Why does the unemployment rate tend to underestimate the level of labor market challenges?

Unemployment rates

Occupational unemployment rates can provide indications of skills imbalances, but it would be naive to think that any positive unemployment rate is an indication of a surplus. There are numerous reasons why the observed and equilibrium occupational unemployment rates are never zero. The challenge with this type of analysis is in determining, at the occupational level, what is the normal (equilibrium) unemployment rate, above which would be considered a surplus situation. Furthermore, among the unemployed may be persons who are not qualified to work in the occupation and, therefore, the number of unemployed overestimates supply. Conversely, if there are persons qualified to work in the occupation but who are employed elsewhere, then unemployment numbers may underestimate supply.

KSA

The main objective of importing foreign workers to the Gulf region is to satisfy the large demand for labor. Most expatriates are destined for the workforce. Figure 27.4 shows the labor force participation rate in KSA by nationality (Saudi versus expatriate) from 2005 to 2011. As expected the participation of expatriates in the labor force is strong and hovers between 70% and 80%. Surprisingly, the labor force participation of Saudis is somewhat stable at a low 10%. While it would have been ideal to have more detailed information, such as participation rate by gender, the data are not publicly available. Figure 27.5 shows the share of employed Saudi males who are working in the private sector to average (between 2005 and 2011) around 55%, while that of Saudi females is less than 20%. For expatriates, regardless of gender, more than 70% work in the private sector.

Figure 27.6 outlines the unemployment rate for both Saudis and expatriates over a period of 13 years.19 The unemployment rate of Saudis is actually steadily growing, surpassing the 12% mark since 2011. Due to the nature of the labor and migration laws in the Gulf, one would expect a low unemployment rate for foreign workers. The unemployment rate of expatriates in KSA is in most years very low and almost always below 1%.

Figure 27.7 displays the distribution of gender by economic activity. The largest share of Saudi females is concentrated in EA9, which stands for community, social, and personal activities. For female expatriates, they are found in mines, oil and gas, and also community, social, and personal activities.

Figure 27.8 mirrors Figure 27.7 but presents gender share by occupations rather than economic activities. The most common occupations for women are scientific, technical and human specialists, clerical jobs, industrial and chemical processes, and food industries regardless of nationality (Saudi versus expatriate). Saudi women are also more involved in sales jobs.

The next two figures (27.9 and 27.10) present average monthly wages of manpower in the private sector between 2005 and 2012. Figure 27.9 shows that females actually make more money, as the average monthly wage of females was 580 USD in 2012 while it was 380 USD for men. Both wage series have the same dynamics, being relatively stable until 2008, with a large drop before wages go back to almost the same level (actually slightly higher) after 2010. Figure 27.10 quantifies the drop as more than 10% for females and more than 20% for males. The period of wage instability seems to be a direct consequence of the financial crisis of 2008.

Finally, Figure 27.11 shows the mean age and education level (2005–10) of those job seekers in KSA. Job seekers who have a diploma or a higher degree (bachelor or graduate school) constitute slightly less than 20% of the total pool of unemployed. On the other hand, almost 50% of them are between 20 and 24 years old, although that share is in decline and has reached 35% in 2010. If one includes the 25- to 34-year-old job seekers, the total share would surpass the 75% mark for those between 20 and 34 looking for a job. Figure 27.11 points to a very young pool of job seekers who are not very highly educated. The next subsection presents data on the UAE.

Unemployment and employment

If the unemployment rate is a simple index of employment insecurity, the Indonesian performance is rather disappointing. Clearly, from 1990 to 2006, the unemployment rates in Indonesia depicted an increasing trend, both for males and females (Figure 7.3). Using the standard definition, the unemployment rate in 1990 was only 3.2 per cent. In 1999 it grew to 6.4 per cent and to 8.1 per cent in 2001. In 2005 the unemployment rate soared to 10.3 per cent, and again slightly increased to 10.5 per cent in 2006 (CBSa, various years). The rise in unemployment after 2001 was partly due to the application of a relaxed definition of unemployment, which includes discouraged jobseekers.6 However, as shown by Sugiyarto, Oey-Gardiner and Triaswati (2006), open unemployment rates based on the standard definition continued to demonstrate increasing trends. It is also important to note that unemployment tends to be concentrated in vulnerable groups of workers, especially youth (e.g. 15 to 24 years old). For instance, the Indonesian male youth unemployment rate was 3.4 per cent in 1990 and soared to 27.8 per cent in 2006. The NLFS data showed that, in 2006, the ratio of youth to adult unemployment rate had reached 5.9 per cent, which means the youth unemployment rate was nearly six times the adult rate. Workers with low education attainment comprise another major group of unemployment: in 2005, more than 57 per cent of jobseekers had a junior high school background or below. Many of these jobseekers have ended up in the informal economy.

