Final Report

The Employment Effect of Subsidies

Report to the Directorate General Employment , Industrial Relations and Social Affairs

Commission of the European Communities

SOC 94 100018 05A01

J K Swales*
D R Holden
G Beacon

November 1995

* Fraser of Allander Institute and Department of Economics, University of Strathclyde
Department of Economics, University of Strathclyde
Beacon Dodsworth Ltd


High levels of unemployment, particularly amongst unskilled groups, is a serious structural problem for many European countries. In this report we build on previous work by Jackman and Layard, Johnson and Beacon and Monk to investigate the effectiveness of general labour subsidies in tackling this problem. We find that the government can influence long-run employment levels by introducing an appropriate tax and subsidy system, even where the economy is working in a perfectly competitive way.

The specific policy package which we considered in detail involves the introduction of a fixed per capita labour subsidy, equal to 5% of the average wage, financed by an increase in VAT. The tax/subsidy scheme works by allowing for some substitution of capital for labour, but more generally by pricing workers into jobs through subsidisation and increasing the incentive to work, especially amongst lower paid workers. Estimates are made of the output and dissagregated employment effects under various assumptions about the capital market and the nature of labour supply. Total employment and total output always increase. The proportionate expansion in total output lies within the range 0.8% and 2.8% and the increase in employment in the range 1.45% to 4.0%. Low paid sectors of the labour force are stimulated the most so the policy has favourable distributional aspects. The existance of savings on unemployment benefits acts to reinforce employment effects.

Governments are generally concerned about the overall level of taxation within the economy and therefore question the desirability of automatic subsidy programmes. However, the type of subsidy and tax plan that we outline could, in principle, be operated as an integrated tax scheme in which the change in the firm's tax bill is calculated as the net difference between the additional VAT and the per capita subsidy. In so far as the scheme increases total employment, and thereby reduces payments of unemployment benefit, it would be associated with a reduction in the overall tax rate. That is to say, the introduction of a new tax scheme would simultaneously increase employment and reduce taxation.

There is at present an increased faith in "market forces" and a general desire to reduce subsidies that artificially maintain inefficient or inappropriate industries. However, where there are high levels of structural employment amongst primarily low skilled workers, and where these unemployed are supported by welfare payments which lower the real income of workers and reduce their incentive to work, the possibility of long-term labour subsidies should be considered. Such subsidies improve, rather than impair, productive efficiency by offsetting market failures in other parts of the economy. They restore, rather than distort, appropriate price signals. They do not rob the private sector of resources but reallocate resources within that sector. As we have seen, such subsidies generate an expansion, not contraction, of private sector economic activity. Further if such subsidies can be packaged as tax rebates the possibility occurs of a simultaneous fall in taxation and increase in employment.


High levels of unemployment, particularly amongst unskilled groups, is a serious structural problem for many European countries. In this report we make a theoretical investigation of the use of labour subsidies to tackle this problem. In Section 2 we review previous work in this area. In Section 3 we outline a simple general equilibrium model for identifying the impacts of a labour subsidy, give a specific form of the model which can be used for simulation purposes and then report the results of some illustrative simulations. Attention is drawn in this section to the balanced budget requirement for the financing of such a subsidy and the interaction with the unemployment benefit system. In Section 4 we discuss the appropriateness of the model for investigating this problem. In Section 5, we consider political issues.. Section 6 is a short conclusion.


In two articles in Economica, Jackman and Layard (1980) and Johnson (1980) argue for government intervention in the labour market in order to reduce total unemployment. One of the policies that they consider is a subsidy on employment for particular segments of the labour market. Whilst these articles are rather abstract and technical, they have a number of important strengths. These are that

  • They explicitly consider neo-classical models. That is to say, systems in which competitive product markets clear. When labour-market distortions are investigated, they are of the most straightforward type. The most common is the existence of unemployment benefit, which might reduce the labour supply for some low-wage groups. However, both papers also deal with cases where there is a degree of wage rigidity, either in the form of minimum wages or fixed wage differentials.
  • They adopt a general, rather than partial, equilibrium approach.
  • They impose a neutral budget condition so that the financing of any subsidies must be made explicit.

Even though both papers adopt rather sparse models, which are solved analytically, the solutions that they adduce are quite complex.

