Employing Youth: Promoting employment-intensive growth

Report for the Interregional Symposium on Strategies to Combat Youth Unemployment and Marginalization

13-14 December 1999, Geneva

ISBN 92-2-111924-6
First published 2000

Download the document in Adobe Acrobat's Portable Document Format (PDF)

Table of Contents

Preface

The large, and perhaps growing, number of unemployed youth is one of the most daunting problems faced by developed and developing countries alike. On average, and almost everywhere, for every one unemployed adult, two young persons find themselves without a job. The social distress caused by this situation is well known. The long-term effects of youth joblessness are equally important. The unemployment spells over a worker's life cycle are related to the ease of transition between school and work. Furthermore, it is disappointing to observe that the unprecedented expansion of investment in youth education in many regions of the world is not being matched by higher employment levels for this population group.

Building upon the work initiated in 1996-97 under the Action Programme on Youth Unemployment, and in response to the resolution on Youth Employment adopted by the International Labour Conference at its 86 th Session (June 1998), the ILO Development Policies Department launched a number of studies on the effectiveness of strategies and programmes implemented in different countries of the world to cope with youth marginalization and unemployment. The present paper, which will be presented at the Interregional Symposium on Strategies to Combat Youth Unemployment and Marginalization (Geneva, December 1999) provides a synthesis of these reports.

Our challenge has been to identify practical and effective strategies to eradicate youth unemployment. While much has been learnt, we must admit that more has yet to be discovered. As the present employment situation throughout the world shows, only few countries have managed to generate decent employment for their youth, let alone their adult population. We are still searching for practicable solutions to this global problem. One thing is clear: sustainable, job-enhancing economic growth remains an indispensable component of any strategy to eradicate youth unemployment. Experience has shown that employment-friendly growth is essential, and targeted programmes can only provide complementary resources and cross-gap actions. Starting from this premise, the present paper examines the effects of policies and programmes on employment, and suggests some viable strategies for the inclusion of youth in the process of development.

The debate about education versus training needs to be revived. In an increasingly mobile world, the need to acquire the ability to learn is, often, more important than the acquisition of a specific skill. There is, however some evidence to indicate that effective apprenticeship systems ease the transition from education to work. Those systems must rely on the growth of enterprises in the formal sector. Investment in better, earlier and longer education might be more effective in promoting the attitudes and competencies required for the world of work. It is questionable that large untargeted youth employment or training programmes have positive rates of return. There is little evidence that such programmes improve either the employment prospects or earnings for the young and especially so for the disadvantaged among them. Narrowly targeted and carefully evaluated programmes can, however, ease the plight of specific youth categories. The effective use of public resources can only be achieved if there are ways to measure the short-, medium- and long-term outcomes of specific strategies. It is strongly argued that detailed evaluations must be conducted as a pre-requisite for the design of any job-creating strategy for the young.

It is hoped that the present study has provided a synthesis of the experience with youth employment, and will thus contribute to a more realistic approach for future strategies.

Samir Radwan
Director
Development Policies Department

This document is primarily based on David Blanchflower™s forthcoming paper: What Can Be Done to Reduce the High Levels of Youth Joblessness in the World? which drew upon contributions by G.K. Chadha, W. van Eekelen, N. Ismail, G. Kanyenze, S. Lijtestein, L. de Luca, S. Mamder, M. Matsumoto, G. Mhone, J. Ramirez, T. Sparreboom, and H. Tabatabai. R. Islam, N. Majid and R. Zachmann, as members of the team implementing the Action Programme, that led to the preparation of this document, provided comments and modifications to the final manuscript. This publication would not have been possible without the dedication and attention to detail demonstrated by Ms. G. Ople.

I. Executive Summary ¹

1. The aim of the ILO's action programme on Strategies to Combat Youth Marginalization and Unemployment is to identify effective policies and interventions to combat youth unemployment and exclusion. To do this, it is necessary to assess both the mechanisms that regulate the demand for labour and identify the strategies that can increase the possibilities of employment and the wages of the young. This, however, must be done in different economic structures and within different business cycles.

2. The work carried out under this action programme has been largely based on the resolution on Youth Employment adopted by the International Labour Conference at its 86th Session, June 1998. It is also a continuation of the work initiated in the 1996-97 biennium under the Action Programme on Youth Unemployment.

4. This paper, which summarizes the work carried out under the Action Programme, describes youth joblessness in the world, assesses some experiences to deal with this problem and suggests what can be done to improve it. In a very summarized manner, research indicates that:

a. There are many similarities in youth labour markets. In particular, it has been found that youth unemployment is approximately double the adult unemployment in most countries of the world. This ratio appears to decline as unemployment increases.

b. In examining possible causes of youth joblessness it has been found that, in general, wages, minimum wages, cohort size, shifts in industry composition, trade, technology and increased female participation are not related to youth unemployment. Reaffirming the fundamental message of the resolution concerning youth employment adopted at the 86th Session of the International Labour Conference, the level of aggregate demand in the economy does appear to play an important role. Contrary to widespread belief, unemployment makes young people very unhappy, which suggests it is not a conscious choice.

c. It is noted that there are substantial supply responses to economic incentives in the youth labour market. High unemployment encourages young people to stay on longer at school and acquire more education. The young are more likely to continue living with their parents these days than was true in the past. It is also observed that there is a number of worrying responses which include increased drug taking, more participation in crime and increased suicide.

d. It has been concluded that increased youth wage flexibility does not seem to be an effective tool to deal with youth joblessness: there is little evidence to suggest that the young are being priced out of jobs. There has been a decline in the wages of the young relative to adults over the last decade in many countries and youth unemployment has not improved. Schemes to encourage self-employment may have some value. Active labour market policies have generally not been very successful in improving the situation of the disadvantaged young. A series of recommendations for narrow targeting and careful monitoring are made in this report.

e. Finally, as it appears that solutions to youth unemployment are driven by what happens to overall unemployment, the effectiveness of the suggested macroeconomic policies in decreasing unemployment is examined. Unfortunately, we are a long way from understanding why aggregate unemployment is so high and why it has trended upwards over the last couple of decades. High unemployment does not seem to be primarily the result of job protection, trade union power or wage 'inflexibility'. There is some evidence that overly generous benefits do tend to raise the level and duration of unemployment by making work less attractive. However, quantitatively, the impact of benefits is small. There are two components of the aggregate unemployment problem to be understood. First, cyclical movements in joblessness - why does unemployment in general and youth unemployment in particular fluctuate up and down in large, irregular cycles? Second, why in so many countries has unemployment trended secularly upwards over the last few decades? It appears that the main explanation for the cyclicality rests with changes in commodity prices in general and the oil price in particular, while explanations of the upward trend are related to aggregated demand, unemployment benefits and labour taxes, the internal mobility of the population, home ownership and the existence of a well-functioning private rental sector.

5. On the basis of present knowledge about the policies adopted and experiences evaluated, it is concluded that:

a. Economic strategies that boost aggregate demand must be adopted;

b. Carefully targeted dual apprenticeship - education systems should be implemented;

c. These systems will be successful if the formal sector in the economy grows, and if active tripartite participation in these schemes is achieved; and

d. Self-employment and small enterprises in the formal sector, backed up with financial services and training, are promoted within carefully targeted population groups.

II. Youth and the world of work: An economic and social overview

7. To place our analysis of youth in context, an examination of some background information on demographics and various measures of living standards across countries illustrates a marked gap between the developed and developing countries (Table 1). These statistical profiles show the brutal disparities between countries. The proportion of the population under the age of 18 was as high as a remarkable 55 per cent (Benin, Niger and Zambia) to a low of 18 per cent in Italy. Among these 192 nations, annual per capita GNP is as low as US$80 (Mozambique) and as high as US$45,360 (Switzerland).

8. The under-5 mortality rate varies from 4 to 320 deaths per 1,000 live births (Singapore and Niger). The percentage of under 5's under weight is as low as 1 (Chile, Croatia, Czech Republic and the United States) and as high as 56 per cent (Bangladesh). The maternal death rate ranges from single figures to 1,800 deaths per 100,000 live births. The fertility rates vary from a high of 7.2 (Niger) to 1.2 (Italy and Spain) while the maternal mortality rate was highest in Sierra Leone at 1,800 compared with a low of 6 in Canada, Norway and Switzerland.

9. The primary school enrolment rate varies from 24 per cent (Afghanistan) to 100 per cent (many OECD countries) of young people. The share of youth in the total labour force, as seen below, ⁵ is continuing to decrease as educational enrolments have increased. The proportion of 12-17 year olds in the world enrolled in secondary school increased from 7.1 per cent in 1970 to 88.3 per cent in 1990. This contrasts with 35.8 per cent and 48.2 per cent respectively in developing countries. ⁶ In all regions, except eastern Asia, the young female labour force is increasing more rapidly than the male labour force.

10. Enrolment in higher education is growing rapidly in all regions and is greater in all countries among young women than among young men, especially in Latin America and the Caribbean. Even with this growth in schooling, which has been substantial in many countries in the developing world, gross enrolment ratios ⁷ vary considerably by degree of development. They range from 10.7 per cent in developing countries to 44.7 per cent in developed countries and economies in transition. ⁸ In most developing countries, education is compulsory for between four and eight years, while in the developing countries it is compulsory for at least eight years. Only a few developing countries have been able to close this gap (e.g. Gabon, Malaysia, Namibia, Peru, South Africa, Sri Lanka and Venezuela). Developing countries have increased public expenditure on education as a share of GNP since 1980. On a per capita basis, East Asia and Latin America increased their public expenditures on education more rapidly than the other developing regions. East Asia more than doubled public expenditures on education per inhabitant, while Latin American countries raised outlays by 30 per cent between 1982 and 1992. ⁹ Despite this growth in spending, the overall gap between the developing and developed countries in per capita public expenditures on education widened between 1930 and 1993.

Source: Calculated by the United Nations Statistics Division from the ILO, Economically-Active Population 1950-2010, 4th edition, 1996 (Paris).

11. The extremely low per capita incomes in some countries have made it hard for communities to find the resources to contribute further to the education of their children. Many developing countries have attempted to extend public primary education by hiring teachers with less formal education but more in-service training and hence lowering salary costs (e.g. Colombia, Senegal, Zimbabwe). Others, such as Zambia and Bangladesh, have raised their pupil-teacher ratios and introduced double shifts which can reduce costs. Many developing countries have moved to favour primary education. In Chile, the share of secondary and higher education expenditures was reduced from 18 per cent and 33 per cent in 1980 to 13 per cent and 21 per cent in 1993 respectively. Bangladesh lowered the share of higher education from 13 per cent in 1980 to 8 per cent in 1992. According to UNESCO, illiteracy rates have continued to decline around the world, falling from 30.5 per cent in 1980 to 22.6 per cent in 1995. They are higher in South Asia (49.8 per cent) than in sub-Saharan Africa (43.2 per cent) or in the Arab States (43.4 per cent). They are especially low in Latin America and the Caribbean (13.4 per cent) and East Asia and Oceania (16.4 per cent).