The unemployment figure is higher if the definition is expanded to include underemployment – people who are technically employed, but work fewer hours either voluntarily or involuntarily.7 To a certain extent, the NLFS data allows one to distinguish between these two groups. Our concern here is those who are involuntarily working fewer hours, but are still looking for work and willing to accept another job. Using this definition, the data reveal that in 1997 there were 10.7 million involuntarily underemployed people, or 38 per cent of the employed population, still looking for jobs. By 2006 the figure reached 14.2 million, or 47.5 per cent, involuntarily underemployed people. If we sum open unemployment and the involuntary underemployed, there were 25.2 million and 25.3 million people or 23.8 per cent of the total labour force in both 2005 and 2006 looking for jobs (CBSa, various years). Table 7.1 shows that the number of involuntary underemployed was higher for males than that for females, and higher in rural than urban areas. It is interesting to observe that male involuntary underemployment showed an increasing trend, while the female trend has reversed from positive to negative in recent years. However, the rate of involuntary underemployment in rural areas also increased steadily.

Table 7.1. Involuntary underemployment by sex and residence, Indonesia, 2002–2006 (thousands)

Sources: CBS, NLSF, 2002–2006.

In so far as those in employment are concerned, three sectoral activities have been the source of livelihood for the majority of the employed population in Indonesia: agriculture, trade/hotels/restaurants and community/social/personal services. The share of agriculture as an employment provider exhibited a declining trend until the crisis. In 1985 around 55 per cent of employed people worked in agriculture. This declined to 50 per cent in 1990, and then declined to 43 per cent in 1995. However, after the financial crisis, more workers returned to agriculture, which drove the agricultural share of jobs to 46 per cent in 2003. It declined to 43 per cent in 2004, but again increased to 44 per cent in 2005. By 2006, 44.47 per cent of workers in Indonesia relied on agriculture (Figure 7.4). People employed in agriculture, due to its nature, have tended to work fewer hours and have often been identified as self-employed or unpaid family workers (see Sugiyarto, Oey-Gardiner and Triaswati, 2006). The increasing proportion of workers in agriculture during and after the crisis has suggested that this sector has been a buffer when other economic sectors could no longer accommodate workers.

Trade/hotels/restaurants was the second largest sector to employ people in Indonesia, with about 15 per cent of workers in 1985, and employment in trade/hotels/restaurants has been increasing gradually. In 2005 it accommodated almost 20 per cent of the employed population in Indonesia, but it then declined to 19.5 per cent in 2006. Many people in this sector were working as self-employed. Much of this sector is portrayed as an informal economy: many establishments in the trade sector are non-legal, unregistered entities. In 2001, for instance, as many as 99.7 per cent of all entities in the trade sector were small businesses. In terms of employment, 91.6 per cent of workers in the trade sector were working in micro and small trading establishments (Widarti, 2004).

The community/social/personal services sector in Indonesia also accounted for a substantial proportion of employed people. In 1985 around 13 per cent of employed people worked in these services. This declined in 1990, and then increased again in 1995. Since 1999, however, the role of the services sector as a job provider has been declining. In 2006, community/social/personal services work provided jobs for 11 per cent of all workers. The manufacturing sector, on the other hand, has played a larger role in employment absorption. Again, employed people in manufacturing are concentrated in micro establishments or SMEs. In 2001 as many as 65.4 per cent of workers in manufacturing were employed in micro and small industries, but these small-scale establishments only contributed to 17 per cent of the manufacturing sector GDP. In contrast, large-scale manufacturing returned a high value- added, but did so with a low absorption of labour (ibid.).

