Essays on testing for speculative bubbles in the stock market

This dissertation proposes two empirical studies regarding tests for speculative bubbles in the stock market. The first essay tests the speculative bubble hypothesis assuming constant discount rates. I focus on a model that defines market fundamentals to be the sum of the expected present value of dividends and an omitted variable, called unobservable fundamentals, which are discounted at a constant rate. Testable implications for detecting bubbles and/or unobservable fundamentals are explored for constructed flow and stock specifications. The statistical analyses are performed on MSCI monthly stock market data for Germany, Canada and the United States, and I find evidence that unobservable fundamentals are serially uncorrelated with the stock price data when using the Hansen test. The empirical results also provide evidence to support the hypothesis of no speculative bubbles.; To compare West's procedure with the method that I suggest in this essay, West's approach is also evaluated using the MSCI data. With different information sets, the empirical results show a similar conclusion between both methods. The evidence shows that this testing procedure accepts the model based on the hypothesis that the model is correctly specified even when stock prices are added to the instrument set, which is the same finding as done using the procedure that I suggest. In particular, West's method provides evidence suggesting that the price process is misspecified.; In the constant discount rate model of the first essay, the question of whether the nonstationary of stock specifications can be explained by the variation in discount rates is not formally addressed. To overcome this limitation, in the second essay I propose a present-value model of stock prices with time-varying discount factors. Two testable specifications (stock and flow) are derived with a linearized present-value model. In particular, these two specifications can capture rational bubbles and/or model misspecifications in both time-varying and constant discount rate models. They thus allow me to examine the effect of time-varying discount factors on the behavior of price by simply comparing the unit-root test statistic estimates for both models. Using the same data set as in the first essay, there is evidence suggesting model misspecification, which can be interpreted as a failure of the no arbitrage condition in the stock market. The test results also support the hypothesis of no rational linear bubbles. Nevertheless, variations in discount factors can roughly explain the movements of stock prices for Canada and the United States, but not for Germany.