12  Does the supply of skills create its own demand?

From a policy perspective, it is crucial to know whether the skills structure of jobs on offer from employers is responsive to the availability of skills. If it is not, then it is the structure of demand for skills (as it emerges from industry structure and choices of production methods) that determines the quantity of high, middle and low paid jobs: it is not the quantity of skilled/educated workers. Should this be the case, an increase in the average levels of education will merely lead to more educated workers doing the same jobs as were previously done by less educated workers.

The single most compelling reason for not believing that the supply of skills creates its own demand is the juxtaposition of changes in levels of education (equated with skills) with changes in inequality of the distribution of earnings. In Canada, the UK, the US, Australia and New Zealand, average education levels have been increasing over the past two decades. So too has the proportion of the workforce that has completed secondary schooling and obtained some post-school qualification. Nonetheless, in all of these countries inequality in the distribution of earnings has risen, the real pay of low wage earners has fallen and, at least for the US and the UK, mobility from low to higher wages has fallen. These trends have been strongest in the US, the country that has the highest levels of formal education in the world. Table 2 shows that education, especially in the US and Canada, does not protect people against low wages. Over one third of low wage workers in those two countries have post-school education: demand has not arisen to meet their supply of skills. These macro facts make it difficult to argue that the solution to inequality in earnings and continuing low wage employment is to be found in raising average levels of education/skills.

One concept that can explain this combination of rising average levels of education, a reduction in the size of the tail of low education, and rising earnings inequality is credentialism. This is allied to the concept of signaling. Both imply that workers’ levels of education are used by employers to rank job applicants. The formal qualification may not contain any content that is relevant to the job, but it is valued by the employer as a simple and inexpensive way of deciding whom to interview among a number of job applicants. On this view, formal education signals attributes that the worker already has, such as intelligence and persistence, rather than creating new skills. Education then reduces the costs of hiring and probably improves the quality of job matches, but it does not fundamentally alter the structure of skills available to or demanded by employers. In a world of signaling, more education will be profitable for the individual worker. But its contribution to the social good will only be through its role in improving the quality of the job match. One fact in the US experience that credentialism cannot reconcile, however, is that there has been a sharp rise in earnings differences within each educational group (eg, Gottschalk, 1997:33). This is commonly interpreted to mean that some attributes of workers that are not easily observed, such as customer skills, motivation and flexibility, have become more highly valued by employers. It can also partly be explained by a rise in short term fluctuations in people’s earnings. “Jobs were becoming less stable as well as less equal.” (Gottschalk, 1997:33).

In support of the demand-determined view, Blau (1999) argues that “an abundance of labor has never spurred employment: the unemployment rate is consistent proof of that.” Yet the illusion persists

. . .if only the unemployed consisted more rather than less skilled workers, unemployment would disappear. This view romanticises why workers get hired. They are not hired because they are available; they are hired because employers in the private sector believe that the value of their contribution to the business exceeds their cost. Nothing in the history of economics suggests that a change in the skills workers possess is sufficient to alter this equation.

Blau, 1999:132

Prior and Schaffer (1999) show convincingly that in the US the growth of average levels of education has caused people with a given level of education to take increasingly lower skilled/paid jobs. This has caused a considerable downward occupational mobility at each education level. “In brief, university graduates are taking high-school jobs.” (p 3) This has been accompanied by a rising joblessness of prime-age males, particularly among the less educated, which has persisted even with the strong growth in employment in the 1990s. The labour market is increasingly sorting workers by their cognitive skills (not just education, which is why there is an increasing dispersion of pay within educational categories). “Workers experiencing downward occupational mobility generally have lower cognitive skills than others with the same educational credentials.” (p 4).

Prior and Schaffer argue that credentialism is a very important phenomenon, and increasing the formal education of some people will not increase total employment, just change who has the jobs, making it even harder for those with little formal education. They believe bumping down the educational ladder has been a major explanation for the fall in wages for those with low education, and the lowest educated (especially men) have fallen off the bottom rung into non-employment. Their views are supported by Pigeon and Wray (1998), who report that over the course of the 1990s, fewer than 500,000 extra jobs in the US were taken by people on the bottom half of the education ladder (ie, no more than high school). The 11.3 million other new jobs went to people with at least some college education. Many of these jobs were low skilled and could have been done by those with only high school education. They also report an interesting survey of employers of production labour, asking what they look for when recruiting. The answer is, in order of importance, attitude, communication skills, previous work experience, views of co-workers and previous employer, industry-based credentials, years of schooling. (p 207). Thus, adding to schooling may not do much, if does not affect attitudes and communication skills. This is hard to do if home and neighbourhood are against it.

The explanation of the rise in inequality of earnings that is based around the idea of credentialism is strongly challenged by an alternative school of thought that argues instead for the role of skill-biased technological change. In brief, this perspective argues that the simultaneous rise in the number of more highly educated workers and the return to education (strongly apparent in the US since the mid-1970s) can only be explained by a large increase in the demand for highly educated workers. Greater integration of the world economy and reduced protections for the working conditions of low wage workers are judged to explain part but by no means all of the rise in inequality (or return to education and experience). This leaves skill-biased technological change as the remaining major candidate to explain the observed facts. In an impressive review of the evidence and arguments for the role of skill-biased technological change, Acemoglu (2002) argues that it was the dominant influence over the preceding three decades in the US. This view is widely shared among American economists (eg, Katz and Autor (1999), Autor, Katz and Kreuger (1998)). He goes further to argue that the rise in the rate of skill-biased technological change was caused by a sudden expansion of college enrolments in the late 1960s, mainly as a way of deferring conscription for the Vietnam war. This view is directly contrary to that of Prior and Schaeffer. It argues that the rapid expansion in college level education made it more profitable to develop and introduce technologies that required college level skills to utilise them. That is, the expansion of skills did call forth demand for those skills.

While the skill-biased technological change story has much empirical and theoretical support, there remain several puzzles. One is that over the relevant period, wages of low skill workers fell substantially and the rate of productivity growth was low in the US. Another is that other parts of the developed world did not experience the same rise in inequality and often had faster rates of growth in productivity, yet faced broadly comparable technological frontiers. The final story is yet to be told on the links between education, inequality in wages, technological change, globalisation and institutional and regulatory changes. In the context on concern about the role of low wage jobs, the key question is whether a strategy to increase average levels of education, or the levels of education of lower skilled workers, is likely to be effective in improving the employment outcomes for those at the bottom of the skill ladder. While there can be no doubt that improvement in the level of education of any particular person with low skills will improve their job prospects, it is more controversial to assert that the same is true for an increase in education for the whole class of low skill workers.