EI, Economists and Unemployment

I note from a CP wire story  that Ottawa U economist David Gray is weighing in behind Harper’s argument that lowering eligibility for EI would be a “disaster”, and I suspect strongly he will not be the last to do so.      http://thechronicleherald.ca/Canada/1135526.html The received wisdom among mainstream neo liberal economists is that a “generous” EI program creates disincentives to work, and many apparently believe that is what the economic research demonstrates. I’ll give David Gray the benefit of the doubt, but I do think there is an ethical issue when economists over-state what the research has found in order to publicly oppose policy changes which would benefit the disadvantaged.

I looked into this issue fairly closely during the last major round of EI “reform” in the mid 1990s, when HRSDC commissioned a lot of academic research on the economic impacts of the program (which indeed did let some people collect 40 weeks of benefits for 10 weeks of work.) My major conclusion then was that most of the research in fact showed that EI “generosity” had only negligible impacts on unemployment.

I’ve pasted in below my 1995 CLC Research Paper.

THE ECONOMIC IMPACTS OF UNEMPLOYMENT INSURANCE — THE CASE FOR THE DEFENCE

1. Introduction

In recent key policy documents, the federal government has levelled serious charges against Canada’s Unemployment Insurance system. The system, we have been told, creates important “disincentives” to work which significantly raise the national unemployment rate. At the same time, the current system of UI financing — premiums paid by employers and workers — is held to destroy jobs. These alleged negative impacts of Unemployment Insurance on the economy have, in turn, been used by the government to justify cutbacks to benefit entitlements and durations for unemployed workers. Perversely, attacking the benefits of the unemployed has been held out as part of the solution to high unemployment.

The purpose of this paper is to draw on the recent academic literature in order to marshall a case for the defense. Many will be surprised to learn that the consensus of experts is that the negative economic impacts of Unemployment Insurance were greatly exaggerated in earlier studies, and that any negative economic impacts of the program are, in overall terms, minimal. No program as large and complex as UI can fail to have a complex range of impacts upon the economy and the labour market. But the overall evidence does not support the key charges which have been levelled in recent major policy documents. Further, research has pointed to significant positive economic impacts of the program.

The view that significant numbers of working people organize their lives around the UI system in order to maximize their UI benefits while minimizing their weeks of work is widespread, but it is almost completely wrong. It is wrong mainly because it rests on a caricature of the determinants of human behaviour which is deeply embedded in the assumptions of the economics profession, but which has no relevance to the real world in which working people live. This has been increasingly understood by economists, but editorial writers and too many politicians accept the old caricatures.

Unemployment Insurance is by no means perfect, but it is an essential income security program for working people in a high unemployment economy which also yields important positive economic impacts.

The argument that UI “causes” unemployment is one key thread in the more general argument for labour market deregulation which was detailed and criticized in a recent C.L.C. Research Paper “Jobs, Jobs, Jobs —The Liberal Plan and its Consequences for Workers”. This paper argues that the key driving force between rising unemployment has been inadequate demand in the economy, not the way in which the labour market operates.

2. The Claims of the Prosecution

The current federal government released three key policy papers in late 1994 with the common title Agenda: Jobs and Growth. These documents pointed the finger to UI as a major cause of unemployment, and called for UI “reform” (ie cuts to benefit durations and entitlements) as part of a strategy to create jobs.

The “Purple Book” of the Department of Finance (A New Framework for Economic Policy) levelled three key charges against UI.

First, it was argued that “disincentives” have been an important cause of Canada’s rising rate of unemployment. UI was held to “harbour features that can discourage the active search for work” (p.21) and it was argued that “the rules of the programs have encouraged chronic, repeat use … thus explaining some of the increase in the core rate of joblessness.” (p.52)

Second, it was argued that “the program has unquestionably fostered a dependency in certain regions and industries .. inhibit(ing) the adjustment of both people and businesses to more productive opportunities and has thus perpetuated job and income stagnation.” (p.52)

Finally, it was argued that “the payroll tax that finances UI has its own perverse effect on job creation which has contributed to Canada’s rising core rate of unemployment.” (p.52) The argument was that the taxes used to finance UI “drive an increasingly thick ‘wedge’ between what the potential employer is prepared to pay and what the employee wants to receive.” (p.53) By increasing employer wage costs and reducing employee take home pay, UI is alleged to lower employer demand for workers (ie. jobs), as well as incentives to work.

These three key arguments are developed at much greater length in the Supplementary Paper to the “Green Book” — From Unemployment Insurance to Employment Insurance — released by the Department of Human Resources Development.

Again, the argument is made that UI has increased unemployment by creating “persistent disincentives” to work (p.12) It is argued that UI increases the duration of unemployment (the length of a particular spell of unemployment) (1) by reducing the incentive of unemployed workers to search for work and (2) by raising the “reservation wage” of unemployed workers, that is the wage they would be prepared to accept in a new job. The fundamental view — to be found in most introductory textbooks written by “free market” economists — is that unemployment is, in significant part, the result of voluntary, individual decisions to choose UI subsidized leisure over paid work. These “disincentive” effects are held to increase the national unemployment rate by between 1% and 2%. The appropriate policy response, it is argued, is to reduce the generosity of UI benefits, both in terms of duration and in terms of the proportion of the previous wage which is replaced by UI during periods of unemployment.

