Unemployment Efficiency wages Borjas ch 12 not temporary

Unemployment Efficiency wages Borjas, ch. 12 (not ”temporary lay-offs”, 12. 6, 12. 7, 12. 9) + article by Yellen In the simple market equilibrium model, there can be no involuntary unemployment. One of the most difficult tasks for economists has been (and is) to explain how unemployment can exist.

Because unemployment does exist: The Great Depression – one quarter to a third of all workers unemployed in the industrialised countries. In Sweden in 1933 when 22% of trade-union members were unemployed. After 1933, unemployment fell almost steadily and in the 1950 s stabilised at about 2% and remained near that level until the early 1980 s. It did not reach 4 % before the depression of the 1990 s.

In other European countries unemployment was also low in the 1950 s and 1960 s but increased sharply after the ”oil-shock” of the early 1970 s. ◦ UE in ”OECD-Europe” in 1980 was 6% (2% in Sweden) ◦ in 1986, it was 10%. During the 1990 s depression Swedish unemployment rates were only a little below the European average. But when they started to fall (1998) they fell more rapidly in Sweden.

Unemployment as a social and economic issue. Unemployment has been shown to have substantial negative effects on the health and well-being of the unemployed and their families. It can lead to social exclusion, especially for the long-term unemployed. There is an opportunity cost to society of what the unemployed could have produced. A long period of unemployment leads to loss of skills and experience. This means lower future earnings for the individual and lower productivity for the economy. Unemployment leads to increased differences in living standards and to social conflicts and tension, particularly if some groups are more affected than others. Unemployment insurance and social security to the unemployed have to be paid by the public sector.

The US and Europe In 1960 s unemployment was twice as high in the US as in most European countries and remained higher until the late 1970 s. Since the early 1980 s, US unemployment rates have decreased (with one minor upturn) while European mostly have not.

Unemployment rates in Sweden, the EU 15 and the US Source: Laborsta (ILO)

Unemployment 1998 -2009 10. 0 9. 0 8. 0 7. 0 6. 0 Sweden 5. 0 EU 27 4. 0 US 3. 0 2. 0 1. 0 0. 0 199819992000200120022003200420052006200720082009 Age 15 -74

A few more countries (compare Borjas fig. 12 -19) 12. 0 10. 0 8. 0 Sweden Denmark Netherlands 6. 0 France Germany Italy 4. 0 US 2. 0 0. 0 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

Is the ILO-unemployment rate the only measure of ”success”? In Sweden (2006) one could take into account : UE (old definition) was 5. 4% Students wanting to work 2 %. Other latent unemployment 2 %. Persons in labour market programs. Early retirement ”for labour market reasons”. Involuntary part-time (partial unemployment)

Unemployed and in labour market programs in % of LF 1990 -2002 (Source: Labour Market Board, AMS) UE + in LMP = total

In the US one could take into account (but not simply add): Discouraged job-seekers (about 1% of LF) Involuntary part-time (Borjas fig. 12 -5) Prison population (2. 3 m in 2006, eq. to 1. 5 % of the labour force). Working poor.

Who is more/less likely to be unemployed? Having said this: 1. Gender Traditionally women had higher unemployment rates than men. That is not necessarily true today. In Sweden, female UE was a little higher than the male up to 1980, then ≈ equal. In the 1990 s, more men were unemployed. (Manufacturing reduced workforce more than the public sector. ) In some other countries too, like Norway & UK, there is lower UE among women. In others, there is higher. (In Spain, Italy & Greece about twice as high as that of men. )

2. Age In most countries, UE is higher among young people. Countries with apprentice systems tend to have more young people with jobs. But youth unemployment figures should be interpreted with care – they depend on: ◦ Numbers unemployed. ◦ Participation rates of young people.