2.1 Jackman and Layard (1980)

Their basic model has the following characteristics:

  • There is only one commodity which is produced from two inputs only. These are different types of labour, which the authors label "skilled" and "unskilled".
  • The production function is linear homogeneous.
  • The output is sold and labour is bought in perfectly competitive markets.
  • The supply of each labour type is an increasing function of the real take-home wage.
  • There is no international trade.
  • Government expenditure on public goods is fixed.
  • Initially the PSBR is zero, with taxes on labour income covering expenditure on public goods and unemployment benefit.

This model is used to analyse the impact on total employment, output and welfare of two labour tax and subsidy schemes.

First, the authors consider a "self-financing" tax and subsidy regime where one group of workers is taxed and the other subsidised and the total value of the taxes on one group just equals the total value of the subsidies on the other labour group. The results here are very straightforward. If group 1 is taxed and group 2 subsidised, total output, employment and welfare will rise as long as

respectively, where:

i is the labour supply elasticity of group i with respect to the relevant gross wage,

Wi is the gross wage of an individual in group i, and

mi is the labour market distortion which is defined as

where ti is the average tax rate payed by workers in group i and Bi is the unemployment benefit which such a worker would qualify for.

Although the schemes here are called "self-financing", this is a little misleading. In general such a scheme will lead to an increase in employment in unskilled (type 2) labour and a reduction in employment in skilled (type 1) labour. There will therefore be foregone taxation on the income from newly unemployed skilled workers and increased unemployment benefit payments to these workers. On the other hand, there will be increased tax receipts from newly employed unskilled workers and a reduction in the benefit payments for this group.

These changes will have effects on the overall public sector budget which will lead to variations in the general taxation levied to meet the financing of public goods (whose provision is assumed constant) and unemployment benefit. Jackman and Layard take these general taxes to be levied at a constant rate on (wage) income, so that changing the general level of taxation implies varying this tax rate. Clearly it could be the case, on the criteria above, that the introduction of the "self-financing" tax-subsidy regime will lead to an increase in employment whilst generating a public sector deficit. This would occur if the increase in employment were modest and there was a big difference in the average wage and the unemployment benefits paid to the two groups, the levels of both being assumed to be higher for skilled than unskilled workers. Such an increase in general taxation would reduce the net returns from work, decrease the supply of both types of labour and thereby reduce employment. Such reductions would partly, and might wholly, offset any gains predicted from the earlier analysis.

However, Jackman and Layard show that where there are positive welfare gains from the introduction of the "self-financing" subsidy, the public sector finances will improve also. Under these circumstances, the introduction of the self-financing tax-subsidy scheme will allow a reduction in general taxation, which will generate further increases in employment. We therefore concentrate on the conditions both for an increase in "self-financing" employment and welfare. Therefore from the analysis presented in this paper a sufficient condition for employment to increase is

We would normally assume that the wages of unskilled workers are lower that skilled (W1 > W2) and that the distortion in the labour market caused by taxation and unemployment benefit is higher for lower waged workers (m2 > m1), this implies

so that a sufficient requirement for employment to unambiguously increase is simply that the elasticity of labour supply is greater for unskilled labour than skilled labour.

Note that this condition also ensures that economic welfare will rise and that there will be a more even distribution of income as employment and wage rates for unskilled workers will rise, as against skilled workers.

Although Jackman and Layard are primarily concerned with a situation in which the wage rates for both types of labour are flexible, they do also analyse the case where there is some wage rigidity. In particular, they consider a set up where there are a set of rigid wage differentials. Specifically:

  • the relative gross pay of skilled to unskilled workers is fixed.
  • skilled workers are on their labour supply curve but there is excess supply of unskilled workers at the relevant wage, so that the unskilled labour supply curve is suspended.

Under these conditions the potential benefits from wage subsidies, financed by a tax on employers for each skilled worker, generates an even greater welfare gain.

2.2 Johnson

Johnson is concerned with a very similar analytical framework to that adopted by Jackman and Layard. However, his work differs in a number of important respects.

  • There are an undetermined number of inputs (n).
  • Two are not fully employed, youth labour and unskilled labour.
  • The causes of unemployment in these two sectors of the labour market are as follows. In the youth market, there is some form of minimum real wage which prohibits labour market clearing. In the market for unskilled labour, a similar sort of arrangement might occur or the presence of unemployment benefit might reduce effective labour supply below the registered labour force.
  • All other inputs are taken to be in fixed (totally inelastic) supply and their price set at the market clearing level.
  • A central concern is the distribution of costs and benefits between different factor owners.

