VIEWS: 47 PAGES: 40 POSTED ON: 8/9/2011
Unemployment Some Questions to think about How many people here are in their last year of school? How many are looking for full time work? For those who are, do you care what the unemployment rate is right now? Why? How many people think low unemployment is correlated with low wages in the economy? How many people think low unemployment is correlated with high wages? What in the long-term? If you graduate in a recession (when unemployment is high) or a boom (when unemployment is low), would that affect your career 10 years down the road? What does economic theory suggest? What does the data suggest? Say you are working full time and your firm goes bankrupt and you are displaced. Clearly you will be unemployed, at least for a little while you look for another job at another firm. But would you expect your displacement to affect your career 5 years later? Ten years later? Stick around and I will tell you why know is a pretty good time to be looking for work. On Friday, the latest unemployment figures were released by Statistics Canada. Does anyone know what the current unemployment rate is? Statistics Canada announced that the unemployment rate had reached a 30-year low of 6.4%. Recent Unemployment Statistics The unemployment rate differs markedly by province too: UR by province for February 2006: Canada 6.4% Newfoundland and Labrador 15.1% Prince Edward Island 10.9% Noval Scotial 8.4% New Brunswick 9.4% Quebec 8.2% Ontario 6.2% Manitoba 4.4% Saskatchewan 5.3% Alberta 3.1% British Columbia 4.8% Alberta’s unemployment is 3.1%! Different ways that Statistics Canada Measures Unemployment (and Employment) Unemployment is obviously a huge topic in labour economics. What it means to be unemployed is not easy to define. But it is a situation for which most of us do not want to find ourselves in. The amount of unemployed is often used to measure the pulse of the economy and as an indicator of hardship in society. Let’s first examine the official way to measure unemployment. The labour force (LF) is defined as the set of people who are either employed (E) or are unemployed but would like to be employed (U): LF = E + U Statistics Canada classifies someone as unemployed if he or she is on a temporary layoff waiting to be recalled by their previous employer or he or she is without a job and actively searched for work in the previous month. So most people who are unemployed are not working but looking for work. Each month, Statistics Canada conducts the Labour Force Survey, which is used primarily for calculating the extent of unemployment in Canada. The agency randomly selects and visits or calls about 35,000 households across Canada. Among the many responses they collect from each household member is whether or not they are working and if not, whether they are looking for work. The unemployment rate is simply calculated as the fraction of people in the labour force unemployed: Unemployment rate = U / LF One of the need things with statistics is that you only need a random sample of all Canadians to get an unbiased estimate of the unemployment rate in all of Canada (you don’t have to check with everyone all the time). With 35,000 households and more than 100,000 individuals surveyed each month, we get a pretty accurate estimate of the true population level of unemployment that month. There are multiple measure of unemployment, and it is not obvious which one provides the best description of the state of the economy or the degree of hardship in the labour market. The unemployment rate reflects the proportion of a group that, at a point in time, actively want to work but are not employed. However, it does not necessarily provide an accurate reflection of the economic hardship that members of a group are suffering. Why? First, individuals who are not actively searching for work, including those who searched unsuccessfully and then gave up, are not counted among the unemployed. Individuals who were looking for work but stopped actively looking because they figured there’s nothing out there are not classified as unemployed by Statistics Canada, and are known as discouraged workers. It is difficult to identify discouraged workers. For this reason, some economists prefer to focus on the employment rate, which is the fraction of people in the population (P) employed (E): Employment Rate = E / P One minus the employment rate will give the fraction of people either 1) unemployed, 2) discouraged workers, or 3) not working, not looking for work, and not discouraged (e.g. people who at a point in time do not want to work). Second, the unemployment rate does not tell us anything about the earnings levels of those who are employed, and does not distinguish between part time and full time work. An individual who is working part-time but looking for full-time work is not counted as unemployed (Why? – look at the definition again). To account for this, some economists pay closer attention to the labour force statistics that come out each month and differentiate between part time and full time work. Journalists will often note what fraction of the employment growth that occurred over the last month was due to full-time employment or part-time. For example, from the CBC on Friday, March 10, 2006: “Canada's unemployment rate fell two-tenths of a percentage point to 6.4 per cent in February, matching the 30-year low set last November, Statistics Canada said Friday. The country churned out 24,700 new jobs last month, matching analysts' expectations. The job creation was due entirely to growth in part-time work. More than 56,000 part- time jobs were added, while more than 31,000 full-time positions disappeared.” Sometimes journalists will view individual’s unemployed taking on part-time work as having to ‘settle’. Like discouraged workers, they are no longer classified as unemployed, but the outcome may not be reason to