Although the three economic sectors (agriculture, trade/hotels/restaurants and community/social/personal services) were the largest employers of the Indonesian population, these sectors have had the lowest labour productivity (BALITFO, 2005) and, as Sugiyarto, Oey-Gardiner and Triaswati (2006) pointed out, the value added of these three sectors was low. Meanwhile, such economic sectors as finance/business service, mining/quarrying, electricity/gas and water have demonstrated the highest labour productivity, but have relatively low shares of employment.

2.3.1.2 Unemployment and Poverty in Egypt

Egypt’s unemployment rate was estimated at 12.5% in 2016, an increase of between 2.5% and 3.5% from its level before 2011. Highest rates of unemployment have traditionally been encountered among youth and women. Thus, youth unemployment was reported at around 25%, even before the events of 2011. However, estimates for 2014 point to youth unemployment rates of around 42%.44 The country’s youthful population is essentially a product of the high fertility and population growth rates that prevailed in the late 1980s and early 1990s. The former was above 4.3 per woman till 1992, whereas the latter stood above 2.5% between 1980 and 1989. Naturally, at 2.1% per annum, Egypt’s present population growth rate is still rather high, placing added pressures upon the country’s present and future infrastructure and services.

Another shortcoming of public policies in Egypt is manifested in the country’s inability to train its young citizens. Thus, while university education is free, the quality of higher education and training institutions is poorly matched to actual needs. Medical and engineering graduates might be an exception, however, a large proportion of these graduates end up working in the Gulf region rather than in their home country.

Official data indicate that 28% of the population lived below the poverty line in 2015. However, it is estimated that poverty rates rise to levels as high as 60% in rural Upper Egypt.

Prison Education in the United Kingdom and Ireland: Partnerships between Government Training Agencies and Prison Systems

Findings about the high unemployment rate among ex-prisoners and subsequent higher reconviction rate among those who remain unemployed have influenced the development of major policies and programs dealing with prisoner education in the United Kingdom. In 2006, the government published its three priorities for reducing reoffending through skills and employment. These include engaging employers through a corporate alliance to employ ex-offenders; building on the offender learning and skills service; and reinforcing the emphasis on skills and jobs (Department for Education and Skills, 2006).

In practice, this means a partnership between the National Offender Management Service and the Learning and Skills Council to design and fund the Integrated Offender Learning and Skills Services program, which caters to offenders in custody and in the community. The program aims to ensure that prison industries and workshops provide more meaningful work and prepare prisoners more effectively for jobs in the community. (Most prisoners are required to work while in prison: prison work usually refers to those activities which supply services to the prison such as laundry, kitchen and bakery duties, cleaning, gardening, and maintenance. This may also include horticulture, e.g., growing fruit and vegetables, or farming and animal husbandry such as dairy to provide food for prisons. Prison industries and workshops usually refer to specific industry work which supply goods or services to the outside community or customers on a commercial basis. These may include metal work production, plastic products, repairing electronic goods, or other manufacturing or catering services depending on the resources available in the prison.) This includes working more closely with employers to meet their needs. The hope is that employers would employ the prisoner on completion of their sentence. Prior to the integrated Offenders Learning and Skills Services program being implemented in August 2005, researchers from the Learning and Skills Development Agency conducted an interim evaluation of prototype activities in the three development regions. (The interim report titled: Evaluation of Regional Plans for the New Integrated Offender Learning and Skills Service, highlights findings and lessons learned from evaluation workshops and interviews with key stakeholders in each region (Walker et al., 2005).) The evaluation noted that successful transition to the community can be helped by one-to-one support and motivation for prisoners who move through the gate.

2 Stock Equilibrium

Milton Friedmans's celebrated ‘natural unemployment rate’ is a stock-equilibrium concept of structural unemployment: ‘the level that would be ground out by the Walrasian system of general equilibrium equations, provided there is imbedded in them the actual structural characteristics of the labor and commodity markets, including market imperfections, stochastic variability in demand and supplies, the costs of gathering information about job vacancies and labor availabilities, the costs of mobility and so on’ (Friedman 1968). Apparently, this concept of equilibrium unemployment covers structural and frictional unemployment as described earlier.

