Employment and Non-Employment
2002-04-15
Employment and Non-Employment A Study of the Swedish Labour Market
Erik Jonasson and Lars Pettersson Jönköping International Business School P.O. Box 1026 551 11 Jönköping Sweden e-mail: [email protected], [email protected]
Abstract This study focuses on the structural shift in the labour market in Sweden in the early 1990s, and has its emphasis on the non-employed people in the working age population. Persistence in unemployment is a frequently recognised phenomenon in the countries of the Western Europe from the 1970s and onward. Sweden has to a large extent been spared from this problem until the 1990s. The traditional way to investigate persistent unemployment and hysteresis effects has been to use the reported statistics of open unemployment. However, this measurement does not capture every aspect of the problem and we argue that one should instead study all groups of individuals that are not employed. The method engaged in the empirical part of the study is a decomposed analysis of the part of the population in working ages that does not have a regular job. According to the statistical results there is significant structural break for the size of not employed in Sweden in the 1990s. The group of not employed increased with around 630,000 people between 1990 and 1994. The study support the point of view that the persistence in increased non-employment has been present in Sweden from the middle of 1990s. Prepared for presentation at the conference “European Integration in Swedish Economic Research”, Grand Hotel, Mölle, May 14-17, 2002. ___________________________________________________________________________ This paper is written as part of the research project “Globalisation and the Welfare State”, which receives financial support from the Bank of Sweden Tercentenary Foundation.
1 Introduction During the last decades there has been a growing interest in analysing the labour market and a substantial attention has been put on studies concerning labour force participation and unemployment. From the 1970s and onwards the issue of hysteresis in unemployment has been in focus, in both research and policymaking. The problem with unemployment and nonemployment is associated with substantial negative economic effects for the society. Behrenz and Delander (1998) analysed the economic effects from unemployment and labour market programs in Sweden for the year 1995, a period with 11 percent total unemployment. The authors estimated the fiscal costs for the public sector, as well as real resource costs (opportunity costs associated with e.g. the potential loss in production output) to the society as a whole, and economic costs for the unemployed. According to Behrenz and Delander, the sum of these effects could amount to 165 billion SEK (around 10 per cent of GDP in 1995). According to research in this field it is clear that once the open unemployment has risen to a higher level it is most likely that it will remain on that higher level, or slightly decrease, and not move back to the previous level. Layard, Nickell and Jackman (1991) presents an overview of a number of empirical studies of Post-War unemployment in OECD countries and show that, in general, unemployment rates were higher in the 1980s compared to the 1960s. This observation is particularly strong concerning the European countries. The empirical studies confirm that the hysteresis in unemployment is very much a phenomenon that is related to the period from 1970s and onwards. Sweden had for a long time a record of very low unemployment. Unlike the rest of Western Europe, Sweden and a few other countries were able to avoid persistent high unemployment until the 1990s.1 However, a break occurred in the first years of the 1990s, when the rate of open unemployment increased from around 2 per cent to 8 per cent. Even if the rate of open unemployment in 2001 has declined compared to the level in 1993-94, one can still argue that there appears to be a structural break in the early 1990s.
1
Other countries in Western Europe that also had comparatively low figures of open unemployment were the other EFTA-countries (Austria, Finland, Norway, Sweden and Switzerland).
2
When the labour market situation were getting worse in Sweden in the early 1990s a massive amount of labour market policy efforts were implemented in the labour market. Different types of active labour market policy tools were used in purpose to deal with the increasing unemployment. This policy comprises for example labour market training, job broking and various forms of job creation etc. It is also interesting to note that policy making of this kind is frequently recommended by for example the European Commission (2000). As regards the effectiveness of different forms of active labour market policy programs there are different opinions between different scholars. Martin and Grubb (2001) presents an overview of experiences from active labour market policy programs in the OECD countries, using OECD data primarily from the 1980s and 1990s. Martin and Grubb also survey the evaluation literature in the field. Their general conclusion is that active labour market programmes appears to have little success in terms of “increasing future employment and earnings prospects” and that this is a problem in particular for disadvantaged youngsters. The same kind of negative view of the problem with unemployment for youngsters respond to labour market programs is also found in Heckman, Lalonde and Smith (1999) concerning the US labour market. At the same time job-search assistance, wage subsidies in the private sector, and labour market training programs seem to work for some “target” groups, in so far that they improve the post-programme employment prospects for the participants. Concerning evaluations of the labour market policy strategies in Sweden, IFAU (the Institution for Labour Market Policy Evaluation) has made a number of studies, on both micro and macro level. Ackum Agell and Lundin (2001) state that one can draw several conclusions from the experiences of previous labour market policy with respect to the formation of the policy of today. One interesting result in this research is that it is not possible to state that it is significantly better to take part in active labour market programs than to be open unemployed. Calmfors, Forslund and Hemström (2001) questions in particular the volume of active labour market programs in Sweden from the beginning of the 1990s. The authors’ conclusion is that the Swedish practice in this field has shortcomings with respect to use active labour market programs to treat problems with unemployment. An interesting issue related to the dramatic increase in the volume of active labour market policy tools in the 1990s is that it has important effects on how to interpret the related data 3
series. As an example, the open unemployment has decreased significantly between 1993 and 2000. However, this is not the whole picture. The unemployed and the employed do not constitute the entire working age population. In fact, the working age population (20-64) can be divided into several groups, and in order to get the total picture of the changes in the labour market one has to analyse the total population. In order to get a total representation of the changes in the labour market in Sweden during the 1990s is important to scrutinize all different form of occupations, both regular employment and various forms of policy tools and non-employment. In this paper we consider the whole working age population. This means that we, to some extent, use a reversed concept of analysing the labour market, compared to the traditional one. The argument for applying this method is that unemployment does not constitute any perfect measure of problems concerning non-working groups in the population. From the official statistics of open unemployment in Sweden one receives the picture that the labour market situation has improved during the last years in the 1990s, with an unemployment rate that decreased from 8 per cent to around 4 per cent between 1994 and 2000. By applying this method we are able to contribute with some developed arguments on certain issues regarding the labour market and labour force participation, and to give a more complete picture than the ordinary unemployment statistics do.