celebrate entirely. However, whether it is appropriate to view part-time employment like this depends obviously on each individual’s circumstances. My sense is that most part-time employment is likely driven by people who want to work part-time and not full time. Third, a substantial fraction of the unemployed come from families in which other earners are present – for example, many unemployed are teenagers who are not the primary source of their family’s support. Another case is if one spouse is unemployed while another is not. The unemployed spouse is not the single source of income for the family. Fourth, length of unemployment matters a great deal in determining the extent of an individual’s economic hardship. You would probably consider the circumstances of an individual who is unemployed for 1 month differently than for an individual who is unemployed for more than 1 year. Note that unemployment is measured at one point in time (with cross-section data). The unemployment rate does not distinguish someone just starting to look for work from someone who has been looking for work for a long time. However, the Labour Force Survey does ask people how long they have been looking, and we can use that information to construct alternative measures of unemployment. For example, some additional measure of unemployment measured by statistics Canada are: R1 unemployment rate for those unemployed one year or more R2 unemployment rate for those unemployed 3 months or more R5 Official Rate plus official discouraged workers R5 Official Rate plus official discouraged workers R7 official rate plus those working part time who say they want to work full time R4 is the official rate. R1 and R2 focus on groups for whom unemployment may represent particular economic hardship. Discouraged workers in R5 are measured in the labour force survey by identifying persons who want work, are available for work, but are not seeking work for ‘personal’ or ‘economic’ reasons. In 2004 (free access data is only available up to this year), the official unemployment rate including all 15 year olds and older not working but looking for work was 7.2%. R1 was 0.7%, and R2 was 2.2%.R5 was 7.3% Most respondents are unemployed for less than 3 months (however, this may mask people who would work but feel they have no chance at finding a job so do not look for work). This is also deceiving because we do not yet know how long these people will be unemployed: the question only asks how long have you been unemployed SO FAR. (see below). Changes in these measures are highly correlated. This suggests that, while no one measure of unemployment can accurately describe the degree of economic hardship in the economy, changes or trends in the unemployment rate can adequately capture whether hardship is increasing or decreasing. You can easily look up these rates, for any year, for different age groups, by gender, by province at: http://www.chass.utoronto.ca/datalib/codebooks/cstdli/cansim.htm Click on E-stat And make your way to CANSIM Table 282-0086 Or try using this url: http://estat.statcan.ca/cgi- win/CNSMCGI.EXE?regtkt=&C2Sub=&ARRAYID=2820086&C2DB=&VEC=&LANG=E& SDDSLOC=http%3A%2F%2Fdissemination.statcan.ca%2Fenglish%2Fsdds%2F*.htm& ROOTDIR=ESTAT%2F&RESULTTEMPLATE=ESTAT%2FCII_PICK&ARRAY_PICK=1 &SDDSID=&SDDSDESC= Stocks and Flows of Unemployment and the Labour Market Although net monthly changes in unemployment and employment tend to be relatively small, so that the overall totals of the employed and unemployed are quite stable, these net figurers mask the important fact that Canadian labour markets are very dynamic. Even in stable economic times, there are large monthly flows into and out of employment. For example, flows which roughly cancel in net terms but which in gross terms amount to a high degree of labour market mobility. The Labour Force Survey asks every surveyed individual questions on whether they are in the labour force, whether they are working, no working, looking for work, and also what they were doing the month before. Steve Jones from McMaster calculated the average flows into and out of 1) Being not in the Labour Force, 2) Being Employed, and 3) Being Unemployed. (The textbook on page 515 shows the same information). E stands for employment, U employment and N not in the labour force. EU stands for going from Employed to Unemployed in one month, etc… PEU stands for the conditional probability of going from Employed to Unemployed (conditional on being Employed) in one month, etc… The most striking aspect of the economy-wide figures is the size of the flows, with about 200,000 individuals moving from any one state to any other in each month (over this period). On a mean base on 11.1 million employed, this results in a monthly probability of moving from employed to unemployed of 2 percent. On average, 20 percent of people who are unemployed find employment in the next month. Young are much more likely to be unemployed. Women are much more likely to leave unemployment from leaving the labour force. Somewhat more formally, we can show that if labour markets are roughly in balance, with the flows into and out of unemployment and the labour force equal, the unemployment rate for any group depends upon the various labour market flows in the following manner: UR F ( PEN , PNE , PUN , PNU , PEU , PUE ) In this equation, F means “a function of” PEN is the fraction of Employed who move to ‘Not in the Labour Force’ (who leave the labour force). PNE is the fraction of those not in the labour force who become employed PUN is the fraction of unemployed who move to ‘Not in the Labour Force’ (who leave the labour force). Etc… A plus sign over a variable in the equation means that an increase in that variable will increase the unemployment rate, while a minus sign means that an increase in the variable will decrease the unemployment rate. You should think carefully why a variable