Again, the argument is made that the UI system “retards adjustment”. The argument is that patterns of UI are not consistent with a productive and dynamic economy because they establish patterns of “dependency” for certain individuals, regions and industries which hold back necessary change. As in the first argument, the basic view is that UI has a major negative influence on the behaviour of workers. Because UI subsidizes unemployment, it undermines the willingness of individual workers to make necessary adjustments to structural changes in the economy, such as moving to another part of the country, or reducing their wage expectations, or improving their skills. Part of the solution is argued to be reduced benefits for frequent users of the system and higher thresholds to access the system in terms of weeks of work. Another part of the solution is held to be to reallocate UI spending away from “passive” spending on income support towards “active” programs (which promote and speed adjustment), such as training and assistance to workers.

Finally, this paper repeats the argument that high payroll taxes, such as UI premiums, have raised the unemployment rate by reducing employer incentives to hire and workers’ incentives to accept jobs. (p.12.)

It is important to underscore the fact that the “disincentives” argument levelled against UI in the Purple Paper of the Department of Finance has both a “micro” and a “macro” economic dimension. At the “micro” or individual level, the argument is that UI subsidizes periods of unemployment and thus results in some voluntary unemployment.

At the “macro” or economy wide level, it is argued that UI has contributed to a higher rate of unemployment by raising the so called “Non Accelerating Inflation Rate of Unemployment” or NAIRU. The Purple Paper argues that Canada’s core rate of unemployment is “at least” 8%, and that unemployment cannot fall below this level without triggering inflation. (p.20)

The link drawn by many economists between UI and the level of the NAIRU is as follows. UI insulates workers to some degree from the negative income impacts of unemployment, and thus makes unemployed workers more choosey when it comes to accepting or rejecting job offers. UI thus counters downward pressures on wages in a high unemployment economy, preventing the economy from generating the needed extra jobs which would result if only wages were to fall to a “market clearing” level. Further, because UI undermines the disciplining effect of unemployment on wages, central banks must squeeze the economy even at relatively high rates of unemployment in order to keep inflation in check. (The Purple Paper footnotes Technical Report Number 50 of the Bank of Canada — The NAIRU in Canada: Concepts, Determinants and Estimates — which puts forward this argument in detail.)

3. “Disincentive Effects” of UI:

The Case for the Defence

An important recent paper by Dalhousie University economist Lars Osberg has drawn attention to some key weaknesses of the economics textbook model in which workers choose between work (seen as a “disutility” or bad thing) and leisure (seen as a “utility” or good thing.) In fact, common sense and the psychological literature tell us that, for the great majority of working people, paid work is desired for reasons over and above the wage received, and that there is an enormous and profound difference between unemployment and leisure. Far from being sought after or chosen, unemployment is usually a cause of great distress and anxiety. And, far from being avoided, steady employment is frantically sought by the great majority of unemployed workers.

Osberg reminds us that the textbook model of workers choosing between work and leisure simply assumes that work is available when an unemployed worker runs out of benefits, and simply assumes that periods of unemployment have no consequences for individuals in terms of future earnings, future promotion prospects, future pension benefits etc. Both assumptions are at complete odds with the realities of a high unemployment economy. Today, jobs are very hard, if not impossible, to find, and even very low pay, temporary jobs are attractive because of the opportunity they provide to accumulate experience and skills and thus to gain a toehold in the labour market.

It could be added that the “worker choice” model of unemployment ignores the key administrative rules of UI — that a claim can only be established by a lay-off — an employer decision — and that, to maintain a claim, an unemployed worker must actively seek work and must accept job offers within certain parameters.

Osberg also makes the important point that any UI “disincentive” effects are likely to be much more pronounced and relevant at very low rates of unemployment. The simple fact that jobs are hard to find when the rate of unemployment is high means two things. First, individuals are much less likely to reject available jobs when they are unemployed. Second, even if one unemployed person does turn down an available job, there will be no shortage of other unemployed workers willing to take that job, and thus no net effect on the unemployment rate.

Osberg and other prominent labour market economists such as Miles Corak of Statistics Canada and Shelley Phipps of Dalhousie University have also drawn academic attention to the seemingly obvious point that unemployment is a function not just of the willingness of workers to accept and keep jobs, but also of the willingness of employers to offer jobs to unemployed workers. In other words, unemployment is a function of not just the “supply side” of the labour market — the decisions of workers — but also of the “demand side” — the decisions of employers. Surprisingly, economists have devoted enormous attention to the first side of the question, and very little to the second. (In part, this may reflect the fact that there is a lot of data available on the labour market behaviour of individuals, and very little data on the hiring and layoff decisions of employers.)

In an important recent article, Does Unemployment Insurance Increase Unemployment?, (Canadian Business Economics. Spring. 1993) Shelley Phipps develops the point that any assumed individual choice made by individuals between paid work and UI subsidized leisure is “demand constrained.” In other words, not enough weeks of work are generally made available by employers to satisfy the desires of workers. Phipps cites survey evidence showing that, even in the mid 1980s, when the unemployment rate was relatively low, the majority of people who experienced unemployment wanted additional weeks of work, and that a very high proportion of those in work (eg. 41% of adult men) wanted more hours of work than they were able to obtain. (One might add that the Statistics Canada Labour Force Survey shows, not only that close to 10% of the labour force are actively seeking work today, but also that 1 in 3 part-timers — another 6% of the labour force — report that they are working part-time only because they cannot find more hours of work.)