The nonaccelerating inflation rate of unemployment (NAIRU) is close to Friedman's natural rate, though it emphasizes (more than Friedman does) the nonclearing character of the labor market. The NAIRU is defined explicitly as the unemployment rate at which inflation is constant. It is derived from an expectations-augmented Phillips curve, i.e., a function that assumes that the price-inflation rate is a decreasing function of the unemployment rate and an increasing function of the expected rate of future inflation (Phelps, 1967). In the simplest possible terms (with linear relations):

(1)π=−au+b+cπe

where π is actual inflation, u is the unemployment rate, and πe is expected inflation. For given values of the (positive) parameters a, b, and c, the equation may be depicted as a set of downward sloping short-term Phillips curves (SRi), one curve for each set and for each value of πe (see Fig. 1).

In the context of this model, usually (and realistically) it is assumed that economic agents ask for full compensation for expected inflation in the long run; thus c=1 in that time perspectives. It is usually also assumed that economic agents do not make systematic expectational errors in the long run; thus π=πe. The only unemployment rate consistent with these two assumptions is, from Equation (1), u*=b/a, where * denotes the equilibrium unemployment rate. It is often also assumed that inflation in this situation is constant over time (and not just that actual and expected inflation coincide), because agents have no reason to revise their price and wage setting behavior when inflation is exactly what they had expected when they formed their decisions. u* is then the NAIRU (the unemployment rate at which inflation is constant), geometrically represented as a vertical long-run Phillips curve, LR in Fig. 1. Inflation increases to the left of that level and falls to the right. Because inflation in this framework in the long run is independent of unemployment, there is no long-run trade-off between these variables. But it has been argued by Lucas (1972) that there is not even a systematic short-term trade-off if economic agents have ‘rational expectations’ (thus, if they never make systematic expectational errors), and if various agents are free to revise their prices and wages in response to changes in expectations. Such a trade-off will then only arise due to random (non-systematic) expectational errors, e.g., due to random shocks or randomized government policy.

An alternative exposition of stock-equilibrium in the labor market, and hence structural unemployment, gained ground in the 1980s and 1990s; it emphasizes the requirement of consistency of price- and wage-setting behavior (Shapiro and Stiglitz, 1984, Layard and Nickell, 1986). Figure 2 gives a simple diagrammatic illustration, with aggregate employment (N) on the horizontal axis and the aggregate real-wage level (w) on the vertical. The LS curve is the traditional aggregate labor supply curve, i.e., the sum of labor supplied by households. The PS curve is the price-setting curve of firms: for alternative unemployment levels and given nominal wages, the curve defines the price level desired by profit-maximizing firms that operate in imperfectly competitive product markets. Or the PS curve can be described as a labor-demand relation, because it also expresses the combination of employment and real wages desired by profit-maximizing firms. Unemployment is then the horizontal difference between the PS and LS curves.

The WS curve defines wage-setting behavior. It depicts the influence of the (un)employment situation on the real-wage rate that the wage-setting process generates. The intuition is that firms feel compelled to offer higher real wages and that workers and unions demand higher real wages when the labor market is tight (low unemployment) than when it is slack. So by contrast to models with perfectly competitive labor markets, a distinction is made between labor supply (by households) and wage setting (by firms, unions, or more realistically a bargaining process).

The equilibrium real-wage rate is now w0, equilibrium employment N0 and equilibrium unemployment U0, the latter being the only unemployment level at which there are no incentives for agents to change prices relative wages and vice versa (corresponding to the equilibrium unemployment rate u* in Fig. 1). This model will be called the PS–WS model of equilibrium (structural) unemployment.

As an illustration, suppose that the real-wage rate in Fig. 2 is initially w0 but that aggregate employment happens to be above N0, e.g., as a result of positive demand shocks to which nominal magnitudes have not yet adjusted. Firms and workers are then dissatisfied with the existing real wage w0. Workers and their unions try to raise nominal wages (at existing nominal prices), as can be read off from the WS curve, and firms try to raise prices (at existing nominal wages), as can be read off from the PS curve. The result is a wage-price spiral, as described by the expectation-augmented Phillips-curves model. Firms also start cutting their work force because their desired level is lower than the actual one. Aggregate employment continues to fall until it reaches the unique equilibrium level N0. The same reasoning, mutatis mutandi, holds if aggregate employment is initially lower than the equilibrium level.