is listed here as a plus or a minus. The equation thus asserts, for example, that, other things being equal, increases in the proportions of individuals who voluntarily or involuntarily leave their jobs and become unemployed or leave the labour force will increase a group’s unemployment rate. The equation and an appreciation of labour market stocks and flows make clear that social concern over any given level of unemployment should focus on both the INCIDENCE of unemployment (the fraction of people in a group who become unemployed) and the DURATION of unemployment. Society is probably more concerned if small groups of individuals are unemployed for long periods of time than if many individuals rapidly pass through unemployment status. When the proportions of people flowing in and out of each state are constant, the unemployment rate can also be expressed exactly as the product of the average incidence rate and average duration rate: Unemployment rate = Average Incidence Rate * Average Duration Rate For example, in this case of a steady state labour market, when 2 percent of people each month become unemployed and the average duration spent unemployed is 5 months, the unemployment rate will be 10%. To see this more clearly, if 5 percent of the population each month become unemployed for only one month, in steady state, the unemployment rate will be 5%. We can also go backwards: If the unemployment rate is 10% and 2 percent of the population each month become unemployed, then the average duration rate, in steady state, must be 5 months. Our data suggests that while many people do flow quickly through the unemployed state, at any given time, a number of individuals experiencing prolonged spells of unemployment characterize those found in the stock of the unemployed. How do Labour Economists View Unemployment?: the distinction between voluntary and involuntary unemployment Think back to when we were studying the theory of labour supply. Individuals were maximizing utility by choosing between how much to work and how much not to work, given the wage offer. In a perfectly competitive labour market, what is the wage equal to? It is equal to a worker’s marginal product. Our theory predicts that any individual who is equally productive to a firm than another individual could offer to work for one penny less than the wage that other individual is working for. The two would get into a bidding war until the wage is lowered to a level that one of them prefers to look elsewhere, or not to work at all. So in a perfectly competitive labour market, unemployment is, for the most part, voluntary. • If there is always someone out there willing to pay 50 cents if you shine their shoes, can you ever be involuntarily unemployed? • Orley Ashenfelter’s view: “A worker is involuntarily unemployed if he or she has identical preferences and skills as other workers and yet is unable to find the number of hours work that others have both chosen and managed to find”. Ashenfelter, 1978. This is required reading and I’m going to put it up on the web page. You’ll like it. Involuntary unemployment Other workers with same You are offered w2 : your preferences and skills utility would be higher if offered w1 you were offered w 1 40w 1 40w 2 0 40 0 0 0 40 Or, an individual accepts a job (e.g. paying w2) while there exists another job where another equally productive worker is being paid w1. Although both are employed, one worker is paid more than another. Why is unemployment viewed as a social concern? Outcomes unequal: some able to choose optimal leisure/consumption bundle, others not “Why don’t you accept the work available at lower wage rates?” Response 1: “Why don’t all other similarly situated workers have to do the same?” Response 2: I would work for a lower wage, but now employer seems to be willing to lower the wage, even though I would do just as good of a job as the person currently working. These 2 responses portray the involuntary unemployed. The big question here is why would these situations arise in the first place? Why would some workers with the same skills be offered one wage, and others a different wage? Why don’t employers lower the wage when similar workers are willing to work for less? First, let me just say briefly, there is empirical evidence that wages DO differ across industries and firms, in which similarly skilled individuals are paid different wages. There is also empirical evidence that employers DO NOT tend to lower nominal wages. The rate at which wages increase may differ over time, but very few firms actually lower (nominal) wages. Economists have thought about this long and hard to explain reasons for involuntary unemployment. There are many theories. I will present three of the main theories here (my favourites and probably the three most popular ones): Efficiency Wages Efficiency wage theory focuses on the effect that wages have on incentives, motivation, and worker productivity. Suppose, for example, I hire one of you for $20 an hour to enter data in for me. I don’t really know how hard you’re working, because I cannot look over your shoulder to see how fast you’re typing. But $20 is a pretty good rate, and you may feel happy with that – happy enough that you are willing to work at a good pace because if I ever thought you were slacking off, you would be out of a good paying job. But now suppose another one of you comes up to me and says you’d be willing to work for $10 an hour. That would save me a lot of money if you work as hard as the other guy. But at $10, you have less at stake if I catch you slacking off. At $10 you may work less. So I may be willing to pay someone $20 an hour as a way of encouraging effort, even though other people are willing to work for less but I will not hire them (and so they are involuntarily unemployed). Let’s look at this more formally. A firm is trying to choose how much labour to hire to maximize