Phipps concludes that “mounting evidence about the importance of quantity constraints on individual labour supply suggests that we should focus on the lack of jobs rather than on the lack of incentives to take paid work when we seek to understand continuing high rates of unemployment in Canada.” (p.47.)

Phipps also surveys all recent Canadian studies to refute the argument that UI premiums paid by workers — which lower take-home pay — reduce the willingness of workers to accept jobs. The demand of Canadian men for paid work has been found to be “highly inelastic” — that is to say, men will work at a lower wage rather than not work at all. It used to be thought that women would work less if wages were lower, but this picture has changed very substantially in recent years. Only women from higher income families will choose in significant numbers to work less hours if wages are lower. Thus research does not confirm the Department of Finance view that UI premiums paid by workers have any significant impact on incentives to work.

In an extremely important overview of the economic literature on Unemployment Insurance published by the C.D. Howe Institute , Statistics Canada economist Miles Corak generally endorses the arguments of Osberg and Phipps that the generosity of the UI system will have only very marginal impacts on the level of unemployment in a high unemployment economy. In such an economy, the vast majority of unemployment is not voluntary — the result of someone “choosing” between work and UI subsidized leisure. Rather the great majority of unemployment is involuntary, the result of a shortage of jobs and hours of work relative to the demand for work.

Corak’s survey of the research undermines the myth of the 10/40 worker — the worker who allegedly works for 10 weeks in order to collect 40 weeks of benefits. In fact, the duration of only a tiny proportion of jobs which end with a layoff are influenced by the generosity of a UI program. Perhaps 2% of all jobs do end because workers become eligible for UI benefits, but even here the effect is to slightly reduce the length of jobs which would have been temporary in any case. And the job comes to an end because of an employer decision to lay off rather than because of a worker decision to quit. Research has confirmed that some small firms in seasonal industries in rural areas (mainly in Atlantic Canada) do — out of “community solidarity” — alter their hiring and layoff decisions in order to qualify slightly more workers for UI. And provincial governments have been known to create temporary jobs in order to “stamp up” workers. But the impact of this behaviour on overall levels of unemployment is trivial and insignificant.

One study that bears out this general conclusion even though it found that changes in entrance requirements do have some impacts on unemployment was conducted for the Department of Human Resources Development (HRD) by David Green and Craig Riddell. (“Qualifying for Unemployment Insurance”). They attempted to assess the impacts of a legislative delay in 1990 which meant that the entrance requirement in high unemployment/maximum entitlement regions temporarily rose from 10 to 14 weeks for most of the year. The authors found that this led to a very small — 0.4% drop — in the unemployment rate in very high unemployment regions because some firms extended the length of temporary jobs in order to qualify workers for benefits. A similar study by Christofides and McKenna, also commissioned by HRD, confirmed the conclusion that small impacts of changes in entrance requirements on the unemployment rate result mainly from the decisions of firms, not workers. Yet another study for HRD by David Green and Timothy Sargent (“Unemployment Insurance and Employment Durations: Seasonal and Non Seasonal Jobs”) finds that “as many as” 2% of all seasonal jobs end (suspiciously) at the point at which one extra week of work generates no increase in UI benefits.

It is notable that these studies for the Department of Human Resources Development — which, at worst, find very small impacts in very specific circumstances — are those which are drawn upon to support sweeping statements in government policy documents around the alleged “work disincentive effects” of UI.

The research as summarized by Corak in his C.D. Howe study finds that there is no relationship at all between the generosity of UI benefits (defined as the portion of the previous wage replaced by UI benefits) and the length of unemployment spells for men. Also, there is no relationship between benefit generosity and the likelihood of being a repeat claimant. There is a relationship between the length of a spell of unemployment and eligibility for UI benefits, but this does not mean that we have to accept the “worker choice” model. Rather, eligibility for UI may be used to mount a more effective search for an appropriate job, as argued below.

The view that UI creates “moral hazards” which result in workers playing the system in order to “choose” UI subsidized leisure over work was a major factor behind recent cuts. Bill C-113, passed in April, 1993, disentitled from UI benefits workers who quit their jobs without just cause. (Previously, a substantial penalty had been imposed but workers who quit still qualified for benefits after a delay.) The Bill also reduced the benefit rate from 60% to 57% of insurable earnings.

A study for HRD by Stephen Jones of McMaster University (“Effects of Benefit Rate Reduction and Changes in Entitlement on Unemployment, Job Search Behaviour and New Job Quality”) found that Bill C-113 had “negligible effects on individual’s re-employment chances and on job duration.” In fact, holding economic factors constant, the change was the opposite of what was predicted. More people found work within six months of being laid off from their jobs before benefits were cut than was the case afterwards, and there was absolutely no difference found in terms of the intensity with which unemployed workers looked for jobs.

In yet another important study for HRD, Miles Corak (“Unemployment Insurance, Temporary Layoffs and Recall Expectations”) thoroughly documents the extremely important point that the great majority of UI claims are driven by employer decisions to lay off workers, rather than by worker decisions to quit jobs. (Of course, workers who quit without cause are now ineligible to collect benefits.) This study also shows that the theory that workers collecting UI benefits may “choose” to remain unemployed rather than to actively search for work and to accept available jobs has little basis.