2 Persistence in Unemployment There is an extensive literature in the field of labour economics, which, among other things, focus on the unemployment persistence problem. This problem has attracted a substantial degree of attention during the last decades. One explanation to this is (most) likely due to the problem with hysteresis effects in the labour market in the Western Europe from the 1970s and onwards. A common theoretical approach to explain why persistence unemployment occurs relates to the way prices and wages are determined in the economy. Basic models in this field combine both short-run and long run (NAIRU) interactions in the labour market to explain how unemployment and inflation are determined.2 2
An overview of labour market theory is, for example, presented in Layard, Nickell and Jackman (1991). The phenomenon of unemployment and persistence in unemployment can, according to Layard (et al. ) be stressed
4
The typical questions of interest are why excess supply occurs and why the labour market does not clear. This phenomenon is evident since there are queues with applicants for jobs. Related to these questions is why wages do not fall when there exist an excess of supply of workers (and unemployment). In the literature, the answers to these questions are usually that (i) firms are not free to set the wages (usually because of wage bargaining) and/or (ii) firms choose to pay high wages because it is in their interest to do so. The latter relates to the theory of efficiency wages that states that firms will freely choose to pay high wages in order to motivate large efforts and to maintain their labour (Layard et al., 1991). As concerns the wage bargaining and the formation of wages the performance in the labour market is usually explained by the way unions are acting. A central issue for the European economies is the role of central wage negotiations that we can find in many countries. Such wage bargaining is present in Sweden and has for a long time constituted a central feature of the so-called Swedish model. A particular way to stress the importance of the wage-setting procedure when central wage negotiations are present is the insider-outsider theory, which has been launched by for instance Blanchard and Summers (1986) and Lindbeck and Snower (1988). In light of this theory it is also possible get an explanation to why persistence in unemployment may be likely to occur related to the different phases of the business cycle. According to the insider-outsider theory there is a risk that the number of outsiders in the economy is likely to increase during times of depressions and, hence, the number of persistent unemployed increases at the same time. This can be explained from the difficulties outsiders face when they try to enter the labour market that is governed by the behaviour of insiders as they defend their positions as insiders. The way wages are settled, with the use of central negotiations between employers and employees, can be assumed to generate difficulties for outsiders. If wages are set with respect to marginal productivity conditions among insiders, outsiders will face problems to enter the labour market if their marginal productivity is on a lower level than for insiders. This situation can be the case if an outsider position over time using the NAIRU concept and the Phillips curve relation. This approach is slightly different from the one used in “new classical marcroeconomics” (Lucas, 1972). However, both approaches make use of an equilibrium level of unemployment. Another way of studying the occurrence of unemployment would be from the concept of real business cycles, which is not dealt with in this paper.
5
also is related to a loss in the human capital for individuals. This means that people who get unemployed during times of recession are likely to face difficulties to compete for jobs when the economy is booming due to the insider-outsider relation that is settled on the labour market. A problem that has a possible explanation in the insider-outsider theory concerns immigrants that tries to enter the labour market, and another example is youngster that are in position to enter the job market. In empirical studies that analyse the presence of hysteresis and persistence in unemployment in the labour market it is common to search for structural changes and to decompose unemployment into cyclical, and natural rate of unemployment. An example of such a study is Jaeger and Parkinson (1994), where unemployment is decomposed into a nonstationary natural rate of unemployment (UN), and a stationary cyclical rate of unemployment (UC).3 The sum of these two types of unemployment constitutes the “total rate of unemployment” (UT). Following Jaeger and Parkinson (1994), total unemployment can be formulated as equation (1). U tT = U tN + U tC
(1)
Jaeger and Parkinson performed an empirical analysis where they investigated evidence of hysteresis effects in the labour markets in the U.S. and Germany, using seasonally adjusted quarterly data. The data they used was for the U.S. from the sampling period of 1954:1 to 1989:1, and for Germany from the period of 1962:1 to 1989:1. The method used by the authors is an unobserved components approach where they test for appearance of hysteresis in unemployment. The purpose with their study is to separate the non-stationary natural rate of unemployment and the stationary cyclical rate of unemployment. Jaeger and Parkinson applies a unit root test (augmented Dickey-Fuller test) in order to test whether the timeseries are stationary or not. The results from the statistical analysis imply that hysteresis effects are not present in the U.S. economy on any higher level of significance. However, for Germany the statistical results suggest a different interpretation. The hysteresis effect is significant for the German economy, and of substantial magnitude. 3
This approach is also suggested in Jaeger and Parkinson (1990).
6
A problem that is common in empirical studies of unemployment is that the labour force is often treated as a constant. In other words, the cyclical changes in labour force participation is often neglected or not paid much attention to. Of course, many scholars are aware of the problem, and carry out their analysis while having this problem in mind. For example Barro (1988) explored the persistence problem in the labour market for a number of matured Western countries, using a time-series approach. Barro also included variables such as degree of unionisation and the size of the government sector in order to explain the performance in the labour market. Barro argue that the higher degree of unionisation during the time after World War II should be connected with the higher degree of persistence in the labour market. This consideration was based on the theoretical considerations that are launched by the insider-outsider theory. The focus of the empirical analysis in our study is on the non-employment in Sweden during the past three decades. In this respect we divide the population in working ages that do not have a “regular job” into different categories (based on the data we have access to). By doing so we also incorporate various “intermediate” types of unemployment, and instead of using the traditional rate of unemployment we use a definition of not employed, defined as: NEt = U t + LMPt + WH t + STt + DWt + OTH t .
(2)
Hence, we define non-employment (NE) as the sum of open unemployment (U), people involved in labour market policy programs (LMP), people working in the own household (WH), full-time students (ST), people disabled to work (DW), and other people in working age not participating in the labour force (OTH).4 Adding the sub-population that is employed (E) and not employed (NE) gives the total working age population. Thus Et + NEt = 1 if E and NE are reported as shares of the total population. The relation stated in equation (2) is not commonly found in studies in this field and thereby important dynamic aspects are left out. As shown in Figure 1, non-employment, as a share of the population in working age (20-64 years), decreased until the early 1990s. At the time for the deep economic recession in the first years in the 1990s there appears to be a shift in the 4
Closer definitions of these groups are found in connection to Figure 3a-f.
7
trend, and from the middle of the 1990s and onwards the non-employment has been on a significantly higher level, compared to the situation in the early 1970s. Non-employment, as share of working age population (20-64). 5
Figure 1
27% 25% 23% 21% 19% 17% 15%
98
00 20
19
94
92
96 19
19
90
19
19
86
84
82
88 19
19
19
19
78
76
80 19
19
72
74
19
19
19
19
70
13%
Source: SCB Labour Force Surveys.
When we make a comparison in a European perspective we find that the labour market in Sweden was in a tremendous peak position in 1990. In Table 1 below a comparison is presented for the open unemployment rate and non-employment in a number of European countries. The data in Table 1 includes persons aged 15-64 years (not the same definition as used in the analysis in the other sections of this study), and shows the situation for the years 1990 and 2000. In 1990 the open unemployment was very low in Sweden, less than two per cent. At the same time the non-employment also was on a very low level. The average rate of unemployment for the countries in the European Union was on the same level, both in 1990 and 2000, and the rate of non-employment decreased by approximately two percent. This means that on average in the EU-countries a larger share of the population aged 15-64 years was working the year 2000 compared to 1990. A number of countries, such as Belgium, Ireland, the Netherlands, Luxemburg and Norway, decreased their rate of non-employment substantially
5
Non-employment refers to the sum of unemployment and people outside the labour market as suggested in equation (2), in practice the inverse of the employment rate. For closer definitions see footnotes to Figure 3. Source: SCB Labour Force Surveys and The National Labour Market Board.
