RBA Annual Conference – 1998 Discussion
1. Glenn Stevens
Any conference on unemployment has to devote a substantial part of its time to examining the international evidence and in particular, the exceptional performance of the US labour market. If we are interested in the generation of jobs, there is no other country which matches the US over the past fifteen years or so. As such, the US labour market is the best international benchmark for a discussion of Australian issues, and Katz's paper gets our discussion off to a good start, covering a lot of ground in an authoritative way.
The paper begins by summarising the long-term time-series evidence across countries. It is commonplace to compare the US with Europe, and such comparisons are frequent throughout both the papers for this session. Table 1 of Katz's paper is a useful reminder in that context that while US unemployment rates are widely admired today, this has not always been so. Through the 1950s, 1960s and 1970s, the US was a high unemployment country (and Australia a low unemployment country) by the standards of the OECD average. Indeed, I am told that many years ago, it was not unknown at international meetings for the US to be berated by other countries for its unemployment performance. This only really began to change in the 1980s, when US unemployment began to fall after the early 1980s recession; it did not rise much in the 1990 recession, and has fallen considerably after it. The contrast with Europe is striking.
It is perhaps worth noting – and Katz does note in passing – that the comparisons of end-point unemployment rates may flatter the US to some extent, because the early 1990s recession was mild in the US but severe in Germany due to the unique shock of unification and, by the wonders of the ERM, therefore severe in much of continental Europe. The US is also more or less at the end of an exceptional period of above-trend growth, whereas Europe has barely begun.
Nonetheless, the likelihood of structural differences in unemployment rates remains. The paper utilises a simple, standard NAIRU framework to analyse the time series for the two regions, and reaches the conclusion that the US NAIRU has not changed much over three decades, but that the NAIRU in Europe increased in the 1980s and increased again in the 1990s. This is a pretty standard result, and will not find much disagreement around the table. On the quantities dimension of labour market performance, then the US is clearly, to use the current adjective of choice, triumphant.
This point can been made even more clearly using statistics on labour force participation rates and employment to population ratios (Figure 1). What is striking here is partly the level the US employment/population ratio has reached, which is about equal to the highest amongst the OECD group, but more importantly, the size of the increase in this ratio – about 10 full percentage points over twenty-five years. Virtually no other country in the OECD group has seen an increase of this magnitude. In the big continental European economies, E/P ratios were higher than in the US in the mid 1970s, but have actually fallen and are now a long way below American levels. Those in the UK have shown no net change.
When I look at these, it seems to me that comparing unemployment rates may understate the degree of contrast between the US and Europe. Is it too crude to say that the US economy has seen a massive increase in the supply of labour, mainly in the form of women wanting to work, and that it has been remarkably effective in finding ways of utilising that increased supply to grow the economy's output? And that European labour markets, for whatever reason, have been stagnant or declining for the best part of twenty-five years? And, finally, that the Australian experience is somewhere between these two extremes (as it seems to be on most international comparisons)?
The paper considers various candidates for explaining the performance of the US vis-à-vis Europe on unemployment. Amongst these candidates are the usual ones like generosity of unemployment benefits and the extent of general labour market intervention which is less in the US than in Europe. As Katz (and others) point out, however, this was always true – so to explain the change in relative positions on unemployment over the past twenty years requires some additional hypothesis. One might be that European NAIRUs were always higher, but actual unemployment was held down by some other factor for a long time, though not permanently. Richard Jackman's paper contains an intriguing idea of this kind, which I'm sure discussion will want to take up.
Katz points out the way that various others have tried to handle this, which is by focusing not just on the labour market and regulatory structures, but also on how those structures interact with the shocks to which economies have been subject.[1] One set of shocks specifically mentioned is the rise in inflation in the 1970s, followed by efforts at disinflation, and the possibility that there are hysteresis effects as a result of this. The paper notes that evidence is pretty mixed across countries on hysteresis, but cites Larry Ball's work which suggests that longer or larger disinflations, combined with longer durations of unemployment benefits, seem to be associated with bigger rises in NAIRUs. Hence it was not until inflation rose in the 1970s and countries needed to have major disinflations, that we found out the real importance of these supply-side characteristics.
In Katz's models of the US and European Phillips curves, the sensitivity of inflation to unemployment is about the same in Europe as in the US. So Europe would not need a larger unemployment gap for any given amount of disinflation than the US, unless something else – like inflation expectations for example – were moving adversely. And I find it hard to believe that European central bankers set out to disinflate more slowly than the Fed. This seems to me to put any hysteresis back on to the labour market structures or other features of the economies in question, rather than different choices in disinflation strategies per se. There is still, furthermore, a question as to why it was that the 1950s and 1960s were apparently so conducive to such low unemployment in Europe (and Australia). Was it really just luck, with an absence of large adverse shocks?