Stock Equilibrium

Milton Friedman's celebrated ‘natural unemployment rate’ is a stock-equilibrium concept of structural unemployment: ‘the level that would be ground out by the Walrasian system of general equilibrium equations, provided there is imbedded in them the actual structural characteristics of the labor and commodity markets, including market imperfections, stochastic variability in demand and supplies, the costs of gathering information about job vacancies and labor availabilities, the costs of mobility and so on’ (Friedman, 1968). Apparently, this concept of equilibrium unemployment covers structural and frictional unemployment as described earlier.

The nonaccelerating inflation rate of unemployment (NAIRU) is close to Friedman's natural rate, though it emphasizes (more than Friedman does) the nonclearing character of the labor market. The NAIRU is defined explicitly as the unemployment rate at which inflation is constant. It is derived from an expectations-augmented Phillips curve, i.e., a function that assumes that the price-inflation rate is a decreasing function of the unemployment rate and an increasing function of the expected rate of future inflation (Phelps, 1967). In the simplest possible terms (with linear relations):

[1]π=−au+b+cπe

where π is actual inflation, u is the unemployment rate, and πe is expected inflation. For given values of the (positive) parameters a, b, and c, the equation may be depicted as a set of downward sloping short-term Phillips curves (SRi), one curve for each set and for each value of πe (see Figure 1).

In the context of this model, usually (and realistically) it is assumed that economic agents ask for full compensation for expected inflation in the long run; thus c = 1 in that time perspectives. It is usually also assumed that economic agents do not make systematic expectational errors in the long run; thus π = πe. The only unemployment rate consistent with these two assumptions is, from eqn [1], u∗ = b/a, where ∗ denotes the equilibrium unemployment rate. It is often also assumed that inflation in this situation is constant over time (and not just that actual and expected inflation coincide), because agents have no reason to revise their price and wage setting behavior when inflation is exactly what they had expected when they formed their decisions. u∗ is then the NAIRU (the unemployment rate at which inflation is constant), geometrically represented as a vertical long-run Phillips curve, LR in Figure 1. Inflation increases to the left of that level and falls to the right. Because inflation in this framework in the long run is independent of unemployment, there is no long-run trade-off between these variables. But it has been argued by Lucas (1972) that there is not even a systematic short-term trade-off if economic agents have ‘rational expectations’ (thus, if they never make systematic expectational errors), and if various agents are free to revise their prices and wages in response to changes in expectations. Such a trade-off will then only arise due to random (nonsystematic) expectational errors, e.g., due to random shocks or randomized government policy.

An alternative exposition of stock-equilibrium in the labor market, and hence structural unemployment, gained ground in the 1980s and 1990s; it emphasizes the requirement of consistency of price- and wage-setting behavior (Shapiro and Stiglitz, 1984; Layard and Nickell, 1986). Figure 2 gives a simple diagrammatic illustration, with aggregate employment (N) on the horizontal axis and the aggregate real-wage level (w) on the vertical. The LS curve is the traditional aggregate labor supply curve, i.e., the sum of labor supplied by households. The PS curve is the price-setting curve of firms: for alternative unemployment levels and given nominal wages, the curve defines the price level desired by profit-maximizing firms that operate in imperfectly competitive product markets. Or the PS curve can be described as a labor-demand relation, because it also expresses the combination of employment and real wages desired by profit-maximizing firms. Unemployment is then the horizontal difference between the PS and LS curves.

The WS curve defines wage-setting behavior. It depicts the influence of the (un)employment situation on the real-wage rate that the wage-setting process generates. The intuition is that firms feel compelled to offer higher real wages and that workers and unions demand higher real wages when the labor market is tight (low unemployment) than when it is slack. So by contrast to models with perfectly competitive labor markets, a distinction is made between labor supply (by households) and wage setting (by firms, unions, or more realistically a bargaining process).