profits. The production function that determines how much output, Q is produced, is a function of labour hired (L) and the amount of effort labour puts into their work (e): Q F (eL ) Now, before, the production function was just a function of how much labour is hired. The wage paid to workers only affected the cost of hiring. But support the wage you offer also affects the level of effort each worker decides to commit. So effort is a function of the wage: e (w) e Profit for the firm is: profit F ( eL) wL (1) profit F ( e( w) L) wL Before, the firm would just choose how much labour to hire, given the wage. The wage was taken as given because, in a perfectly competitive labour market, firms would offer the lowest wage possible and workers would accept the highest wage possible and the equilibrium wage was what the amount that reflected the worker’s marginal product: if profit is G ( L) , and the equilibrium wage is w ' ( L) . wL G Now, a firm has to take into account the impact of the wage on effort, and its implications for profit. A firm may want to offer a higher wage. The first order conditions from equation (1) above are: eF ' (eL) w 0 e ' ( w) LF ' (eL) 0 L We can rearrange these two equations to imply that: w e 1 L e' ( w) L or w e' ( w) 1 e So, when profits are maximized, the elasticity of effort with respect to the wage, should be one. The intuition (as discussed more in the text) is straightforward. An increase in the wage by 1 percent will increase labour costs by 1 percent, no matter what happens to output. So the firm should increase the wage as long as the gain from extra effort per worker is greater than the extra cost per worker. At some point, the extra cost is not worth it. This occurs when the condition above holds. Some economists believe the persistence of unemployment is the result from the widespread payment of above-market wages. Further, persistently different wage rates paid to quanlitatively similar workers in different industries are the hypothesized result of efficiency wage considerations. One of the most important implications of efficiency wages relates to their effects on productivity. Those firms that raise wages above what other firms are willing to pay for similar workers should be the ones that 1) stand to gain the most from enhancing worker reliability, or 2) find it most difficult to properly motivate their workers through output based pay or supervision. Social Norms on the Worker Side Even if some firms pay higher than market wages, why can’t individuals who would like to work for these better paying firms not settle for jobs at other paying firms that pay only market wages (where wages can adjust)? Some economists believe that the failure of unemployed workers to flock to low-wage jobs derives from their sense of status (their relative standing in society). Individuals may prefer unemployment in a good job to employment in an inferior one, at least for a while. This is not involuntary unemployment. Social Norms on the Worker Side Unemployed workers may not be able to undercut wages of employed (bid the wage down). Worker moral and motivation of existing workers when the wage is reduced would drop dramatically. The disruption cost of having to encourage motivation, or hire new workers at a lower wage is too high to be willing to lower wages. Some economists have suggested that inflation helps ‘grease the wheels’ of the economy by allowing for real wage decreases even though firms are reluctant to lower nominal wages. Thus, inflation may actually help reduce unemployment not because of the traditional Phillips Curve relationship (through an increase in money demand) but because it gives firms more flexibility to lower real wages. Job Search Search models relax the assumption that workers always know wages offered at all other firms. Frictional unemployment is when jobs are available and unemployed individuals want them, but a match hasn’t occurred yet. Finding a job involves search. Allowing for the more realistic assumption that it takes time for workers to find a job and for firms to find someone who wants to work alters the perfectly competitive model in dramatic ways. Set-up of Search Model: Workers: • Many workers, equally productive • New job offers come about to both unemployed and employed workers at a constant rate (e.g. every 2 months) • A worker accepts another job if the wage offer from the new one is above her current wage and stays otherwise Firms: • Many firms that want to maximize profits • all workers in the firm are paid the same wage • In this set-up where it takes time for workers to find another job, firm can change it’s wage without losing all workers • An unemployed individual accepts a job if the wage offer is above her reservation wage Why is perfectly competitive wage not the equilibrium wage for all firms in this model? • With all firms charging competitive wage and making zero profit, one firm could lower its wage • Workers currently at that firm will move if they get another offer, but this takes time • Unemployed workers receiving an offer at this firm but not at another firm will prefer to work there if wage above reservation w. • This firm then can make a profit! If one firm can make a profit, the other firms will want to as well. What can they do? • Firms face a trade-off: • The lower the wage offered, the bigger the difference between productivity and wage paid, but the higher the quit rate • What is the highest wage offered? The lowest? In equilibrium, all firms making same profit, but charging different wages Distribution of workers by wages paid by firm P = VMP Reservation wage w r p Note how a slight change in assumptions leads to very different result • Firms offer different wages to workers of equal productivity • Because workers take time to find alternative jobs, firms can pay less than a worker’s marginal product, make a profit, and retain the