0. 25 unemployed aged 15/16 -24 in 2008 0. 2 0. 15 unemployment unemployed/population 0. 1 0. 05 0 Denmark 15 -24 Germany 15 -24 Sweden 15 -24 United Kingdom United States 1616 -24 24 Source: Laborsta

Youth unemployment in Sweden and Denmark (Q 4 2009) In percent Sweden 15 Sweden Denmark -24 years 15 -34 years 15 -29 years Unemployment in age group Share unemployed in age group Labour force participation in age group 24, 0 13, 9 10, 6 11, 1 9, 2 7, 7 46, 1 66, 0 72, 9 Sources: webpages of Statistics Sweden and Statistics Denmark

3. Education. UE-rate in LFS 2010 (%): Less than upper known secondary 18. 2 (Average was 8. 4) Upper secondary 7. 9 Post-secondary Not 5. 0 31. 4 Usually: higher education lower risk of unemployment. But short upper secondary schools gave vocational training in occupations with a shortage of labour supply.

Unemployment = proportion who become unemployed * the length of time they stay unemployed. The analysis of the unemployment rate can be divided into analysis of ◦ The inflow into unemployment ◦ The duration of unemployment Note that duration is not the average time those who are now unemployed have been employed. It is the average length of unemployment of those who get unemployed at a particular time not those who are unemployed at particular time.

How do people get unemployed? The labour market is a dynamic system (Lecture 1) Employed Outside LF Unemployed

US (Borjas fig. 12. 3): Of the unemployed ◦ Roughly half have lost their jobs ◦ Roughly a third have quit a job ◦ About one tenth each of new entrants and reentrants to the LF. Sweden (Björklund et. al. Arbetsmarknaden based on LFS): Inflow to unemloyment 2003: ◦ ◦ 26 16 46 12 % % new or re-entrants redundancies contracted work finished* other * Was only 29% in 1980

Steady state unemployment For simplicity, consider only flows between employment and unemploytment. Let l = the proportion of workers who lose their jobs in each period h = the proportion of the unemployed who find a job in each period LF = the labour force, E = the employed, U=the unemployed In steady state inflow=outflow l. E = h. U l(LF-U) = h. U l. LF=(h+l)U Unemployment rate =

Different types of unemployment Frictional unemployment Seasonal unemployment Structural unemployment Cyclical unemployment

Frictional unemployment – Firms go in and out of business, workers go in and out of jobs (or in and out of the labour market). Information is not perfect. Workers and firms need time to ”match up”. Policy action: Improve information to workers about job openings, and information to employers about job-searchers.

Seasonal unemployment Employers whose need for labour varies may lay workers off with the mutual understanding that they will be re-hired. Examples: Winter/summer tourism, construction If the workers receive unemployment benefits ”the public purse” subsidises these firms. (In Sweden employers are not taxed according to their propensity to lay workers off. ) But in Sweden too, a substantial minority of the unemployed who get work, are re-hired by a previous employer. (Partly as an effect of legislation, LAS. )

Structural unemployment There is no reason to assume that people will choose to learn precisely the occupations in which employers will want to hire or live in the regions where jobs are. In a changing economy with sectoral shifts chances are big that skills supplied do not match skills demanded. Policy action: Provide and subsidise training that improves the match between supply and demand. Decrease the cost of moving to where the jobs are.

Cyclical unemployment Even if there was no mismatch, there may be more supply of than demand for labour. According to the supply/demand market equilibrium model, wages should fall when there is unemployment, until the market clears. In reality, wages are sticky, at least downwards. Government expenditure, taxes and benefits and interest rates influence aggregate demand therefore unemployment. They can be used counter-cyclically.

Three directions in economic theories of unemployment: Search theories of unemployment Inflation and unemployment Issues in the determination of wages. (Why do wages not fall to the market clearing level? ).

1. Search models of unemployment Search models start from incomplete information – it takes time and money for workers/employers to find good worker/job matches. Assume that unemployed workers do not know where to find a job with a particular wage but they do know the distribution of wage offers “out there” in the market. For example, the worker knows that the best paying job/s he/she could get have a monthly wage of 30 000 SEK but there also jobs paying only 15 000 – and many levels in between.