12. The size of the youth labour force is declining in agriculture and industry and increasing in services in developing regions, northern Africa and western Asia, Latin America and the Caribbean. In south-central Asia, increases in the labour force are roughly equally distributed among the three sectors, but in sub-Saharan Africa, eastern and south-eastern Asia and Oceania, about half of the increase of youth in the labour force is still in agriculture. From 1980 to 1990, services absorbed all of the increase in the youth labour force in developed countries and over half the increase in northern Africa, Latin America and the Caribbean and western Asia. In 1995 nearly two-thirds of the world's youth lived in countries with per capita GDP's of less than US$1,000 per year.

III. Youth in Labour Markets

13. We now turn to a series of statistics (Tables 1 and 2) that describe the extent to which the young are jobless by country and over time. First, changes in the relative size of the youth population aged 15-24 compared with the older age group 25-54 are reported (Table 1). The size of the youth population aged 15-24 years relative to the numbers aged 25-64 has fallen in all developed countries and most developing countries - the main exceptions are Niger, El Salvador, Nicaragua, Paraguay and Pakistan. The youth population is increasing in most of the transition economies.

14. From Table 1 it can be observed that: (1) the youth participation rates are higher in developed countries than in developing countries; (2) the participation rates for young men have declined in almost all countries; and (3) they have declined for women aged 15-19 in most countries, but increased for the older age group of 20-24 year old for most countries, except for the eastern European countries where the rates have declined.

16. Data on current unemployment rates across countries for the 15-24 and 25+ age groups reveals that the youth unemployment rates are approximately twice as high as adult unemployment rates across both developed and developing countries. In a number of developing countries the ratio is considerably higher (Egypt, Colombia, Chile, Honduras, Indonesia, Republic of Korea and Sri Lanka). The ratio is also above two in several southern European countries (Greece, Italy and Turkey), as well as in a number of Eastern European countries (e.g. Bulgaria, Estonia, Romania and Slovenia). ¹¹

17. An examination of male and female unemployment rates for our two age groups shows that the unemployment rates for young men aged 15-24 are lower than the equivalent rates for young women in virtually all of the developing countries and transition economies (Table 2). The pattern is more mixed in the developed world - male youth rates are higher in the English speaking countries (Australia, Canada, Ireland, New Zealand, the UK and the US) plus Japan, Sweden and Turkey, but lower in countries such as Austria, Belgium, Denmark, France, Germany, Italy, Spain and Portugal.

18. Youth unemployment rates are significantly higher in the 1990s than they were in the 1980s in a number of countries. Even where they have not increased significantly they have generally remained high (e.g. Spain at 36 per cent and Italy 32 per cent). The rate has increased markedly in Brazil, Colombia, El Salvador, Nicaragua, Canada, Italy, France, Finland, Sweden and New Zealand but declined sharply in Chile, the Netherlands and Portugal.

19. The average annual changes in the proportion of 20-24 year olds in higher education between 1990 and 1995 show a rise. It is likely that this is in part a labour supply response on the part of the young to the lack of unskilled job opportunities, as well as to the worldwide increase in the demand for the skills. A lack of jobs causes young people to defer their entry into the world of work. The transition from school to work appears to be sensitive to aggregate economic conditions, with the employment and unemployment of youths highly dependent on the rate of unemployment, particularly for younger youths and for those out of school (see Blanchflower and Freeman, 1996b and OECD, 1996).

IV. Youth Unemployment: Causes and Effects

20. A number of possibilities suggest themselves to explain patterns of high and persistent unemployment amongst the young: aggregate demand; youth wages; the size of the youth cohort and a lack of skills. Clearly, in finding solutions to the youth unemployment problem, it is crucial to determine the relative importance of these factors. Many studies have shown the importance of aggregate demand. As argued above, youth unemployment rates are approximately twice as high as adult unemployment rates. It is also the case that young people are not only more likely to quit their jobs voluntarily but are more likely to be fired as well (last in, first out). However, the opportunity cost to firms of firing young workers appears to be lower than firing older workers. Young workers are also less likely to be subject to employment protection legislation. It is often held that the wages of young workers are too high because of the existence of minimum wage legislation, which raises the wage of the young making them uncompetitive, especially compared to married women who, around the world, have entered the labour force in large numbers over the last two decades. There is a good deal of evidence that youth wages relative to adult wages have declined considerably in recent times in many countries. ¹² Furthermore, there is growing literature that suggests that the employment reducing effects of the minimum wage have been greatly exaggerated, especially in the United States where its level is very low. ¹³ The evidence does not seem to suggest that youths are being priced out of jobs in any major way.

21. A further explanation for high and persistent youth unemployment is the size of the youth cohort. The higher the number of young people, the more jobs that will be required to accommodate them. This explanation does not fit the data well, as was discussed earlier, since the size of the youth cohort has been in decline in most countries.

22. Finally, it is argued that in this new technological age the young do not possess the skills that firms need. There is less demand than in the past for unskilled jobs, particularly because of new technology, and this substantially affects the young. ¹⁴

23. No matter what the cause of youth unemployment is, it does seem to have serious consequences especially if the unemployment spells are long or if an individual experiences numerous spells of it. The duration of unemployment spells tends to be shorter for the young than for older workers. There is, however, some evidence across countries that although youth unemployment is of shorter duration than that of adults, the difference is not substantial. ¹⁵ The longer an unemployment spell, the more difficult it is for that person to find work because of the loss of skills, morale, psychological damage etc. There are three major reasons why unemployment while young, especially for frequent or long periods, can be particularly harmful:

a. Early unemployment in a person's career may permanently impair his or her future productive capacity.

b. Barriers to employment can block young people in the passage from adolescence to adulthood, which involves setting up a household and forming a family. There is some connection between youth joblessness and serious social problems such as drug abuse, petty crime and single parent families.

c. High levels of youth unemployment may, at an aggregate level, lead to alienation from society and from democratic political processes, which may give rise to social unrest. Unemployment makes people unhappy (Blanchflower and Oswald, 1999a, Winkelmann and Winkelmann, 1999; Oswald,1997b).

24. In the next three sections, evidence that exists on the causes and consequences of youth unemployment is examined for the OECD, Transition Economies and finally the developing countries of Latin America, Asia and Africa. As much more is known about the workings of the youth labour market in the OECD, the first section on the OECD lays out the ground for other regions of the world. In the following sections, an attempt is made to compare and contrast the situation in the developing countries with that found in the OECD. Interestingly, there seem to be important similarities.

V. Regional perspectives

Youth Labour Markets in OECD Countries ^{16 17}

Changes in the Transition from School to Work

26. Perhaps the most important and positive way in which young persons can respond to poor labour market conditions is by postponing entry into the job market and remaining in school. Without a family to support, youths can invest in human capital rather than struggling to make a living in a difficult market. In virtually all OECD countries, enrolments in school rose from the 1980s through the 1990s. The deterioration in the youth job market seems to have contributed to particularly large increases in enrolments in higher education. The increase has been more rapid in OECD countries, other than the US. In addition to enrolling in school, young persons shifted among fields of study and occupations. In the US students rejected sciences and liberal arts in favour of business related areas, and PhD degrees in favour of professional degrees. The flow of students toward relatively higher paying fields should have increased the earnings of young workers relative to the earnings of older workers, but such a pattern is not found in the data.

27. Is the extension of the period of schooling and the delay of working the result of the state of the macroeconomy, or is it the result of other factors? To what extent is the schooling-employment status of youths sensitive to aggregate economic forces? Blanchflower and Freeman (1999b) examined the data for 15 countries ¹⁹ for the period 1983-94. In addition, data was available for the US from 1970 to 1993 and for Canada for 1976-94, making an overall total of 8,000 observations. The findings on the relationship between schooling and unemployment were mixed. ²⁰ Pooling all of the countries together, schooling was positively related to unemployment, but the diverse country results prevent any broad generalization. By contrast, there is no ambiguity in the effect of aggregate economic conditions on the proportion of a cohort that is neither in school nor working or that was employed. The proportion neither in school nor working, sometimes called iaidlele, falls with unemployment in nearly all countries. In the pooled OECD sample, an increase in aggregate unemployment raised the proportion idle by 0.73 percentage points. Contrarily, unemployment reduced the employment rate of youths by 1.13 percentage points.

28. An examination of education and labour market status of 18 and 22 year olds in 1984 and 1997 shows that education and employment are quite separate activities for many young people. In 1997, on average, 56 per cent of the cohort of 18 year olds were in full-time education, while among 22 year olds, 46 per cent were solely in employment. In terms of the OECD average, the trendsover the period are the same for both: a strong increase of youth in full-time education and a much smaller increase of those combining education with employment; conversely, there has been a strong decrease of those in employment without studying and a much smaller decrease of those neither in education nor in employment.

29. In several countries, a high proportion of young people combine education and work, while in others this is rarely the case. For example, in Belgium youths rarely work while in school. Belgian teenagers are nearly all full-time students. ²¹ In contrast, in the United Kingdom, quite a large proportion of teenagers are working. ²² Interestingly, in both countries, there exists a sizeable group at risk of social marginalization which is composed of teenagers not in education nor in employment; this group accounted for over 20 per cent of the teenage population in the United Kingdom in 1997, compared with about 12 per cent in Belgium. It is also noticeable that the relative size of this at-risk group has shown little change in both countries over the past 15 years (OECD, 1998).

30. The data on the schooling and labour market status of young persons aged 18 and 22 by gender in 1984 and in 1997 (Table 3) show large variations across countries in transition patterns. The high proportion of young persons in vocational training/ apprenticeships in Austria, Denmark and Germany contrasts markedly with those in schooling in other countries. In general, however, they show a general pattern of increases in school attendance and of declines in employment/population ratios and high rates of unemployment in most countries for youths of both genders. The rise in school enrolments is most marked outside the US. ²³

31. The proportion of young men that are idle - that are neither in school nor in the labour force - has increased over the period 1984- 97, especially so in the UK and the US, although the level is considerably higher in the former case - 11.4 per cent and 6.8 per cent for 18 year olds and 8.4 per cent and 5.6 per cent for 22 year olds respectively. The proportion of young women that are idle decreased in the OECD as a whole but increased, as it did for men, in Germany, the US and the UK.