worker until that worker finds a firm willing to pay them more. • The distribution of wages varies between the reservation wage and the perfectly competitive wage (w = VMP) • As the rate of obtaining a new offer gets smaller and smaller, the equilibrium approaches the competitive one (once all firms offering a job at the same time to workers, firm profit goes to zero) Summary of Search Model • With market friction, workers must search employers for a job and don’t get offers from everyone at once • This causes workers to stay at a firm while searching for a job that pays a higher wage • Each employer chooses a wage given this search behavior • The labour force available to an employer is constantly changing • Trade-off between offering a higher wage and having a more stable labour force (fewer quits) and lower profit Some Important Implications of the Search Model • Good jobs and bad jobs out there (for similar workers) Otherwise identical individuals will wind up receiving different wages. Two unemployed individuals with the same skill level could choose the same reservation wage and have the expected post-unemployment wage. However, the wages they actually wind up with will depend upon pure luck. • Because workers only switch firms for higher wages, the longer workers are employed the more likely they are at a higher paying firm • There may be ways of jumping the que (connections) • Discrimination may lead to certain groups more likely to end up in lower paying jobs If the cost to an individual being unemployed were to fall, the person should be led to increase his or her reservation wage (that is, the person would become more choosy about the offers deemed to be acceptable). A higher reservation wage, of course, would increase both the expected duration of unemployment and the expected post- unemployment wage. One important influence on the cost of being unemployed, and hence on the reservation wages of unemployed workers is the presence and generosity of governmental unemployment insurance programs. Unemployment and Search • To switch jobs, workers must ‘search’ by definition: those not working do not find jobs immediately. • Costs to switching jobs may lead to unemployed waiting longer before accepting offer (trying to find best job) Two predictions of the search model, in contrast to the perfectly competitive model of the labour market. 1) You are displaced (lose your job from a firm going bankrupt or having to layoff workers because of slower demand for its product). What are the long-run (say, 5 or 10 years out) implications to your wages under the perfectly competitive model? Under the search model? In the search model, workers are moving from one job (or firm) to the next in search of a higher paying wage. Workers that lose their high paying job must start over, searching while unemployed and willing to work at any firm that offers higher than the reservation wage. If you had a top paying job (compared to other similar workers) and were laid off because of bad luck, you may experience prolonged lower wages than if you were not laid off. Let’s look at the evidence. Jacobson Lalonde, and Sullivan (AER 93) exploit employer-employee data to examine the wage dynamics from unexpected job loss. The Burdett-Moretensen model predicts job loss lower wages for those that lose their jobs, as they end up trying to climb the wage ladder again from the bottom. In a perfectly competitive setting, workers should be able to obtain similar paying jobs at other firms, and we shouldn’t observe a long run impact. There are other reasons to expect a prolonged impact from job loss, especially in the presence of specific firm capital. But the analysis is informative in that mild and temporary impact from job loss shocks would be evidence against the Burdett- Moretensen model. JLS have quarterly data over 15 years from government administrative records in Pennsylvania. Sample is: workers at same firm for 6 years between 1974-80. ‘Treated’ Group: workers that left firm in quarter where at least 30% of firm’s labor force left (e.g. mass layoff or firm closure). This is an attempt to identify exogenous job loss. ‘Control’ Group: The rest of the sample • In recession, more individuals searching (e.g. from lower output demand) • This may lead to longer search times and makes individuals more likely to ‘settle’ for lower paid jobs Could there be long-run effects from graduating during a recession or boom? • In a competitive market, shouldn’t matter • Initial conditions (e.g. lower labour demand) may increase search time and lower initial reservation wage • But once recession over, should ‘bounce-back’ • Workers with equal productivity should be paid same Possible reasons why graduating in recession might matter • Jobs harder to get during recessions may offer more opportunities for human capital accumulation – productivity grows faster • Jobs harder to get during recessions may make search for higher paying jobs easier • Firms may use initial job as ‘signal’ for a workers’ productivity (stigma model) Let’s look at the data what happens, on average, when university students graduate during a recession? • University Student Information System (USIS) tracks university enrollment status and graduate status for universities in Canada (1976 – 2000) • USIS data matched to administrative records tracking earnings data from 1982 - 2000 Data matched to youth unemployment rate conditions by province Regression Set-up: log w p ,c,exp p c exp exp unemp _ 1524 p ,c unemp _ current p ,c,exp p ,c,exp e log w p ,c,exp average log wages for workers from province p grad. year c, and with years since grad. exp unemp _ 1524 p ,c youth unemployment rate at time of graduating, g, from prov. p, unemp _ current p ,c ,exp current unemployment rate (set to zero when exp 0) The Effects of Unemployment Insurance Benefits Virtually every advanced economy offers its workers who