In Johnson's analysis, ad valorem subsidies are paid at different rates to employers of youth and unskilled workers. These subsidies, together with all government expenditure on public goods, are financed by a proportionate tax on all other factor inputs. That is to say, it is assumed that youths and unskilled workers pay no tax, so that the burden of all government expenditure is borne by the other factors of production.

The impact of the subsidies on total employment depends solely on their impact on the two "target" groups. This is because the supply of all other factors is fixed and their market price is set at their market clearing rate. The effect of a subsidy on one labour group is always to increase the employment of that group. Therefore if there is only a subsidy on youths, youth employment will rise. However, the effect on the other group suffering unemployment (in this case, unskilled workers) is ambiguous and depends on whether they are complements or substitutes for the subsidised group. If they are complements, the employment of the unsubsidised labour group will rise: if they are substitutes it will fall. Three points are important here.

  • Where both labour groups are subsidised, the effect of their own subsidy on their own employment is positive.
  • Conventional production functions have inputs as complements: the productivity of one input rises as the supply of a second input rises. In this case, the subsidy on one input increases the demand for the second subsidised input.
  • Even where the two inputs are substitutes, the negative cross-effect between the two inputs have to dominate the own effects if employment is not to rise.

The employment effects identified by Johnson are rather more complex than those in Jackman and Layard. The reason is that he allows the factors to be substitutes, whereas in a two-factor well-behaved production function, inputs must be complements. However, Johnson's main interest is in the effect of the subsidy on the gross and net-of-tax earnings of the other factor owners. His argument is essentially that if the income of one group is adversely effected by a subsidy, that group might mobilise to block such a subsidy.

If we start with the introduction of subsidies where the wages of both youths and the unskilled are fixed in real terms (by minimum-wage legislation, for example), the gross earnings of the other n-2 factor owners are increased by the amount spent on the subsidies. Moreover, if the employment of unskilled workers is increased, expenditure on unemployment benefit for that group is reduced. (Youths are assumed not to qualify for unemployment benefit). This implies that for taxpayers as a whole, net after-tax income rises. As Johnson remarks, this implies that if the target labour groups have inflexible wages set too high to clear their segments of the labour market, subsidies are an excellent deal for taxpayers as a whole. One caveat here is that whilst taxpayers in general benefit, the after-tax income of certain groups of factor owners might fall, if their factors are substitutes for the subsidised labour.

In the more complex case, the youth labour market again has a fixed real wage, but the labour market for unskilled workers clears.1 However, this does not mean that there is no unskilled unemployment. In this segment of the labour market, the position is similar to that analysed in Jackman and Layard: the supply of unskilled labour is not perfectly inelastic, and there is a gap between the level of employment and the number of unskilled workers registered for employment. Changes in the demand for unskilled workers will therefore effect both the wage and employment of unskilled workers.

Under these conditions, the subsidy programme for low-skilled adults increases the post-tax income of taxpayers as a whole as long as

where e is the elasticity of supply of unskilled workers and r is the replacement ratio in the unskilled labour market, that is the ratio of the unemployment benefit to the unskilled wage. Where this condition holds, the youth subsidy will increase the taxpayer's real income as long as youth and unskilled employment are complements. Finally, it is important to state that even where other taxpayers are made worse off by the introduction of labour subsidies for some labour groups, that does not necessarily imply that there are not welfare gains to such subsidies.

Whilst these two papers are rather abstract, they are important in that they show:

  • In a long-run equilibrium perfectly-competitive system, a self-financing tax-subsidy regime can generally increase employment. This has nothing to do with traditional Keynesian macroeconomic influences, non-governmentally induced labour market imperfections or terms-of-trade effects.
  • Where the government pays unemployment benefit, there are straightforward conditions under which the introduction of a subsidy will increase economic welfare, even where economic welfare is defined in a very conventional (and conservative) manner. That is to say, we need not take into consideration issues such as a link between unemployment and ill health, crime or other forms of social breakdown, for this argument to go through. This is not, of course, to assert that these social issues are unimportant. Howeverr, the UK, at least, seems extremely reluctant to accept any evidence of a relationship between these forms of social problems and unemployment.
  • Where the government pays unemployment benefit, there are clear conditions under which a subsidy will increase the post-tax income of the average owners of non-subsidised factors. These conditions are not arcane of extreme. Moreover, the notion of income here is the straightforward one of command over goods and takes no account of any real increase in welfare that employed workers of one type may gain from reducing the level of unemployment in other sectors of the labour market, or their own risk of becoming unemployed.