Corak tells us that, in 1986 through 1988, 80% of laid off workers had a legitimate expectation of being recalled to work by their former employer, since this was what was reported by the employer on the Record of Employment form. 60% of all workers on claim were, in fact, recalled by their former employer. In short, the majority of use of UI was for the classic purpose of providing income support to laid off workers during a period of temporary, involuntary unemployment. These findings help explain the (not terribly surprising) finding of others that unemployed workers tend to search harder for jobs as the exhaustion of benefits is approached. While the “search model” of economists would have us believe that workers simply “choose” to be idle until the money threatens to run out, Corak reminds us that it is quite reasonable for most workers to expect in the early period of a claim that they will be recalled to work by their former employer. The fundamental difference between recall layoffs and permanent lay-offs is well-known to administrators of the UI program, but it is a distinction which is lost on too many economists.

Corak also notes in this study that a large proportion of “frequent claimants” — 40% of those with 5 or more claims over 12 years — had 3 or fewer employers. Another study by Corak co-authored by Pyper (“Workers, Firms and Unemployment Insurance” Statistics Canada. 1995) comprehensively documents the employer driven dimension of Unemployment Insurance. As has become increasingly understood, frequent use of the UI system by some workers is largely a function of the frequent layoff decisions of some firms. Corak and Pyper report that 1 in 7 jobs in the late 1980s (1986-89) were in firms which laid off workers every year. While these firms were heavily, and unsurprisingly, concentrated in seasonal industries — ie the primary resource industries, tourism and construction — Corak and Pyper also found that there are substantial differences in the behaviour of firms within the same industry.

Recognition of the fact that frequent use of UI is heavily associated with employer layoff decisions, and only weakly related to worker “choice”, has led to proposals to “experience rate” employers which frequently resort to layoffs. In short, the argument for change shifts from the “workers as abusers of UI” model to the “firm as abuser of UI” model. But there is very little evidence that the UI system has a significant influence on employer decisions to lay off workers, as opposed to keeping them on the payroll through a temporary downturn. A Betcherman and Leckie study for HRD (“Employer Responses to UI Experience Rating: Evidence from Canadian and American Establishments”) compared employer layoff decisions in Ontario and in US border states, where employers’ UI premiums are experience rated at least in part. They found that, holding other conditions as constant as possible, UI experience rating has minimal influence on employer layoff decisions. They also add the important point that it is far from clear that it would be desirable to structure the UI system so as to “punish” firms which layoff. If “experience rating” did result in fewer layoffs, it would almost certainly have another effect — it would make firms much more reluctant to hire new workers in an upturn. Again, the more general conclusion is that UI has only a marginal impact on the functioning of the labour market.

4. Macro Economic Studies

It was once all but received wisdom among Canadian economists that — as is still argued by the Department of Finance and Human Resources Development — the major reforms to UI in 1971 which significantly increased UI eligibility and benefits also raised Canada’s core unemployment rate. Indeed studies making this link became virtually a cottage industry in the 1970s. [The results of such studies, themselves suspect, have been repeatedly cited, even though UI benefits and entitlement have been greatly cut since 1971.]

Studies tried to link some measure of UI “generosity” to the unemployment rate, holding everything else constant. However, as the Evaluation Branch of HRD has itself noted, benefit levels are extremely complex and varied while “academic economists have paid little attention to how UI actually works. … once the UI system is modelled more carefully, the estimates of the benefit effect become insignificant.” (“HRD Evaluation Brief: UI and the Unemployment Rate.”)

While most academic studies have used a single measure of generosity (such as average UI benefits as a proportion of the average wage) there are a range of possible ways to measure UI “generosity”. And they can change depending upon overall economic conditions, irrespective of program changes. For example, the system becomes more “generous” as the unemployment rate rises, since eligibility for benefits is broadened and benefits are extended as the local unemployment rate rises. Also, “generosity” is affected by changes in the composition of the unemployed population. In deep recessions, “generosity ” increases as more relatively well-paid, normally full time, full year workers are affected by unemployment, thus raising the average benefit rate. The administrative complexities of UI have also not been adequately recognized in many of the academic studies, and this is also true of many of the “micro” studies summarized above.

Lars Osberg has also argued that most macro-economic studies not only simplistically link “generosity” by some measure to the unemployment rate, but also have failed to take into account the fact that any macro-economic impacts of UI could be expected to be quite different when the economy is expanding, as opposed to when it is in recession. It has also been pointed out that there are numerous fallacies involved in thinking that any “disincentive” effects of UI add up to a major aggregate impact on unemployment. Many have pointed out, as noted above, that the effect in a high unemployment economy may be to simply reshuffle unemployment among different groups of workers. Others have pointed out, as argued at length below, that UI plays an important stabilizing role in the economy, and thus saves jobs and reduces unemployment during downturns.

In any case, studies of the effects of the UI system on the national unemployment rate are highly inconclusive. Of the 7 studies of the effect of UI generosity on the unemployment rate completed since 1987, 3 found no effect at all, while 2 of the 4 which did find some effect found an impact of 1% or less on the national unemployment rate. Corak’s study for the C.D. Howe concludes that “macro-level research conducted since the mid 1980s does not appear to offer conclusive evidence that changes in the UI program since 1970 have increased the aggregate unemployment rate or altered its dynamics” (p.147.) Osberg cites a survey of 14 studies by Myatt who found that “more evenly divided results could not be imagined.” As noted above, earlier studies of the impact of UI “generosity” on the unemployment rate are suspect in the sense that the system has become markedly less generous while unemployment has spiralled upwards. Indeed, only half of all unemployed workers now qualify for UI benefits — massively below the levels of the 1970s.

