The evidence clearly shows that the rise in inequality has been principally driven by technological change. But governments can make a difference. Education and training policies to assist the adjustment to new technologies can help preserve the distribution of income and an egalitarian society.
In a recent article (Pearls & Irritations, 30 December 2020), I argued that the two priorities for economic reform should be action to reduce inequality and carbon emissions. Both would increase economic growth and jobs, as well as obviously being desirable for other reasons.
Although there is general agreement about the need to reduce inequality, some of the comments on my article suggest that there is substantial disagreement about the causes of the rise in inequality.
This article, therefore, considers the evidence regarding the causes of increased income inequality over the last forty years, noting that any effective policy response must be based on this evidence, and not just on assertions or whims.
The distribution of labour market earnings
Labour market earnings are the principal source of household disposable incomes and in fact, labour market earnings are highly correlated with household incomes. In searching for the causes of rising inequality this article, therefore, starts by examining the changes in the distribution of earnings.
In principle, there are two possible ways that the distribution of labour market earnings for full-time workers can change:
· Change in the occupational structure
· Change in relative wage rates
Changes in the occupational structure
Australia, in common with most other industrial countries, has experienced a significant hollowing out of middle-level jobs since the beginning of the 1980s. This fall in the proportion of middle-level jobs has been balanced by a relatively fast increase in the most skilled jobs, but also to a lesser extent, by an increase in relatively unskilled jobs mostly providing various personal services.
In a comprehensive review of the evidence, Coelli and Borland (2016) found that ‘Between 1966 and 2011 the shares of employment in low- and high-paid jobs increased respectively by 2.2 and 17.0 percentage points, and the share in the middle decreased by 19.2 percentage points’. Furthermore, most of this job polarisation occurred primarily in the 1980s and 1990s, whereas in the 1970s and 2000s changes in the composition of employment were more consistent with general upskilling.
As a matter of arithmetic, these changes in the distribution of jobs automatically show up as an increase in earnings inequality.
Changes in relative wage rates
Changes in the demand for labour in favour of the more highly skilled jobs could be expected to impact relative wage rates between the different occupations, and this appears to have happened to some extent. But what is perhaps most interesting about the Australian experience is that the wage premium for post-secondary qualifications has tended to narrow, thus reducing inequality – a point to be further discussed below.
Changes in occupational composition and inequality
Together these changes in the occupational composition of employment and in the average earnings between occupations account for most of the increase in income inequality over the last few decades.
According to the research by Borland and Coelli (2016): ‘For males, these occupational factors explain one-half of the growth in inequality between 1986 and 1995 and three-quarters of the growth from 1996 to 2006. For females, the occupational factors explain essentially all the growth in inequality over both periods.’
Non-labour income and inequality
The return to capital
A plausible hypothesis is that the increase in inequality has been partly caused by an increase in profits relative to labour income. However, this is not supported by the data. In fact, employee compensation as a share of national income was higher in 2019-20 (48.9%) than in 1959-60 (46.9%).
It is true that this wage share was as much as 8.5 percentage points higher in the mid-1970s, but subsequently, there was widespread agreement that the wage share in the mid-1970s was excessive and a cause of unemployment. Consequently, the initial purpose of the Accord between the Hawke Labor Government and the trade unions was to reduce unit labour costs while still maintaining real household incomes.
Again, in the last few years, wage increases have not kept pace with productivity, and the wage share has fallen slightly. However, this may well reflect an increase in the share of the most capital-intensive industries, and in any case, it does not signify a serious increase in inequality.
Government taxes and transfers
One of the purposes of government taxes and transfers is to achieve a more egalitarian distribution of income than results from market activity.
The record shows that in Australia between 1981-82 and 1996-97 (the period of the Labor Government), overall the amount of redistribution achieved solely through the tax-transfer system stayed constant, reducing inequality by 34 per cent at the beginning, and by just over 35 per cent at the end (Johnson and Wilkins, 2006). In the 2000s, however, fiscal restraint has led to a decline in the amount of redistribution achieved, with the combined effect of taxes and transfers on income inequality declining by nearly 28 per cent; mostly because of a reduction in benefits (Herault and Azpitarte, 2015).
Nevertheless, the evidence confirms that most of the increase in inequality of household incomes reflects the shift in the distribution of jobs, and changes in government taxes and transfers have been much less significant.
Causes of the increase in job polarisation
The next question therefore is: what caused this increase in job polarisation?
Significantly there is wide agreement that the main explanation for this increasing job polarisation is the impact of computer technology over the four decades starting in the 1970s but impacting particularly in the 1980s and 1990s.
Specifically, the principal impact of computers on the labour market has been to replace routine cognitive and manual tasks which were previously undertaken by middle-skill workers. At the same time, computer technology has been complementary to the non-routine cognitive and interactive tasks undertaken by high skill workers, and this skill-biased technological change can raise the demand for high skilled workers. While the non-routine manual tasks undertaken by low skill workers are, so far, not easily computerised.