have lost jobs some form of unemployment compensation, although these systems vary widely in their structure and generosity. In Canada, the system is operated by the federal government and, since 1996, has been called Employment Insurance instead of Unemployment Insurance. In contrast, in the United States the unemployment insurance system is actually composed of individual state systems. You can go up on Human Resource and Development Canada (HRDC)’s website and discover how it works: To qualify for EI, generally you must have worked about 12 to 20 consecutive weeks in the same job, which translates to roughly 12 to 20 weeks at 35 hours per week. Someone on EI can receive transfers from ranging from 14 to 45 weeks, depending on the unemployment rate in the region: if the unemployment is high, you are eligible for EI for longer. The basic payment is 55% of your previous earnings, although this amount is capped at $413 per week ($21,476). So if you have an annualized salary more than about $40,000, you will get less than 55%. The way it works is when an individual is laid off or dismissed, he obtains a copy of the Record of Employment (ROE) which shows evidence of the length of time working, and the reason that the individual left the firm. With this, the individual fills out the application, including SIN, bank information, details of dismissal, and salary, and sends it off to HRDC. If everything is in order, HRDC starts sending checks within about a month. You can’t just go on EI if you stop work. You have to convince HRDC that you were laid off or dismissed. If you voluntarily quit or you are fired, generally you are not eligible for EI. Also, while on EI, if you do find work, you’re supposed to let HRDC know, otherwise they might send you a check you’re not supposed to have. Payments generally stop after the number of weeks you’re eligible for. If you still haven’t found a job by then, you’re in trouble, because welfare might be the only remaining option for non-labor income. We can go through the effects of EI using our standard labor supply model. Suppose the wage rate is $500 per week. Suppose EI pays out 55% of annualized earning, so that’s $275 per week. To qualify for the program, an individual must work at least 10 weeks, and wait for 2 weeks. This was a similar set up to the UI program before 1994. Assume an individual can receive UI for up to 40 weeks. In analyzing EI over just 52 weeks, for every hour worked more than 10, that’s one less hour available to be on EI. This is a awkward assumption that the textbook makes also. It does work well for seasonal workers that have steady employment for at least part of the year. But in general, perhaps a more realistic assumption is to assume the benefit amount (annualized), stays the same. Note that in this graphical analysis, we set it up as though an individual automatically qualifies for EI if they work more than 10 weeks. This is not a realistic assumption. Also note this initial analysis ignores the fact that EI is paid for may employer and employee contributions. The textbook generally does not discuss how contributions alter this discussion, but we will look at that in just a bit. An individual maximizes utility by choosing the number of weeks to work. After working 10 weeks, and waiting 2 weeks, an individual is eligible to receive EI. The wage rate is 500. For every hour worked after 10, the individual gives up 1 hour of unemployment at .55(500) = 275 per hour. For someone working, what do you suppose this program will do to labor supply? The way we’ve specified the program here, the implicit wage after introducing the EI program is lower than before for those already working more than 10 hours. So what direction is the substitution effect? It leads to decreasing the number of weeks worked because the relative price of leisure has fallen. What about the income effect for those already working more than 10 hours? EI provides more income, and so the income effect is to reduce work if leisure is a normal good. Thus, BOTH the income and substitution effects lead to reducing work among those already working. Here is an example where we can predict the response in labor supply from a change in the budget constraint, but note that this is only because of our assumption that the individual is limited to looking at just these 52 weeks. For someone not working, the EI program could actually increase the number of hours worked. The logic is simple: someone not working would be able to receive a lot more income if they just work 10 hour more, because not only do they gain from the additional income, they also then qualify for EI. But this need not be the case for everyone. We could easily draw the indifference curves for individuals not working still being better off not working after the program. This analysis supposes the individual automatically qualifies for EI if they stop working which, of course, is incorrect, unless an individual is able to co-ordinate the reason for dismissal with the employer. One reason for lowering the limit of EI in 1994 was that many seasonal workers (e.g. fisherman, tree planters, …) were using EI to prop up their salaries while not really looking for work in between periods. But most people are able to receive UI only if they lose their job from unpredictable reasons. EI can’t make anyone worse off unless we take into the costs of the program because the same budget constraint as before is available, plus more if the new portion of the budget constraint, that requires going on EI, is considered. In ECO200 you examined how insurance can make risk averse individuals better off because a loss in income generates a larger fall in utility relative to the gain from the same increase in income. EI is like other insurance plans, except everyone must participate in EI when they work, whether they are better or worse off from doing so. Let’s compare the