8
during the 1990s. Of course, one may argue that some of these changes might depend on demographical changes etc. Although, the size of the effects in these countries is of interest, in particular when we compare with the development in Sweden. Sweden, together with Finland experiences a sharp increase in the rate of non-employment. In this way these two countries are special cases. No other country among the ones included in table 1 shows a similar development pattern. Table 1
Country
Open unemployment and non-employment in European countries for persons aged 15-64 years in percent, 1990 and 2000. Open Unemployment 1990 7.3 8.5 3.2 9.2 6.2 13.2 9.9 1.6 7.7 5.3 4.8 16.1 1.8 6.8
Open Unemployment 2000 6.6 4.5 9.9 10.1 8.1 4.4 11.0 2.4 2.7 3.5 4.1 14.1 5.9 5.6
Belgium Denmark Finland France Germany Ireland Italy Luxemburg Netherlands Norway Portugal Spain Sweden United Kingdom OECD Europe 8.1 8.9 European Union 8.3 8.4 Source: OECD Employment Outlook, June 2001.
NonEmployment 1990 48.6 26.1 26.6 43.2 37.8 53.0 50.1 41.5 41.5 28.2 36.0 55.2 17.2 29.0
NonEmployment 2000 41.4 24.5 35.6 42.1 35.9 37.0 51.1 38.2 27.8 22.8 33.1 48.8 27.0 29.0
41.3 40.9
41.7 38.9
There could be different explanations to why Sweden and Finland has been developing different compared to the other countries in Europe. First, one could argue that there is a socalled “catch-up effect”, that Sweden and Finland was on higher level in the first place, and that the other countries naturally have narrowed the gap. This explanation could be valid for Sweden to some degree, but Finland was on the same level as Norway, Denmark and also United Kingdom. The “catch-up” explanation to similar phenomenon’s regarding growth patterns has also been questioned in other studies since for example countries with comparatively high purchasing power have remained on a position in the top for a long time (for example the US).
9
In the following parts of this study, we will concentrate the analysis on the different decomposed parts of the non-employment that is defined in equation (2), and the way different categories of the population are sensitive to labour demand fluctuations with respect to employment.
3 The Labour Market and the Labour Force Participation in Sweden In Sweden, a far-reaching intervention policy with an active public sector has been used for a long time-period. It was implemented at the time of the Second World War and onwards (during the 1950s and 1960s) and is known as the “Swedish model”. The basic targets for the economic- and social policy have been (1) economic security and (2) egalitarianism (Lindbeck 1997). The policy has been guided by Keynesian-type of ideas, and consequently economic security includes full employment. The “golden years” for this particular policy is usually addressed to the first years of the 1950s to the middle of the 1970s when both inflation and unemployment in Sweden were kept on comparatively low levels. At the same time Sweden also had a strong economic growth. In the middle of the 1970s (at the time for the first oil crises) the Swedish economy started to perform not as good as before. The currency was devaluated several times in the late 1970s and in the early 1980s. In the middle of the 1980s the trade balance improved and the economic performance also recovered. However, the problems continued and in the early 1990s a severe crisis hit the economy. The banking and real estate sector played a central role in the crisis and the currency was converted from a fix exchange rate regime to a floating regime. During the first years of the 1990s when the fixed exchange rate was defended, the economy suffered from a high interest rate level. A most significant chock did of course affect the economy, with decreased demand. Thereby the very strong recession of the first half of the 1990s was triggered. The recession in the early 1990s had to be handled differently from previous recessions. The government debt had to be cut down and various reforms had to be adopted in order to adjust
10
markets towards international competition criteria. The stated intention to fulfil the criteria of the European Monetary Union and the global trend of deregulating markets are often mentioned as two important factors behind the changes that were implemented. Deeper questions were also raised about the viability of the welfare arrangements in Sweden (for example the pension system etc) (Björklund, Edebalk, Ohlsson and Söderström 1998). As a consequence of the deep recession and the structural problems in the economy at this time, the unemployment rate increased sharply.6 Table 2 shows the Swedish population in working age (20-64) divided into people inside and outside the labour market during 1970–2000. Our measure of employment in this case does not include people involved in labour market programs. Our motivation to exclude this group from the regular measure of employment is because the employment is related to a labour market policy tool. From Figure 1 it is possible to recognise the structural change that took place in the first part of the 1990s. Table 2
The working age population, 1970-2000.
Working age Labour forcea Employmentb Unemploymentc Outside labour force population, 20–64 3,540,000 3,494,000 46,000 4,715,000 1,160,000 1970 (75.4%) (74.1%) (1.3%) 3,975,000 3,909,000 66,000 4,767,000 769,000 1980 (83.9%) (82.0%) (1.7%) 4,322,000 4,258,000 64,000 4,941,000 610,000 1990 (87.7%) (87.7%) (1.5%) 4,147,000 3,830,000 317,000 5,116,000 913,000 1995 (82.2%) (74.9%) (7.5%) 4,196,000 4,005,000 191,000 5,200,000 965,000 2000 (81.4%) (77.0%) (4.5%) a) Labour force participation rate, as share of working age population, is given in brackets. b) Employment rate, as share of working age population, is given in brackets. c) Unemployment rate, as share of labour force, is given in brackets. Source: SCB Labour Force Surveys.
The increase in employment rate, defined as employment as a share of the working age population, is to a substantial degree explained by women who entered the labour market during the time period of 1970-1990.7 The change in overall employment rate is presented in 6
The sharp increase in unemployment in the 1990s was very much a Nordic phenomena. The Nordic countries reached a level that other EC-countries had had from the early 1980s and onwards (Elmeskov, 1994). 7 Note that employment rate differs from labour force participation rate, which is defined as the labour force (employed + unemployed) as a share of the total population in working age.
11
Figure 2. As we can see from Figure 2 the employment rate increased from 74 per cent in 1970 to 86 per cent in 1990. The difference in labour force participation rates between men and women is discussed more in detail in section 4 below. From the exceptionally high rate in 1990 the employment started to decrease during the recession in the early 1990s. From Figure 2 we find that the period of change occurs mainly between the years 1990 and 1994. From 1994 and onwards the employment rate remains more or less constant (74-76 per cent). In the latter half of the 1990s the employment rate seems to have shifted down to a substantially lower rate compared with the situation from the 1980s. Still, the employment rate is on a comparatively high level in an international perspective; the OECD average is below 70 per cent (OECD, 2001). Figure 2
Employment rate, as share of working age population (20-64 years).
88% 86% 84% 82% 80% 78% 76% 74% 72% 2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
1974
1972
1970
70%
Source: SCB Labour Force Surveys
In section 2 we defined non-employment as the sum of six non-working groups in the working age population. The timeseries, showing the magnitude of these groups, are plotted in Figure 3a-f. As can be seen, open unemployment and the number of people engaged in labour market policy programs increased dramatically in the beginning of the 1990s. Even if the unemployment rate has fallen during the latter part of the 1990s, the number of people employed was still 250 000 lower in 2000 than in 1990. This means that the decrease in unemployment does not cover the whole picture, including every aspect of dynamic changes in the labour market.
