• 3. Empirical evidence on the relationship between labour market deregulation and unemployment
  
  • Micro-economic evidence
  • Aggregate Evidence

3. Empirical evidence on the relationship between labour market deregulation and unemployment

This section reviews, in a selective but fair way, the voluminous empirical evidence brought forward to support the case that labour market deregulation in the UK has been responsible for increased labour market flexibility and has generated more employment. The reform process began almost immediately after the 1979 Conservative election victory, and cumulated with the abolition of the Wages Councils in 1993 and the recent additional shake-up of the benefits system. Thus, as well as documenting the whole period, we focus heavily on the recent recovery when the most significant effects of the reforms would be expected to appear.

Micro-economic evidence

I. Wage Flexibility: Reform policies have clearly had the desired effect on wages in terms of micro wage flexibility. The UK has seen the most rapid increase in wage inequality of any OECD nation over the last 18 years. Figure 1 shows the evolution of relative wages since the early 1970s. After a period of wage compression prior to 1978, there has been an explosion in the gap between the top and the middle wages and between the bottom and the middle. The collapse at the bottom is easier to attribute in some part to the reform package. However, unlike the situation in the US, this increase in wage dispersion did not imply falling real wages at the bottom. Instead real wages were growing on average. However, since 1992 the pace of real wage growth has slowed while the trend toward inequality has continued. Thus for the first time, real wages at the bottom have fallen (see Table 1). Over the 17 year period, 1979-1995, real wages for the 10th percentile of full-time male workers have risen by 9% whereas for the 90th they have increased by 61%. For women the gaps are similar but with faster wage growth at all points in the distribution.

Schmitt (1995) shows that the rise in inequality includes increasing returns to education and tenure. Analysis of the sensitivity of wages to market conditions, according to Blanchflower and Oswald (1993), has not brought about a substantive increase in the sensitivity of wages to regional unemployment rates. Rather over much of the 1980s, regional disparities in unemployment became more acute. However, Blanchflower and Freeman (1994) found that wages at the plant level became more sensitive to product market conditions (profitability) and that there was also a strong increase in the impact of firm size on wages.

II. Joblessness: Despite the flexibility in relative wages which the advocates of the policy wanted, the predicted result of reduced unemployment is harder to detect. Table 2 provides the broad picture of employment and unemployment since 1979. Employment recovered strongly after 1983, passing the peak 1979 employment-to-population ratio in 1989. (For the population of working age, these two peaks were approximately the same.) Unemployment remained higher than pre-1981 recession levels because the rise in female labour supply more than offset the declining numbers of men seeking work. Hence more people were seeking work among the jobless population. This has declined somewhat since 1992. Since then unemployment has fallen by around 400,000, but both employment and inactivity have risen by 400,000. So the recent unemployment pattern is somewhat deceiving because of rising inactivity among working age men. Despite evidence of hidden unemployment, the UK labour market compares well with continental European countries. The peak in 1993 was well below that in 1984. At 8%, the unemployment rate is now among the lowest for EC countries. (The exceptions being Luxembourg, Portugal and the Netherlands.) In terms of recent improvement the UK is perhaps second to Ireland, which has reduced its unemployment rate by nearly 5 percentage points from 18% to 13% since 1987.

At the micro level there is abundant evidence of a worsening concentration of unemployment among similar categories of individuals and households. The annual entry rate into work for unemployed men fell from 46% in 1979 to 32% in 1990 when the economy was at the top of an upswing. For women the rate was 43% in both periods. Thus between the cyclical peaks of 1979 and 1990 the rate of outflow from unemployment into work had declined sharply for men and the aggregate. This implies a lengthening duration of unemployment spells. Table 3 shows the pattern of duration of such spells since data for unemployment according to the ILO/OECD definition became available in 1984. Whilst the share of the unemployed with over one year's duration fell in the strong upswing of the 1980s, by 1995 the picture was little worse than that in 1988 with almost identical unemployment levels. Thus, despite the many additional initiatives directly targeted at the long-term unemployed, there is no evidence that deregulation reduced the duration of unemployment. This holds for a comparison of the mid-1990s with the 1980s when similar unemployment levels prevailed. Compared with the pre-Thatcher era, long-term unemployment was far worse.

The structure of joblessness across society is dramatically changing especially between men and women. In Table 2 the employment rates in 1979, 1989 and 1996 are in the same ball park. For men they have fallen from 75% in 1979 to 70% in 1989 and 65% in 1996. For women they have risen from 44% to 50% over the same period. Men, especially less educated men, are increasingly becoming inactive rather tahn unemployed. These men (and a smaller number of women) are not searching for work but are rather accepting semi-enforced retirement or long-term sickness. For men over 25 the UK employment and unemployment record is among the worst in Europe. Only in female employment has the UK excelled.

