RBA Annual Conference – 1995 Discussion

1. John Quiggin

One of the strongest motives in life is the desire to ‘keep up with the Joneses’. In economic debate, this tendency is reflected in a concern with international rankings of various kinds. These rankings are frequently used to bolster a variety of spurious arguments. A typical form of the argument is ‘Australia has declined relative to country X on scale Y, and we should therefore adopt policy Z (claimed to be practised in X)’. With an appropriate choice of criterion, almost any country can be used as the model. For example, New Zealand's per capita income has declined by 20 per cent relative to that of Australia over the past decade or so – hardly a promising basis for an argument based on relative performance. However, if attention is focused instead on the current account deficit, New Zealand can be made to appear as a model worthy of emulation. Ian Castles has made a number of useful contributions pointing out this kind of argument, and this paper is no exception.

First, Castles makes a number of judicious points about the general notion of ‘league tables’, and the idea that Australia's supposed slide from the top of the international table last century to a mediocre position today justifies some or other program of radical reform. In the 1980s, this idea was combined with observation of the performance of the ‘tiger economies’ to justify the prediction that Australians would soon become ‘the poor white trash of Australia’ in the absence of free-market reforms. (The same claim was occasionally used to justify interventionist policy as well.) In fact, the notion that Australia was exceptionally wealthy in the late 19th century is a statistical artifact arising from peculiar features of a frontier economy (Quiggin 1987). It is, in any case, entirely irrelevant to our current situation: with what possible policy decisions could a comparative advantage in wool production and the existence of some modest gold deposits have been made an engine of world-beating economic growth?

As Castles observes, on the current standard league table, all the major industrial countries (except the US) have per capita incomes within 20 per cent of that estimated for Australia. Given the wide range of uncertainty associated with the construction of such tables, this fact alone ought to be enough to dispel the idea that maintenance or improvement of our relative ranking is a justification for any particular policy agenda. Of course, there is nothing new about this observation. Castles made many of the same points nearly ten years ago at a conference in honour of Fred Gruen, who had himself made similar observations in Gruen (1986). But the myth that our relative growth performance is so poor as to justify radical reform refuses to die. For example, Clark (1995), presents the myth in its full glory, from our alleged world-beating status in the late 19th century to the rise of the Asian tigers as a reason ‘why microeconomic reform is unavoidable’. The only sign of progress is the absence of any reference to ‘poor white trash’.

In other work on this topic, Steve Dowrick and I have looked at a number of factors which make the standard World Bank league table an inappropriate basis for comparison, including differences in working hours, ‘quality of life’ variables and the range of uncertainty inevitably involved in making revealed-preference judgments from price-quantity data. Castles, however, challenges the ICP data on its own terms.

Castles' first point relates to the construction of the price indices. Because the world's consumption basket contains millions of different items, it is necessary to choose some representative subset to derive the index. If the items omitted are cheaper in some countries than in others, the relative price level in the first group of countries will be overstated relatively to that in the second. Obviously this is likely to happen if the items used in the index are chosen because they are heavily consumed in the second group of countries (e.g. small cars and small fridges in Europe). Hence, Castles argues, the PPP indices overstate the real income of European countries relative to that of Australia, US, Canada and New Zealand.

Castles next discusses the index number problem – that is, the irreducible degree of ambiguity associated with comparing consumption bundles at different sets of relative prices. In Dowrick and Quiggin (1993), we found that this problem led to the existence of several sets of ten countries, within which pairwise comparisons were generally ambiguous.

Can the idea of examining ‘the actual content of typical family budgets’ can tell us anything that the revealed-preference concept cannot? In principal I doubt it, but with imperfect data, it is possible that careful inspection of the household budget can give an indication of the standard of living. Certainly, in the Australian context, a food expenditure share of 50 per cent would be indicative of severe poverty (or highly idiosyncratic tastes). However, in a country with a higher relative food price, the food expenditure share will be higher, even for expenditure bundles that pass a revealed-preference test relative to some given Australian consumption bundle. I would conclude that we need to proceed with care.

The issue of ‘comparison-resistant’ items is a critical one. More precisely, it is a set of critical problems, one or more for each of the different comparison-resistant items. Castles focuses on housing, and the difficulties associated with valuing locations. The ICP data show Australia (and also the US) as having low levels of housing consumption relative to the OECD average, and, in particular, relative to Japan. Castles begins by considering the possibility that this is due to the failure to take account of location effects. However, as Castles finally concludes the ICP data appear quite simply, to be erroneous in its own terms. It is worth looking at the example of a representative housing unit cited by Castles ‘an apartment in a 20-year old multi-storeyed building, of 120 square metres, with central heating and one bathroom’.[1] This would hardly be appealing to the average middle-class Australian family, who would expect a separate house with a floor area of 150 to 200 m2 on a block of 800 m2 or more, and would regard ‘living in a flat’ as a serious come-down. On the other hand, for a flat, the stated specifications are relatively luxurious, particularly the requirement for central heating. I surmise that very few apartments meeting the ICP specifications could be found in Australia and that those few would be mostly up-market inner-city residences. Thus the problems with cars and fridges reappear, writ large, in the case of housing.