12
If we compare values over two periods, 1976-1990 and 1991-2000, we find that the nonemployment, NE, has a mean that is 350,000 higher in the 1990s than in 1976–1990. The explanation is to be found both inside and outside the labour market: on average, 180,000 more were unemployed, while the number of people outside the labour market (in working age) was 170,000 higher during the 1990s. Looking at the five sub-groups of people outside the labour market, the number of people working in the own household has fallen steadily, whereas the number of people involved in labour market programs (LMPs), students, people disabled to work, and those in the category of “others” have all increased. Due to changes in statistical methods and definitions, the Labour Force Surveys do not allow meaningful comparison over time of the different groups outside the labour market (at least not before the 1990s). However, among the 440,000 “disabled to work”, there is a group of approximately 100,000 people who have chosen early retirement by non-health-related reasons. This group has been fairly stable during the 1990s. The remaining part of this group of people, those reported ill on long term or early retired by health-related reasons, has increased from about 130,000 in the end of the 1980s to 320,000 in the end of the 1990s. Among the people outside the labour market (most likely to be found in the non-categorized group “others”), 80,000 are considered as discouraged workers, i. e. people who are willing to work but have not tried to get a job. In the 1970s and 1980s this group ranged from 20,000 to 30,000 people; in the middle of the 1990s it was as many as 140,000. The majority of this group is likely to be engaged in labour market policy programs.
13
Figure 3 Non-employment Figure 3a shows open unemployment, whereas Figures 3b-f provides a subdivision of the people outside the labour market. Figure 3a: Open unemployment as per cent of labour force. 8 9% 8% 7% 6% 5% 4% 3% 2% 1% 2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970
0%
Figure 3b: Labour market policy programs, number of persons involved. 9
250 000 200 000 150 000 100 000 50 000 2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
1974
1972
1970
0
Figure 3c: Students 10
350 000 300 000 250 000 200 000 150 000 100 000 50 000
Full-time students exkluding those willing to work
8
2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
0
All full-time students
Standard definition used. Source: SCB Labour Force Surveys. Some of the labour market policy programs (LMPs), such as Relief work and Trainee replacement scheme, are considered as employment in the Labour Force Surveys. In this study, all LMPs but handicap programs are considered as non-employment. Source: The National Labour Market Board. 10 Refers to full-time students. The majority of the “students willing to work” is involved in LMPs such as Work experience scheme, Employment training, and Work experience for youngsters. Source: SCB Labour Force Surveys. 9
14
Figure 3d: People working in the own household. 11
600 000 500 000 400 000 300 000 200 000 100 000 2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
0
Figure 3e: People disabled to work. 12
450 000 400 000 350 000 300 000 250 000 200 000 150 000 1990
1992
1994
1996
1998
2000
1990
1992
1994
1996
1998
2000
1988
1986
1984
1982
1980
1978
1976
100 000
Figure: 3f: Other people outside the labour market. 13
11
1988
1986
1984
1982
1980
1978
1976
200 000 175 000 150 000 125 000 100 000 75 000 50 000 25 000 0
Unpaid assistants in firms owned by a member of the same household are considered as employed. See note 12 also. Source: SCB Labour Force Surveys. 12 The variable includes early retirement, people reported ill on long term, and a smaller amount of people being/working abroad. The time series break in 1987 is due to a change in statistical method in the Labour Force Surveys. From 1987 early retirement has been reported separately. By including this group among people disabled to work, a one-time upward-shift of 100,000 is created to the expense of downward-shifts in the time series covering people working in the household, even though other groups were affected also. From 1998 the number of people working/being abroad is reported separately in the Labour Force Surveys and amounts about 50,000. Source: SCB Labour Force Surveys. 13 The group covers those persons who are reported as being outside the labour market but does not fit in to the groups above. Source: SCB Labour Force Surveys.
15
One plausible explanation, not mentioned so far, that might partly explain the structural break in the first part of the 1990s is the increased number of immigrants that came to Sweden. There are mainly two factors that can explain a relation with the structural break from this point of view: (1) the stream of immigrants was very much consisting of refugees originating from countries in warfare, and (2) the stream of immigrant increased substantially at the same time the economy went into a severe recession. Figure 4 shows that the yearly number of immigrants that entered Sweden increased sharply in the latter half of the 1980s. Rolf Ohlsson (1995) compares the level of vacancies in the manufacturing industry and the stream of immigrants to Sweden. The author points out the relationship between demand for labour and immigration and argues that there appears to exist a relation between these two variables. In particular the stream of immigrants before 1975 was to be considered as labour force migration. After 1975 we cannot state that immigration is to be characterised as labour force migration in the same way, although the pattern of the stream of immigrants appears to follow business cycles trends (at least until the end of the 1980s). Figure 4
Immigration to Sweden 1972-1999
90000 80000 70000 60000 50000 40000 30000 20000 10000
98 19
96 19
94 19
92 19
90 19
88 19
86 19
84 19
82 19
80 19
78 19
76 19
74 19
19
72
0
Source: Statistics Sweden.
The way the stream of immigrants cause effects on the labour market figures is from the observations of a lower level of employment compared to other groups in the population. According to Ohlsson (1995) this is a significant problem for several nationalities. In general Ohlsson finds that the labour participation rates were lower for all nationalities in 1993. 16
However, this problem was more significant for nationalities outside Europe than for closer neighbours (Finland, Denmark and Norway). Ohlsson also shows that, in 1993, the labour force participation was significantly lower for women than for to men, in particular for individuals born outside Europe. This also means that there are differences between nationalities with respect to employment rates, a fact that is also confirmed by Rooth (1999). In a study by Ekberg (2000) it is shown that employment rates among immigrants are very different with respect to which region they live in. Ekberg analysed employment rates for immigrants from Bosnia for the year 1997, and found that the employment rate for this group in Gnosjö, Vaggeryd, Gislaved and Värnamo (a labour market region in the mid south of Sweden) was around 75 per cent for men. The same figure was 30 per cent for the average in Sweden, and in Stockholm 35 per cent, in Göteborg 20 per cent, in Malmö 14 per cent.
4 Labour Force Participation and the Influence of Labour Market Conditions We have seen from Table 1 and Figure 2 that the employment, and consequently also the labour force participation, has fallen since the end of the 1980s. Whether this drop will be recovered remains to see. If the hysteresis effect in employment is fairly strong, we should not expect the participation rate to rise much above 80 per cent, or the employment rate to rise above the 75 per cent level. It is likely that different demographic groups among the labour force respond differently to changes in labour market conditions. In a study of the influence of employment conditions on labour force participation rates in OECD countries, Elmeskov and Pichelmann (1993) find evidence for most countries that total participation rates are significantly affected by employment rates. In times of economic recession with declining labour demand, individuals loosing their job face the options either to remain on the labour market as unemployed or to leave the labour market for other activities. This choice may be affected by several factors, e g the individual’s perceived chances of getting employed again, personal finance, study plans etc. Social norms may also play a major role: the lower the acceptance of unemployment, the more likely it is that the individual chooses non-participation rather than unemployment.
