However displacement may occur when the government offers assistance targeted towards .... workers hired is considerably negatively responsive to the wage.
Displacement effects of active labour market policy
by Dr Simon Chapple
Report for the Department of Labour
NZ INSTITUTE OF ECONOMIC RESEARCH (INC) 8 Halswell Street, Thorndon P O BOX 3479, WELLINGTON
The Institute, its contributors, employees and Board shall not be liable for any loss or damage sustained by any
TABLE OF CONTENTS
1.2 Treasury’s 1984 position on active labour market policies
1.3 Types of active labour market policies
2. DISPLACEMENT AND ITS FORMS
2.1 Defining displacement
2.2 Views on the definition of displacement
2.3 Economic approaches to considering the issue
2.4 Macroeconomic displacement
2.5 Displacement and education and training
2.6 Displacement and job placement & information subsidies
2.7 Displacement and private sector job subsidies
2.8 Inter-industry effects
2.9 Displacement & direct public job creation
2.10 Displacement and neighbourhood effects
2.11 Factors influencing the degree of displacement
3. EMPIRICAL EVIDENCE ON DISPLACEMENT
3.1 Introduction & summary
3.2 Evaluation methodologies
3.3 Microeconomic studies
3.4 Cross-country OECD studies
3.5 Time series studies
3.6 Macro or general equilibrium model simulations
3.7 Other approaches
4. POLICY DESIGN TO MINIMISE DISPLACEMENT
4.1 Microeconomic issues
4.2 Effective policy design and macroeconomic policy
4.3 Measurement and monitoring
5. TRADING-OFF DISPLACEMENT AND EQUITY OBJECTIVES
TABLE OF FIGURES Figure 1: Displacement in the Layard and Nickell model ______________________________ 13 Figure 2: Displacement and training ______________________________________________ 15 Figure 3: Displacement and subsidies: a partial equilibrium analysis ____________________ 22
• In theory active labour market policy promises a possible way out of the strong wage growthlow employment or weak wage growth-high employment scenarios suggested respectively by the European and US labour markets.
• However displacement may occur when the government offers assistance targeted towards those it wishes to directly encourage into employment. While there may be an increase in employment for the targeted groups, it may be at the expense of jobs for others.
• In theory, displacement may not occur. Replacement, or additions to employment of nontarget groups, are theoretically possible.
• There is no single generally accepted or persuasive framework within which to theoretically consider displacement. A variety of ad hoc approaches can be used, each of which deals with certain aspects of the problem.
• Because determining the empirical extent of displacement (or replacement) involves an assessment of the indirect effects of active labour market policy, any complete evaluation of active labour market policy must consider economy wide effects.
• Because most evaluations of active labour market policies are microeconomic, they do not usually satisfactorily address the issue of displacement. Rather at best they generally address the impact of the programme on participants’ earnings and employment probabilities. The few microeconomic evaluations which do address displacement typically do not use robust methodologies in measuring all possible displacement and replacement effects.
• The macroeconomic evidence on the effects of active labour market policies yields ambiguous results. None of this macroeconomic evidence grapples with the problems of design. It simply uses spending or numbers of persons on programmes to investigate programme effectiveness.
• The literature provides only a broad set of design guidelines. These include: •
Enforcing additionality criteria at a firm level.
Putting in place a broad and tailored portfolio of programmes to avoid problems of diminishing returns.
Designing programmes to shift workers from geographical or skill defined labour markets of labour surplus to those of labour shortage.
Designing programmes to shift workers from labour markets with inelastic labour demand and elastic labour supply to those with elastic labour demand and inelastic labour supply.
Providing programmes complementing private sector activity rather than directly substituting for it.
Concentrating more on training rather than direct job creation.
• In terms of measuring and monitoring the extent of displacement in employment initiatives, information is likely to be costly and, even if obtained, highly uncertain. One relatively low
cost source worthy of further investigation is employment register duration data from local employment offices.
• While displacement or replacement may occur in theory and in practice, economists remain a long way from obtaining reliable and robust estimates of the size and significance of displacement and the institutional and other circumstances under which it varies.