32. Employment to population rates fell between 1984 and 1997 in virtually all the OECD countries, as demonstrated in Table 3. The unweighted average shows that 35.4 per cent of 18 year old men were employed in 1997, compared to 43.8 per cent employed in 1984; and that 29.9 per cent of 18 year old women were employed in 1997, compared to 36.6 per cent in 1984. The comparable figures for 22 year olds show a drop in employment rates for men of 7.0 percentage points, compared to 4.0 percentage points for women. Interestingly, unemployment as a proportion of population declined in most countries for both men and women. Major exceptions to this are found in Australia, France and Canada.

33. Successful transition into the world of work varies considerably by education attainment in every country. In general, the burden of joblessness among the young falls on the least educated and the least skilled. In a number of countries, amongst young men - much less so for young women - the most educated have to wait the longest to find work. Examples where this occurs are Germany, Greece, Italy, Portugal and Spain. What does stand out, however, is how low the unemployment rate amongst the new school leavers is in Germany for the least educated (9.7 per cent for men and 13 per cent for women). This contrasts dramatically with most other countries where more than one-third of such individuals were unemployed one year after completing education. What is perhaps surprising is the similarity in the degree of concentration of unemployment in Germany and the US. Among all Germans 1.6 per cent of the population who experienced at least two years of unemployment accounted for 25 per cent of all weeks of unemployment over the five year period examined. Analogously, in the US 1.8 per cent of the population with at least two years of unemployment accounted for around 20 per cent of total unemployment. This evidence is inconsistent with the view that the transition from school to work is dominated by short spells.

34. A case in which young people do relatively well in booms is examined by Freeman and Rodgers (1999) who analysed the 1990s boom in the United States and found that it substantially improved the position of non-college educated young men, especially young African Americans who are the most disadvantaged group in the US. Young men in tight labour markets experienced a substantial boost in both employment and earnings. Adult men had no gains, and their earnings barely changed even in areas where unemployment rates were below 4 per cent. Youths did particularly well in areas that started the boom at lower jobless rates, suggesting that they would iobenefit especially from consistent full employmentl.. ²⁴ The earnings of youths in the US appear to be especially responsive to changes in the unemployment rate. Similar results have been found in other countries such as the UK, Australia and Canada. ²⁵ A virtuous cycle is set in place as unemployment amongst the young decreases and as the attractiveness of work increases.

35. A number of OECD countries have experimented with labour market programmes designed to help youths in the job market. On the supply side are programmes that link schooling to work before youths encounter difficulties in the market; and second chance programmes that try to increase the skills of youths who have trouble in the job market. On the demand side are programmes that raise youth wages, for instance through minimum wage, or that target some employment opportunities at youths. On the basis of aggregate outcomes, the German apprenticeships seem to be a highly successful supply-side programme. Less educated young workers have lower unemployment rates and higher relative earnings in Germany than in the US. In the first five or so years of work, fewer young Germans are jobless than young Americans. Apprenticeships offer a good return for most young persons. However, the German apprenticeship system has its own problems. The number of apprenticeship contracts has fallen as more youths have chosen higher education. Youths who do not find a job immediately after their apprenticeship face a comparatively long period of non-employment, and those who fail an apprenticeship programme suffer long-term reductions in earnings. The apprenticeship system does not improve the effects of family background; children of blue-collar and white-collar employees were more likely to be employed subsequently than children of non- employed parents. ²⁶

36. By contrast, second chance programmes, including Sweden's much heralded active labour market programmes, do not seem to be overly effective. ²⁷ There is also considerable evidence that large-scale programmes designed to move young people from unemployment to work, such as Youth Opportunities Programme (YOP) and Youth Training Scheme (YTS) programmes that operated in the UK in the 1990s, were ineffectual. For example Dolton et al. (1994) found that YTS lowered the probability of subsequent employment. Some studies such as O'Higgins (1994) have found more positive effects on employment but he finds no significant employment effects for the disabled and ethnic minorities. The substantial variation in magnitudes of such estimates has lead Ryan and Buchtemann (1996) to question the reliability of these studies. Moreover, research on the effect of YTS on earnings by Green et al. (1996) have generally found a negative effect (see O'Higgins, 1997).

37. For many years Sweden was viewed as having solved the problem of joblessness and economic inequality. During the 1970s and 1980s, young workers fared reasonably despite sharp increases in the youth's relative wages. However, the recession of the early 1990s proved that Sweden was not immune to substantial unemployment nor to a major youth joblessness problem. In the 1990s, youth unemployment has risen sharply, and the state has expanded youth participation in active labour market programmes. This has reduced unemployment somewhat without solving the joblessness problem. Indeed, the increase in unemployment has been roughly proportional by age and education, implying that these programmes have not altered the relative distribution of unemployment. The proportional growth of joblessness suggests that aggregate factors were more important in Sweden's joblessness than disaggregate shifts in demand for labour among different skill groups. ²⁸

38. France has a wide variety of youth programmes and indeed leads the advanced countries in the proportion of youths employed under some special programme. France also has relatively high minimum wages which, in contrast to the minimum in the United States, has increased quite strongly in recent years, possibly leading to an adverse impact on youth employment. To some extent, these two factors offset one another. The real minimum hourly wage in France (the Salaire Minimum Interprofessionel de Croissance [SMIC]) has risen steadily since 1967 whereas in the US, the Federal Minimum Wage has declined. In 1990 approximately 28 per cent of French workers were at or below or within 5 Francs per hour of the minimum. In 1987 in the US, only 18 per cent of employed persons had hourly wage rates at or below the minimum or within an additional US$1.00 of the minimum. Young workers paid around the minimum wage in France were more likely to become unemployed or move out of the labour force than those paid over the minimum wage. ²⁹ While an analogous similar pattern is found in the US, where a larger share of workers employed at or around the minimum wage were either unemployed or out of the labour force in the previous period than was true among workers above the minimum wage, the smaller scope for the minimum implies less of an impact on the youth market. The employment effects of the minimum wages in France are mitigated somewhat by participation in employment promotion programmes that shield workers from some of the effects of the increasing real SMIC. When this eligibility ends, the probability of subsequent non-employment rises sharply.

Evaluating the causes: wages, cohort size, changing industry structure, the rise in female participation rates, aggregate demand or other factors.

39. The country which seems to have most successfully dealt with the youth problem is Germany. While some German youths have great trouble in the job market, young, less educated Germans have done markedly better in terms of both employment and wages than comparable youths elsewhere. The situation for young women is less worrying, as young women have continued to move into the job market in increasing numbers and as female pay has improved relative to male pay. Still, in the late 1990s, young women earned less than seemingly comparable young men and experienced a similar twist in the age-earnings profile against them. The unemployment rate for young women workers has risen in most countries, and in the US and the UK, poverty has become increasingly concentrated among single parent female-headed households. What is puzzling about the deterioration of the job market for young workers is that economic forces operate to raise the relative position of youths. Movements of wages, cohort size and industry all appear to work in youth's favour. In addition, the increased years of schooling and skill of younger workers relative to that of older workers should have raised their relative pay and employment. In short, things did not work out as expected in the youth job market. We will consider each of these factors in turn:

b. Theminimum wage does not seem to play a major part in explaining the poor economic performance of the young in OECD countries. The findings on the extent of the impact of the minimum wages on employment are mixed. ³⁴ Regardless of the wage experience, however, youth unemployment rates rose substantially everywhere except in Germany.

c. Cohort size. Korenman and Neumark (1999) have documented that the youth proportion of the population declined in virtually every OECD country in the 1980s and 1990s. Declining youth cohort size should lead to lower unemployment rates for youth and higher relative earnings for youth. This should be particularly marked in countries like Japan, Ireland, Italy, Spain and Portugal, where the fall in the relative size of youth cohorts was exceptional. However, the economic position of youths worsened rather than improved. That the demographic changes failed to improve the position of youths much does not mean that the shifts in supply have no effect on the youth job market - the elasticity of youth unemployment rates with respect to relative youth cohort size may be moderately large (Korenman and Neumark, 1999). Rather, it means that other factors, such as the aggregate rates of unemployment or the technological changes, dominated youth job market outcomes.

d. Industry, technology and trade. In addition to the demographically induced decline in the number of young workers and the declines in youth wages relative to adult wages documented above, there was a shift in the industrial composition of employment toward sectors that hire relatively many workers - retail trade and services like hotels and restaurants. ³⁵ This should have increased employment if not the wages of young workers. Furthermore, the technological factor that many analysts cite as underlying the long run rise of inequality and higher premium to skill - computerization - should have benefited the young, who have grown up with computers, relative to older workers who have not. Increased trade with the Third World countries is another potential determinant of the deteriorated economic position of young workers. On a world scale, the share of youths in the working age population is much larger than in advanced countries. Thus, trade with LDCs might be expected to reduce the relative position of young workers. Nevertheless, the sectors which compete most with less developed countries are those such as apparel that traditionally employ women workers; so one would expect trade to have lowered their wages or employment rather that of young men.

e. Increased female participation. The influx of women into the job market may also have affected the economic position of young workers. Many women workers are new entrants or re- entrants into the job market who might fill jobs that younger workers would otherwise hold. Such a story is difficult to sustain as female pay has increased with the growth in the supply of female workforce and since the effects of such increase in the supply of women are expected to be greater on women than on substituting young workers.

f. Other explanations. While the increased supply of competitive workers due to women or trade may have affected the position of young workers, these forces do not seem to be sufficiently powerful to counteract the demographic and demand factors that favoured young workers. To explain the observed deterioration in terms of labour supply, we must argue that workers in the baby boom generation are highly substitutable with younger workers such that the baby boom cohort reduced not only their earnings but those in the ensuing smaller cohorts as well. As the baby boom cohort gets older and older, however, and as the economic position of young workers remains depressed, this becomes an increasingly tenuous claim. There is yet another supply side possibility: the young workers are simply not as good as older workers. However, the 1994 survey results for all countries except the US and Ireland show that younger workers are more skilled than older workers (OECD, 1997a). ³⁶ It is difficult to argue the case that the shifts in demand or supply or for deterioration of youth skills caused the worsening job market for young workers.

g. Aggregate demand. If demographic factors and long-term demand worked to improve the situation of young workers, why did their economic position deteriorate? The main reason appears to be that the aggregate unemployment was relatively high in OECD countries in the 1980s and 1990s. The demand for young workers is highly sensitive to aggregate economic conditions (Blanchflower and Freeman, 1996; Clark and Summers, 1982). As new entrants to the job market, young workers lack the specific training or seniority that buffers older workers from swings in market conditions. Their employment is highly dependent on the aggregate state of the labour market. High rates of unemployment in the EU thus go a long way to explaining the prevailing rate of youth joblessness. ³⁷ The fall in joblessness in the US in the late 1990s produced some rise in youth wages, as well as in employment, after two or so decades of decline even though it did not come close to restoring the relative position of young workers.