This still leaves unresolved the question of location and the closely related issue of ‘defensive expenditures’. Do we, for example, regard high levels of expenditure on heating fuel as evidence of a high level of comfort or as partial compensation for a bitterly cold climate. In this case, to paraphrase Lenin, households everywhere are voting with their feet.[2] Once defensive expenditures relating to a relatively severe climate and high crime levels are taken into account, the unambiguously superior measured performance of the US is likely to be replaced by a set of ambiguous comparisons with other wealthy countries.

Another set of problems arises with health care. Castles observes the measurement problems associated with public provision. A more fundamental difficulty is that the ICP measures inputs. It would seem preferable to measure outcomes, such as longevity and health status. The conceptual difficulties associated with incorporating outcome measures into an index compatible with revealed preference are substantial, but they can be overcome (Dowrick, Dunlop and Quiggin 1994).

Finally, a considerable part of the debate about high-growth economies, such as Hong Kong, Singapore and (until recently) Japan, concerns the role of factor inputs. One aspect of this is relatively long working hours. Castles stresses the Australian preference for leisure. On the available evidence (Dowrick and Quiggin 1993), taking account of leisure yields a slight improvement in Australia's standard of living relative to that of other OECD countries. However, in the OECD context, it is the apparent East-Asian appetite for work that is unusual. When account is taken of leisure, Hong Kong and Japan drop sharply, both in league table measures and in terms of revealed preference.


This is one of many examples of housing, but it appears that the ICP rental price measures are dominated by apartments. [1]

A counter claim is that the mass migration towards the equator observed in the residential choices of Americans and Australians, and in the holiday destinations of other OECD citizens, is a lagged response to the invention of air conditioning, another form of defensive expenditure. [2]


Clark, D. (1995), ‘Microeconomic Reform’, in P. Kriesler (ed.), The Australian Economy: The Essential Guide, Allen and Unwin, Sydney, pp. 142–169.

Dowrick, S. and J. Quiggin (1993), ‘Australia, Japan and the OECD: International GDP Rankings and Revealed Preference’, Australian Economic Review, 1st quarter, pp. 21–34.

Dowrick, S. and J. Quiggin (1994), ‘International Comparisons of Living Standards and Tastes: A Revealed Preference Analysis’, American Economic Review, 84(1), pp. 332–341.

Dowrick, S., Y. Dunlop and J. Quiggin (1994), ‘Living Standards and Social Welfare Indicators: A Revealed Preference Approach’, CEPR Discussion Paper No. 312.

Gruen, F. (1986), ‘How Bad is Australia's Economic Performance and Why?’, Economic Record, 62(177), pp. 180–193.

Quiggin, J. (1987), ‘White Trash of Asia?’, Current Affairs Bulletin, 64(2), pp. 18–25.

2. General Discussion

Discussion centred on two main questions about the use of international league tables:

  • What do relative real income levels imply about economic progress and standards of living?
  • What are the implications of changes in relative real income for policy makers?

There was general agreement that the relative levels of real income reported in conventional league tables did not adequately capture economic progress or relative living standards. In contrast, there was considerable disagreement about the role of such summary measures of economic progress as devices for gauging the benefits of alternative policy regimes.

One reason why relative real income levels may not adequately capture standards of living is that some societies make greater ‘defensive expenditures’. It was noted that such expenditures account for about 15 per cent of GDP in the United States. Examples of defensive expenditure include: spending to make up for a miserable climate (air conditioning in summer, heating in winter); spending to make up for a dispersed metropolitan area (automobiles); spending related to crime (police, prisons, private security), and so on. If one peels all this away, most of the difference in relative real per capita income levels between the US and Europe is removed.

In addition to the types of expenditure made, issues about the way in which these expenditures are denominated were raised. In particular, the possibility of measurement techniques creating a systematic bias against particular types of countries was discussed. The prices used to convert national income levels to a common unit were described as ‘Eurocentric’, and to the disadvantage of non-European countries.

Given the difficulties involved in identifying a comprehensive price set of representative goods, the merit of choosing a small basket of representative goods was debated. After all, in principle, only one price is necessary to denominate real product, as shown by the Big Mac league table of relative real income. Choosing a small representative basket would have the advantage of transparency. It was mentioned that one reason for doing this is that whatever the shortcomings of attempts to measure relative real income, there must be some ‘grain of truth’ in major changes in relativities between countries that are sustained over time.

Alternatively, it was argued that if one's basic concern is with relative living standards, ‘social indicators’ could be given greater attention. Social indicators were considered to be very important, especially in areas such as health where no policy specialist could glean relevant information from the ranking of real per capita income. The main disadvantage of such an approach, though, is that these indicators must be prepared for each social issue in question. In contrast, the appeal of the conventional league table is that it provides a single summary measure.

Nonetheless, the fact remains that what is left out of measures of living standards matters. For this reason, we must be circumspect about the meaning of the relativities reported. As one discussant put it, there is a greater need to develop a theory of how to raise growth and welfare than to explain existing relativities in measured real income.