|THE MACROECONOMICS OF FISCAL POLICY
FEDERAL RESERVE BANK OF BOSTON ECONOMIC CONFERENCE
June 14-16, 2004
IS THE US PRONE TO “OVER-CONSUMPTION”?
Jean-Philippe Cotis, Jonathan Coppel and Luiz de Mello
OECD, Economics Department1
“I can get no remedy against this consumption of the purse: borrowing only lingers and lingers it out, but the disease is incurable” William Shakespeare
In 2003 US national saving amounted only to 13 ½ per cent of GDP. This was the lowest rate in US history since the Great Depression and also one of the weakest among OECD countries. Following a relatively brief upward trend over the mid-1990s, the saving rate started to decline during the penultimate years of the investment boom, which continued during the subsequent investment slump. The weakening of national saving has been associated with a parallel shift of the external current account, with the deficit reaching 5 per cent of GDP, a level widely seen as unsustainable.
In this context, commentators and policymakers outside the US have become increasingly worried about a “structural” lack of saving in America, with potentially adverse consequences for capital accumulation world-wide. A major source of concern in the short run is a steep back-up of long-term interest rates crowding out private investment at a premature stage of the recovery. From a longer-term perspective too, the substantial fall of US national saving is not seen as an optimal response to the acceleration in total factor productivity and potential output that the American economy has enjoyed since the mid-1990s. It is often felt that the US investment boom has led to a costly diversion of foreign saving with detrimental consequences on potential growth elsewhere. What would have been needed instead is a stronger contribution from US domestic saving to the financing of the investment surge.
Whether these concerns are legitimate is an open question as reflected in this session’s title “is the US prone to ‘over-consumption’?” The issues deserve indeed a thorough examination, with a clear distinction between the past behaviour of the US economy and future prospects. In a nutshell, this paper will argue that “prone to over-consumption” is an excessively severe description of the past, but may well prove a real issue in the future.
Looking at past behaviour in a purely descriptive way, it seems obvious that the US has been consuming over and above what historical and cross-country standards would suggest. The American situation is not unique, however, and bears many similarities with developments in the UK and other English-speaking countries, such as Canada and Australia, where either national or household saving have fallen to very low levels. It should also be added that these countries have in common a very robust macroeconomic performance in recent years: their output growth has been strong and their resilience during the past slowdown was impressive.
This common pattern of performance and behaviour could be taken to suggest that falling saving rates are the desirable response in a variety of circumstances. For instance, during the past economic slowdown, falling public or private saving in English-speaking countries may have been key to stabilising the world economy. Conversely, an overhang of precautionary savings may have proved detrimental to growth, as suggested by the lacklustre performance of continental Europe.
In a world where capital has become very mobile, falling saving may also be appropriate for countries experiencing above-average growth and profitability. As foreign investors seek to share in these gains, home-country asset prices are driven up, generating substantial wealth effects and lower competition from domestic savers. In principle this decline in domestic saving is welfare enhancing: it allows profitable portfolio diversification for foreign investors and consumption smoothing for domestic households without necessarily harming world capital accumulation.
With a bit of creativity one can therefore find good reasons why apparent “over-consumption” was in fact desirable. Given the very steep fall of US national saving it may nonetheless be worth assessing the extent to which such a decline could be due to suboptimal public policies and “abnormal” behaviour on the part of private agents and what this may imply for the sustainability of output and consumption growth both in the US and world-wide.
The objective of this paper is therefore to survey recent empirical work at the OECD and elsewhere that can shed some light, in a tentative and crude way, on a possible US “over-consumption” problem. The paper is organised as follows. The next section briefly reviews recent and longer-term developments in US saving, and considers how these compare internationally. It also addresses a number of important measurement issues. Section 3 discusses the concept of “prone to over-consumption” adopted in this paper and how it relates to the theoretical notion of optimum consumption. In section 4 we try to identify periods of unusually high consumption using econometric methods for ex-post analysis and then assess judgementally whether these situations represent cases of over-consumption in a normative sense. A fifth section considers the likely consequences for US national saving and long-term growth of a lasting shift towards high structural public deficits. Section 6 summarises our main conclusions.
2. Past saving trends
2.1 Measurement of saving2
Saving can be conceptualized as output produced in a period that is not consumed in the same period; instead, it is available for future consumption. In principle, a straightforward operational definition of saving is income less consumption. But in practice saving statistics are fraught with measurement problems (see Box 1). These relate to the measurement of income, the classification of expenditures between consumption and investment, and the allocation of saving between households, businesses and government sectors.
The effect of these measurement problems not only show up at the aggregate level, where the effects can be large3 but, sometimes more importantly, also influence the sectoral composition of saving. The household saving rate, for instance, is biased by revaluation effects due to inflation, the partial treatment of capital gains in personal income, and the treatment of unincorporated companies. Focussing on the level of saving in one or the other sector is thus likely to be misleading.