40. Many analysts would expect the relative employment of youths to vary inversely over time with their relative wages. Perhaps greater declines in youth wages generated more jobs for them in some countries, but the declines that did occur, including the large drops in youth wages in the US, did not suffice to stabilize, much less raise, youth employment to population rates. One interpretation is that the wage and employment numbers lie along labour supply curves, due to massively declining labour demand for young workers. Another interpretation is that the concordance of joblessness and falling pay reflects disequilibrium in the labour market, also the result of declining demand for young workers. Whichever way, we have identified one basic pattern in the worsened job market for young workers: the disproportionately large response of youth employment or unemployment to changes in overall unemployment. Unless overall rates of unemployment are reduced, there is little prospect for improvements in youth outcomes in the OECD, even if youth shares of the population continue to fall or remain relatively small or even if the composition of employment shifts modestly toward service sectors that hire relatively many more youths.

41. To summarize, there is little evidence then that the size of the youth cohort (which is in relative decline almost everywhere), or the level of youth wages (which have been falling) or the existence of minimum wages (which are low) explain the rise in youth unemployment over the last couple of decades. Changes in aggregate demand, increased demand for skilled workers and the rising participation of women who compete with the young for jobs appear to constitute the main explanations for the increase in youth unemployment. Youths in the OECD appear to have responded to the worsening job market by deferring entry and undertaking more education.

Consequences of youth joblessness

42. A number of other important changes in society accompanying the high and rising levels of youth unemployment are correlated with a number of other social outcomes:

a. Unemployed youths are increasingly concentrated in workless households. Of considerable concern is the fact that the proportion of teenagers and young adults (20-24) living in households in which nobody else is employed has risen in the EU as a whole and especially in Belgium, France, Germany, Ireland and the UK. The share of unemployed youth living in workless households is, at over 40 per cent, highest in Finland, Ireland, and the UK and lowest in the southern European countries, Austria, Luxembourg and Switzerland.

b. Increasing proportions of young people are living with their parents. In Canada, France, Greece, Italy, Portugal and Spain there has been a strong increase between 1985 and 1996 in the proportion of young people living with their parents. In Canada and the US, low youth wages increased the likelihood that young women would remain living with their parents and that they would attend school; while low employment rates raised the chances that women would remain in their parents' home with only a marginal impact on their rate of school attendance. The proportion of young people living with their parents are especially high in Spain. Interestingly enough, Spain has the highest rate of home ownership in the OECD. ³⁸

c. The young are increasingly involved in crime. Large numbers of young American men committed sufficiently serious crimes in the 1980s and 1990s to make 'prisoner' just about the fastest growing occupation among the young. This incarceration rate is approximately ten times higher than other western countries. In the UK, which has the highest rate of incarceration in western Europe, just under 50,000 people were in prison in 1995 or 0.13 per cent of the population aged 15-64 (48,983 men and 1,979 women). ³⁹ Many young persons involved in crime are employed before their arrest, suggesting that they have reservation wages for both legal and illegal work. ⁴⁰ The reaction of youths to the deteriorated jobmarket in terms of enrolments, residence in parental homes, and crime suggests substantial supply responsiveness to economic incentives, which may augur well for the future.

d. Increasing numbers of young people are committing suicide. The gender disaggregated death rates per 100,000 by suicide and self-inflicted injury for young and older persons for 22 countries for 1970, 1980 and 1992 show that the suicide rates are in all cases higher for men than for women. Across the countries, there is a wide variation in both the adult and youth rates and considerable variation in the pattern of change. In English-speaking countries - the US, Canada, UK, Australia, New Zealand and Ireland - rates of suicide rose sharply, which could potentially reflect the rising problems for youths in the job market in those countries, in particular, the increase in inequality that marked the 1980s. However, the rates of suicide also rose among young men in Norway, where earnings inequality is small and the social safety net high. That youths in these countries report themselves as being happier or more satisfied with their lives ⁴¹ further complicates any simple interpretation of these patterns and their link with the increasingly elongated transition from school to work.

Youth Labour Markets in the Transition Economies of the Former Soviet Union

43. Are the patterns observed in the OECD repeated in the group of countries that were members of the former Soviet Union? The answer is a qualified yes. There has been some work on the labour market situation of the young in the ex-communist countries of the Soviet Union since the fall of the Berlin wall in 1990. Lack of quality data has made such an analysis difficult, but the situation appears to be improving. The countries for which the best data exists include former East Germany; there is also some good published data available on Hungary, Poland and the Czech Republic, who have recently joined the OECD. There are also a number of micro-data files that have recently become available. Their analysis has shed some light on the situation in these countries. Examples of such surveys include the German Socio-Economic Panel, the East Europe Eurobarometer Survey series, the International Social Survey programme (ISSP), the Polish Social Survey, the Russian Longitudinal Monitoring Survey and the Russian Survey of Employment, Income and Attitudes. Analysis of these labour market surveys for eastern Europe is reported in Blanchflower and Freeman (1996), Blanchflower and Oswald (1999), Hunt (1999), Krueger and Pischke (1995), Kollo (1998), Blanchflower (1999b).

44. It is apparent from the data on general indicators of demographics and living standards and of labour market (Tables 1-2) that:

a. The Central Asian Republics such as Armenia, Azerbaijan, and Tajikistan are poor and similar to many other South Asian countries. The transition economies in Eastern Europe are much richer. Compared to many developing countries, maternal and child mortality rates are relatively low in countries like Hungary, Poland and the Czech Republic.

b. By 1997, approximately a quarter of the population of the European transition economies was under the age of 18. This is a significantly higher proportion than is found in most Western European economies in which closer to 20 per cent of the population is under the age of 18.

c. In contrast to most Western European economies, the relative size of the youth population aged 15-24 has been rising in most of these countries in the 1990s.

d. The level of aggregate demand varies a lot across the transition economies for which we have data. It is particularly high in the Former Yugoslav Republic of Macedonia (1997=38.8 per cent) and very low in Tajikistan (2.7 per cent), Uzbekistan (0.4 per cent) and the Czech Republic (1997=4.7 per cent). It is unclear what to make of the unemployment rate in many of these countries. Does it indicate a well-functioning labour market (Czech Republic) or one where little adjustment to capitalism has occurred (Russia)?

e. As was found in OECD countries, the level of youth unemployment is generally between two and four times higher than for older workers. The ratio of youth to adult unemployment was highest in Romania at 4.7 times.

f. Based on the most recent estimates we have (mainly 1997), there is no obvious pattern in youth unemployment rates between men and women. Female rates are higher in Belarus, Croatia, Czech Republic, Macedonia, Lithuania, Poland, Romania and Slovakia; but are lower in Hungary, Latvia, Russia and Ukraine. They are approximately the same for the two sexes in Slovenia.

g. The proportion of 20-24 year olds in the 'third-level' or higher education has declined in Belarus, Latvia, Russia and Ukraine but increased strongly in the Czech Republic, Hungary and Estonia and has increased but to a lesser extent elsewhere.

45. Adjustment from a centrally planned economy to a market economy has been difficult. Not only have there been dramatic increases in unemployment rates, but also declines in real wages and considerable widening of the overall wage distribution. Year to year swings are very pronounced in many of these transition economies. Declines in real wages have been dramatic in many countries, especially from the CIS. The annual changes in real wages (1989-95) are presented in Figure 2 and Figure 3. ⁴²

Source: Report on the World Social Situation, 1997, United Nations, New York, p. 124.

46. A recent review ⁴³ of micro-data for randomly sampled workers in both Eastern and Western Europe reached three conclusions:

a. Microeconometric studies demonstrate similar behaviours of unemployment both in Eastern Europe as in the industrialized West;

Source: Report on the World Social Situation, 1997, United Nations, New York, p. 124.

b. Unemployed people in the transition economies are as unhappy, relative to the employed, as those who are jobless in the industrialized Western countries. Such a result sheds doubt on the idea that voluntary or benefit-induced unemployment is worse in the East; and

c. Estimating a 'wage curve', (the unemployment elasticity of pay following Blanchflower and Oswald, 1994 - see text box) using pooled data from five East European nations, produces a local unemployment elasticity of pay fairly close to - 0.1, which is the figure commonly found for the rest of the OECD. Such findings cast doubt on the argument that wages are inherently less flexible in the East. Keune (1998) examined youth unemployment in Hungary and Poland and argues that "In general, the main factors explaining youth unemployment are the same as the ones explaining the general unemployment".

47. The broad conclusion from the analysis conducted to date on the transition economies is that the workings of the labour markets of East and West appear surprisingly similar. To understand youth unemployment, it is necessary to understand adult unemployment. There is another side to this coin: youth employment policy is unlikely to be successful if it ignores the aggregate unemployment picture. There is probably no distinctive solution, therefore, to Eastern Europe's unemployment (whether youth or adult). Instead, there is only a single problem - joblessness in Europe.

Youth Labour Markets in Developing Countries

Latin America

48. The countries of Latin America show considerable variations in wealth, degree of development and labour market performance. ⁴⁴ In terms of income, the country with the highest GDP per capita is Argentina (US$8,380) and the lowest is Nicaragua (US$380) (Table 1). Similar differences are found on other dimensions, such as maternal and under 5 mortality rates. The differences in the proportion of the population that is young (<18) shows enormous variations from a high of 50 per cent in Nicaragua to a low of 28 per cent in Uruguay. Although the size of the youth population is declining relative to the adult population (Table 2) in most countries in Latin America (e.g. Brazil, Chile and Peru), it has been increasing in Bolivia, Nicaragua and Paraguay.

49. The growth of employment in the informal sector is of particular concern in the region. It is estimated that 85 per cent of all new jobs are created in the informal economy. ⁴⁵ Only in Chile and Colombia did employment in this sector diminish during the 1990s.

Source: ILO News, Latin America and the Caribbean: 1998 Labour Overview.

Note: a/ Greater Buenos Aires, May 1998; c/ 6 metropolitan areas, Jan-Sep 1998 average; d/ national total, Jun 1998; f/ national urban; g/ 41 urban areas, 1998 1st quarter; j/ Metropolitan Lima, as of 1996, national urban, 1998 2nd quarter; k/ Montevideo, Jan-Sep 1998 average; l/ national urban, national total, 1st semester 1998.