GLOSSARY OF KEY ECONOMIC TERMS USED IN THE REPORT Additionality: The net job gain from active labour market policies. This may be at a micro level (net gain for the target group, say the long term unemployed, described as target group unemployment additionality, or at the firm level) or at a macro level (net gain overall, described as total unemployment additionality). Aggregate demand: The sum of consumer, business, government and net foreign demand for goods and services produced by the national economy. Active Labour Market Policies: These are measures targeted at unemployed and disadvantaged workers aimed at improving labour market functioning. There are three broad types of active labour market policy: supply-side policies which provide education and training to raise the skills of the target group, demand-side policies which stimulate employment through direct job creation in the public sector or the subsidisation of private sector jobs, and subsidies to labour market information and provision of job placement services which increase efficiency of matching job seekers to jobs. Often known in the literature by their acronym ALMPs. Beveridge curve: A curve tracing out an inverse relationship between the rate of unemployment and the number of job vacancies per member of the labour force. As the vacancy rate goes up, the unemployment rate goes down. Deadweight: Hires made of people who take part in an active labour market programme which would have naturally occurred in the absence of the programme. One survey of OECD wage subsidy schemes estimated deadweight to be between 16 and 91 percent of jobs created, with estimates in excess of 50 percent common. Displacement: Displacement refers to any changes in employment elsewhere in the economy, in addition to the direct job creation effect, as a consequence of the active labour market programmes. While it is normally assumed that the indirect impact of active labour market policies reduces jobs, it can be shown that indirectly more jobs may be created indirectly than destroyed (see replacement). Displacement is sometimes also known as “crowding out” of jobs. Equilibrium: A situation where the economy is at rest, the desired quantity supplied in all markets equals the desired quantity demanded, and no one can be made better off without another being made worse off. Elastic labour demand/supply: Broadly speaking, the responsiveness of labour demand and supply to changes in the wage rate. An elastic labour demand curve is one where the numbers of workers hired is considerably negatively responsive to the wage. An elastic labour supply curve is one where the number of workers seeking work is strongly positively responsive to the wage. More formally, the elasticity is the proportional change in labour demand (or supply) relative to the proportional change in wages. Fiscal substitution: If the output produced by workers hired into the public sector on an active labour market programme directly reduces jobs elsewhere in the public sector, jobs will be displaced by fiscal substitution. Half life: In the context used below, the length of the period of time over which half the deviation from the equilibrium price is eliminated. Hysteresis: The idea that the final outcome of a process is not independent of the starting position. Where the economy is in the short run influences the ultimate position of long run equilibrium to which it eventually settles down. Lucas critique: The argument that deriving policy conclusions from a model not explicitly derived from individual behaviours may be inaccurate. This is because the policy change causes
individuals to alter their behaviour and thus the outcomes will differ from those predicted by the model. Maximand: The objective which is to be maximised. Replacement: Negative displacement or “crowding in” of jobs as a consequence of active labour market policies. Replacement occurs when the net number of jobs created as a consequence of the active labour market policy exceeds the number of jobs directly created. Replacement rate: The ratio of income on a welfare benefit to the wage that could be expected by the beneficiary in the labour market (usually proxied in practice by the average wage). Say’s law: The idea that an increase in supply generates sufficient demand to absorb the increased production. Say’s law implies there can be no general problems of a lack of aggregate demand. Substitution: A term for displacement within a firm which hires workers from an active labour market policy. These workers may be substituted by firing workers already in the firm or giving them fewer hours of work than otherwise. Also described as “intra-firm” displacement in this study.
1.1 Introduction The recent historical experience of OECD economies offers policy makers two rather unpalatable labour market role models. One possible model is the US where the positives of strong job growth and stable, relatively low unemployment rates must be balanced against poor real wage growth. Another possibility is the European experience, where wage growth has been stronger, but job expansion has been muted and unemployment rates have risen and stubbornly stuck at historically high levels. Neither model appears wholly satisfactory. Is there a third way which yields both job and wage growth? In recent times many economists and others have explored active labour market policies of achieving the twin outcomes of wage and job growth. Active labour market policies have an ongoing history in New Zealand. Support for them has followed a cyclical pattern of popularity. Despite growing unemployment between 1984 and 1992 and a current unemployment rate still well above 1984 levels, policy support for active labour market policies in New Zealand has been relatively low. The strong viewpoints expressed by Treasury (1984) against the active labour market policies practiced during the late seventies and early eighties seem to have been very influential in this regard. Recently however, government is again seriously considering greater efforts in the direction of active labour market policy (see Kilroy 1997 for example). Given this history, it is as well to summarise the influential Treasury position of 1984.
