Results and Discussion
Here, we present the regression results of Equation (4). We employ panel regression techniques with fixed and random effects. Since a Hausman test revealed that the random effects and regressors were correlated, we have discarded the random effects model and report only the results of the fixed effects model. Specification (1) in Table 3 comprises the entire sample for the whole observation period, while specification (2) is limited to the bubble period pre-1990. This particular period deserves special attention, because stock market valuations of firms might have been irrationally exaggerated without reference to real economic indicators. The concept of Tobin's q which acts as our dependent variable, however, assumes a rational assessment of the asset base of a firm. Rational means that a firm's market value can be derived by its future cash-flow stream. Finally, in specification (3), an auto regression term was introduced into the equation in order to account for possible autocorrelation, which typically occurs in panel data. Finally, we experimented with some control variables such as size, as measured by the number of employees, and profitability, as measured by profits on total assets, that may potentially influence Tobin's q. Following Hirshey and Weygandt (1985), we also included sales growth in the equation. As the results were not altered much by the control variables, we report only the outcome of the latter specification in column (4).