The equilibrium real-wage rate is now w0, equilibrium employment N0 and equilibrium unemployment U0, the latter being the only unemployment level at which there are no incentives for agents to change prices relative wages and vice versa (corresponding to the equilibrium unemployment rate u∗ in Figure 1). This model will be called the PS–WS model of equilibrium (structural) unemployment.

As an illustration, suppose that the real-wage rate in Figure 2 is initially w0 but that aggregate employment happens to be above N0, e.g., as a result of positive demand shocks to which nominal magnitudes have not yet adjusted. Firms and workers are then dissatisfied with the existing real wage w0 Workers and their unions try to raise nominal wages (at existing nominal prices), as can be read off from the WS curve, and firms try to raise prices (at existing nominal wages), as can be read off from the PS curve. The result is a wage-price spiral, as described by the expectation-augmented Phillips-curves model. Firms also start cutting their work force because their desired level is lower than the actual one. Aggregate employment continues to fall until it reaches the unique equilibrium level N0. The same reasoning, mutatis mutandi, holds if aggregate employment is initially lower than the equilibrium level.

4.2.1 Unemployment Risk

Idiosyncratic unemployment risk is completely determined by the four 2 by 2 transition matrices π(s′|s,Z′,Z) summarizing the probabilities of transiting in and out of unemployment for each (Z,Z′) combination. Thus π(s′|s,Z′,Z) has the form

(7)πu,uZ,Z′πu,eZ,Z′πe,uZ,Z′πe,eZ,Z′,

where, for example, πe,uZ,Z′ is the probability that an unemployed individual finds a job between one period and the next, when aggregate productivity transits from Z to Z′. Evidently each row of this matrix has to sum to 1. Note that, in addition, the restriction that the aggregate unemployment rate only depends on the aggregate state of the economy imposes one additional restriction on each of these 2 by 2 matrices, of the form

(8) ΠZ′(u)=πu ,uZ,Z′×ΠZ(u)+πe,uZ,Z′×(1−ΠZ (u)).

Thus, conditional on targeted unemployment rates in recessions and expansions, (Πl, Πh) this equation imposes a joint restriction on (πu,uZ,Z′,πe,u Z,Z′) for each (Z,Z′) pair. With these restrictions, the idiosyncratic transition matrices are uniquely pinned down by πu,eZ,Z′, ie, the job-finding rates.ac

We compute the job finding rate for a quarter as follows. We consider an individual that starts the quarter as unemployed and compute the probability that at the end of the quarter that individual is still unemployed. The possible ways that this can happen are (denoting as f1, f2, f3 and as s1, s2, s3 the job-finding and job-separation rates in months 1,2, and 3 of the quarter):

Does not find a job in month 1, 2, or 3, with probability (1 − f1) × (1 − f2) × (1 − f3).

Finds a job in month 1, loses it in month 2, does not find in month 3, with probability f1 × s2 × (1 − f3).

Finds a job in month 1, keeps it in month 2, loses it in month 3, with probability f1 × (1 − s2) × s3.

Finds a job in month 2, loses it in month 3, with probability (1 − f1) × f2 × s3.

Thus the probability that someone that was unemployed at the beginning of the quarter is not unemployed at the end of the quarter is:

(9)f=1−((1−f1)(1−f2)(1−f3)+ f1s2(1−f3)+f1(1−s2)s3+(1− f1)f2s3)

We follow Shimer (2005) to measure the job-finding and separation rates from CPS data as averages for periods corresponding to specific Z,Z′ transitions.ad Equating these with πu,e Z,Z′ delivers the following employment-unemployment transition matrices:

Aggregate economy is and remains in a recession: Z=Zl.Z′=Zl

(10)0.33780.66220.06060.9394

Aggregate economy is and remains in normal times: Z=Zh.Z′=Zh

(11)0.18900.81100.04570.9543

Aggregate economy slips into recession: Z=Zh.Z′=Zl

(12)0.33820.66180.06960.9304

Aggregate economy emerges from recession: Z=Zl.Z′= Zh

(13)0.22200.77800.03780.9622

We observe that the resulting matrices make intuitive sense. One possible (but quantitatively minor) exception is that the job-finding rate is higher if the economy remains in normal times than if it emerges from a recession. On the other hand, the lower job-finding rate is consistent with the experience during the Great Recession per our definition, as job-finding rates did not recover until well into 2014, whereas by our calibration the recession ended in 2013.