17
To trace these differences in behaviour, we may test for the sensitivity to labour market conditions among different groups. Assume the following simple model: LFit* = α + βERt + ε t ,
(4)
where LFit LFit* ≡ ln 1 − LFit
, LFi denoting the labour force participation rate (0
Employment and Non-Employment A Study of the Swedish Labour Market
Erik Jonasson and Lars Pettersson Jönköping International Business School P.O. Box 1026 551 11 Jönköping Sweden e-mail: [email protected], [email protected]
Abstract This study focuses on the structural shift in the labour market in Sweden in the early 1990s, and has its emphasis on the non-employed people in the working age population. Persistence in unemployment is a frequently recognised phenomenon in the countries of the Western Europe from the 1970s and onward. Sweden has to a large extent been spared from this problem until the 1990s. The traditional way to investigate persistent unemployment and hysteresis effects has been to use the reported statistics of open unemployment. However, this measurement does not capture every aspect of the problem and we argue that one should instead study all groups of individuals that are not employed. The method engaged in the empirical part of the study is a decomposed analysis of the part of the population in working ages that does not have a regular job. According to the statistical results there is significant structural break for the size of not employed in Sweden in the 1990s. The group of not employed increased with around 630,000 people between 1990 and 1994. The study support the point of view that the persistence in increased non-employment has been present in Sweden from the middle of 1990s. Prepared for presentation at the conference “European Integration in Swedish Economic Research”, Grand Hotel, Mölle, May 14-17, 2002. ___________________________________________________________________________ This paper is written as part of the research project “Globalisation and the Welfare State”, which receives financial support from the Bank of Sweden Tercentenary Foundation.
1 Introduction During the last decades there has been a growing interest in analysing the labour market and a substantial attention has been put on studies concerning labour force participation and unemployment. From the 1970s and onwards the issue of hysteresis in unemployment has been in focus, in both research and policymaking. The problem with unemployment and nonemployment is associated with substantial negative economic effects for the society. Behrenz and Delander (1998) analysed the economic effects from unemployment and labour market programs in Sweden for the year 1995, a period with 11 percent total unemployment. The authors estimated the fiscal costs for the public sector, as well as real resource costs (opportunity costs associated with e.g. the potential loss in production output) to the society as a whole, and economic costs for the unemployed. According to Behrenz and Delander, the sum of these effects could amount to 165 billion SEK (around 10 per cent of GDP in 1995). According to research in this field it is clear that once the open unemployment has risen to a higher level it is most likely that it will remain on that higher level, or slightly decrease, and not move back to the previous level. Layard, Nickell and Jackman (1991) presents an overview of a number of empirical studies of Post-War unemployment in OECD countries and show that, in general, unemployment rates were higher in the 1980s compared to the 1960s. This observation is particularly strong concerning the European countries. The empirical studies confirm that the hysteresis in unemployment is very much a phenomenon that is related to the period from 1970s and onwards. Sweden had for a long time a record of very low unemployment. Unlike the rest of Western Europe, Sweden and a few other countries were able to avoid persistent high unemployment until the 1990s.1 However, a break occurred in the first years of the 1990s, when the rate of open unemployment increased from around 2 per cent to 8 per cent. Even if the rate of open unemployment in 2001 has declined compared to the level in 1993-94, one can still argue that there appears to be a structural break in the early 1990s.
1
Other countries in Western Europe that also had comparatively low figures of open unemployment were the other EFTA-countries (Austria, Finland, Norway, Sweden and Switzerland).
2
When the labour market situation were getting worse in Sweden in the early 1990s a massive amount of labour market policy efforts were implemented in the labour market. Different types of active labour market policy tools were used in purpose to deal with the increasing unemployment. This policy comprises for example labour market training, job broking and various forms of job creation etc. It is also interesting to note that policy making of this kind is frequently recommended by for example the European Commission (2000). As regards the effectiveness of different forms of active labour market policy programs there are different opinions between different scholars. Martin and Grubb (2001) presents an overview of experiences from active labour market policy programs in the OECD countries, using OECD data primarily from the 1980s and 1990s. Martin and Grubb also survey the evaluation literature in the field. Their general conclusion is that active labour market programmes appears to have little success in terms of “increasing future employment and earnings prospects” and that this is a problem in particular for disadvantaged youngsters. The same kind of negative view of the problem with unemployment for youngsters respond to labour market programs is also found in Heckman, Lalonde and Smith (1999) concerning the US labour market. At the same time job-search assistance, wage subsidies in the private sector, and labour market training programs seem to work for some “target” groups, in so far that they improve the post-programme employment prospects for the participants. Concerning evaluations of the labour market policy strategies in Sweden, IFAU (the Institution for Labour Market Policy Evaluation) has made a number of studies, on both micro and macro level. Ackum Agell and Lundin (2001) state that one can draw several conclusions from the experiences of previous labour market policy with respect to the formation of the policy of today. One interesting result in this research is that it is not possible to state that it is significantly better to take part in active labour market programs than to be open unemployed. Calmfors, Forslund and Hemström (2001) questions in particular the volume of active labour market programs in Sweden from the beginning of the 1990s. The authors’ conclusion is that the Swedish practice in this field has shortcomings with respect to use active labour market programs to treat problems with unemployment. An interesting issue related to the dramatic increase in the volume of active labour market policy tools in the 1990s is that it has important effects on how to interpret the related data 3
series. As an example, the open unemployment has decreased significantly between 1993 and 2000. However, this is not the whole picture. The unemployed and the employed do not constitute the entire working age population. In fact, the working age population (20-64) can be divided into several groups, and in order to get the total picture of the changes in the labour market one has to analyse the total population. In order to get a total representation of the changes in the labour market in Sweden during the 1990s is important to scrutinize all different form of occupations, both regular employment and various forms of policy tools and non-employment. In this paper we consider the whole working age population. This means that we, to some extent, use a reversed concept of analysing the labour market, compared to the traditional one. The argument for applying this method is that unemployment does not constitute any perfect measure of problems concerning non-working groups in the population. From the official statistics of open unemployment in Sweden one receives the picture that the labour market situation has improved during the last years in the 1990s, with an unemployment rate that decreased from 8 per cent to around 4 per cent between 1994 and 2000. By applying this method we are able to contribute with some developed arguments on certain issues regarding the labour market and labour force participation, and to give a more complete picture than the ordinary unemployment statistics do.