50. There are large differences in labour market performance. Tables 1-2 also suggest the following:

a. Unemployment rates are generally higher in the 1990s than they were in the 1980s (Table 2). The main exception to this is Chile, where unemployment has declined dramatically. In Chile the unemployment rate was as high as 19.6 per cent in 1982; by 1997 the rate had fallen to 5.3 per cent. Unemployment rates are especially high in Argentina (1995=16 per cent), Colombia (1997=12.1 per cent) and Panama (1996=14.3 per cent). They are low in Bolivia (1996=4.2 per cent) and Mexico (1997=3.5 per cent).

b. Higher unemployment rates for the least educated are found in the more developed of the Latin American countries (e.g. Argentina, Brazil, Chile, Mexico and Uruguay). Urban unemployment rates in 1994-95 for those with 13 or more years of schooling compared with those with 0-5 years are presented in Figure 5. ⁴⁶ The most educated have higher unemployment rates than the least educated in Bolivia, Ecuador and Venezuela. It must be recalled that in most OECD countries unemployment rates are highest for the least educated - the two major exceptions are among the poorest member nations, Greece and Republic of Korea. ⁴⁷

c. Over the last 15 years or so, the size of the youth cohort relative to that of adults has declined in most Latin American countries and dramatically so in some, such as Colombia, Costa Rica, Honduras, Puerto Rico, Trinidad & Tobago and Venezuela. It has remained roughly the same in Bolivia, El Salvador and Uruguay but has increased substantially in Nicaragua and Paraguay.

Growth per annum in % in higher education, 1990-95
	Men	Women
Bahamas	-7.1	5.4
Brazil	2.4	3.0
Chile	4.9	4.3
Colombia	3.8	3.3
Dominica	0.8	12.4
El Salvador	1.7	16.2
Guyana	5.6	10.6
Honduras	5.3	6.2
Mexico	2.9	6.0
Nicaragua	9.7	5.4
Trinidad & Tobago	4.3	8

d. Youth unemployment rates are approximately twice as high as adult rates in most countries. Youth unemployment seems to be a particularly serious problem during the late 1990s in Argentina (24.6 per cent), Colombia (35.1 per cent), Panama (27.3 per cent) and Uruguay (24.6 per cent). As can be seen from the Figure 4, ⁴⁸ youth unemployment rates for the youngest age groups in a number of countries are worryingly high, even in Chile which has experienced strong declines in aggregate unemployment.

e. There has been a considerable growth in the proportion of those aged 20-24 going to college in most Latin American countries in the 1990s. The table above lists average growth rates of growth of students in tertiary education by gender. ⁴⁹ The rates for women are particularly noteworthy.

Source: Social Panorama of Latin America, 1997, Economic Commission for Latin America and the Caribbean, United Nations, Santiago, Chile.

51. Urban minimum wages exist in most Latin American countries. The coverage of minimum wage laws is incomplete, and their enforcement is weak and likely to vary considerably across countries. As can be seen in Figure 6, there is considerable variation across the countries in how the minimum wage has changed. ⁵⁰ There appears to be no simple relation between changes in the minimum wage (bars measured in left hand side axis) and the levels (not graphed) or changes (depicted as circles and measured in the right hand axis -note where the zero is) in unemployment. In a few countries, growing real minimum wages are associated with increases in unemployment, especially in Argentina, Brazil and Ecuador and vice versa in Colombia, Costa Rica, Guatemala, Honduras and Paraguay. The pattern is more mixed elsewhere. There seems to be little empirical work done on the role of minimum wage in Latin America, with the main exception being the work of Reynolds and Gregory (1965); Card and Krueger (1995); and Castillo-Freeman and Freeman (1992) on Puerto Rico. Minimum wages in Mexico were found by Feliciano (1998) to have little effect on male employment but a small negative employment effect on females. Lustig and Mcleod (1997) found minimum wages increased unemployment but lowered poverty.

52. There has been rising income inequality in many Latin American countries, especially in Brazil, Chile and Colombia. Poverty rose between 1980 and 1989 from just over a quarter to just under one-third of the population. ⁵¹ It was highly localized to a subset of countries - in 1989 Brazil had 45 per cent of the continent's poor but only 33 per cent of its population. At the same time, there has been declining income inequality in Bolivia, Chile, Colombia and Uruguay. ⁵²

Source: Tardanico, 1997.

Labour Market Programmes in Latin America

53. Since the beginning of the current decade, training programmes focussed on unemployed young have spread rapidly through several Latin American countries. These were pioneered in Chile in the late eighties under the name of "Chile Joven" (CHJ). The programme is specifically directed at young people in a situation of "social risk and/or structural unemployment". The CHJ makes use of short training and apprenticeship programmes that help young people to acquire basic skills to be eligible for work. The scheme, originally implemented with the intention that it should last for four years (1991-95), was extended for a further four years. The training and occupational practice process normally lasts about six months (from 200 to 300 hours' training and two to three months of work practice, although in some variants of the Chilean case training reaches up to 420 hours). The programme is highly decentralized and relies on around 1,000 private training centres which undertake public bidding for training contracts. While they are on the programme, the users get a maintenance and transportation subsidy to encourage them not to drop out. Usually, the subsidy is about 50 per cent of the minimum wage in force. Firms who take trainees are not obliged to remunerate trainees or give them employment subsequently.

54. There are some interesting features that must be underlined in the Chile Joven scheme:

a. Programmes must include both training and practical apprenticeships in the private sector; this aims at ensuring that the training and experience acquired correspond to the market needs;

b. Considerable attention has been given to targeting; special programmes have been designed for specially marginalized population segments; and

c. Assessment mechanisms to evaluate the performance of the programmes were introduced as an integral part of the exercise.

55. Evaluation studies have stressed the success of the CHJ in promoting employment. The programme enrolled more than 128,000 young people, well above its target. In the first three years, almost 60 per cent of the young people found jobs at the end of the programme, compared to 40 per cent for those who did not get the training. Around 55 per cent of the participants were employed in the company where they took the traineeship compared with 41.3 per cent in the control group which was made up of youths residing in the same neighbourhood with the same socio-economic characteristics as the programme participants. The difference was even greater for women - 45.5 per cent and 27.0 per cent respectively, and those who were relatively younger had the greatest problems in finding a job. Of these, the overwhelming majority came from the target population: 95.6 per cent came from low income sectors, and 79 per cent were below the age of 24. ⁵³

56. Unfortunately, it is unclear from these studies: a) the extent to which the trainees are cheap substitutes for existing workers who could potentially lose their jobs to the subsidized (lower cost) trainees; b) whether the jobs the trainees obtain are long-lasting - evaluation was generally focused on what happened six months after completion; c) whether the schemes teach real skills; and d) whether the training actually represents a route out of poverty. "Evaluation must at least make an attempt to take into account what would have happened in the absence of such programmes". ⁵⁴ Moreover, it must be noted that the Chilean Labour market was quite tight during the first phases of the interventions. The issues of evaluation are examined in greater detail later in this document.

57. In the rest of Latin America, there have been other ambitious schemes to improve the economic situation of the young, particularly in Argentina and Brazil, but also in Colombia, Peru and Uruguay. ⁵⁵ In Argentina "Proyecto Joven",which is a variant of Chile Joven, has been in existence since 1994. It is addressed at young persons with employment problems from low-income homes, low educational levels and little or no occupational experience. By training youth, the programme attempts to increase their productivity and instill in them values and attitudes which are thought to improve their chances of obtaining and keeping a job. Courses are completely free of charge, including teaching materials, tools, transportation, inputs, safety and hygiene implements and other elements required for training purposes. The programme has set itself the goal of training 280,000 people. The first phase, which started in 1994 and which recently came to an end, had over 100,000 young participants; another 180,000 will be trained in the following three years.

58. There have been some successes in Argentina, but they have been smaller than found in Chile, especially for women. The overwhelming majority of participants are poor (80 per cent belong to low-income families) and only 7 per cent had finished secondary education. Males increased their employment rates from 43.7 per cent to 61.3 per cent over the 11 month interval between the training and the survey, compared with 51 per cent and 51.9 per cent for the control. Results for females are less convincing - employment went up from 35.4 per cent to 38.6 per cent whereas the control group did better, increasing from 35.3 per cent to 41.5 per cent. There are some concerns about the nature of the control groups as they were generated ex-post and do not appear to have matched characteristics to the treatment group. ⁵⁶

59. In 1996, Brazil implemented PLANFOR targeted towards youths, the unemployed and the dispossessed. During the first year of operation the programme trained close to 1.2 million workers. The programme targets rural areas: blacks and non-whites are over- represented. The programme is much more heterogeneous than the Chile Joven but does appear to be having some success. There is some evidence of statistically significant impacts on employment and wages of men and older workers but lesser effects for women and younger workers (de Moura, Castro and Verdisco, 1998) Colombia has implemented the programme of Occupational Training for Young People. Its general objective is to help low income young people from 17 to 25 years of age that are unemployed and have not completed their secondary education, by giving them semi-skilled training in occupations for which there is evidence of demand in the productive sectors. Peru has implemente their own Pro-Joven programme of Youth Occupational Training. The purpose of Pro-Joven is to supply semi-skilled training and labour experience to low-income young people in specific trades in demand in the productive sector. It thus endeavours to face the problem of access of deprived young persons into the labour market. It is intended to have an intake of 150,000 youngsters over a period of five years. Uruguay's PROJOVEN is much smaller than the programmes in Brazil, Chile, Argentina, Colombia and Peru. An initial set of pilot schemes organized training courses with a coverage of only 4,090 young people between 1995 and 1996. To date, they have trained 1,200 to 1,500 young persons a year.

60. Castro and Verdisco (1999) compare the projects in Chile, Argentina and Brazil regarding the quality of the training they offer and the targeting mechanisms used, and come to a conclusion that the two iwJovenlo schemes are strong in targeting but weak in quality, while the courses sponsored by PLANFOR tend to be of good quality but poorly focused. Those in charge of the Chile Joven programme point out that the growth experienced by the Chilean economy is the key factor in the success ascribed to the project. According to Messina (1995), the only positive effect in the Chilean case would be the opportunity of a temporary labour experience for beneficiaries.

61. Just as in Europe and the US, the evidence on the effectiveness of training schemes in Latin America is mixed at best (reference is made to paragraphs 35 to 38 above).