Such divergent employment opportunities have implications for employment across households. Table 4 highlights how the decline in male employment and rise in female employment are not occurring in the same families. It shows that the proportion of households with no working adults has grown from 6.5% in 1975 to 19.1% in 1995 (excluding students and those beyond retirement age). This tripling of "work-poor" households coincides with a tripling of social security payments paid to people of working age lacking jobs. This pattern of jobless households is among the worst in Europe (see Table 5). Only a small part of this rise (less than one fifth) is due to changes in family composition, e.g. toward single parents among whom employment rates are low. The main explanation is that entry into work by those in jobless households has fallen. Men and women with working partners are almost twice as likely to enter a job when out of work than those without a partner or with one who is also not in work. The prime culprit is the changing nature of work offered to those out of work. It is increasingly low-paid, part-time and temporary. The loss of means-tested benefits upon entry into low-paid job reduces the bulk of the earnings received (typically 80-90%). This raises the intriguing possibility that wage inequality and so-called flexible employment forms, in combination with a greater reliance on means-tested benefits, actually generate more, not less, jobless families.

III. Changing patterns of employment and job stability: The composition of employment has been evolving throughout the last 15 years. However, the pace of change has often been exaggerated. It is broadly true to say that the bulk of jobs lost in the early 1980s were full-time and that the bulk of the jobs created after 1984 have been either self-employed or part-time. Somewhat surprisingly during the last recession, full-time employment contracted by 1.3 million (including full-time self-employment) and part-time employment actually rose by 50,000. The share of part-time employment rose from 21% in 1984 to 24% in 1995. The number of full-time permanent jobs fell from around 67% to 61%. Temporary jobs have increased from around 4% of all employment to 6% since 1990. Table 6 lays out the changes in the composition of employment through this recovery period. Since 1992 the quantity of full-time permanent positions declined by over 200,000 and full-time self-employment fell by 33,000. The big growth areas are part-time (450,000 including part-time self-employment) and especially temporary jobs (317,000) with around 40% of the latter also being part-time. It is interesting to note that the expansion of part-time and temporary posts and the contraction of full-time posts is heavily concentrated on men. It is no longer appropriate to consider these flexible forms of labour to be a female domain. These changes thus far have had a limited impact on the stock of jobs, but they are important to the unemployed. Job turnover among the so-called flexible forms of employment is far higher than among full-time permanent types of employment.

Full-time self-employment has an expected tenure of 90% of a full-time employee post; for part-time permanent jobs the figure is just over half whilst for temporary jobs tenure is nearer one-seventh. The consequence of this greater turnover and overall expansion is that the vacancies arising, and new hires of those out of work, are dominated by these types of employment. The lower panel of Table 5 gives the composition of all jobs and those taken by people previously out of work. Whereas 62% of all jobs were full-time and permanent, only 35% of entry-jobs were. Similarly, 50% of entry-level jobs were permanent and/or part time but only 26% in the stock of all jobs were. The missing category is government created positions. Thus the modest changes in the stock of jobs away from the full-time permanent employee posts that dominated the post-war period are magnified among entry-level jobs available to the unemployed. These entry-level jobs also pay only 45% of the typical job in the economy as a whole (median weekly earnings). Such low wages place these jobs right in the zone where disincentive effects hurt those on means-tested benefits. Deregulation and the growth of inequality profoundly have changed the pattern of work and incomes available to those out of work.

Blanchflower and Freeman sumarize the recent UK experience: "the observed outcomes raise the disheartening possibility that the reforms in fact brought the UK a mixture of the worst of two possible worlds: the massive wage inequality of the decentralised US labour market together with high and lengthy spells of unemployment, European-style" (1994, p.75). Rather than wage inequality leading to reduced inequality among the jobless, the two have moved hand in hand producing one of the greatest numbers of benefit dependent families in Europe. Yet the Blanchflower and Freeman summary is a little harsh. UK unemployment and long-term unemployment are substantially better than the European norm as are opportunities for women and the young.

Aggregate Evidence

The relatively good performance in the UK as compared to continental Europe over the last 3 years raises the question whether or not the UK micro reforms are responsible or wether divergent macroeconomic policy is. To put it another way, is it UK micro success or continental European macro failure? One answer determines whether the major European economies are in equilibrium at 10-13% unemployment rates. Is this unemployment "structural" and would lowering it result in rising inflation? It is beyond the scope of this paper to address this in detail. Suffice it to say that there were no signs of excess inflationary pressures with unemployment at least 3 points lower a few years back. Nor has there been a major increase in regulation in Europe likely to give rise to more "structural" unemployment.

This does not rule out the possibility that the UK reforms have lowered its NAIRU, i.e. the amount of unemployment required to keep wage inflation in check. There is a considerable amount of econometric work on macroeconomic data designed to shed light on the determinants of unemployment (Layard, Nickel and Jackman, 1991 is the most influential). Typically these studies involve a regression of the unemployment rate on a set of variables thought to influence unemployment or the estimation of a system of equations from which the determinants of unemployment can be inferred. Variables thought to influence unemployment include the measures of labour market regulation or interferences discussed in this paper.

Some commentators argue that deregulation has had the desired effect but this has been masked by the failure of policy makers to realise that inflationary pressures have eased and thus are pursuing over-stringent monetary policy. However, 17 years is a long time for people to believe in such flimsy evidence of success in reducing unemployment in the face of such overpowering evidence of growing inequality, poverty and social dislocation.

The most fashionable current explanation is that the labour market opportunities for the less-skilled relative to the more-skilled have been declining. This type of argument has its origins in the United States where it seems to fit that data very well. In Europe, including the UK, however, the picture is less obvious.