1.2 Treasury’s 1984 position on active labour market policies Treasury’s 1984 Economic Management contained an assessment of the impact of active employment policies on the economy. Since that discussion probably encompassed the conventional wisdom of the Public Service of the time, a brief consideration of its contents is useful. This provides a spring board to see how much new information has become available in the past thirteen years. In 1984 Treasury was writing in the context of New Zealand’s experience with a wide range of active employment programmes in both the public and private sector from the late seventies onwards. After pointing out that a number of other government interventions (such as subsidies to capital) were pulling against the aim of the employment subsidies, Treasury (1984, pp. 245-6) argue that subsidies tend to displace employment from other activities. They list the following difficulties with active labour market policies:
• Some new hires taken on under the subsidy would have occurred without a subsidy (typically these are known in the literature as the deadweight costs of a scheme).
• Growing firms hire subsidised labour while declining firms shed unsubsidised labour. Thus a rule that subsidies only be available to new hires is incapable of ensuring that a marginal employment subsidy does not become a general subsidy in the longer run. (Treasury do not analyse the impact of time dependent subsidies or subsidies which slowly phase out on this conclusion).
• Firms hiring subsidised labour produce goods and services which compete with unsubsidised labour, reducing employment elsewhere.
• The expansion of labour market programmes involves either a reduction in other government spending and a loss in other government employment, or a rise in taxes or borrowing from the private sector which displaces activity there. Treasury refer to a Department of Labour study which showed a deadweight cost of 65 percent of the subsidised jobs (65 percent of the jobs would have been created in the absence of the programme). They conclude that job creation by employment subsidies is a means of redistributing employment. It does not create it. Its only justification, therefore, is on equity grounds if jobs get re-distributed from those who in some sense need them less to those who need them more. To what extent are these conclusions still valid? What can theoretical and empirical developments since 1984 tell us about their robustness? These are questions I will attempt to answer below. Before this is done, some preliminary remarks on the 1984 Treasury position are in order (for another such discussion see Bertram 1988, pp. 701-702). Treasury’s 1984 approach concentrates on negative spillover effects. It does not consider any potentially positive impact that higher output and employment in the subsidised sectors have for the demand for labour in other sectors. It also neglects the higher incomes which result from lower prices. This, it needs to be emphasised, is not a Keynesian style multiplier effect. Rather it reflects Say’s law that the creation of more incomes (profits and wages) via expansion of output in one sector brings into being at the same time an expansion in demand and employment elsewhere. Treasury’s implicit position is that aggregate demand determines the numbers of jobs available. In addition aggregate demand is invariant to extra employment generated by job creation. In terms of their discussion of the government budgetary implications Treasury focus on the negative fiscal consequences of spending on subsidies. They do not consider possible savings on benefit payments and the additional taxes paid on the earnings of subsidised workers. They also do not explicitly acknowledge the possibility that subsidies not only lower costs for final output, causing shifts in the pattern of final consumption spending towards subsidised goods, they also lower costs of intermediates. As this flows through the input-output system, costs are reduced for everyone, potentially enhancing employment. These sketchy arguments, to be fleshed out in more detail below, are not intended to suggest that on empirical grounds that Treasury are incorrect in their conclusions. Rather they demonstrate that the conventional wisdom in 1984 looks only at a sub-set of the issues involved in considering deadweight costs and displacement issues. Furthermore, the 1984 focus concentrated on the negative effects. Potential positive effects are not discussed in detail.
1.3 Types of active labour market policies What are active labour market policies? Katz (1994, p. 259) defines active labour market policies as “measures targeted at the unemployed and disadvantaged workers with the intent of improving the functioning of the labour market”.1 Katz divides them into three types:
• Supply-side policies which provide education and training to raise the skills of the target groups.
• Demand-side policies which stimulate employment through direct job creation in the public sector or the subsidisation of private sector jobs, and;
See also OECD (1994, Part II, p. 100).
• Subsidies to labour market information and job placement which increase efficiency of matching job seekers to job slots. From the point of view of this paper, I divide active labour market policies in the following way:
• Publicly subsidised job search activity and employment counselling. (In New Zealand the New Zealand Employment Service provides free information and advice to job seekers).
• Labour market training for broad or narrowly defined groups who are unemployed or out of the labour force.