Structures of the complete set of markets in the economy are no doubt relevant to employment and unemployment outcomes of the kind delivered by the US over the years; this point is touched on when Katz says that the labour market structures combined with liberal product markets ‘may add up to more than their individual parts’ in creating strong employment growth. My feeling is that there is quite a lot in this observation: the general dynamism of the US economy seems an important part of the economic outcomes there across the board. Flexibility of the labour market – with its hard edges and all – is part of the very essence of that dynamism, but there are other elements of the US system which are able to take advantage of that flexibility to produce the outcomes.
On the price dimension of US labour market experience, the paper is careful to acknowledge the downside to which many have pointed, namely the decline in real wages of US workers and the increased dispersion of wages across the earnings distribution and the resulting income inequality. I found the section on understanding the changes in the US wage structure instructive. This summary of the US literature suggests that demand shifts favouring skilled workers have been an important part of the story behind the changes in relative wages, with changes within industries and firms more important than changes between industries. Foreign trade has made some difference, but more importance is attached to technological change. In other words, the nature of work in virtually every part of the modern American economy is changing. Surely this is or will soon be true elsewhere as well.
The extent to which greater dispersion of wages is ‘bad’ may depend on whether those earning the lowest rates of pay stay permanently in that position, or whether they are earning low incomes simply at one stage of their working lives and moving up thereafter. If ‘McJobs’ were confined largely to teenage students who subsequently became skilled employees on higher incomes, perhaps we should worry less about this inequality than if they were a lifetime experience. So one question would be about the extent of such mobility in the US economy and whether it is changing. It might be worth us considering the evidence for this in other countries as well. I'm told that OECD work suggests differences in mobility are not that large between countries, though the US appears to have a bit more than others.
A related question which I am not clear on, is what light the US experience casts on the question of whether it is relative wage flexibility or aggregate wage flexibility which is most important in delivering strong employment outcomes. I think it is widely accepted that large changes in aggregate real wages mattered a great deal for aggregate employment outcomes in the 1970s and 1980s in Australia. But it seems to me that the unemployment we presently observe in Australia is decidedly unequal by broad skill classification. Is the US experience teaching us that there are substantial shifts in relative demands for skills, and that relative wage changes help to send signals which induce the necessary adjustments in labour supply?
If so, we will have to re-examine some cherished notions about the wage-setting system in Australia. Given historical concerns about equity, this will raise difficult questions about the nature of other support which might be given to the less well-off. Discussion of this in the paper is relatively brief. There are some tantalising references to various measures, including subsidies, training, and tax credit schemes right at the end of the paper. These issues of how to design the tax-welfare-wage interactions are at the heart of current thinking about better labour market outcomes in Australia. They will be taken up by other papers in some detail.
What questions for discussion can we take away from this paper?
I think from an Australian point of view one natural question to ask is what aspects observed in the US labour market experience might be expected here in future (to the extent they are not seen already). We have already seen a significant degree of increase in wage dispersion, though the Australian social security system has softened the edges of this trend to some extent if we look at household disposable incomes. Will this continue? What would be the costs of trying to resist it? We appear to be seeing changes in the relative demand for skilled versus unskilled labour. Will this result in higher relative returns to education? My understanding is that the available Australian data do not show an increase in returns to education thus far, perhaps because increased supply is keeping up with demand, and perhaps because the data are out of date.
Second, there is the relationship between the structure of labour markets and that of product markets. We have had considerable liberalisation of product markets in Australia, and this is driving labour market outcomes, generally in the direction of forcing considerable gains in productivity. But at that point, I think we need to include the rather vague notion of ‘entrepreneurship’ – the capacity and willingness to take and manage risks in pursuit of new opportunities, utilising the flexibility of markets and responding to their incentives. The US economy seems to have a lot of this. Does Australia? Australian firms since the end of the 1980s have pursued quite substantial productivity improvements involving reductions in workforce numbers, including in previously very secure areas such as banking, and middle management areas of large organisations. Is the supply of these displaced individuals to the employment market resulting in opportunities for ‘entrepreneurship’ to create new products and jobs? Or are those resources remaining underutilised?
Third, the obvious and probably most fundamental question for the conference: is there any way of combining the undoubted capacity of the US system in generating jobs and quickly re-locating displaced people into other forms of employment with the degree of equity in incomes which, other things equal, most would prefer to see? Or are we left with what some of the papers refer to as the ‘diabolical choice’ between a high employment, low average wage, high wage dispersion equilibrium and one characterised by a more compressed wage structure and low employment?
Footnote
This would then raise the question of whether all economies have been subject to the same shocks of the same intensity. One can think of some differences – such as German unification for example – but the other major shocks, such as the OPEC shocks and the productivity slowing in the mid 1970s, seem pretty common. [1]
2. General Discussion
See the general discussion following the paper by Richard Jackman (p. 67).