Thus, it is technological change that has altered the distribution of jobs and these shifts in the pattern of demand for labour have also altered wage differentials to some extent. Together, in both these ways, technological change is the principal source of changes in the distribution of labour market earnings income.
An alternative hypothesis is that the main impact of globalisation on employment has been to also hollow-out middle-level jobs. But typically the evidence suggests that the impact of technological change has been more important. For example, in Australia, the volume of manufacturing production only peaked as recently as 2007-08 while the hours worked in manufacturing started falling almost twenty years earlier after 1988-89. In other words, this decline in manufacturing jobs – many of which were middle-level jobs – was mainly due to technological progress increasing productivity and not to a loss of markets because of globalisation.
Other possible sources of increased inequality
A loss of union power
Another hypothesis is that the less equal wage distribution reflects a loss of union power. But that theory begs the question of why unions have become less powerful.
The main reason is that the transition to an increasingly skills-biased knowledge economy and services employment in recent decades has led to greater labour market segmentation. This has, in turn, weakened the earlier forms of collectivism and collaboration which underpinned trade union membership and power.
Thus, insofar as the loss of union power has played a part in the increased inequality of wage rates, this too is mainly a result of the changes in employment patterns caused by technological change. This technologically driven decline in trade union membership and power is a worldwide phenomenon and is likely to have had a much greater impact on relative wage rates than the impact of legislative changes governing trade union operations which are unique to Australia.
A neo-liberal policy agenda
A quite common view is that the increase in inequality represents the outcome of the government’s adopting “neo-liberal” values that are deemed to favour the top incomes. Typically no evidence is cited to support this view, but the adoption of a neo-liberal culture might help explain the exceptionally large increases in the incomes of the very rich.
Since 1980 the top 1 per cent of incomes have exploded in some countries. Previously, in the 1970s the top percentile’s share of national income was around 6 to 8 per cent in most developed economies, but by the early 2010s, the top percentile’s share in the US was nearly 20 per cent, 14-15 per cent in Britain and Canada, and 9-10 per cent in Australia and New Zealand. By comparison in both northern and southern Europe, the top percentile’s share only increased by 2 or 3 percentage points over the same period.
As Piketty concludes, it is unlikely that technologies that are common to all advanced economies could produce such huge and different increases in the incomes at the top. Instead, the change in income distribution in favour of the very top must reflect a combination of cultural factors, affecting the subjectively determined ‘pay norms’ for executives, and/or monopoly rent-seeking.
The cultural factors could reflect the influence of neo-liberalism, but increased rent-seeking opportunities are exactly what market liberals most oppose.
Summary of causes
The evidence strongly supports the conclusion that it is technological progress that has been principally responsible for the increase in inequality of incomes over the last few decades.
This conclusion is supported by the major international reviews, with the OECD (2011) finding that neither rising trade integration nor financial openness had a significant impact on either wage inequality or employment trends within OECD countries. However, increased financial flows and technological change did have an impact on inequality, with technological change having a more comprehensive effect.
Similarly, the IMF (2017) concluded that technological progress has been the key driver of a decline in labour’s share of production income in advanced economies, along with high exposure to routine occupations that could be automated, with global integration playing a smaller role.
The preferred policy response
Although all advanced economies face much the same challenge in responding to the challenge to their income distribution from technological change, some have done much better than others. And the key reason lies in their education and training policies.
Thus, as Thomas Piketty concluded in his widely acclaimed study of inequality: The best way to increase wages and reduce inequalities, in the long run, is to invest in education and skills.
Interestingly the Australian experience supports this conclusion. Inequality has increased less in Australia than the average for the OECD and the most likely reason is the rapid increase in the supply of people with higher qualifications.
As noted above, the wage premium for graduates fell slightly in Australia, while in the US it doubled. But in Australia, the proportion of graduates doubled whereas in the US it stayed about the same. The net result was a much bigger increase in inequality of incomes in America.
But Australia too could do better. The number one priority is to improve the equality of educational outcomes between and within schools.
Second, much more could be done through vocational education and training and active labour market programs to assist people to adjust to technological change. At present Australia spends about 0.3 per cent of GDP on active labour market programs to assist unemployed or underemployed people; by comparison, the Danish system of ‘flexisecurity’ spends 1.9 per cent of GDP on retraining programs in combination with more generous income support.
Railing against ‘neo-liberalism’ will achieve nothing in terms of a more equal distribution of income. But governments can make a difference by fully embracing education and training.
Of course, governments also should offer adequate support for those who victims of structural changes in the labour market accompanying technological change. But in a sense, this is equivalent to palliative care and the highest priority should be to prevent increasing inequality from arising in the first place, by helping people to help themselves through education and training.
The results from this reform agenda will then not only help preserve the character of our society and politics, it will also help our economic performance.