The Effects of Education on Subsequent Crime Among Adults

There is a relatively large literature linking wages and unemployment rates to criminal behavior. Recent studies conclude that crime is increasing in local unemployment rates and decreasing in wage rates (e.g., Raphael and Winter-Ebmer, 2001; Gould et al., 2002; Machin and Meghir, 2004). To the extent that education increases wage rates (and reduces the likelihood of unemployment), it increases the opportunity costs of crime and will tend to reduce postschool criminal activity. Higher wages raise the opportunity costs of crime in two distinct ways. First, since crime may require time to commit, that time cannot be used for other productive purposes like work. Here, it is useful to think of all of the time involved in planning a crime, locating a target, and, potentially, evading detection and arrest. Second, each crime committed entails an expected period of incarceration, which is more costly for individuals with better labor-market opportunities and wages.

On the one hand, property crimes such as burglary, auto theft, and drug dealing can involve significant planning or time spent on the actual activity itself. On the other hand, violent crimes such as assault would appear to require less time for planning and execution but are associated with higher expected probabilities of arrest, conviction, and incarceration as well as longer sentence lengths conditional on incarceration. For example, Lochner (2004) calculates that for each assault, the perpetrator can expect to spend 63 days incarcerated; however, the expected incarceration period for a burglary is only 13 days. These time costs would appear to exceed the direct time costs associated with committing most crimes. Thus, changes in wages or unemployment rates could have greater effects on violent crimes than on property crimes. (The estimated effects of low skill wages on violent crime are larger than on property crime for some specifications in Gould et al. (2002).)

Education may also affect the rewards from crime. This is most likely to be true for white-collar crimes such as fraud, forgery, and embezzlement. Education may actually increase these types of crime if it increases the rewards from crime more than it increases legitimate wages. Lochner (2004) finds some evidence that white-collar crime rates are increasing in average education levels as discussed below. To the extent that schools socialize students to become better citizens and to treat others better, education may also reduce the psychic returns to crime causing individuals to forego lucrative criminal opportunities.

Education may also teach individuals to be more patient. This will discourage crime, since forward-looking individuals place greater weight on any expected punishment associated with their criminal activities. Education may also affect preferences toward risk. To the extent that schooling makes individuals more risk averse, it will tend to discourage crime.

The Recent Evidence

Table 1 presents the most recent data we have available on youth unemployment rates, here defined as being for those under the age of 25. It is apparent, as noted above, that they are especially high in Southern Europe, with over half of the youth workforce unemployed in Spain (54%) and Greece (53%), but also very high in Italy (43%) and Croatia (42%), and averaging 22% in the EU and 23% in the Euro Area. We also present unemployment rates for older adults and in the final column the ratio of the two. The United States has a relatively low youth unemployment rate of 13% well below that of Europe (22%), although the ratio compared to the older age group is about the same – 2.7 and 2.4 – respectively. Interestingly, the four countries with the highest ratios around four – Romania; Sweden, Italy, and the UK – are not those with the highest rates of youth unemployment. The concern is that high relative rates might be a problem when youth compare themselves to adults, given we know from happiness research that relative things matter (Luttmer, 2005). It is instructive that there have been riots in both the UK and Sweden. (In May 2013, there were riots in the suburbs of Stockholm, apparently stoked by long-term unemployment, //www.theguardian.com/world/2013/may/23/swedish-riots-stockholm and riots in London, which spread to other UK cities in the summer of 2011 //www.theguardian.com/uk/series/reading-the-riots.) Figure 1 reports time series of youth–adult unemployment rates for four countries with high ratios – Italy, Sweden, and the UK with the United States plotted for comparison purposes. The extremely high ratio of around seven is especially notable in Italy, along with its subsequent decline, whereas the UK and Sweden saw a steady increase, which was not seen in the United States.

Table 1. Youth and adult unemployment rates, 2014 ranked by youth unemployment rates

Notes: data are for May for Estonia, Greece, Hungary, and Norway; for April for the UK and Turkey; and March for Latvia and Romania.