Box 1: Some issues in measuring saving in the national accounts
The economic definition of saving differs from the one followed in the System of National Accounts (SNA). The SNA only treats as income those revenues that are generated from the current production flow, ignoring revaluation effects on the stock of wealth. Thus, when, for instance, capital gains are realised, they are not included in personal income, even though taxes paid on them are fully deducted from income. This implies a shift of income, and thereby saving, from the household to the public sector.
A period of sustained capital gains can also lead to an artificial shift of saving from the household to the corporate sector, because of the treatment of saving through defined-benefit pension plans. Under such schemes, capital gains may enable employers to reduce their direct contributions to employee pension funds while keeping the system fully funded. In this case, since employers’ contributions are counted as “other” labour income, measured wages and salaries decline, even though the benefits to beneficiaries, and hence consumption plans, remain unchanged. Likewise, revaluation effects due to inflation likewise cause a downward bias in the measure of household saving, since nominal interest payments are included fully as a component of current income, but the erosion of the real value of monetary assets and gains on debt are not.
The measurement of saving also depends on how certain expenditure items are classified. By convention spending on education and on R&D funded by the public sector are treated as consumption, though from an economic point of view they could be considered as investment, given they contribute to raising future levels of potential output. It could also be argued that the purchase of a durable good by households should be treated as investment, as is currently the practice when the buyer is a firm.
The methodological difficulties in measuring saving are most evident at the sectoral level. Apart from the measurement problems already mentioned, this is due to the inconsistent treatment of like transactions across sectors and more generally the arbitrary boundary between household and corporate saving. For example, although dividends and corporate repurchases both involve shifting funds from the corporate to the household sector, they have different effects on the sectoral composition of private saving. The trend in corporate finance away from dividend payments towards capital gains exerts downward pressure on the measured household saving rate. Similarly, incorporations of unincorporated companies result in shifting saving from the household to the corporate sector. Another example is the treatment of saving in pension plans. Transactions between households and government social security systems are considered as current, while those with private schemes are treated as capital transactions.1
1. The rationale for this difference reflects the view that households generally regard contributions to private pension schemes as financial investment, but consider social security contributions as being of the same nature as income taxes.
Statistical uncertainties surrounding the measurement of savings may be less pervasive, however, when it comes to assessing long-term trends. Such a possibility is suggested, for instance, by Gale and Sabelhaus (1999) in the US case. The authors adjust savings statistics for the classification of consumer durables in investment, the inflation tax and the treatment of saving in government retirement plans and other federal social insurance trust funds. They find that, except for the inflation tax, adjustments to total private saving remain broadly constant over time, thus adding to the level, but not changing the profile of saving4. The adjustment for the inflation tax has fallen markedly, however, over the past few decades, moving from -2 per cent of GDP during the 1970s to -0.3 per cent of GDP only in the late 1990s.
In sum, on the basis of US experience, it appears that focussing on the level of private saving is potentially misleading, but the broad trends in private saving rates, as measured in the SNA, are probably more reliable.5 Indeed, in the US, the fall of the national account savings measure over the period reflected for a large part a fall of the “true savings rate” alongside an easily identifiable decrease in the inflation tax,6 while other adjustment factors played a relatively minor part.
Bearing in mind measurement caveats, historical trends in US saving by sector relative to GDP are shown in Figure 1. The key features to observe, abstracting from cyclical variations, include:
The household saving rate was broadly steady through the 1960s and 1970s. However, since the early 1980s it has exhibited a downward trend, falling by the equivalent of some 7 per cent of GDP to reach 3 per cent of GDP, an historical low.
In contrast, corporate saving, which is closely related to profit developments, has remained essentially constant relative to GDP over the past 4 decades, averaging about 11 per cent.
Trends in private saving, therefore, have mirrored those of households. Moreover, the private sector saving rate is more stable than its household and corporate components. This seems to suggest that households pierce the “corporate veil”, although empirical work usually indicates offsets are incomplete.7
Movements in the public saving rate have frequently coincided with opposite changes in the private saving rate. Thus, the national saving rate has tended to be more stable, although it has declined somewhat over the past three decades.
Measures of gross saving have fallen by less than net measures (not shown in Figure 1), in line with the shift in the composition of capital towards ICT equipment, which has a higher depreciation rate.8
Figure 1. Historical trends in US saving rates
Source: US Bureau of Economic Analysis.
Certain features of trends in US saving are equally evident in other countries. For instance, changes in fiscal stance have often taken place concomitantly with opposite co-movements in private saving, thus smoothing fluctuations in national saving (Figure 2), as discussed further in Section 5.9 A decline in domestic saving has also been evident in Japan, France and Italy over the past three decades, although the initial rate of saving in these countries was higher. And household saving rates have dropped considerably in most G-7 countries. There are also important differences among OECD countries. The German and Canadian national saving rates have been remarkably stable, while developments in corporate saving rates have been diverse, rising in some countries and falling in others. Overall, however, the US domestic saving rate is lower than in all the other G-7 countries, except the UK.
Figure 2. Private and public saving: deviations from averages