Frequency $5 45 15 $8 18 $22 27 $25 30 Wage The chance that a wage offer will be in the tails – say above 27 000 or below 18 000 is small. If the offer the worker receives is 17 000, the probability that next offer will be better is large. If the offer is 27 000 the chances that the next will be better is small. Wage offers in 1000 SEK

In a non-sequential search, the stopping rule is to stop searching after a certain number of offers (decided in advance). Not optimal – does not take into account what the offers are. In a sequential search, the worker’s stopping rule is to stop searching when there is an offer above a certain level – the asking wage. If the worker receives an offer at or above the asking wage, searching stops. If the offer is less than the asking wage, he/she does another search. The expected duration of unemployment is longer, the higher the worker’s asking wage.

What determines the asking wage? There are costs connected with continuing the search – direct costs of travel etc. and the opportunity cost of the rejected wage offer. The higher the offer that is rejected, the larger the MC of another search round. The higher the wage offer, the larger the marginal benefit of accepting it. The asking wage will be the wage at which MC = MB

Determinants of the asking wage The asking wage depends on the distribution of job offers (the probability of getting job offers at different levels. ) A higher time discount rate lower asking wage. To continue the search involves a cost now and a gain later. Higher unemployment insurance higher asking wage. A long unemployment period tends to reduce the asking wage.

Factors that influence the length on unemployment in a search model: A well functioning Employment Service (Arbetsförmedling) makes search quicker. Lower unemployment rate ◦ makes the chance of a new offer greater ◦ increases the asking wage. High unemployment insurance increases the duration of unemployment Exits from unemployment tend to increase just before a timelimit for unemployment insurance. The asking wage tends to decrease over time A long unemployment period reduces the chances of job offers – depreciation of skills, stigma effect, discouragement. Search efforts may increase or decrease with time in unemployment.

Duration dependence Thus, there are reasons to expect the probability of leaving unemployment during a certain time (week, month) to vary over a period of unemployment. This is called duration dependence. Without duration dependence for each person, observed decreasing exit rate may be due to heterogeneity – those who have the greatest difficulty finding employment will be overrepresented among long-time unemployed. Probability of Leaving unempl. Positive dur. dep. No dur. dep. Negative dur. dep. Unemployment in weeks

So what is a good unemployment insurance (UI) system? With generous UI we can expect: ◦ Longer unemployment periods. ◦ Higher wages after unemployment. If employers (US) or workers/unions (Sweden) pay more towards financing the UI, more account may be taken of unemployment risks in wage-setting. Unemployment insurance and social welfare: ◦ Ethical reasons why unemployed people who want to work should have a “decent” standard of living. ◦ It can be economically efficient to allow the unemployed time to search for a good match.

Unemployment insurance in Sweden Entitlement to unemployment benefit depends on the following conditions. The jobseeker must: 1. be fit and available to work for an employer at least three hours every working day and for an average of at least 17 hours per week 2. be prepared to accept an offer of appropriate work during the period, where no impediment has been accepted as admissible by the unemployment insurance fund 3. be registered as a jobseeker with the Employment Service, in compliance with government regulations or those applied by one of its agencies 4. cooperate in drawing up an individual back-to-work plan in consultation with the Employment Service 5. actively seek but be unable to find appropriate work. 6. fulfil the work requirement (prior to becoming unemployed, you must have worked for at least six months and at least 80 hours in every calendar month, or for at least 480 hours over six consecutive calendar months and for at least 50 hours in each of these months. )

If you are a member of an unemployment insurance fund you receive: ◦ 80 % of previous income up to a maximum of 680 SEK/day for 200 days. (For parents with children under 18 it is 450 days. ) ◦ 70 % of previous income (up to SEK 680/day) for another 100 days. If you are not a member you receive 320 SEK/day. The compensation is reduced for a period of time if you refuse one or two job offers. If you refuse three, you are cut off from compensation. If you do not fulfil the requirements you do not receive UI, only social assistance (försörjningsstöd) which is much less and requires that you do not have major assets like a house, car, bonds, equity etc). Source: Arbetslöshetskassornas samorganisation. http: //www. samorg. org/so/Index. aspx? id=105

