ABSTRACT
The transition from time-series to cross-section regression analysis removes some problems, such as autocorrelation and trend, but others remain, such as heteroscedasticity. Before computing the cross-section, double-logarithmic regression of business saving on profit, it is necessary to consider one important institutional influence during the period 1949–63 which might qualify the regression analysis, namely the existence of dividend restraint. After 1955 the ratio of the gross business savings to profit decreased continuously. It is true that the profits tax burden of dividends was reduced slightly in 1956 but this cannot explain why the aggregate business propensity to save continued to fall after 1957. A macro-economic time-series does not measure the typical reaction of the companies because it is heavily influenced by the reactions of a comparatively small number of the giant companies: a micro-economic analysis is necessary to overcome the problem.