Asia and Africa ⁵⁷

62. The lack of good data makes it difficult to evaluate the extent of the youth labour market problem in many Asian and African countries. Tables 1 and 2 suggest considerable variation in the levels of GDP per capita in both Asia and Africa (cf. Gabon and Eritrea; Singapore and Nepal) and similarly for other variables such as infant death rates and primary school attendance. There are no consistent patterns of youth unemployment - they are very high in some countries (e.g. Algeria, Egypt, Mauritius, Sri Lanka) and very low in others (e.g. Burkina Faso, Central African Republic, Ethiopia, Thailand). Female unemployment rates are generally higher than male rates. There has been considerable growth in Asia and Africa, but presumably from a very low base, in the proportion of young people in 'third-level' or higher education, more prominently in general in Africa than in Asia and also more notably so for females rather than males.

63. The size of the agricultural sector in these countries tends to be large and in Africa the majority of the young live in rural areas. In contrast, for some Asian countries, ⁵⁸ the proportion of youth in the urban areas is uniformly higher than in rural areas. As a measure of labour market slack, the unemployment rate is generally not useful in rural areas, especially in countries where no unemployment benefit system exists, since unemployment in the developing world is primarily an urban phenomena. Where unemployment rates for rural areas exist they are almost always much lower than in urban areas. Youth unemployment rates tend to be higher than adult rates in these countries being, on average, twice as high as overall unemployment rate in some African countries. ⁵⁹

64. The relationship between youth unemployment and educational attainment is mixed. In Africa, the unemployment rates for the least and the most educated tend to be lower than for those with intermediate level of education (e.g. South Africa). ⁶⁰ On the other hand, in Asia, they are generally higher for the more educated than the less educated. This is partly due to the increased participation in education, resulting in a greater number of educated youths. For example, in Indonesia, among the rural male unemployed adolescents (15-19 years old), more than a third have completed primary schooling, more than a third have had a junior high school degree and about a fifth possess senior high school degree; their female counterparts are slightly better placed in this regard.

65. Underemployment is the main problem in rural areas. For example, a much higher proportion of underemployed young men, women and persons are found: (1) in rural Indonesia, 45.2, 60.5, and 50.9 per cent respectively, compared with 23.5, 23.8 and 23.7 per cent for their urban counterparts; and (2) in agriculture and allied activities in the Philippines, 52.0, 70.0, and 55.0 per cent respectively, compared with 9.0, 15.0 and 13.0 per cent among service workers, 13.0, 12.0 and 12.0 per cent among production and related workers. ⁶¹ In general, for some Asian countries, the proportion of underemployed youth in the rural areas is more than twice as high as in the urban areas.

66. In comparison with the OECD as well as the transition economies and the Latin American countries, much less empirical work on wages and/or youth unemployment has been undertaken on other developing countires. The main published exceptions are Hoddinott (1996) on Côte d'Ivoire; Dickens and Lang (1995) on Sri Lanka and Schultz and Mwabu (1998) and Moll (1993) on South Africa; Blanchflower (1999b) on the Philippines as well as a new crop of unpublished papers by Rodgers and Nataraj (1998) on Taiwan and Kingdon and Knight (1998) on South Africa.

67. A number of generalized facts emerge from a set of country studies commissioned by the ILO ⁶² on the nature of the youth labour market in Asian and African countries:

a. Youth unemployment in most developing countries is perceived as a major problem. In many countries young people constitute a very high proportion of both the total population and total unemployment, comprising more than half the total number of the unemployed in Africa. In many countries the young unemployed are looking for their first job (e.g. India).

b. Inadequate and incomplete data make it hard to know exactly the scale of the problem. Some of the features of the problems related to the data for the purpose of making cross-country comparisons are: (1) cross-country variations in the definition of youth; (2) cross-country variations in measurement of employment, unemployment and underemployment; and (3) country-specific data gathering and survey systems. ⁶³

68. In general, Africa is thought to be caught in a number of self-reinforcing, vicious circles:

a. African economies are unable to generate adequate growth rates in GDP and enough employment and income generating opportunities to absorb the majority of their labour forces;

b. The inability of the private sector to generate sustainable livelihoods has given prominence to rivalry over control of the state as a primary means for attempting to share in whatever fruits of the economy that there may be, further exacerbating the possibility of unrest; and

c. The capacity of the state to govern and deliver with respect to social services, such as education, and the security is compromised. ⁶⁴

69. High inflation and restrictive macroeconomic policies have especially harmed youth. The main feature of the African labour market is the slow growth of employment in the formal sector and the retrenchment of labour in the course of implementing structural adjustment. In response, the non-formal sectors have not only acted as residual sectors but also as labour absorbers of last resort. Such structure of wages and unemployment can have some perverse effects on youth:

a. High unemployment rates may discourage youth from investing in education and training as the investment appears wasted;

b. The association of increasing age with increasing probability of employment may result in a passive approach to job search; and

c. Youth who have relatives in wage employment may develop a dependency that makes them have a high reservation wage for entry into formal employment (ILO/SAMAT, 1999).

70. In a different context, the south-east Asian experience has demonstrated that youths are more vulnerable to external shocks such as the financial and the resulting socio-economic crisis since they are the first ones to be retrenched and face greater difficulties in finding employment. Overall, there seems to be some recognition that the problem is closely linked with adult unemployment and overall economic performance.

71. Various strategies have been tried unsuccessfully, and the public sector job generation has not worked. The World Bank and IMF have come to the aid of countries in financial crisis but forced them to reduce public sector employment (e.g. United Republic of Tanzania, Vietnam). There have been some attempts at reforming the existing educational and training system in Africa (e.g. Zambia). The main aim of such reforms is to increase the relevance of education and training system, making it better geared towards the demands of the labour market. ⁶⁵

72. Minimum wages exist in many countries (e.g. Indonesia, United Republic of Tanzania, Mali, Côte d'Ivoire, India, Mauritius, Zimbabwe, South Africa). High levels of the minimum wage apparently exist in Mauritius, Zimbabwe and South Africa which may reduce employment. ⁶⁶ The absence of a youth sub-minimum is likely to have the largest employment consequences on the young who tend to have the least skills, if the minimum is actually implemented. Lustig and Mcleod (1997) found minimum wages increased unemployment but lowered poverty in four African countries (Ghana, Mauritius, Morocco, Tunisia) and five Asian countries (India, Indonesia, Philippines, Sri Lanka, Thailand).

73. A number of countries have experimented with some success with policies to increase self-employment in both urban (e.g. PMRY in India) and rural areas (e.g. TRYSEM also in India). These include workshops on how to set up in business (e.g. Mali and Zimbabwe), provision of lines of credit and advisory activities to help the creation and survival of small businesses (e.g. Philippines, Indonesia, Sri Lanka, Zimbabwe, Mali, Côte d'Ivoire and Cameroon). It is increasingly the case that reforms of the existing educational and training institutions are integrated into programmes for entrepreneurship development and enterprise promotion. ⁶⁷

India: Training of Rural Youth for Self-Employment (TRYSEM) TRYSEM is a facilitating component of a poverty eradication programme which aims at providing basic technical and entrepreneurial skills to the rural poor in the age group of 18 to 35 years to enable them to take up self- or wage- employment. At least 40 per cent of the beneficiaries have to be women. Train- ing is imparted both through training institutions and through the non- institutionalized mode, e.g. master craftsmen functioning from their own place of work. Every TRYSEM trainee becomes eligible to avail of assistance for setting up a self-employment enterprise. During the Eighth Plan (1992-97), 1.528 million youths were trained under TRYSEM, of whom 34.2 per cent took up self- employment and 15.0 per cent wage-employment, while the remaining 51.8 per cent remained unemployed (Govt. of India, 1999, p. 14). The situation was not satisfactory during the earlier years either. For example, over the 16 years between 1980 and 1996, nearly 3.9 million rural youth were trained but only 53.0 per cent of them were employed; almost a quarter of them had found work as wage employees rather than as self-employed 'entrepreneurs' (Visaria, 1998, p. 40).

74. There seems to be broad recognition of the importance of formal schooling ⁶⁸ and the need to improve its quality (e.g. Indonesia, United Republic of Tanzania, Zambia, Zimbabwe). Raising the school leaving age, reducing child labour and raising the quality as well as instituting 'appropriate' curricula of education are important objectives. These findings reaffirm the need to "increase investment in basic education targeted at improving the quality of education and access to further and higher education for disadvantaged categories of young people" as noted in the resolution concerning youth employment adopted by the ILO's General Conference at its 86th Session.

75. There is little evidence that job training schemes work, especially in difficult times, partly because they are biased towards preparation for formal sector jobs which simply do not exist on any adequate scale (Zambia Ministry of Sport, Youth and Child Development - MSYCD - 1996). One exception perhaps is Egypt which has adopted the Mubarak-Kohl initiative which attempts to adapt the successful German dual education and training system to the Egyptian context. Several pilot schemes suggest some success. ⁶⁹ The rate of return to programmes is likely to be a function of the state of the labour market (national and local). It is easier to place programme participants when unemployment is low and vice versa. The displacement effects for non-participants are potentially serious.

The pilot project in the 10th of Ramadan City In Ramadan City, Egypt, the modern private sector is organized in the Investors' Association (IA). Many of its members use relatively sophisticated production methods, and have serious difficulties recruiting adequately qualified workers. Therefore, the IA was willing to invest in the Mubarak-Kohl Initiative. In 1995, a dual vocational education and training centre became operational. The youths, who were to become mechanics, electricians or textile workers, were selected by the factories themselves. Each week for a period of three years, these students receive two days of general and technical education (arranged by the Ministry of Education) and four days of practical work experience. The Regional Unit of the Dual System (RUDS), with specialists from the private sector and the German GTZ, supervised the internships and designed some specialized courses. Together, the Ministry and RUDS designed and evaluated the exams. Almost all factories that participated in 1995 are still participating four years later. This confirms the viability of the dual training system. But the pilot project also showed that even under ideal circumstances (an enthusiastic and organized private sector, with modern factories and a clear need for qualified workers), a dual training system is not established easily. Some individual factories used their interns as cheap labour (and were consequently excluded from the project), and participating factories are concerned about their trainees moving to competitors after graduation (a fear that is not confirmed by dual training system experiences in Germany). To some extent, these issues were solved by recruiting students 'the Egyptian way': through people who already work for the company and who supported the candidacy of friends or relatives. While this type of favouritism is generally considered harmful, impeding fair and effective recruitment processes, the principle proved useful in the case of the dual training system. Youngsters whose parents or relatives work in the same factory are unlikely to be exploited, as their relative will keep a keen eye on their duties and learning process. After graduation, the family's loyalty towards the factory keeps students committed to their employers.

76. It will be hard to 'solve' the problem of urban youth unemployment as this is likely to induce a flow to the cities from the land in countries where the reserve of rural youth labour force is large. Programmes need to be developed to slow the flow from the land and deal with the underemployment of the young in rural areas.