To conclude this section, a comprehensive overview of the academic research does not support the argument that UI has significant negative economic effects because of work “disincentives.”

5. Positive Economic Impacts

of Unemployment Insurance

A focus on negative impacts of UI has tended to distract attention from the many positive, economic and social benefits of the UI program. Yet these are substantial, more than outweighing such small costs as have been documented.

[A] The Income Stabilization Role of UI

Unemployment Insurance was developed on the basis of the view that most unemployment is involuntary, the result of a failure of the market economy and of governments to ensure that sufficient jobs are available to meet the need and desire of workers for paid employment. Keynesian economists have pointed to the need for UI to stabilize individual worker incomes and the economy as a whole during cyclical downturns. More recently, economists have often pointed to the need for programs to assist workers who are the victims of structural change in the economy. For example, it has been recognized that trade liberalization results in job losses as well as gains, and that there is a social obligation to assist the victims of policies designed to increase overall economic efficiency. Income support is clearly one important part of overall adjustment policies.

The Purple Book and the Green Book make the argument that UI has become less relevant than before as unemployment has become more “structural” and less “cyclical.” But a worker who is involuntarily unemployed for “structural” reasons still needs income support to assist her adjustment to changing circumstances. It may be the case that such workers need retraining and adjustment assistance, but it does not follow that the classic income maintenance role of UI has become outmoded. If anything, it has become more relevant than ever.

Further, it is highly questionable if unemployment has indeed become less “cyclical” and more “structural.” There was, indeed, a significant increase in long term unemployment in Canada over the past decade, but this period also saw by far the two worst recessions the country has experienced since the 1930s. Leading economists such as Pierre Fortin have argued that unemployment has not risen because of some inability or unwillingness on the part of workers to satisfactorily adapt to structural change, but rather because the Canadian economy has been operating at well below its potential for most of the past decade, and continues to do so today. Mario Seccareccia of Ottawa University points out that there is no basis for the argument that more unemployment is “structural” in the sense that a higher proportion of unemployed workers lack the skills needed in available jobs. The fundamental problem is that not enough jobs are available.

Unemployment Insurance is a social insurance program which effectively pools the risks of being unemployed, and transfers income from the employed to the unemployed. It has sometimes been argued that UI violates insurance principles, since the “bad risks” — workers who are frequently unemployed — do not pay higher premiums than workers who are rarely, if ever, unemployed. And it is true that the incidence of unemployment is not wholly random and unpredictable, but rather follows some regular patterns. However, few if any workers are immune from the threat of unemployment, and the great majority of working people value UI for its classic insurance role of providing some degree of income security in return for the payment of premiums.

Such insurance would be highly unlikely to be offered by private insurance companies on a broad basis if UI were to be abolished, since the risks involved would be so large and unpredictable. Even if “bad risks” such as seasonal workers were not covered, events like the two recessions of the past 15 years would have likely bankrupted any private unemployment insurance schemes which were not highly selective. As has been pointed out, selling unemployment insurance in an economy such as the one we live in today would be like selling fire insurance in a crowded wooden city with no fire department.

Unemployment Insurance is, by design, a social insurance program which includes the great majority of workers, though part-timers working less than 15 hours per week are excluded, and entrance requirements do exclude significant numbers of temporary, contract and seasonal workers. By design, the program provides benefits to insured workers who are involuntarily unemployed and are actively seeking work. Because involuntary unemployment affects a large proportion of the labour force at some time in each year — between about 25% and 33% in recent years — large numbers of people draw benefits at some time. (Though the proportion of the unemployed who qualify for benefits has been falling sharply in the 1990s.)

UI has sometimes been criticized for transferring income to people who are not poor or even live in higher income households. This follows from the fact that the major purpose of the program is to stabilize the incomes of workers who are unemployed, irrespective of their level of income. Well paid workers are, in fact, treated less generously by the program in that there is a maximum benefit (so high income workers with employment incomes above maximum insurable earnings receive a lower proportion of their previous income than do lower income workers). Also, very high income recipients have a proportion of their UI benefits taxed back. Still, UI benefits do go, by design, to middle and higher income earners as well as to low income earners.

Fred Lazar (“UI as a Redistributive Scheme and Fiscal Stabilizer” in Unemployment Insurance: How to Make it Work. C.D. Howe Institute) shows that UI has been an important source of income support for lower income individuals and households. In recent years, regular UI benefits have amounted to between 4% and 5% of total wage and salary income. In the late 1980s (1986-89), about $3 in every $4 of benefits (74%) went to workers earning less than $20,000 per year, and 1 in 3 low income workers (earning between $10,000 and $20,000) received UI benefits. HRD reports that, in 1991, 80% of frequent UI claimants lived in families with total incomes of less than $50,000 (which is about average family income) and that the majority of frequent claimants had family incomes of less than $30,000. (“From Unemployment Insurance to Employment Insurance” p.47.) 18% of frequent claimants had family incomes below $15,000, and UI benefits contributed 43% of the total family income of this group. Clearly, UI is an important source of income for lower income workers and households.

A study for HRD by Erksoy, Osberg and Phipps (“The Distributional Implications of Unemployment Insurance”) argues that cuts to Unemployment Insurance significantly increase income inequality, and that those directly affected by cuts (particularly those who lose benefits completely because of raised entrance requirements) can suffer very large income losses. Another study for HRD by Martin Browning (“Income and Living Standards During Unemployment”) similarly concludes that “UI benefit rates significantly influence net personal incomes of UI recipients. UI benefits form the major portion of income for those who receive it.”