In addition to the state system of UI, the labour market organisations (unions & employers) have agreed to provide support to workers who are laid off: ◦ Help in search for new jobs ◦ Severance (redundancy) pay ◦ Support during unemployment “topping up” the UI. This influence the actual effect of state decisions on UI (time limits, replacement ratios)

2. Inflation and unemployment Is there a ”trade-off” between inflation and unemployment? A famous study (Phillips, 1958) found a negative correlation between unemployment and inflation in Britain over a century. Infl. rate UE-rate

What could be behind the Phillips curve? When there is high unemployment: ◦ Smaller wage increases ◦ Low aggregate demand Low inflationary pressure. When there is a high level of employment: ◦ High aggregate demand high level of production increased marginal costs price increases ◦ With low unemployment workers have good ”outside options” and can make higher wage demands High inflationary pressure.

In Sweden in the 1950 s and early 1960 s, data on inflation and unemployment fit a convex Phillips curve. To have both low unemployment and low inflation, the Phillips curve has to be ”pushed” inwards. This was the idea behind Swedish economic and labour market policy in the post-war According to labour economists Gösta Rehn och Rudolf Meidner it could be done with a combination of: ◦ Restrictive fiscal policies ◦ Solidaristic wage policy (uniting modernisation of the economy and egalitarianism) ◦ Targeted measures to create jobs and active labour market policies

Stagflation – no more Phillips curve? From late 60 s both inflation and unemployment increased. After 1973 many countries experienced stagflation – high UE AND high inflation. In Sweden in the 1980 s UE and inflation moved in opposite directions between years but the ”curve” was farther from the origin. Eithere was no Phillips-curve, or it had shifted outwards. (Compare Borjas fig. 12 -17 for the US. ) Stagflation increased the influence of neo-liberal economists like Milton Friedman and Edmund Phelps.

NAIRU They claimed that in the short run there might be a negatively sloped Phillips curve but in the long run the Phillips curve was VERTICAL. There was no trade-off between unemployment and inflation. Only one level of unemployment was compatible with a stable rate of inflation - the NAIRU – Non. Accelerating Inflation Rate of Unemployment below NAIRU ACCELERATING inflation. Why? People learn to expect inflation and adjust to that. They include compensation for expected inflation in their wage claims and in price setting.

Long and short run Phillips curve (graphically) If inflation – including wage infl. increases from 3 to 4 %, more workers will find jobs above their reservation wage and firms will increase production and employment. The economy moves from A to B. When firms and workers realise that the new nominal wages/prices are not real, employment will fall again and the economy moves to C. Rate of Inflation 4 3 B C A Unemployment Rate

Long and short run Phillips curve (mathematically) Model 1 Assume: ◦ 1. Productivity is constant. ◦ 2. Firms increase prices in proportion to wage increases. ◦ 3. Workers are able to get wage increases equal to the expected inflation rate + x percentage points ◦ 4. Expected inflation = last year’s inflation It follows that: ΔW/W= x + expected inflation = x + last year’s inflation ΔP/P = ΔW/W = x + last year’s inflation x=0 ΔP/P = last year’s inflation constant rate of inflation

Model 2 with increase in productivity Assume that ◦ productivity increases by y percent each year ◦ firms increases prices by (percent) wage increase minus y. ◦ Wages increase by expected inflation + x % ◦ X depends on the level of unemployment. ΔW/W= x + expected inflation = x + last year’s inflation Inflation = ΔP/P = ΔW/W - y = x-y + last year’s inflation Constant inflation x=y If UE < NAIRU, wages will increase faster than productivity + last year’s inflation. Price increases will be larger than last year (in percent). Inflation will increase from year to year. We get ACCELERATING inflation.