77. The situation in sub-Saharan Africa (SSA) has been somewhat different to that in Asia - modern economic growth has succeeded in increasing the welfare of many developing countries, but it has left most of Africa behind. Thirty six per cent of the region's population live in economies that in 1995 had not regained the per capita income levels first achieved before 1960. The evidence on whether the openness to international trade enhances economic growth is also rather mixed (Sachs and Warner, 1995; Harrison, 1998). Inequality does not seem to be a major factor inhibiting growth in SSA to low levels of human capital. However, recent work has produced equivocal results on the effects of education on productivity (see Krueger and Lindahl, 1998; Topel, 1999; Pritchett, 1997). Freeman and Lindauer (1999) argue that the main limit on African growth is political turmoil, corruption and the lack of security of property; by implication, fast economic growth can only be achieved in a peaceful, stable environment where property rights are guaranteed: "The return to schooling requires stable property relations and a safe economic environment, which have been lacking in most African states. Wars, corruption, revolutions and other instabilities that disturb or distort the normal functioning of markets may make the value of schooling less than it would be in a more stable world. If your country is riven with strife, better to pick up a gun than a book". ⁷⁰

78. The good news is that investment in physical capital is well correlated with economic growth, but for this to occur, property rights have to be guaranteed. Namibia and Uganda, who ranked as 'mostly free' have experienced rates of growth in investment in excess of 7 per cent per annum. Lowly ranked Nigeria (95th) and Togo (134th) experienced investment declines of nearly 10 per cent a year.

VI. Youth Unemployment: The Policy Agenda

Introduction

79. While every effort has been made to review policies and experiences in different regions of the world, it is clear that high quality labour market data, especially micro-data, is now available mainly across OECD countries, often collected in the same way with the same variables. Aggregation matters in the labour market; hence disaggregated data must be drawn from surveys of individuals, households and firms. Panel surveys where the same respondent is surveyed on more than one occasion, are especially useful; repeating a cross-section survey year after year is another valuable resource for researchers. Many such surveys in similar form are available for OECD countries: the panels include the National Longitudinal Survey (NLS) in the US, the British Household Panel Study (BHPS), the Swedish HUS and the German Socio-Economic Panel (GSOEP).

80. More econometric analysis have been conducted on OECD countries about the workings of the labour market than is true for developing countries. Even though many developing countries have implemented labour market programmes, a large body of technical literature has shown that it is virtually impossible to measure the success of any programme without carefully conducted evaluations with appropriate control groups. Experience with these types of evaluations in the United States suggests zero or even negative rates of return for programmes aimed at the disadvantaged young, although there is some evidence that they work for adults, especially so for women.

81. Econometric work that is available suggests that there are many patterns in the data that are similar across the member countries of the OECD which appear to extend over to the transition economies and the developing world. This runs contrary to the conventional wisdom that emphasizes differences and not similarities. Regression equations estimated on various countries - for the OECD and elsewhere - to explain wages and earnings, self- employment, unemployment, union density, happiness, life satisfaction and job satisfaction, for example, tend to have the same significant variables with the same signs, although with differences in their coefficients, no matter for which countries they are estimated. ⁷¹ There is little, if any, systematic variation in the coefficients by country in any of these variables that is correlated with any macro variable of interest, such as economic performance. Mincerian ⁷² earnings equations estimated for the US, the UK and Germany look broadly similar to those estimated for the countries in transition and the developing world.

82. Not only patterns in data appear to be similar. Some of labour market fundamentals seem to be quite similar throughout the world:

a. Youth unemployment is higher than adult unemployment. Double is a good rule of thumb;

b. The relative size of the youth cohort is down;

c. There is a growing move to increased formal education among the young;

d. Young people are deferring marriage until later in life;

e. Female participation rates are rising;

f. Labour market programmes for the young generally have low rates of return; and

g. There is, moreover, evidence from a series of econometric studies that there are patterns in the data that hold in the developed countries and are repeated elsewhere. Institutions may matter less than we think.

The wage curve: patterns in the data Recent research has established that there is a negatively sloped curve linking pay to unemployment. The nature of this relationship - "the wage curve" - is almost identical across the countries of the world. This curve was first found in micro-data for 11 OECD countries: Austria, Canada, Holland, Ireland, Italy, Norway, Republic of Korea, Sweden, Switzerland, the UK and the US. It is also present, within nations, across different periods of time. In the countries studied, the estimated unemployment elasticity of pay was approximately -0.1: A doubling of unemployment is then associated with a 10 per cent fall in pay. Blanchflower and Oswald further reported on work by other researchers which established the same pattern in some non-European countries (Japan, Côte d'Ivoire, and India), making 14 in all. Since the study was undertaken, other researchers have confirmed these results for the group of countries Blanchflower and Oswald analysed and a number of new papers have established similar results in a number of new countries across all continents: Europe (Belgium, Denmark, France and Spain), Latin America (Argentina and Brazil), the transition economies (Bulgaria, Czech Republic, East Germany, Hungary, Poland and Russia), Africa (Ghana, South Africa, Burkina Faso) and Asia (Taiwan). The wage curve slopes down in all 30 countries studied with an approximate elasticity of -0.1. Kingdon and Knight (1998), for example, concluded that "when we use the definition of unemployment that is most plausible for South Africa, i.e. the broad definition, there is evidence of a remarkable OECD-type wage curve in South Africa, a country with several times the typical unemployment rate of OECD countries. The relationship between broad unemployment and wages is downward sloping, becomes flat at high unemployment rates and yields a wage unemployment elasticity of -0.11" (1998, p.21). Other examples of wage curves outside the OECD are Hoddinott (1998) for the Côte d'Ivoire; Galiani (1999) for Argentina; Amadeo and Camargo (1997) and Barros and Mendonca (1994) for Brazil; Rodgers and Nataraj (1998) for Taiwan; and Blanchflower and Oswald (1999b) for the transition economies. The degree of wage flexibility may be more similar across countries, whether in the OECD or otherwise, than has been previously believed.

Macroeconomic Policies

83. It does appear that the solutions to youth unemployment are inextricably linked to, and cannot be separated from, the difficulties countries face in reducing overall unemployment. Unfortunately, we are a long way from understanding why aggregate unemployment is so high and why it has trended upwards over the last couple of decades. Our knowledge about means to enhance employment-intensive economic growth is also scant. There is some evidence that overly generous benefits can increase unemployment, but the correlation is rather weak in the data - Italy has high unemployment and low benefits. Despite conventional wisdom, ⁷³ high unemployment does not seem to be primarily the result of job protection, labour taxes, trade union power or wage 'inflexibility'. ⁷⁴ Econometric support for the importance of home ownership and mobility in explaining unemployment in Europe is provided in OECD (1999) who model unemployment across countries and find ownership to be the only significant influence; job protection, benefits and unions play no role at all. ⁷⁵, ⁷⁶

84. Cross-country comparisons show that unemployment in Europe is higher than in the US and Europe has more job protection, higher unemployment benefits, more union power and a more generous welfare state. However, such comparison tells us little or nothing about changes over time. In many OECD countries unemployment has increased substantially over the last decade or so; ⁷⁷ but the unemployment benefits have been cut; the union density has fallen and the union power has weakened; ⁷⁸ job protection has changed little; and there is said to be a new flexibility in wage bargaining than there was in the past. If these were the causes, then unemployment ought to have decreased. If not, why not? Current research on unemployment has been unable to find a convincing answer. Recent research suggests some promising new candidates that merit consideration: changes in commodity prices in general and the oil price in particular seem to predict reasonably well the cyclical movements in unemployment. One promising line of inquiry is the interregional mobility of the population and the role of home ownership which seem able to explain at least some of the upward trend for the high levels of youth unemployment prevailing today.

85. Home ownership. The large rise in the level of European home ownership may well be the 'missing piece of the unemployment puzzle because it impairs people's mobility'. ⁷⁹ Oswald (1999) makes it clear that the economies need to be adaptable. They require workers to be able to move around to find new jobs. Private rental housing helps as it allows people to be mobile. In the period from 1950 to 1960, most European nations had low owner-occupation rates and low unemployment rates. The link between housing and jobs appears to hold across space within a country as well as across different countries.

86. Internal migration. Data on interregional migration on a number of OECD countries in the 1970s and 1980s shows that there was relatively little cyclicality in the numbers of migration, suggesting that the scale of migration is not simply a response to unemployment rate which was highly cyclical over the period. The low unemployment countries of Norway, Sweden, Japan and the US all have high proportions of their populations that move across regions. Mobility is low and unemployment, especially youth unemployment, is particularly high in Italy (32 per cent in 1998). Faini et al. (1997) have shown that migrations between Northern and Southern Italy declined steadily from 1970 to 1990. Over this period, the unemployment differential between north and south doubled from just under 7 per cent to just under 14 per cent.

87. Wage flexibility. It is sometimes alleged that the trouble with many European economies is that wages are inflexible. However, recent research seems to indicate that the degree of local wage flexibility is approximately the same in all of the countries studied (see above).The earnings of youths appear to be especially responsive to changes in the unemployment rate and have a somewhat higher elasticity of closer to -0.20. This pattern is broadly consistent across countries. ⁸⁰ In brief, as there are very different rates of unemployment across nations in spite of similarity in the degree of wage flexibility across countries than has been believed, making wages more flexible across countries does not seem to be a solution to youth unemployment.

88. Unemployment benefits, taxes and trade unions. There is only a weak positive relation between unemployment rates and unemployment benefits. ⁸¹ A similar picture is seen in a comparison between unemployment rates and the proportion of GDP spent on unemployment compensation. On both measures, Italy has low benefits and high unemployment. In 1996, Italy spent 0.68 per cent of GDP on unemployment compensation, even though it had a 12 per cent unemployment rate. ⁸² Japan and the US are counter-examples, having low unemployment and low spending on benefits (0.40 per cent of GDP and an unemployment rate of 3.4 per cent for Japan, and 0.26 per cent of GDP and an unemployment rate of 5.4 per cent for the US). There is some evidence from Denmark and the US and to a lesser degree from the UK, New Zealand and the Netherlands, that reducing the generosity and the duration of benefits can cut unemployment, even though the degree of responsiveness has been small. This, of course has limited impact in countries with low unemployment coverage.

89. There is no correlation between unemployment and taxes (see Figure 7) ⁸³ and even a weak negative relation with union density. ⁸⁴ Spain and France have very low union density rates and high unemployment while Austria has low unemployment and quite high union density.