• Direct public sector job creation. • Subsidies for private sector employment creation and for self-employment. Each of these active labour market policies has the potential to cause some form of displacement (to be defined more rigorously below). It is intuitively plausible that different types of active labour market policies involve different forms of displacement. Thus displacement will also be discussed in terms of different types of programmes below. Active labour market policies are typically targeted at those who do not have jobs. Sometimes these will be those who appear as unemployed in the Household Labour Force Survey (HLFS), but more often in a strict sense they will have left the labour force in some way, shape or form non-participants. Typically, many active labour market policies may be more tightly targeted than simply at the unemployed or non-participants. They may be targeted on the basis of age, gender, education, duration of time on a welfare benefit, receipt of a certain type of welfare benefit and so on. Or they may be targeted on the basis of combinations of these characteristics. Active labour market policies may have at least three potential objectives:
• A more efficient allocation of resources. • A more equitable distribution of market incomes, and; • A stabilisation of the business cycle. The first two objectives are primarily microeconomic and the last macroeconomic. In New Zealand, current economic policy assigns a stabilisation role solely to monetary policy. Since, roughly speaking, changes in the inflation rate (on the assumption that expected inflation is roughly equal to last period’s inflation) will be induced by deviations of actual from potential output, eliminating output deviations delivers a stable inflation rate. At current low levels of inflation, this enables the Bank to meet its monetary policy target. There has been little impetus from either policy makers and politicians to use active labour market policy in a stabilisation role in New Zealand in the last decade.2 However, active labour market policies are more seriously considered in terms of their potential microeconomic contribution. In this debate the primary concern has been efficient resource allocation. Distributional considerations have also been important in providing a programme
Indeed, according to Grubb (1994, p. 199), only in Sweden of all of the OECD economies has active labour market spending consistently been adjusted counter-cyclically. Grubb reports that in Germany active labour market spending has been pro-cyclical. As a relatively fixed amount is allocated for spending on the unemployed as a whole in Germany, rising unemployment by leading to higher passive labour market spending, automatically crowds out active spending.
rationale. However distributional concerns have been rationalised in terms of quantities reducing the numbers of target groups unemployed. Their influence in raising incomes or compressing wage differentials when participants in programmes obtain work has been neglected.
1.4 Summary This paper focuses on the displacement effects of employment programmes. Displacement will be defined more formally below. A simple definition of displacement is any changes in employment elsewhere in the economy as a consequence of the programmes in addition to the direct job creation effect on those who pass through active labour market programmes. Normally considered to reduce employment elsewhere, it can be shown that such effects, in theory at least, may increase regular employment, depending on the ways the economy is assumed to respond. Often discussion of displacement has focused on the impact of marginal job subsidies, rather than the impact of other forms of active labour market policy. One question which I seek to answer is whether and how other forms of active labour market policy might result in job displacement. Displacement occurs when the government offers assistance targeted towards those it wishes to directly encourage into employment. While there may be an increase in employment for the targeted groups, it may be at the expense of jobs for others. Displacement may occur in a variety of ways, including the following:
• Directly, in terms of substitution of subsidised workers for unsubsidised workers within a given firm.
• Indirectly as firms with subsidised workers out-compete those who do not employ subsidised workers, or;
• Inter-temporally, as work is bought forward in time to take advantage of the subsidies. The project provides:
• A clear definition of displacement and the forms it may take within the context of the available economic models.
• A consideration of the empirical literature on displacement and an assessment of its reliability and relevance.
• A discussion of ways in which policies can be designed to minimise the effects of displacement. This would take into account factors such as scheme design and identifying the crucial parameters. The discussion of these schemes would be based on consideration of the available theoretical and empirical literature.
• Addressing the issue of the trade-off between objectives of the employment scheme and displacement, and;
• A discussion of ways of measuring and monitoring the extent of displacement in employment initiatives.
