Working Paper: CEPR ID: DP5370
Authors: Albrecht Ritschl; Martin Uebele
Abstract: This paper examines the comovement of the stock market and of real activity in Germany before World War I under the efficient market hypothesis. We employ multivariate spectral analysis to compare rivaling national product estimates to stock market behaviour in the frequency domain. Close comovement of one series with the stock market enables us to decide between various rivaling business cycle chronologies. We find that business cycle dates obtained from deflated national product series are severely distorted by interference with the implicit price deflator. Among the nominal series, the income estimate of Hoffmann (1965) correlates best with the stock market, while the tax based estimate of Hoffmann and Müller (1959) is too smooth especially before 1890. We find impressive comovement between the stock market and nominal wages, a sub-series of Hoffmann's income estimate. We can show that a substantial part of this nominal wage series is driven by data on real investment activity. Our findings confirm the traditional business cycle chronology for Germany of Burns and Mitchell (1946), and lead us to discard later attempts to date the business cycle.
Keywords: business cycle chronology; efficient market hypothesis; imperial germany; spectral analysis
JEL Codes: E32; E44; N13
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| real investment activity (E22) | nominal wages (J31) |
| stock market indices (G10) | national product estimates (E01) |
| stock market performance (G10) | real economic activity (E39) |
| Hoffmann's income estimate (J39) | stock market (G10) |