These arguments are important to counter views that labour market intervention reduces employment and welfare and that the issues involved in such policies are the familiar efficiency versus equity ones. Such erroneous views are expressed by the UK Government in HM Treasury (1991, p. 70) in its discussion of the rules to be used in the evaluation of industrial and regional assistance whose central goal is job creation. "Because of crowding out at the macroeconomic level, effects on employment ... should not be included as benefits of projects in an efficiency test." As has been demonstrated by Jackman and Layard (1980) and Johnson (1980), such "crowding out" would not, in general, be expected to occur.

2.3 Beacon and Monk

The work of Jackman and Layard (1980) and Johnson (1980) is purely analytical and each concentrate on two types of labour. In the case of Jackman and Layard, there are only two productive inputs, skilled and unskilled labour: for Johnson, there are n inputs but all but two, youth and low skilled adults, are in completely inelastic supply. Beacon and Monk (1987) approach the problem in a more flexible manner. Their analysis is similar to those of Jackman and Layard (1980) and Johnson (1980) in that they employ a closed-economy, one-sector model and the subsidies investigated are fully funded so that the public sector budget is always balanced. However, their model differs in the following characteristics:

  • There are six factor of production, capital and five separate labour categories.
  • The supply of each factor input is sensitive to the factor's price. That is to say, no factor has a completely inelastic supply.
  • The model is concerned with aggregate unemployment, not simply unemployment within one or two employment groups.
  • This is a numerical model where the results are derived by computer simulation.

Beacon and Monk (1987) simulate the impact of a fixed per capita labour subsidy financed by an increase in VAT. They argue that one would expect that such a subsidy would have two important effects.

  • A substitution of labour for capital.
  • A substitution of lower paid workers for higher paid workers. This effect applies because the proportionate impact of a fixed subsidy is greater where the original wage is lowest. This is seen as important because unemployment is regarded as being a problem primarily for lower paid workers. According to Beacon and Monk, this is because the wages of lower-paid workers are close to the unemployment benefit level, and such workers are, in essence, likely to be "voluntarily" unemployed.

If such substitution effects increase the level of employment, and thereby reduce unemployment, then the employment impact will be further amplified by the downward adjustment which can be made to the VAT rate because government expenditure on unemployment benefit will have fallen.

Because of the differential impact of the fixed per capita subsidy on labour groups with different wages, Beacon and Monk (1987) separate the labour force by wage level. That is to say, the five groups identified in their analysis are simply the five quintiles of the labour force ranked by wage. It is assumed that workers in a given quintile earn the same wage. It is also assumed that workers in different quintiles will have different labour supply characteristics. In particular, the lower the wage, the higher the wage elasticity of supply.

The simulations that Beacon and Monk perform are stylised and indicative. For a fixed per capita labour subsidy equal to 20% of the average wage, there is an increase in total employment of 3.5%. Employment in all segments of the labour market rises, by a maximum of 9.6% in the lowest wage quintile to 0.03% in the highest wage quintile. Capital employment falls by 5.9%. In these simulations all workers benefit, in the form of increased employment and wages and there is a reduction in the payment of unemployment benefit. Owners of capital loose.

The work of Beacon and Monk (1987) in some respects extends the earlier analytical models of Jackman and Layard (1980) and Johnson (1980), but it has a couple of drawbacks. First, it seems to imply a fixed output, so that all the effects come through substitution. From the earlier work, we know that there will be output effects too, which are likely to increase the positive employment impact identified in the simulations. Second, the treatment of capital is rather cavalier. This would be less important were it not the case that the role of capital is crucial in their simulations. Capital is the only factor whose real payments fall as the result of the introduction of the subsidy scheme; both the use of capital and its real return decline, though Beacon and Monk (1987) do not comment on the economic or political implications of these reductions. Moreover, if capital were to be treated differently, it might be that conflict would occur between labour groups. In particular, in their reported simulations, high wage groups experience a very small gain in real income from the labour subsidy. However, such high wage groups might be faced with lower employment and real wages if the supply of capital were more elastic.

To Chapter 3