¹ Van Duijn, J. J., The Long Wave in Economic Life (London, 1983), p. 154.Google Scholar

² The causal perspective that has been emphasized in the literature is the Schumpeterian idea of innovation clusters.Google Scholar

³ Peak-to-peak trade cycle growth measures are used because the peaks are assumed to be more easily comparable than other points of the cycle, since they have the common factor of a capacity constraint.Google Scholar

⁴ A number of significance tests have been undertaken to test this point. First, a dummy variable has been included for one of the subperiods which proved to be statistically insignificant. Secondly, cubic, quadratic, and linear spline tests for structural change were undertaken. These also show insignificant variation. Before undertaking these tests I considered the problems of stationarity and autocorrelation in the data. It should be pointed out that in undertaking such tests I am viewing the problem as being amenable to time series analysis.Google Scholar The usual justification for using time series analysis as just another example from an infinite set of time series which might have been observed (Chatfield, C., The Analysis of Time Series: An Introduction [London and New York, 1980], p. 34).CrossRefGoogle ScholarThis metaphysical abstraction is unconvincing. The only practical justification for using statistical sampling techniques in time series is that they may help us understand and describe regular patterns. For example, given the presumed constancy of natural laws, astronomers use time series methods successfully to describe and predict the Earth-Sun movements. Quite clearly the homogeneity assumption required as a prerequisite for economic time series analysis is not likely to hold for all periods. However, during the period considered here (1850–1913) economic growth can be viewed as taking place within a fairly homogenous modern pattern. Moreover, the econometric problems dealt with here relate to simple descriptive models, not large causal models; thus, the homogeneity requirements are minimized.Google Scholar

⁵ This difference is significant at the 10 percent level with a dummy variable test. The idea of an accelerating growth path over the period 1856–1913 is also reinforced by the fact that the trend for world industrial production is best approximated by a second order polynomial, with both the first and second order coefficients being highly significant.Google Scholar

⁶ A similar conclusion has been reached by Milward, A. S., “Cyclical Fluctuations and Economic Growth in Developed Europe, 1870–1913,” in von Dietmar, H. Petzina and van Roon, G., eds., Konjunktur, Krise, Gesellschaft (Stuttgart, 1981), p. 52.Google Scholar See also Solomou, S. N., “Long Term Growth Phases: Long Waves, Long Swings and Traverses” (Ph.D. diss., University of Cambridge, 1983).Google Scholar

⁷ In mathematical terms, the world growth rate is defined as the weighted sum of the growth of the parts, that is, where g_wt = world growth rate g_it = growth of national parts α_itit = share of national parts, which are a function of relative growth rates. Over time the high g_it will also increase the corresponding α_it and hence, g_wt tends to increase. This is, of course, an abstraction on reality, since the growth paths of the national parts do not follow a steady state trend and the degree of synchronization in the world economy is not constant over time.

⁸ See Gerschenkron, A., Economic Backwardness in Historical Perspective (New York and London, 1962).Google Scholar

⁹ See Maddison, A., “Phases of Capitalist Development” in Matthews, R.C.O., ed., Economic Growth and Resources, vol. 2, Trends and Factors (London, 1980). The sixteen countries are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, Netherlands, Norway, Sweden, Switzerland, United Kingdom, and the United States.Google Scholar

¹⁰ Kravis, I.B. et al. , A System of International Comparisons of Gross Product and Purchasing Power (Baltimore, 1975).Google Scholar

¹¹ The difference in growth between 1872–1890 and 1890–1913 is significant at the 5 percent level.Google Scholar

¹² Similar calculations were undertaken employing the share averages of the cycle of 1872–1882 as the weights. The results are comparable to those employing 1870 weights.Google Scholar