Source: Oswlad, 1999

90. Oil prices. Movements in oil prices appear to cause cyclical changes in unemployment in the US and Europe but seem unable to explain the upward trend in European unemployment. Carruth, Hooker and Oswald (1995, 1998) find that oil prices help to explain unemployment in the UK and Canada and the US. Plotting the unemployment rates for the US and OECD Europe along with the oil price for the years 1970-98, ^{85 86} The oil price hikes of 1973, 1979, 1989 and 1994 all appear to have had harmful effects a short time later on unemployment in both Europe and the US. The steady decline in oil prices which started in 1982 and went on until 1989 predates declines in unemployment in both the US and Europe and similarly for the period 1990-94. The oil price seems to explain the cyclical movements of unemployment.

91. As noted by Schultz, ⁸⁷ "Studies suggest that the structure of output, composition of employment, openness of the economy to trade and technical change all contribute to the derived demand for more educated workers. High technology industries are associated with higher returns to schooling by sector, but this may also reflect selectivity of persons entering into high technology sectors who are more able and better educated, or it could signal the complementarity of on-the-job training and schooling in fields of rapid technological change (Gill, 1989; Gill and Khandker, 1991; Mincer and Higuchi, 1988; Choi, 1993). Increases in the share of the employment in services, commerce, and manufacturing are associated with increased female participation in the labour force, and perhaps enhance market returns to women's education (T.P. Schultz, 1990)." Since the younger populations are increasingly more educated, it stands to reason that strategies that promote technology related employment would enhance the prospects of younger populations. This, however, implies that the enterprises generating new jobs must rely on modern technology which, in turn, has consequences on the sectoral composition of the economy and its degree of informality.

92. There is, thus, considerable evidence backing up the call inviting member States to "adopt and implement policies which improve competitiveness through investment, including investment in technology, human resources development, education and skills, in order to promote economic growth, social development and employment".

How have countries who have tried to make their labour markets more 'flexible' fared?

93. Countries like the UK and New Zealand have attempted to make their labour markets more flexible by, amongst other things, cutting unemployment benefits, tightening regulations, restricting the power of unions and so forth. Has this worked? First, looking at their positions in the OECD country rankings (excluding Luxembourg, Iceland and Switzerland) by unemployment rates and employment/population rates, it seems that such attempts had little impact on these indicators' rankings. Blanchflower and Freeman (1994) analysed the effectiveness of the Thatcherite reforms on the UK labour market. They concluded that they had succeeded in their goals of weakening union power; may have marginally increased employment and wage responsiveness to market conditions and may have increased self-employment. On the positive side, the reforms were accompanied by a substantial improvement in the labour market position of women. On the negative side, the reforms failed to improve the responsiveness of real wages to unemployment; they were associated with a slower transition from non-employment to unemployment for men, a devastating loss in full-time jobs for male workers; and produced substantial, seemingly non-competitive increases in wage inequality.

94. Maloney and Savage (1996) document the labour market reforms that have occurred in New Zealand since 1984. Over the past 15 years or so the economy was made more decentralized; unemployment benefits were cut, welfare eligibility criteria were tightened and industrial relations legislation was passed to restructure the industrial relations system by eliminating national awards and removing compulsory unionism. Union density fell dramatically from 40.8 per cent in 1991 to 24.1 per cent in 1994 . ⁸⁸ Interestingly, product markets were protected and made immune from many competitive pressures. In subsequent work, Maloney (1998, 1999) found that neither changes in unionization nor benefits had any significant effect on unemployment, although they do appear to have some effects on employment and labour force participation. Chapple et al. (1996) concluded at the end of their examination of unemployment in New Zealand that: "despite ten years of stabilization, liberalization and labour market reform, it should be a source of some discomfort that these changes have yet to be reflected in an unemployment rate lower than when the reforms began". ⁸⁹

95. The Netherlands and Denmark have both seen a strong improvement in their position in these rankings, whether measured by unemployment or by the employment to population ratio. The unemployment rate in the Netherlands, for example, fell from 7.1 per cent in 1994 to 4.0 per cent in 1998 while Denmark's unemployment rate fell even faster, from 10.1 per cent in 1993 to 5.1 per cent in 1998. Denmark's rise in the rankings is even more pronounced than that of the Netherlands. It does not appear that this decline in unemployment in either country was brought about by decline in union power, changes in job protection, mismatch or labour taxes. Overall strictness of employment protection measures remained unchanged in both countries ⁹⁰ between the 1980s and 1990s. ⁹¹ What is puzzling is that despite the rapid decline in unemployment over this period in the Netherlands, there was an increase in spending on labour market programmes. ⁹² In the case of Denmark, there is more evidence of declines in benefits being correlated with declining unemployment. In 1994, Denmark introduced a reform package which seemed to work: it reduced the generosity of its unemployment compensation system; job placement interviews were introduced; paid leave schemes were made less generous; the maximum duration of benefits were reduced; and the eligibility criteria were tightened. A tax reform package was also implemented to lower taxes on labour ⁹³ and increase incentives to work. ⁹⁴ This programme appears to have been working as unemployment was halved from 10.1 per cent in 1993 to 5.1 per cent in 1998.

96. Union power does not seem to fit the picture very well either as union membership in the Netherlands increased during the 1990s - the number of members went from 1.4 million in 1990 to 1.87 million in 1995, and union density increased from 26 per cent to 28 per cent over this period . ⁹⁵ Union density in Denmark was the same in 1994 as in 1980. Bargaining coverage and the degree of centralization or coordination of their bargaining remained roughly constant in both countries over this period. ⁹⁶

Minimum wages

98. There is little strong evidence either that youth wages are too high or that the minimum wages have priced young people out of jobs. There is some evidence that youth wages relative to adult wages have been falling in the OECD. It does not appear that youths are being priced out of jobs in this case, and there is much less evidence on movements on relative wages in the developing world. Wage cuts are not a solution.

99. During the last decade or so, the level of the minimum wage in the United States has been at an all time low in real terms. The evidence suggests that its low level and small changes in it had small, if any, effect on employment. ⁹⁷ There is also little evidence from Europe that the minimum wage has had major employment effects. ⁹⁸ Even where evidence has been found in OECD countries for negative effects, it has generally been small and weakly significant. Minimum wages have mostly remained low in most European countries except France, where they appear to have had some employment effects. Neumark and Wascher (1999) report evidence that increases in minimum wages have reduced youth employment rates in Canada, the Netherlands, Luxembourg and to a lesser extent in the US and the UK. They also found evidence that declines in minimum wages were accompanied by declines in youth employment rates in Italy, Belgium, Spain, Greece and Portugal. They could find no relation between these two variables in Germany, Sweden, France and Japan. Both patterns were evident at various times in New Zealand and Denmark. There is stronger evidence for employment reducing effects of minimum wages in a number of Latin American countries.

100. There are, however, strong grounds for operating a youth sub-minima to ensure that youths are not priced out of jobs for lack of skills. It might also make sense to allow some variation within a country to allow for differences in living costs and earnings. Large increases in the minimum wage can significantly worsen young people's relative position. High minimum wages in developing countries do seem to raise unemployment, but overall, they appear to reduce poverty (Lustig and Mcleod, 1997). This result is consistent across high and low poverty lines, alternative measures of poverty and the classification of observations by whether the economy is growing or contracting, by whether the population is urban or rural, and by region (Latin America, Asia or Africa). ⁹⁹ Raising the earnings of some of the young may make things worse for the rest. Lustig and Mcleod argue forcefully that their findings "should not lead to a flat endorsement of minimum wage increases as an effective policy measure to reduce poverty" (1997, p.81). However, their results do suggest that reducing minimum wages in the developing world does hurt the poor, at least in the short term.

Self-employment

101. The persistence of the youth labour market problem seems to demonstrate that standard economic policies have been insufficient. Western governments are searching for new alternatives. One idea is that policy should attempt to create more entrepreneurship among the young. It is not obvious, however, that even a large new supply of young entrepreneurs would solve the job crises. Self-employment presents an opportunity for the individual to set his or her own schedule; they can work when they like; they have to answer to nobody; and ultimately perhaps, it is a way to become rich. Unfortunately, on the downside, if the business fails, it may take with it their job, their savings, their home if, as it often happens, it is used as security on a loan, and perhaps generate a family crisis. If we have learnt anything from portfolio theory, it is that an individual should diversify their portfolio and not to pool their resources into a single, risky activity. Governments on the other hand, frequently see self-employment as a route out of poverty and disadvantage and for this reason, offer aid and assistance for small businesses. The justification for these actions are usually based on a number of 'potential' benefits often discussed by commentators: ¹⁰⁰

a. Entrepreneurship may promote innovation and thus create new jobs;

b. There may be a direct effect on employment if new, young entrepreneurs hire fellow youths from the dole queues;

c. New, small firms raise the degree of competition in the product market, bringing gains to consumers;

d. Young entrepreneurs may be particularly responsive to new economic opportunities and trends; and

e. Greater self-employment among young people may go along with increased self-reliance and well-being.

102. Economists have little evidence, however, on whether these hypothetical benefits exist in practice. Moreover, it is by no means obvious that more self-employment is better than less. Blanchflower (1999a) could find no evidence in OECD countries that increase in the self-employment rate increased the real growth rate of the economy; in fact, there was even evidence of the opposite. It was also found that the overall trend in self-employment, at the economy level in the years since 1966, has been downward in most countries - mostly due to the decline in agricultural self-employment. The end of the twentieth century may mark a particularly appropriate time for young entrepreneurs in developed countries.

103. There is evidence that many more people would like to run their own businesses. On the question of preferences for employment status between being an employee as against self-employment and for working in a small firm as against a large firm in the International Social Survey Programme, a large number of respondents expressed their preference for being self-employed and working in a small firm.

104. A noticeably higher proportion of young people preferred self-employment and this was the case for most countries. The data appears to indicate that there is large, latent demand for a kind of entrepreneurial behaviour: self-employment. People find self-employment intrinsically attractive and those in self-employment tend to be enjoying greater well-being and satisfaction than equivalent employees. This finding suggests that self-employment brings direct microeconomic benefits to people. It raises a puzzle too. If self-employment does this, why are not more individuals running their own business? Economists have amassed considerable evidence that potential entrepreneurs are held back by lack of capital. ¹⁰¹ There is particularly strong evidence in the US that suggests that liquidity constraints bind especially tight on blacks, which may help to explain why the self-employment rate of black males is about one-third that of white males. ¹⁰²

105. The literature on micro-enterprises also identifies a lack of capital as a primary constraint on enterprise development. ¹⁰³ The Grameen Bank in Bangladesh has demonstrated very successfully that the poor will repay small, non-collateralized loans or micro-loans. ¹⁰⁴ Grameen organizes borrowers into self-selected, peer groups usually clustered together in villages; if any member of the group defaults, no member can ever again borrow from the