Expansionary fiscal policy increased nominal purchasing power. If firm’s believe that real demand has increased they increase production and prices. But price increases trigger an inflationary spiral: Inflation expectation of inflation higher wage demands and prices higher production costs firms decrease production decreased demand for labour. Finally: The economy is back at NAIRU. During the adjustment process, there are temporary positive effects on production and employment. High unemployment levels can increase NAIRU, since unemployment tends to be persistent. Those who have been unemployed long have less chance of returning to employment. Policies and institutions which improve the functioning of the economy and of the labour market can achieve a lower NAIRU.

Flexibility: Support/subsidies for labour mobility and training is one side of flexibility Others which employers’ organisations emphasise are ◦ Wage flexibility (differentiation & lowest level) ◦ Employment flexibility (make temporary contracts and lay-offs easier) One example of reforms is the Danish “flexicurity”.

3. Non-market clearing (“sticky”) wages We have seen how collective bargaining can lead to higher wages but less employment than in a perfect competition equilibrium. Another theory of wages that explains this is theory of efficiency wages.

Efficiency wages In the human capital model and the labour demand model, we assume that higher productivity higher wages. It could work the other way around too. Workers could be more productive if they are paid better.

Assume that output Q=Q(e(w) *N) that Q is an increasing function of e*N. In order to maximise Q/w the firm should maximise e/w The derivative of e*N/w with respect to w At a maximum the derivative = zero The profit-maximising firm should pay a wage at which the wage elasticity of production = 1, even if it is > the outside wage.

A firm which pays efficiency wages will attract more workers than it wants to hire but over-supply will not lead to a wage reduction since this wage is profitmaximising. This can explain several things that seemed odd in the neoclassical models.

Why are discriminating firms not competed out of the market? Wages above the market rate will attract workers – if there is a queue of workers wanting to be hired (job-rationing) it is costless to avoid the discriminated group. If the effort functions of different groups are different it can be economically rational to pay them different wages.

How can the same worker face different wage offers in the search models? Not what we would expect from the human capital theory, unless work conditions are different – but even the utility should be the same. Efficiency wages are believed to be a reason for the persistent wage differentials between industries and between small and large firms.

Why do firms not lower the wages when there is an oversupply of labour? Or, in other words, why are wages “sticky”? Or, in other words, how can there be involuntary unemployment? Why does some empirical research find that industries and regions with high wages also tend to have high unemployment? Efficiency wages and the business cycle – sticky wages will cause business cycle fluctuations in employment (but the argument is beyond this course).

The down-ward sloping wage-curve Wage level Unemployment rate

So why does effort depend on pay? 1. Health and nutrition model 2. Shirking model 3. Labour turnover model 4. Adverse selection model 5. Sociological models such as “gift exchange model”. All these approaches lead to a positive relation between wage and effort/productivity.

Shirking model It is often impossible to stipulate exactly how hard or well an employee should work. Workers can “shirk”. Firms could fire shirkers if they are caught. But if the wage in the firm is higher than outside OR if there is unemployment in the “efficiency wage sector”, it is costly to loose a job there. Efficiency wages should be important where direct supervision and control is difficult, where the employer needs to create incentives for the worker to work hard or to take a lot of responsibility.

The Labour Turnover Model A model related to the shirking model. Labour turnover can be costly (hiring costs, training of new workers, need for firm-specific human capital). If all workers are paid the same, new workers will cost more than their marginal productivity. It is important to the employer that they stay long enough to cover the training costs. If all firms pay a wage above the market clearing there will be unemployment and workers will avoid quitting their jobs.

Adverse selection Firms which pay higher wages will attract more able job-seekers – but they need to be able to screen before hiring or monitor afterwards.

Sociological or gift-exchange models Gift exchange models are important in sociology and anthropology. A gift demands reciprocity. Akerlof (1982) gives and example where workers’ effort was above the required level because (not despite) there was no sanction for slow workers. Relations to the firm and within the group of workers created certain ”work norms”. Compare Solow and Marshall (from Introduction”). ”A Fair Day’s Wage for a Fair Day’s Work” but also ”AFair Day’s Work for a Fair Day’s Wage”.