2 Persistence in Unemployment There is an extensive literature in the field of labour economics, which, among other things, focus on the unemployment persistence problem. This problem has attracted a substantial degree of attention during the last decades. One explanation to this is (most) likely due to the problem with hysteresis effects in the labour market in the Western Europe from the 1970s and onwards. A common theoretical approach to explain why persistence unemployment occurs relates to the way prices and wages are determined in the economy. Basic models in this field combine both short-run and long run (NAIRU) interactions in the labour market to explain how unemployment and inflation are determined.2 2
An overview of labour market theory is, for example, presented in Layard, Nickell and Jackman (1991). The phenomenon of unemployment and persistence in unemployment can, according to Layard (et al. ) be stressed
4
The typical questions of interest are why excess supply occurs and why the labour market does not clear. This phenomenon is evident since there are queues with applicants for jobs. Related to these questions is why wages do not fall when there exist an excess of supply of workers (and unemployment). In the literature, the answers to these questions are usually that (i) firms are not free to set the wages (usually because of wage bargaining) and/or (ii) firms choose to pay high wages because it is in their interest to do so. The latter relates to the theory of efficiency wages that states that firms will freely choose to pay high wages in order to motivate large efforts and to maintain their labour (Layard et al., 1991). As concerns the wage bargaining and the formation of wages the performance in the labour market is usually explained by the way unions are acting. A central issue for the European economies is the role of central wage negotiations that we can find in many countries. Such wage bargaining is present in Sweden and has for a long time constituted a central feature of the so-called Swedish model. A particular way to stress the importance of the wage-setting procedure when central wage negotiations are present is the insider-outsider theory, which has been launched by for instance Blanchard and Summers (1986) and Lindbeck and Snower (1988). In light of this theory it is also possible get an explanation to why persistence in unemployment may be likely to occur related to the different phases of the business cycle. According to the insider-outsider theory there is a risk that the number of outsiders in the economy is likely to increase during times of depressions and, hence, the number of persistent unemployed increases at the same time. This can be explained from the difficulties outsiders face when they try to enter the labour market that is governed by the behaviour of insiders as they defend their positions as insiders. The way wages are settled, with the use of central negotiations between employers and employees, can be assumed to generate difficulties for outsiders. If wages are set with respect to marginal productivity conditions among insiders, outsiders will face problems to enter the labour market if their marginal productivity is on a lower level than for insiders. This situation can be the case if an outsider position over time using the NAIRU concept and the Phillips curve relation. This approach is slightly different from the one used in “new classical marcroeconomics” (Lucas, 1972). However, both approaches make use of an equilibrium level of unemployment. Another way of studying the occurrence of unemployment would be from the concept of real business cycles, which is not dealt with in this paper.
5
also is related to a loss in the human capital for individuals. This means that people who get unemployed during times of recession are likely to face difficulties to compete for jobs when the economy is booming due to the insider-outsider relation that is settled on the labour market. A problem that has a possible explanation in the insider-outsider theory concerns immigrants that tries to enter the labour market, and another example is youngster that are in position to enter the job market. In empirical studies that analyse the presence of hysteresis and persistence in unemployment in the labour market it is common to search for structural changes and to decompose unemployment into cyclical, and natural rate of unemployment. An example of such a study is Jaeger and Parkinson (1994), where unemployment is decomposed into a nonstationary natural rate of unemployment (UN), and a stationary cyclical rate of unemployment (UC).3 The sum of these two types of unemployment constitutes the “total rate of unemployment” (UT). Following Jaeger and Parkinson (1994), total unemployment can be formulated as equation (1). U tT = U tN + U tC
(1)
Jaeger and Parkinson performed an empirical analysis where they investigated evidence of hysteresis effects in the labour markets in the U.S. and Germany, using seasonally adjusted quarterly data. The data they used was for the U.S. from the sampling period of 1954:1 to 1989:1, and for Germany from the period of 1962:1 to 1989:1. The method used by the authors is an unobserved components approach where they test for appearance of hysteresis in unemployment. The purpose with their study is to separate the non-stationary natural rate of unemployment and the stationary cyclical rate of unemployment. Jaeger and Parkinson applies a unit root test (augmented Dickey-Fuller test) in order to test whether the timeseries are stationary or not. The results from the statistical analysis imply that hysteresis effects are not present in the U.S. economy on any higher level of significance. However, for Germany the statistical results suggest a different interpretation. The hysteresis effect is significant for the German economy, and of substantial magnitude. 3
This approach is also suggested in Jaeger and Parkinson (1990).
6
A problem that is common in empirical studies of unemployment is that the labour force is often treated as a constant. In other words, the cyclical changes in labour force participation is often neglected or not paid much attention to. Of course, many scholars are aware of the problem, and carry out their analysis while having this problem in mind. For example Barro (1988) explored the persistence problem in the labour market for a number of matured Western countries, using a time-series approach. Barro also included variables such as degree of unionisation and the size of the government sector in order to explain the performance in the labour market. Barro argue that the higher degree of unionisation during the time after World War II should be connected with the higher degree of persistence in the labour market. This consideration was based on the theoretical considerations that are launched by the insider-outsider theory. The focus of the empirical analysis in our study is on the non-employment in Sweden during the past three decades. In this respect we divide the population in working ages that do not have a “regular job” into different categories (based on the data we have access to). By doing so we also incorporate various “intermediate” types of unemployment, and instead of using the traditional rate of unemployment we use a definition of not employed, defined as: NEt = U t + LMPt + WH t + STt + DWt + OTH t .
(2)
Hence, we define non-employment (NE) as the sum of open unemployment (U), people involved in labour market policy programs (LMP), people working in the own household (WH), full-time students (ST), people disabled to work (DW), and other people in working age not participating in the labour force (OTH).4 Adding the sub-population that is employed (E) and not employed (NE) gives the total working age population. Thus Et + NEt = 1 if E and NE are reported as shares of the total population. The relation stated in equation (2) is not commonly found in studies in this field and thereby important dynamic aspects are left out. As shown in Figure 1, non-employment, as a share of the population in working age (20-64 years), decreased until the early 1990s. At the time for the deep economic recession in the first years in the 1990s there appears to be a shift in the 4
Closer definitions of these groups are found in connection to Figure 3a-f.
7
trend, and from the middle of the 1990s and onwards the non-employment has been on a significantly higher level, compared to the situation in the early 1970s. Non-employment, as share of working age population (20-64). 5
Figure 1
27% 25% 23% 21% 19% 17% 15%
98
00 20
19
94
92
96 19
19
90
19
19
86
84
82
88 19
19
19
19
78
76
80 19
19
72
74
19
19
19
19
70
13%
Source: SCB Labour Force Surveys.
When we make a comparison in a European perspective we find that the labour market in Sweden was in a tremendous peak position in 1990. In Table 1 below a comparison is presented for the open unemployment rate and non-employment in a number of European countries. The data in Table 1 includes persons aged 15-64 years (not the same definition as used in the analysis in the other sections of this study), and shows the situation for the years 1990 and 2000. In 1990 the open unemployment was very low in Sweden, less than two per cent. At the same time the non-employment also was on a very low level. The average rate of unemployment for the countries in the European Union was on the same level, both in 1990 and 2000, and the rate of non-employment decreased by approximately two percent. This means that on average in the EU-countries a larger share of the population aged 15-64 years was working the year 2000 compared to 1990. A number of countries, such as Belgium, Ireland, the Netherlands, Luxemburg and Norway, decreased their rate of non-employment substantially
5
Non-employment refers to the sum of unemployment and people outside the labour market as suggested in equation (2), in practice the inverse of the employment rate. For closer definitions see footnotes to Figure 3. Source: SCB Labour Force Surveys and The National Labour Market Board.
