| The thesis consists of two separate but related chapters. In the first chapter, the predictable nature of the financial crises which occurred in the United States from 1886 to 1914 is investigated, and in the second chapter the properties of the U.S. stock market from 1886 to 1928 are studied.; The results of the first chapter are largely negative. Using monthly and weekly data on a wide variety of economic sectors, probit models are incapable of accurately predicting the financial crises even when estimated in-sample. And in vector-autoregressive models based on the same data set the innovations are not reliably unusual prior to the crises. However, there is some significance in the weekly probit models, providing weak evidence that stock market declines (bad news or recessions) and wheat supply deficits (droughts) were components of the set of conditions which led to the crises.; Many interesting results concerning the pre-Great Depression stock market are documented in the second chapter's analyses of monthly data, only the most important of which will be discussed in this abstract. First, stock returns are shown to have been very low (negative) and volatile during the more severe financial crises, and much of the deviation from normality present in stock returns is shown to be accreditable to these events. Second, the results indicate that the advent of personal income taxes did not significantly alter the seasonal component of stock returns; specifically, income taxes do not appear to be responsible for the anomalous returns of small firms in January. Third, it is found that stock volatility was not normally distributed, only weakly seasonal, and reasonably well modelled by a second-order autoregression. And lastly, regressions of stock returns on volatility reveal that returns were strongly negatively related to contemporaneous unexpected volatility and somewhat positively related to next month's expected volatility. As in similar studies using more recent data, these regression results are interpreted as evidence of a positive relationship between returns and anticipated volatility (risk); however, as in the other studies, the low explanatory power of the models depreciates the inferences. |