Research has confirmed that there is a significant overlap between UI and the social assistance system, with a large and growing group of low income Canadians moving from low wage jobs, to UI, to social assistance. A study for HRD found that one third of welfare recipients in BC in 1986 also had a UI claim in that year. (Barrett, Doiron, Green and Riddell, “The Interaction of Unemployment Insurance and Social Assistance.”) UI exhaustees who cannot find work turn to social assistance when family incomes are low, and when their financial assets have been exhausted. This degree of interaction between UI and social assistance and low wage, temporary work is, however, an enormously under-researched area.

The central point to be made is that UI plays an extremely important role in stabilizing and maintaining the income of unemployed workers and that this role is particularly important for low income workers and their families.

[B] The Macro Economic Stabilization Role of UI

In addition to stabilizing incomes for workers affected by involuntary unemployment, the UI system has performed an important economic function in terms of stabilizing the economy as a whole. It used to be, and remains the case to a lesser degree, that the system is an “automatic stabilizer.” By cushioning falls in total wage and salary income during recessions and thus maintaining purchasing power in the economy, UI saves some jobs in downturns. This cushioning effect was particularly large when all of the cost of UI benefits above a low unemployment rate were borne directly by the government. Now that the UI fund has been mandated to be self financing, the cushioning effect in downturns remains, but the need to run a surplus in the UI account during an upturn implies that recoveries will also be cushioned. Hassan Bougrine and Mario Seccareccia (“Unemployment Compensation and Unemployment — Demand Side Effects for Post War Canada.” University of Ottawa Department of Economics Working Paper #9419E) argue that “empirical results indicate that, contrary to common assertions by mainstream economists and conservative policy makers, UI benefits paid out to eligible workers are not a burden on the economy and … unemployment compensation can actually reduce unemployment through its counter cyclical influence.”

HRD recently commissioned two major macro economic studies of UI’s stabilization effects, by Peter Dungan and by Ernie Stokes, which drew on two different large-scale econometric models of the Canadian economy (the FOCUS model maintained at the University of Toronto, and the WEFA model maintained by the WEFA group.) The FOCUS model showed that the UI program prevented between 11% and 14% of potential job losses in the last two recessions, while the WEFA model showed that UI prevented 7% of potential job loses in 1990. Both studies concluded that UI continues to play an important macro-economic stabilization function. This view has, indeed, been acknowledged by the Department of Finance.

[C] UI and Effective Job Search

While right-wing economists have argued that UI “subsidizes” unemployment and the “choice” of leisure over work, other economists have argued that eligibility for UI benefits allows workers to conduct more effective job searches. This in turn contributes to a more productive and efficient economy. The reasoning is that if UI allows workers to search for jobs for a period of time rather than be forced to take the first job that comes along, there will be a better fit between the skills of the unemployed worker and the job, reducing unnecessary turnover in the labour market. Also, UI benefits will enable an unemployed worker to conduct a more efficient job search, perhaps looking for new opportunities outside her or his immediate community and occupation.

While this area has been greatly under-researched, there is evidence that collecting UI benefits does improve an unemployed workers chances of finding a good job. Pierre-Yves Cremieux and colleagues conducted a study for HRD (“Unemployment Insurance and Job Search Productivity”) which found a, positive effect between UI benefits and the wage level of the job eventually found by an unemployed worker. Holding everything else equal, workers ineligible for UI earned 7% to 9% less in the new job they took following a period of unemployment than did UI claimants with 50 weeks of benefits. Most of the beneficial effect on wages came with 30 to 40 weeks of benefits. This study also found that workers drawing UI benefits searched just as hard for jobs as did unemployed workers who were ineligible for benefits.

[D] Incentives to Employment

As argued above, the major emphasis in much of the academic economics literature on UI is on “disincentives” to work. Intellectually consistent economists have long recognized that, on the same assumptions about human behaviour, UI also creates important incentives to work. Since it is necessary to work in order to collect benefits, UI may induce some people who would otherwise remain outside the labour force to accept temporary or seasonal jobs. Research has shown some minor effects of UI on labour force participation rates, mainly for women in higher income households.

The more important point, which has been under-researched, is the impact of UI on levels of employment in inherently seasonal industries. If the entrance requirement for UI in Atlantic Canada was raised to, say, 20 weeks, many workers who qualify for benefits after working in the tourism industry (which has a 2 to 3 month season) would not qualify for benefits. Assuming that wages could not be raised enough by employers to make up for the shortfall, some seasonal workers would likely be forced to leave the communities where they can now at least find seasonal work. As a result, at least some enterprises would have to shut down with wider spin-off impacts on the regional economy. Given a high national unemployment rate and much higher unemployment rates in some regions, it is far from clear that implicit “subsidies” to seasonal and temporary jobs through the UI system are to be condemned as economically damaging and inefficient. Indeed, it is somewhat bizarre that the government proposes to redirect “savings” from UI cuts to subsidies to employers to create jobs, since this may well be a less efficient way of achieving job creation goals. Certainly more research needs to be done on the positive, job creating role of UI in seasonal industries, as pointed to by the Task Force on Seasonal Work and UI set up by the Minister of Human Resource Development.

It is worth adding that the role of UI in “retarding adjustment” in high unemployment regions has been greatly exaggerated. This point is frequently repeated, but few, if any, empirical studies have attempted to determine to what extent, if any, UI reduces worker migration from community to community and province to province.

