective wage rate of the firm’s entire work force; if the subsidy is a marginal one, it measures the effective wage rate of workers hired under the subsidy.


Contrary to the short-run increasing returns to labor reported by several authors, Sims (1974) has demonstrated with monthly data on three manufacturing industries that (a) there is a roughly proportional relationship between labor and output, and (b) most of the labor response to output changes is completed within six months. The relatively speedy and proportional adjustment of labor to output has also been found in a number of other, less rigorous analyses surveyed by Hamermesh (1976).


The employment-wage nexus has not been investigated in the past as thoroughly as the employment-output relationship, particularly insofar as the time path of adjustment is concerned. However, much of the existing evidence indicates that the response to wage changes occurs in one year or less, as shown by Hamermesh (1976). The empirical specification of the distributed lag in equation (11) is analogous to the one used by Lucas (1969).


Note that, by implication, if j = 1, then β0 = 0; if j = 2, then β0 = β1 = 0; and if j = 3, then β0 = β1 = β2 = 0 and β3 = 1.


The measurement of variables is described in the Appendix.


See the labor market model estimated on the aggregate U. S. private nonfarm sector by Black and Kelejian (1970) and Hamermesh (1977).


Estimates of the first-order autocorrelation coefficient using the Hildreth-Liu search technique suggested that the global minimum standard error had been located with the Cochrane-Orcutt iteration.


These dummies reflect the depressing effect on output of the unusually harsh winter in the first quarter of 1963 and of the civil unrest in the second quarter of 1968. They are statistically significant in three industries.


To test the realism of this adjustment speed (notwithstanding the references to outside evidence in footnote 24), additional lagged values of changes in output were included. Unfortunately, in several industries such experiments resulted in explosive serial correlation and severe multicollinearity.


See Hamermesh (1976). The only comparable estimates for France, by Berner (1976, p. 240), indicating very high elasticities (in excess of unity for some industries), seem to be entirely out of line with those found elsewhere in the literature.


Dhrymes (1969) found even greater variation in σ across two-digit Standard Industrial Classification (SIC) industries (ranging between -0.15 and -0.97) in the United States.


Although no simultaneity bias can be readily detected in our results, Hamermesh (1976) attributed the larger wage elasticities estimated by some authors to failure to hold the cost of capital constant.


Assuming that the Bonus creates the same number of jobs at the margin as an equivalent general wage subsidy (i.e., paid in the same proportion to total wages as the proportion of the marginal subsidy in the incremental wage cost) would.


These shifts are reflected in part by the changing weights of various subindustries in each industrial branch—particularly machinery and primary metals, and intermediate goods.


See footnote 34. As compared to the employment effect predicted by the model, the authorities initially expected that the Bonus would provide jobs to 300,000 workers, although “it was hoped realistically” that 100,000 placements would thus be created; see Dumont (1976, pp. 112-13). Note that an increment of 130,000 jobs is approximately equivalent to a one percentage point reduction in the French unemployment rate.


Obviously, any correlation between the two figures can be purely spurious; the actual number of subsidized workers cannot be interpreted as an indication of the subsidy’s effectiveness. The number of subsidized employees includes those who produce a windfall on the one hand and, on the other, excludes employees who, although not directly subsidized, are hoarded by firms to meet the base level of employment necessary for hiring subsidized employees.


Any such delayed response to the subsidy cannot be verified by extending our forecasts because the INSEE quarterly series were discontinued after 1975.


As a first approximation, the net cost can be calculated as follows. Multiplication of 38,000 Bonus payments times F 3,000 per employee results in a maximum gross disbursement of F 114 million. This, in turn, is reduced by the corporate tax (at a 50 per cent rate) liability incurred thereon. The remaining amount is cut further by personal income tax payments generated and unemployment compensation foregone by the estimated 27,000 employees who, according to our forecasts, would not have been hired in the absence of the Bonus. In this connection, it may be noted that the average French worker is eligible, after being laid off, to receive unemployment compensation ranging between 50 and 90 per cent of his previous earnings for a period of up to two years.


This and other administrative difficulties are reflected in the timing of actual disbursements that were begun in September 1975 (three months after the inception of the program). Also, see the discussion by Dumont (1976, p. 113).


See Dumont (1976, pp. 105-107).


This may have disqualified, at most, about 5,000 workers (hired under such contracts through the end of 1975) for the Bonus.


Although, of course, information about the subsidy must be disseminated promptly to all potential users, the base level of employment should not be announced before the date for which it is fixed, in order to avoid a perverse reduction in the eligible firm’s work force in anticipation of the subsidy.


Yet it should be recalled that the main purpose of the subsidy is to mop up cyclical unemployment, leaving the far more cumbersome task of eliminating structural unemployment to manpower training programs to be undertaken directly by the public sector or by the private sector under various fiscal incentives.


A general wage subsidy, provided to the entire employment flow, is analogous to capital cost subsidies (in the form of grants, accelerated depreciation, or tax credits) that are ordinarily applied to the gross investment flow. On the other hand, a marginal wage subsidy is comparable to a subsidy extended only to capital expenditures in excess of replacement investment or depreciation.