Source: Eurostat.

Table 2 reports on levels of youth employment to population rates, which can be a more accurate indicator of youth joblessness than unemployment rates, given low levels of youth participation rates in some countries, which are also reported in the table, for 2008 and 2013. The higher the proportion of youngsters in education, the lower will be the participation rate.

Table 2. Youth ages 15–24 employment and participation rates (%)

Notes: employment rate is employment/population; participation rate is (unemployed + employed)/population or can be defined as workforce/population.

Source: OECD.

It turns out that we get a very similar picture of youth joblessness from the employment rates as we do from unemployment rates. Employment rates for the young fell during the recession in most advanced countries and are lower in 2013 than they were in 2008; the three main exceptions are Sweden, Switzerland, and Germany where they are unchanged. In contrast, employment rates are higher in Chile, Columbia, Israel, and Turkey. Among advanced countries, participation rates of the young are only up in Switzerland (+7ppt); Sweden (+1ppt), suggesting that the young have not rushed to increasingly expensive education as a refuge from joblessness.

So in most European countries, both employment and participation rates are down since the onset of recession for the young and some sharply so. Examples of the differences between 2008 and 2013 rates for some selected countries with participation rates with employment rates in parentheses are as follows as percentage points: Denmark −10 (−12); Canada −4 (−5); the UK −4 (−7); the United States −4 (−5); Germany −2 (0); Finland −5 (−6); Spain −10 (−21); Ireland −13 (−18); France 0 (−2); Portugal −6 (−12); Italy −1 (−6); and Greece −2 (−12). It turns out that the three countries with the highest employment rates are also the ones with the highest participation rates – Iceland, the Netherlands, and Switzerland. The correlation coefficient is 0.96. The three countries with the lowest employment rates – Italy, South Africa, and Greece are close to the bottom in the ranking of participation rates, ranking 33, 38, and 34, respectively. As a consequence, unemployment rates are also up in the advanced countries but less so in the developing world.

A phenomenon that we have also observed in the recession is that not only those without jobs, the jobless, have been impacted, but also the youngsters who have held jobs. Table 3 shows the rise in the share of involuntary part-time employment in total employment of the young, which has more than doubled in many countries since including Greece, Portugal, the Netherlands, the UK, the United States, and especially Spain that has seen nearly a threefold increase. Young workers are disproportionately underemployed (Bell and Blanchflower, 2011c).

Table 3. Share of involuntary part-time employment in total employment, 15- to 24-year-olds

Note: The figure for UK is for 2007 and not 2008

Source: OECD.

These rises in unemployment, and falls in employment and participation rates that have occurred in many countries have come at a time of large youth cohorts. So, a rise in the supply of young people may exacerbate further the evidence found by Kahn (2011) that graduating from college in a recession permanently lowers a cohort's earnings. Table 4 reports on such changes in cohort size for a number of countries since 2000 and reports the change in the size compared with that in 2000 over the period 2004–2020. We can calculate 16–24 cohort size in 2020 by extrapolating from the number of 6- to 14-year-olds 10 years earlier, assuming no further immigration, which is a stretch of course. It is apparent that some countries had big increases in the size of their youth cohorts between 2000 and 2008 when the Great Recession started, including the UK, Sweden, and the United States, whose cohort size was 10% bigger. In contrast, cohort size was sharply smaller in Spain (−30%), Italy (−21%), and Greece (−25%). By 2020, youth cohort size will be at least 10% smaller in most countries except the United States, where it will continue to grow. By 2020, the youth populations of North Africa and Latin America will have grown by over a quarter, whereas that of Western Europe will have shrunk by 10%. The youth population in most East European countries, including Poland and Russia, will likely fall by around two-fifths over a 20-year period from 2000 to 2020. The Arab Spring occurred in countries that had growing youth cohorts and high youth unemployment rates.

Table 4. Cohort size of those aged 16–24 based on current population, compared with size of cohort in 2000 (%)

Source: Eurostat.

We now turn to examine five questions about youth unemployment from the literature. The concern is that spells of unemployment experienced during the Great Recession will create permanent scars for a Lost Generation.