The studies which do exist on mobility show that Canadian workers are, in fact, highly mobile and do move in response to changing employment opportunities. And the UI system assists mobility by providing the income security which is essential for an unemployed worker to travel in search of a job. A study for HRD by Lin [”Interprovincial Labour Mobility in Canada”] points out that UI recipients are much more likely to move between provinces than social assistance recipients, and are more mobile than the labour force as a whole.

[E] UI and Worker Skills

While little research has been done on the issue, it has been suggested that UI may have positive effects on the skills base of the labour force, and thus on productivity at the firm and national level. The existence of UI means that firms can lay off workers during temporary downturns with the expectation that those workers will return to work when business conditions improve. In the absence of a UI system, or if UI benefits were to be significantly cut, workers who were temporarily laid off might leave the community more quickly, particularly in areas with a narrow economic base. Employers would, in that case, have to hire new workers to a greater extent in a recovery, raising training costs and depriving them of the benefits of the experience of workers who had left. Further, firms might be even more reluctant to invest in the training of workers than is already the case if layoffs were to pose a greater threat in terms of losing investments in worker skills.

6. Payroll Taxes

As noted above, government policy documents have pointed to payroll taxes — of which UI premiums are the most important — as a significant cause of higher unemployment. Finance Minister Martin and HRD Minister Axworthy have often referred to payroll taxes as “taxes on jobs” and the former has even described payroll taxes as “a cancer on jobs”. (Globe and Mail. March 7, 1995.) The basic argument is the same as that frequently and loudly made by employers — payroll taxes increase labour costs and thus result in fewer jobs.

As noted by Livio Di Matteo and Michael Shannon (“Payroll Taxation in Canada: An Overview”. Canadian Business Economics. Summer, 1995.) the business view is not widely shared by economists for two reasons. First, economists generally recognize that the business view that higher employer premiums for UI and other programs are borne by them is far too simple, and indeed wrong in most cases. Second, most economists recognize that the economy wide impact of changes in payroll taxes will be different from the impact on any single employer.

With respect to the first issue — who really pays payroll taxes levied on employers — economists generally agree that employers pay only a modest share of the tax in the medium to long run. Indeed, this has even been recognized by the Department of Finance. A footnote in the Purple Book (p.22) heavily qualifies the “tax on jobs” argument as follows:

“The initial impact of a payroll tax is to raise the employer’s total labour cost, usually with no offsetting benefit. This diminishes willingness to hire and raises the unemployment rate, other things being equal. Studies indicate that in the longer term, provided the tax is unchanged, wages will eventually readjust with most of the ultimate tax effect falling on the employee. Thus a constant payroll tax has a diminishing effect over time on the employer’s reluctance to hire.” (Purple Book p.22)

This footnote reflects the widespread consensus among economists that it is not very relevant in the long run if a payroll tax is paid by the employer or by the employee. In the long run, the employer share of the tax will be paid for by workers in the form of lower wages than would otherwise have been the case, particularly in highly competitive markets.

Consistent with this view, the Organization for Economic Co-operation and Development (OECD) reported in the recent Jobs Study that cross country studies show there is no systematic link between the level of payroll taxes as a percent of wages, and labour market performance expressed either in terms of employment creation or unemployment rates. (p.244) Germany, for example, has a relatively low unemployment rate, but payroll taxes amount to some 15% of GDP, double the level in Canada. The OECD also finds that there is a strong negative relation between the wage share in GDP and the rate of employer social security contributions. In other words, as employer premiums go up, wages tend to go down to compensate.

Bev Dahlby of the University of Calgary (“Payroll Taxes” in A. Maslove (Ed.) Business Taxation in Ontario. University of Toronto Press,. 1993) surveyed Canadian and international studies of the “shifting” of payroll taxes, and concluded that, in the long run, 80% of the burden of the employer portion of such taxes falls on workers. A recent study for HRD by Beach, Lin and Picot (“The Employer Payroll Tax and Its Effects on the Demand for Labour”) finds that the evidence is “more consistent with full shifting of the tax back onto labour than with no or only partial shifting.” (p.38)

Most empirical studies do show that all things being equal (an important point considered below) increases in employer payroll taxes may have modest, negative impacts on employment in the short run, mainly because the “shifting” process takes time. An employer will likely recoup all or most of a payroll tax increase by holding the line on future wage increases, or by hiring new workers at slightly lower wages, but there will be a short run impact on wage costs and thus there may be a short run negative effect on jobs. This effect is more severe for low wage workers, since higher UI or other premiums cannot be passed back onto workers who are paid at or near the minimum wage.

Dahlby found that an increase in employer payroll taxes amounting to 1% of the total wage bill may reduce employment by up to 0.2% in the short-run in Canada, while the recent study by Di Matteo and Shannon cited above finds that such an increase will bring about a short-term fall in employment of 0.3%. Note that the employer UI premium would have to increase by about 33% for the wage bill to rise by 1%, since the effective UI payroll tax on employers is 3% of the total wage bill. (This is lower than the premium rate since premiums are only paid up to a maximum earnings level, and since premiums are not paid on benefit costs.) The research conclusion is, then, that a very large premium increase will have a very modest impact on jobs, in the short run.