8
during the 1990s. Of course, one may argue that some of these changes might depend on demographical changes etc. Although, the size of the effects in these countries is of interest, in particular when we compare with the development in Sweden. Sweden, together with Finland experiences a sharp increase in the rate of non-employment. In this way these two countries are special cases. No other country among the ones included in table 1 shows a similar development pattern. Table 1
Country
Open unemployment and non-employment in European countries for persons aged 15-64 years in percent, 1990 and 2000. Open Unemployment 1990 7.3 8.5 3.2 9.2 6.2 13.2 9.9 1.6 7.7 5.3 4.8 16.1 1.8 6.8
Open Unemployment 2000 6.6 4.5 9.9 10.1 8.1 4.4 11.0 2.4 2.7 3.5 4.1 14.1 5.9 5.6
Belgium Denmark Finland France Germany Ireland Italy Luxemburg Netherlands Norway Portugal Spain Sweden United Kingdom OECD Europe 8.1 8.9 European Union 8.3 8.4 Source: OECD Employment Outlook, June 2001.
NonEmployment 1990 48.6 26.1 26.6 43.2 37.8 53.0 50.1 41.5 41.5 28.2 36.0 55.2 17.2 29.0
NonEmployment 2000 41.4 24.5 35.6 42.1 35.9 37.0 51.1 38.2 27.8 22.8 33.1 48.8 27.0 29.0
41.3 40.9
41.7 38.9
There could be different explanations to why Sweden and Finland has been developing different compared to the other countries in Europe. First, one could argue that there is a socalled “catch-up effect”, that Sweden and Finland was on higher level in the first place, and that the other countries naturally have narrowed the gap. This explanation could be valid for Sweden to some degree, but Finland was on the same level as Norway, Denmark and also United Kingdom. The “catch-up” explanation to similar phenomenon’s regarding growth patterns has also been questioned in other studies since for example countries with comparatively high purchasing power have remained on a position in the top for a long time (for example the US).
9
In the following parts of this study, we will concentrate the analysis on the different decomposed parts of the non-employment that is defined in equation (2), and the way different categories of the population are sensitive to labour demand fluctuations with respect to employment.
3 The Labour Market and the Labour Force Participation in Sweden In Sweden, a far-reaching intervention policy with an active public sector has been used for a long time-period. It was implemented at the time of the Second World War and onwards (during the 1950s and 1960s) and is known as the “Swedish model”. The basic targets for the economic- and social policy have been (1) economic security and (2) egalitarianism (Lindbeck 1997). The policy has been guided by Keynesian-type of ideas, and consequently economic security includes full employment. The “golden years” for this particular policy is usually addressed to the first years of the 1950s to the middle of the 1970s when both inflation and unemployment in Sweden were kept on comparatively low levels. At the same time Sweden also had a strong economic growth. In the middle of the 1970s (at the time for the first oil crises) the Swedish economy started to perform not as good as before. The currency was devaluated several times in the late 1970s and in the early 1980s. In the middle of the 1980s the trade balance improved and the economic performance also recovered. However, the problems continued and in the early 1990s a severe crisis hit the economy. The banking and real estate sector played a central role in the crisis and the currency was converted from a fix exchange rate regime to a floating regime. During the first years of the 1990s when the fixed exchange rate was defended, the economy suffered from a high interest rate level. A most significant chock did of course affect the economy, with decreased demand. Thereby the very strong recession of the first half of the 1990s was triggered. The recession in the early 1990s had to be handled differently from previous recessions. The government debt had to be cut down and various reforms had to be adopted in order to adjust
10
markets towards international competition criteria. The stated intention to fulfil the criteria of the European Monetary Union and the global trend of deregulating markets are often mentioned as two important factors behind the changes that were implemented. Deeper questions were also raised about the viability of the welfare arrangements in Sweden (for example the pension system etc) (Björklund, Edebalk, Ohlsson and Söderström 1998). As a consequence of the deep recession and the structural problems in the economy at this time, the unemployment rate increased sharply.6 Table 2 shows the Swedish population in working age (20-64) divided into people inside and outside the labour market during 1970–2000. Our measure of employment in this case does not include people involved in labour market programs. Our motivation to exclude this group from the regular measure of employment is because the employment is related to a labour market policy tool. From Figure 1 it is possible to recognise the structural change that took place in the first part of the 1990s. Table 2
The working age population, 1970-2000.
Working age Labour forcea Employmentb Unemploymentc Outside labour force population, 20–64 3,540,000 3,494,000 46,000 4,715,000 1,160,000 1970 (75.4%) (74.1%) (1.3%) 3,975,000 3,909,000 66,000 4,767,000 769,000 1980 (83.9%) (82.0%) (1.7%) 4,322,000 4,258,000 64,000 4,941,000 610,000 1990 (87.7%) (87.7%) (1.5%) 4,147,000 3,830,000 317,000 5,116,000 913,000 1995 (82.2%) (74.9%) (7.5%) 4,196,000 4,005,000 191,000 5,200,000 965,000 2000 (81.4%) (77.0%) (4.5%) a) Labour force participation rate, as share of working age population, is given in brackets. b) Employment rate, as share of working age population, is given in brackets. c) Unemployment rate, as share of labour force, is given in brackets. Source: SCB Labour Force Surveys.
The increase in employment rate, defined as employment as a share of the working age population, is to a substantial degree explained by women who entered the labour market during the time period of 1970-1990.7 The change in overall employment rate is presented in 6
The sharp increase in unemployment in the 1990s was very much a Nordic phenomena. The Nordic countries reached a level that other EC-countries had had from the early 1980s and onwards (Elmeskov, 1994). 7 Note that employment rate differs from labour force participation rate, which is defined as the labour force (employed + unemployed) as a share of the total population in working age.
11
Figure 2. As we can see from Figure 2 the employment rate increased from 74 per cent in 1970 to 86 per cent in 1990. The difference in labour force participation rates between men and women is discussed more in detail in section 4 below. From the exceptionally high rate in 1990 the employment started to decrease during the recession in the early 1990s. From Figure 2 we find that the period of change occurs mainly between the years 1990 and 1994. From 1994 and onwards the employment rate remains more or less constant (74-76 per cent). In the latter half of the 1990s the employment rate seems to have shifted down to a substantially lower rate compared with the situation from the 1980s. Still, the employment rate is on a comparatively high level in an international perspective; the OECD average is below 70 per cent (OECD, 2001). Figure 2
Employment rate, as share of working age population (20-64 years).
88% 86% 84% 82% 80% 78% 76% 74% 72% 2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
1974
1972
1970
70%
Source: SCB Labour Force Surveys
In section 2 we defined non-employment as the sum of six non-working groups in the working age population. The timeseries, showing the magnitude of these groups, are plotted in Figure 3a-f. As can be seen, open unemployment and the number of people engaged in labour market policy programs increased dramatically in the beginning of the 1990s. Even if the unemployment rate has fallen during the latter part of the 1990s, the number of people employed was still 250 000 lower in 2000 than in 1990. This means that the decrease in unemployment does not cover the whole picture, including every aspect of dynamic changes in the labour market.