gain or loss in EXPECTED utility from the program for a particular individual. Suppose the probability of Katie losing her job is 2% over a 52 week period. If Katie loses her job, she is paid $6,425 for that year. I’m keeping things simple here by assuming a lump sum payment that does not depend on how much Katie works, and that Katie either works or loses her job for the whole year. Katie’s utility is C*L, and her wage rate if 500 per week, so her budget constraint is C+500L = 500*52. Assume she has no control over whether she loses her job. Things would be way more complicated if how much she worked before losing her job affected ho much her income is when she loses her job. Without EI, Katie maximizes her utility when MRS = w. This is when C*=13,000 and L*=26. Her utility at these points is 338000. But note, what is Katie’s utility if she loses her job? 0, because she would have no income to consume. So her expected utility is 331,240. With EI, Katies faces a different budget constraint. Because EI contributions are 10%, her real take home wage is $450 per week, not $500. Going through the same maximization problem under this new constraint, Katie’s utility if she doesn’t lose her job is now lonely $304,200, but her utility if she does lose her job is much higher (than 0), because EI pays her $6435, plus she doesn’t work that year, so her utility would actually be higher than if she worked. But since her chances of losing her job are very small, Katie’s expected utility is 304,808, lower than that if she wasn’t participating in EI. Individual’s can expect to gain from EI if their expected utility under the program is higher than the expected utility without it. Notice I say ‘expect to gain’. If I’m, on average, better off without EI, but chances have it I lose my job, I could turn out to be much worse off without the program. What are some things that will make me more likely to prefer the program? In general the higher the probability of losing one’s job, the higher the expected gain from UI. This creates incentives to ‘manipulate’ that probability, especially if there are few costs with working again. Of course, the lower the payout while unemployed, the lower the expected gain from the program, and the greater the contribution, the lower the gain. Once the program is set up, even if an individual’s expected utility is lower under the program, say because his probability of losing their job is low, he still has an incentive to try and receive EI. Suppose this individual has a zero percent chance of losing their job. He is worse off because the contribution rates lower how much leisure and consumption he can purchase. Note that if he did lose his job, he would be worse off without the program. If he could choose how much of EI to collect (how much not to work), He would work less: the way I’ve drawn the new budget constraint, the substitution effect will lead to less work because the wage rate, with contributions, has gone down. An insurance program is actuarially fair if the total expected payout equals the total expected revenues of the program. Under an actuarially fair program, if the probabilities of losing a job are different for the population, but the contribution rates are generally the same, then the program will end up transferring resources from those with lower probability of losing their job to those with high probability. This is how our EI system works, because, in many cases, workers’ contributions are not equal to their expected payment (some are higher, some are lower). If I had a choice to opt out of EI, I probably would, even though, if I did lose my job, I’d be worse off than before. If the government allowed it, I could probably find an insurance company who is willing to insure me for loss of job for a lot less than 6% of my pay. It’s also interesting to note that the EI program is not even actuarially fair: the government receives more revenues from the program each year than it receives. Those revenues are used for other things (like paying down the debt, or supporting other programs, or keeping taxes low). Thus, when the size of contributions are taken into account, we realize that a large part of the EI program is a redistribution program – not everyone gains from it, plus it creates incentives for people to ‘maniuplate’ unemployment. That said, for those that do unpredictably lose their job, it’s helps a lot. What does the job search model predict from increasing unemployment insurance benefits? Why? Empirical Evidence on the effects of Unemployment Insurance and incentives to encourage employment. Between 1984 and 1985 the state of Illinois conducted a claimant bonus experiment to test whether providing cash bonuses to unemployment insurance recipients who found a new job ‘quickly’ would be an effective way to reduce their durations of unemployment without affecting their post unemployment wages. The idea was that the promise of a cash bonus for rapid reemployment would cause UI recipients to increase the fraction of time they spent searching for work. UI recipients in the experiment were randomly assigned to two groups. The first served as a control group and received regular benefits. Members of the second group were promised an additional cash bonus of $500 if they found a full-time job within 11 weeks and held that job for at least 4 months Given that individuals were randomly assigned to the two groups, one would expect the two groups to exhibit, on average, roughly equivalent durations of unemployment and post-unemployment wages in the absence of any ‘bonus effect’. The incentive created by the $500 bonus, which was actually paid to only 570 out of 4,186 UI claimants assigned to the experiment, reduced state regular benefits paid to the randomly selected treatment group by an average of $158, and reduced average weeks of insured unemployment by more than one week (over the full benefit