The usual all things being equal qualification on these studies is important since the link between payroll taxes and jobs at the economy wide level is much more complicated than studies focussed on wage costs suggest. UI is a program which transfers income from the employed to the unemployed. Thus, if benefits were cut in order to finance a cut in premiums, employer wage costs would fall with positive effects on jobs, but the incomes of many households would fall, with negative effects on jobs. It is true that the process of collecting UI premiums and distributing UI benefits does have economic impacts. But it is unlikely that a program which, in essence, simply transfers purchasing power within the economy will have very large net impacts on employment.

The government could, of course, reduce UI premiums by raising money from other taxes rather than by cutting benefits. But a research paper done by the Applied Research Branch of HRD (“Payroll Taxes and Employment”. July, 1994.) concluded that “because of the effects on aggregate demand, reducing the payroll tax burden by shifting it to personal income or consumer taxes would not necessarily lead to more job creation over the short to medium term.” (p.21). In other words, all taxes have some negative impacts on jobs, and payroll taxes are no worse than other taxes.

It should be noted that the “shifting” argument works in reverse, so reductions in employers’ UI premiums would result, first of all, in an increase in profits. Higher profits do not automatically lead to more jobs. Rather more jobs might result if employers invested increased profits in expansion, or lowered their prices in order to win more orders. Over time, some of the premium reduction would likely be recovered by workers in the form of higher wages.

Today, the overall level of payroll taxes in Canada is low compared to other industrial countries. Statistics Canada reports that payroll taxes in total amounted to 6.3% of GDP in 1992, less than 7.8% in US, and massively below the level in many European countries (eg. 19.6% of GDP in France; 15.5% in Germany. (“Recent Trends in Payroll Taxes.” Canadian Economic Observer. September, 1995.) Statistics Canada calculates that the effective employer payroll tax rate for UI is now 3%, admittedly up from about 2% in the mid to late 1980s, but still a modest levy. Overall, the case that employer UI premiums are a major problem in terms of job creation does not stand up to close scrutiny.

It is worth making the perhaps rather obvious point that the cost to employers — and workers — of UI could and should be reduced by bringing down the national unemployment rate. There is no good case to be made for payroll taxes as such, which are indeed a drain on the real income of working people and impose some unnecessary costs on employers. But the problem is the unemployment crisis, not the system of financing unemployment benefits.

Premiums are an appropriate way of financing an Unemployment Insurance program for three reasons. First and foremost, UI benefits go only to people who have qualified through participation in the labour force. It would be technically possible, but inappropriate, to finance all UI benefits from taxes paid by the whole population. Second, the payment of premiums for benefits means that the costs of the program are paid by beneficiaries, creating a legitimate sense of entitlement and removing the welfare stigma. (The government would add that premium payments are an essential ingredient of a self-financing program, and that this is a source of fiscal discipline since beneficiaries and premium payers are the same group.) Third, it has been argued with some reason that social insurance programs replace in part private insurance programs that would otherwise be jointly purchased by many employers and employees. If there were no UI program, many employers would certainly be pressed at the bargaining table to help fund unemployment plans, or to significantly increase wages to help compensate for increased insecurity of employment.

All that said, the financing of UI is an important issue. The issue is, however, more one of the scale and timing of premium changes than premium financing as such or the overall level of premiums.

Employer anger over UI costs is not hard to understand when it is considered that the UI premium rate was raised by more than half in the recent recession. (The employee premium was raised from $1.95 to $3.00 per $100 of insurable earnings between 1989 and 1992 with two intervening steps; the employer contribution is 1.4 times the employee rate). This major increase took place despite an explicit government promise not to raise premiums after the passage of Bill C-21, so the increases took employers by surprise. Given that, as noted above, there is a short run impact on employer costs from higher premiums, and given that these increases were unexpected, there were undoubtedly negative impacts on profits and on jobs. Premium increases took place at the worst possible time — in the depths of a major recession — in order to prevent too large a deficit accumulation in the UI fund which is now meant to be self-financing.

It would, as argued by the government, be preferable to build up a large surplus in the UI fund in periods of expansion in order to limit premium increases in the next recession. Unfortunately, the “tax on jobs” rhetoric stands in the way of recognition that it is more important to stabilize premiums than it is to lower the overall level of premiums.

To conclude, the “payroll taxes as killers of jobs” argument is not well supported by the economic research. There are some short run negative impacts from premium increases, particularly if they are unanticipated, but even this cost must be set against the alternative of benefit cuts or increases in other taxes. Premiums are an appropriate way of financing the UI system which does not impose significant economic costs. Indeed, all the evidence shows that the cost of the system is almost entirely borne by workers in the long run.

7. Conclusions

The purpose of this paper was to draw extensively upon recent academic research in order to defend the UI system against some of the key charges levelled against it in recent government policy documents. UI is an extremely large and complex program, and it is far from perfect. There are reasonable proposals for change and reform which should be considered. But the key government charges against UI — that it creates major disincentives to work, that it retards necessary adjustments in the economy, and that it kills jobs — have been enormously exaggerated. And the government has generally failed to recognize the major social and economic benefits of a good UI program.

The views of UI held by many key policy makers and editorialists which underpin current proposals for major cuts simply cannot be sustained by reference to the expert research which has been undertaken, and their proposals should be judged accordingly. Attacking income support for the unemployed is not the solution to high unemployment. But job creation is certainly the answer to the high and rising costs of Unemployment Insurance.

AJ:nh*opeiu-225

November 9, 1995

V:\2-Operations\03-Social-and-Economic-Research\20302 – Reports and Papers\03 – Research Papers\003-The-Economic-Impacts-of-UI-EN.wpd