1.5 Approach My approach involves a detailed literature search in the appropriate Journal of Economic Literature categories and associated bibliographies to build up a complete body of literature on
active labour market policies and displacement in theory and in practice. There is a sizeable body of material, both theoretical and empirical, which deals with active labour market policies. Unfortunately there is no one accepted model in the field with which to analyse their impact. A variety of models are used in which active labour market policies are considered in a more or less ad hoc fashion. The literature also indicates a variety of not always consistent definitions of displacement. While displacement issues are regularly referred to, they are typically passed over in a brief discussion of several paragraphs as likely to be important in practice. A rigorous theoretical treatment is generally lacking. I will try and show below that there are good reasons for this avoidance behaviour. Below I provide a description and analysis of the various forms of displacement. This allows isolation of the crucial factors on which the extent of displacement depends. In this vein the project uses a partial equilibrium (Marshallian) microeconomic framework to analyse displacement. In addition, I consider a more general macroeconomic or general equilibrium approach which takes into account a wider range of economic interactions (see both Layard and Nickell 1980 and Calmfors and Skedinger 1995 for various macroeconomic approaches and Bertram 1988 for some previous discussion of both these broad approaches in a New Zealand context). Macroeconomic effects include the crowding out impact of government spending and the deadweight losses of taxes required to fund the programme. Set against this will be positive effects on output via the expansion of government spending and possible positive hysteresis effects which may arise. Finally, following Calmfors and Skedinger (1995) I am interested in the overall impact of active labour market policy on wage setting and hence equilibrium employment and unemployment rates. As indicated above, I also consider inter-temporal aspects of displacement. In addition (but related) to this is the issue of short-, medium- and long-run dimensions of displacement. I utilise both partial and more general equilibrium frameworks to examine whether they deliver similar conclusions regarding the impact of active labour market policies. If, as is likely, the net outcomes depends on parameters, our aim will be to specify the key ones. The study aims to draw upon available econometric estimates to get some idea of the size of displacement. If the approaches come to similar conclusions regarding the effects of active labour market policies, we will be able to put more faith in the resulting conclusions. In addition, for the same reasons, it will be of much interest to see if two different broad approaches come to the same empirical conclusions. The empirical conclusions may provide insights into how to design employment schemes to minimise the cost of displacement. What types of approaches and institutional structures have succeeded overseas in minimising displacement? What have the trade-offs been? What are the factors crucial to success and can they be isolated from broader economic considerations? The amount of robust and detailed literature in this area was found to be quite limited. A further feature of my approach will be an assessment of the relevance of various forms of overseas evidence to New Zealand, given structural differences in our economy. It should also be noted that minimising displacement per se is unlikely to be the aim of the Department of Labour’s active labour market programme. Rather it is minimising displacement subject to (at a minimum):
• The transactions costs of so doing, and; • The fact that there will be a trade-off between displacing existing workers and putting people into jobs from the front, back or middle of the economy’s job queue.
Clearly this part of the overall project will involve the consideration of considerable institutional detail regarding schemes employed overseas. The paper considers the issue of measuring and monitoring displacement within a transactions cost framework. It takes into consideration the costs and benefits of information given the range of available monitoring technologies. This part of the project will have to be specific about the various potential mechanisms for monitoring displacement, the trade-off between costs and accuracy of information, and consider the most appropriate portfolio approach for measurement and monitoring. Thus the paper aims to explicitly consider available systems and methods of monitoring and measuring displacement in existing employment programmes overseas where information is available. Obtaining all the potential information in this area has been constrained to a certain extent by the time-frame of the project.
2. DISPLACEMENT AND ITS FORMS
2.1 Defining displacement Net job creation from active labour market policies (N)3 is equal to gross job creation G less gross job destruction as a consequence of the policy (D) and deadweight losses (jobs that would have been created in the absence of the active labour market policy) (L):4 N = G - (L + D) In many cases with active labour market policies, gross job creation (the take-up for example on an employment subsidy, or the number of job seekers through the job search service who obtain a job) can be readily measured. In addition deadweight is reasonably readily measurable. However, gross job destruction is more difficult to measure. By rearranging the net job creation equation one obtains displacement as: D=G-N-L Several empirical papers in the area (to be examined in some detail below), assess the weakness of particular active labour market policies in terms of the ratio (L+D)/G. A ratio of one for this indicator indicates that the programme has simply redistributed jobs from one group to another. It is also conceivable - at least in theory - that the ratio could be negative. A negative ratio means that there has been crowding in of jobs rather than out. A negative ratio can arise for a variety of equilibrium and disequilibrium supply and demand side reasons. It is also worth noting that many studies present deadweight and displacement losses as a summed total rather than splitting them out.
Net job creation is also known as additionality in some studies (e.g. NERA 1995, p. ii).
Note a similar approach to this is employed by Fay (1996, p. 6). However, his definition of displacement is much narrower than is adopted here and he does not seem to acknowledge the theoretical possibility of negative displacement (“crowding in”) as a consequence of active labour market policy. See also NERA (1995, pp. 2-3), where this is perhaps chauvinistically described as “the standard UK additionality approach”.
The theoretical aim of an active labour market policy may be to maximise net job creation subject to some budget constraint. Alternatively, it may be to achieve a certain rate of net job creation at the lowest possible budget cost. I consider the consequences below if the maximand is not simply net job creation but takes into consideration distributive effects. In other words it may be the case the maximand is: N = G - (1-α)L - (1-β)D