Speculation

This dissertation extends previous research on bubbles by investigating whether changes in the financial asset prices of the S&P500 reflect changes in fundamentals. We propose that if this is not the case the volatility is due to a bubble. Hence, this is the general hypothesis from which several testable hypotheses are developed. A key issue in bubble research is the definition of fundamentals. In this work we assume that, in the long-run, operating revenues are the only source from which any payments can be made, including dividend payments. Therefore, if expectations are formulated correctly, on average, there has to be a relationship between changes in prices and changes in corporate revenues. Thus, we use different accounting variables as proxies for fundamentals. In addition, since the literature points to contagion of opinion as one of the causes for the creation of bubbles, we also examine the contemporaneous relationship between prices and several proxies for herding behavior. OLS, panel data analysis, and quantile regression are used to analyze the contemporaneous relationship between prices and fundamentals or contagion proxies; while cointegration (reconciled to be used with panel data) and the Bonferroni inequality are used to investigate the long-run equilibrium between prices and fundamentals. The results indicate that, overall, company earnings are not explanatory of prices. These findings hold both in the short-run and in the long-run equilibrium scenarios. In addition, we find that investors do not reward an increase of the debt in the capital structure of corporations. In reference to our contagion variables, changes in money flow, volume, and volatility are found explanatory of changes in prices. Nevertheless, the effect of these variables is not homogeneous across price changes. Specifically, Money Flow is significant across all quantiles except for the 30% lowest price changes, Volume is explanatory of the 35% highest price changes, while volatility is explanatory across all the distribution of price changes. An interesting observation is that the three independent variables become increasingly explanatory as we move up to higher quantiles. Taken together our findings are supportive of the bubble and contagion hypotheses.

The research questions of this study were: (1) Can non-randomness
be identified in the movement of commodity futures prices which would indicate
that the random walk theory is not fully applicable? (2) What model or models demonstrate this non-randomness? and (3) Are characteristics
of non-randomness uniformly present in different commodity types?
A data base of commodity futures price history was assembled
covering wheat, corn, oats, and soybeans. Considering the four commodity
types and the various delivery months for each commodity, a total of 215
years of price history was available. The historical data was used to test two methods of technical
analysis: (1) a moving average model and (2) a seasonal model. The models
simulate on a digital computer various decision rules in trading commodity
futures using the price history data as input. The models were evaluated
by comparing the resulting net profits from trading against the net profit
from a simple buy-and-hold policy.
Moving average and seasonal strategies were found which would
produce more profitable results than the buy-and-hold strategy. The moving
average model produced best results with wheat and corn. Oats showed the
poorest results and soybeans was in an intermediate position. Oats,
however, was the only commodity of the four tested which exhibited consistently better results using the seasonal model compared to the moving
average model.
Recommendations were made on areas of possible further study based
on the research results.