Consumption, Wealth And Business Cycles: Why Is Germany Different?
Britta Hamburg
Deutsche Bundesbank – Economic Research Centre
Mathias Hoffmann
Deutsche Bundesbank
Joachim Keller
Deutsche Bundesbank – Economic Research Centre
2005
Bundesbank Series 1 Discussion Paper No. 2005,16
Abstract:
This paper studies the long-run relationship between consumption, asset wealth and income – the consumption-wealth ratio – in Germany, based on data from 1980 to 2003. Earlier papers for the Anglo-Saxon economies have documented that departures of these three variables from their common trend signal future changes in asset prices. We find that for Germany they predict changes in income — the consumption wealth ratio predicts business cycles, not stock market cycles. Asset price changes are found to have virtually no effect on German consumption, both in the short as well as in the long-run. Conversely, German asset prices are predictable from the U.S. consumption-wealth ratio. We offer an explanation of these findings that emphasizes structural differences between the bank-based German financial system and the rather market-based Anglo-American system: stock ownership by private households is much less widespread in Germany than in the Anglo-Saxon economies and the share of publicly traded equity in household wealth is much smaller in Germany than in the U.S., the UK or Australia.
Consumption, Wealth And Business Cycles: Why Is Germany Different? – Introduction
The idea that fluctuations in asset prices can have huge effects on the real economy and notably on consumption has recently obtained renewed and increased attention. In particular during the decline of international stock markets in the first years of this decade it was feared that consumers in countries where stock ownership is relatively widespread, might reduce their spending in response to an abrupt decrease in asset wealth.
Most extant empirical studies document a long-run relation between wealth and consumption, but the evidence on the effects of sudden and abrupt changes in asset prices — those most feared by policymakers — is much less clear cut.1 One important reason why certain asset price busts may lead to pronounced adjustments in consumption whereas others do not is that the prices of financial assets may have transitory components. According to economic theory, consumption should react only to the permanent component of wealth. This could explain the long-run link between consumption and wealth. But to the extent that consumers perceive certain asset price fluctuations, e.g. the bull market of the late 1990s, as a temporary phenomenon, consumption should neither react to a build-up nor to a subsequent correction in stock prices.
If temporary fluctuations of wealth leave consumption unaffected then it should be possible to identify them with fluctuations in the consumption-wealth ratio. This fundamental insight underlies a recent strand of empirical research initiated by Lettau and Ludvigson (2001, 2004) that has demonstrated very convincingly that an empirical characterization of the consumption-wealth ratio predicts capital gains, and in particular excess returns in the stock market.
The results obtained by Lettau and Ludvigson for the United States have been corroborated for other economies (Fernandez-Corugedo et al. (2003) for the UK and Tan and Voss (2003) as well as Fisher and Voss (2004) for Australia), but all of these studies are based on data from Anglo-Saxon countries. To the best of our knowledge, there has, to date, not been any comparable evidence for economies in continental Europe. One reason for this could be that asset wealth data are not readily available for most continental European economies. In this paper, we compile a unique new data set of German household wealth that explicitly accounts for real estate. This allows us to examine the wealth effect on consumption, based on German data, from 1980 to 2003.
Our results — besides being of interest in their own right — provide important differential evidence vis-`a-vis those studies that have concentrated on the Anglo-Saxon economies. Germany’s financial system is one of the main representatives of the continental European type of financial system where private stock ownership is much less widespread than in the Anglo-Saxon countries and households generally hold large shares of their wealth in the form of relatively illiquid assets. The evidence we present here suggests that these differences find their reflection in a very different transmission mechanism between financial markets and the real economy and in particular in a very different role of asset price fluctuations for consumption.
In keeping with Lettau and Ludvigson, we can characterize the consumption-wealth ratio as a cointegrating relationship between consumption, asset wealth and income — the cay residual. But while earlier studies find the consumption-wealth ratio to predict fluctuations in asset wealth and in particular in stock prices, we find that the German cay mainly predicts temporary fluctuations in income — cay signals business cycles rather than stock market cycles. The dynamic analysis we conduct shows virtually no evidence of an effect from asset prices on German consumption, irrespective of whether these asset price changes are permanent or transitory. In German data, shocks to consumption ultimately reflect permanent shocks to income, in line with quite basic permanent income models.
We note that German asset prices and in particular stock markets do have transitory, predictable components; we find the U.S. consumption-wealth ratio to be a very good predictor of excess returns on the German stock market. However, stock price fluctuations hardly affect German household wealth, because households’ direct ownership of stocks in Germany is very limited. This explains why fluctuations in the German consumption-wealth ratio do not help identify these transitory components.
The remainder of the paper is structured as follows: section two discusses recent evidence on stock market predictability and the particular role that the consumption wealth ratio plays in this literature. We build on Lettau and Ludvigson (2001, 2004) to derive the empirical approximation of the consumption-wealth ratio in terms of a cointegrating relationship between consumption, asset wealth and income. Section three offers a preview of our main results and suggests an interpretation. In section four we discuss our data set and our econometric implementation. Section five discusses and concludes.
See full PDF below.