year), compared with the randomly selected control group. These reductions in average benefit payments and weeks of unemployment were achieved over all 4186 eligible claimants whether or not they agreed to participate or acted on the incentive. If the take-up rate in the claimant experiment had been 100 percent, and UI benefit reductions had remained unchanged, then the benefit cost ratio of the experiment would have fallen to 1.26 from 2.32. Thus, even with a 100 percent take-up rate, the claimant experiment would have reduced UI benefit costs by considerably more than the bonus costs incurred. Would take-up rates change in an actual program, and if so, how? Earnings of members of the treatment groups after they became reemployed did not differ from the control. Perhaps the introduction of the program would lead to many more claiming unemployment insurance to begin with just so they can qualify for the $500 bonus. Canada’s Earnings Supplement Project (ESP) The funding of the Social Research and Demonstration Corporation’s Earnings Supplement Project (ESP) by HRSDC is another important development in EI’s evaluation framework. This project began as a demonstration project to test the effectiveness of a financial incentive in encouraging re-employment among unemployed workers. Between March 1995 and June 1996 ESP enrolled over 11,000 claimants from nine local offices across Canada in a randomized trial. Half of the claimants who volunteered for this project were randomly assigned to the program group and offered the financial incentive, and half were assigned to the control group and thus not offered the incentive. ESP participants applying for regular benefits could receive an earnings supplement if, within a specific period of time, they took a new full-time job that paid less than the job from which they had been laid off. ESP would then make up 75 per cent of the earnings loss for up to two years. The earnings supplement was offered to two separate groups of claimants. The first was a group of 8,144 “displaced workers” (claimants who had been working continuously for at least three years before becoming unemployed and who did not expect to be recalled to the job they had lost). The second group consisted of 3,414 “frequent claimants” (claimants who were applying for benefits for at least their fourth consecutive year). The displaced workers were given 26 weeks to find new full-time jobs while frequent claimants were given only 12 weeks to find new jobs. The offer of the earnings supplement was found to have a small and short-lived impact on the re-employment of displaced workers. Those in the program group were four per cent more likely to be re-employed during the first six months after random assignment compared with those in the control group, but all differences in the employment of program and control group members had disappeared by the 11th month. The results for frequent claimants were even more disappointing: fewer than half of eligible frequent claimants were willing to volunteer for the project, and only five per cent of those who volunteered and were assigned to the program group received a supplement payment. Policy Framework Worker displacement is a natural feature of flexible labor markets. Changing technologies, competition, and shifting demand all have the potential to push people out of long-term, continuous, and even well-paying jobs. While unemployment insurance enables such people to hold out for an attractive new employment opportunity, it may also prolong their joblessness by reducing the urgency of finding work. Canada’s Earnings Supplement Project (ESP) was designed to encourage rapid reemployment by offering financial incentives to displaced workers. Sponsored by the Canadian government, the demonstration was managed by the Social Research and Demonstration Corporation (SRDC) and evaluated in a random assignment study by MDRC in collaboration with SRDC. Agenda, Scope, and Goals Two primary benefits, each of which could last as long as two years, characterized ESP’s offer to unemployment benefit claimants who found a full-time job within 26 weeks: If the new job paid less than the old one, ESP enrollees received an earnings supplement equal to 75 percent of the difference between the two wages, up to a maximum of $250 per month. Earnings insurance protected ESP enrollees against a future reemployment earnings loss. By designing its incentives in this way, the Canadian government sought to avoid the largely disappointing results of earlier programs piloted in the U.S. that offered a lump sum bonus to displaced workers who became reemployed quickly. ESP’s goals were threefold: to compensate displaced workers for job loss associated with economic growth that benefits society as a whole to provide displaced workers with a source of temporary financial support that promotes rapid reemployment without lowering their earnings or increasing the unemployment rolls to reduce unemployment insurance costs Design, Sites, and Data Sources About 8,000 displaced workers entered the ESP study over a one-year period from 1995 to 1996. Half were randomly assigned to the supplement group, making them eligible for the earnings supplement or the earnings insurance if they found full-time work within 26 weeks. The other half were randomly assigned to the control group, making them ineligible for the supplement and earnings insurance but eligible for standard unemployment benefits. ESP’s effects on labor market reentry and other outcomes were estimated by tracking and comparing the two groups’ experiences over a 15-month period. The demonstration sites were chosen to reflect a mix of labor market types and geographic areas. The sample — composed of Granby, Quebec; Oshawa, Ontario; Saskatoon, Saskatchewan; Toronto, Ontario; and Winnipeg, Manitoba — was therefore diverse but not fully representative of the Canadian labor market.