12
If we compare values over two periods, 1976-1990 and 1991-2000, we find that the nonemployment, NE, has a mean that is 350,000 higher in the 1990s than in 1976–1990. The explanation is to be found both inside and outside the labour market: on average, 180,000 more were unemployed, while the number of people outside the labour market (in working age) was 170,000 higher during the 1990s. Looking at the five sub-groups of people outside the labour market, the number of people working in the own household has fallen steadily, whereas the number of people involved in labour market programs (LMPs), students, people disabled to work, and those in the category of “others” have all increased. Due to changes in statistical methods and definitions, the Labour Force Surveys do not allow meaningful comparison over time of the different groups outside the labour market (at least not before the 1990s). However, among the 440,000 “disabled to work”, there is a group of approximately 100,000 people who have chosen early retirement by non-health-related reasons. This group has been fairly stable during the 1990s. The remaining part of this group of people, those reported ill on long term or early retired by health-related reasons, has increased from about 130,000 in the end of the 1980s to 320,000 in the end of the 1990s. Among the people outside the labour market (most likely to be found in the non-categorized group “others”), 80,000 are considered as discouraged workers, i. e. people who are willing to work but have not tried to get a job. In the 1970s and 1980s this group ranged from 20,000 to 30,000 people; in the middle of the 1990s it was as many as 140,000. The majority of this group is likely to be engaged in labour market policy programs.
13
Figure 3 Non-employment Figure 3a shows open unemployment, whereas Figures 3b-f provides a subdivision of the people outside the labour market. Figure 3a: Open unemployment as per cent of labour force. 8 9% 8% 7% 6% 5% 4% 3% 2% 1% 2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
1970
0%
Figure 3b: Labour market policy programs, number of persons involved. 9
250 000 200 000 150 000 100 000 50 000 2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
1974
1972
1970
0
Figure 3c: Students 10
350 000 300 000 250 000 200 000 150 000 100 000 50 000
Full-time students exkluding those willing to work
8
2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
0
All full-time students
Standard definition used. Source: SCB Labour Force Surveys. Some of the labour market policy programs (LMPs), such as Relief work and Trainee replacement scheme, are considered as employment in the Labour Force Surveys. In this study, all LMPs but handicap programs are considered as non-employment. Source: The National Labour Market Board. 10 Refers to full-time students. The majority of the “students willing to work” is involved in LMPs such as Work experience scheme, Employment training, and Work experience for youngsters. Source: SCB Labour Force Surveys. 9
14
Figure 3d: People working in the own household. 11
600 000 500 000 400 000 300 000 200 000 100 000 2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
0
Figure 3e: People disabled to work. 12
450 000 400 000 350 000 300 000 250 000 200 000 150 000 1990
1992
1994
1996
1998
2000
1990
1992
1994
1996
1998
2000
1988
1986
1984
1982
1980
1978
1976
100 000
Figure: 3f: Other people outside the labour market. 13
11
1988
1986
1984
1982
1980
1978
1976
200 000 175 000 150 000 125 000 100 000 75 000 50 000 25 000 0
Unpaid assistants in firms owned by a member of the same household are considered as employed. See note 12 also. Source: SCB Labour Force Surveys. 12 The variable includes early retirement, people reported ill on long term, and a smaller amount of people being/working abroad. The time series break in 1987 is due to a change in statistical method in the Labour Force Surveys. From 1987 early retirement has been reported separately. By including this group among people disabled to work, a one-time upward-shift of 100,000 is created to the expense of downward-shifts in the time series covering people working in the household, even though other groups were affected also. From 1998 the number of people working/being abroad is reported separately in the Labour Force Surveys and amounts about 50,000. Source: SCB Labour Force Surveys. 13 The group covers those persons who are reported as being outside the labour market but does not fit in to the groups above. Source: SCB Labour Force Surveys.
15
One plausible explanation, not mentioned so far, that might partly explain the structural break in the first part of the 1990s is the increased number of immigrants that came to Sweden. There are mainly two factors that can explain a relation with the structural break from this point of view: (1) the stream of immigrants was very much consisting of refugees originating from countries in warfare, and (2) the stream of immigrant increased substantially at the same time the economy went into a severe recession. Figure 4 shows that the yearly number of immigrants that entered Sweden increased sharply in the latter half of the 1980s. Rolf Ohlsson (1995) compares the level of vacancies in the manufacturing industry and the stream of immigrants to Sweden. The author points out the relationship between demand for labour and immigration and argues that there appears to exist a relation between these two variables. In particular the stream of immigrants before 1975 was to be considered as labour force migration. After 1975 we cannot state that immigration is to be characterised as labour force migration in the same way, although the pattern of the stream of immigrants appears to follow business cycles trends (at least until the end of the 1980s). Figure 4
Immigration to Sweden 1972-1999
90000 80000 70000 60000 50000 40000 30000 20000 10000
98 19
96 19
94 19
92 19
90 19
88 19
86 19
84 19
82 19
80 19
78 19
76 19
74 19
19
72
0
Source: Statistics Sweden.
The way the stream of immigrants cause effects on the labour market figures is from the observations of a lower level of employment compared to other groups in the population. According to Ohlsson (1995) this is a significant problem for several nationalities. In general Ohlsson finds that the labour participation rates were lower for all nationalities in 1993. 16
However, this problem was more significant for nationalities outside Europe than for closer neighbours (Finland, Denmark and Norway). Ohlsson also shows that, in 1993, the labour force participation was significantly lower for women than for to men, in particular for individuals born outside Europe. This also means that there are differences between nationalities with respect to employment rates, a fact that is also confirmed by Rooth (1999). In a study by Ekberg (2000) it is shown that employment rates among immigrants are very different with respect to which region they live in. Ekberg analysed employment rates for immigrants from Bosnia for the year 1997, and found that the employment rate for this group in Gnosjö, Vaggeryd, Gislaved and Värnamo (a labour market region in the mid south of Sweden) was around 75 per cent for men. The same figure was 30 per cent for the average in Sweden, and in Stockholm 35 per cent, in Göteborg 20 per cent, in Malmö 14 per cent.
4 Labour Force Participation and the Influence of Labour Market Conditions We have seen from Table 1 and Figure 2 that the employment, and consequently also the labour force participation, has fallen since the end of the 1980s. Whether this drop will be recovered remains to see. If the hysteresis effect in employment is fairly strong, we should not expect the participation rate to rise much above 80 per cent, or the employment rate to rise above the 75 per cent level. It is likely that different demographic groups among the labour force respond differently to changes in labour market conditions. In a study of the influence of employment conditions on labour force participation rates in OECD countries, Elmeskov and Pichelmann (1993) find evidence for most countries that total participation rates are significantly affected by employment rates. In times of economic recession with declining labour demand, individuals loosing their job face the options either to remain on the labour market as unemployed or to leave the labour market for other activities. This choice may be affected by several factors, e g the individual’s perceived chances of getting employed again, personal finance, study plans etc. Social norms may also play a major role: the lower the acceptance of unemployment, the more likely it is that the individual chooses non-participation rather than unemployment.
17
To trace these differences in behaviour, we may test for the sensitivity to labour market conditions among different groups. Assume the following simple model: LFit* = α + βERt + ε t ,
(4)
where LFit LFit* ≡ ln 1 − LFit
, LFi denoting the labour force participation rate (0
