-statistic is the ratio between a normal (0,1) distributed variable and the square root of a centrally distributed chi-square variable divided by its expected value, i.e., its degree of freedom.

This paper evaluates empirically some very common theories of the freight rate generating process in the time charter markets. After a review of the most common assumptions made of the way time charter rates are set, the hypotheses are identified as follows: (a) the Zannetos Hypothesis, (b), the Lagged Zannetos Hypothesis, (c) the Koyck Lag Hypothesis, (d) the Rational Expectation Hypothesis, and (e) the Conventional Wisdom Hypothesis. These hypotheses are tested using statistical cointegration analysis that includes both an Augmented Dickey-Fuller (ADF), and a Johansen likelihood ratio test. Confronted with the data, hypotheses (a) and (b) are rejected outrightly. In the Koyck Lag case, the ADF statistic seem to confirm the hypothesis. A closer look at the numbers reveals that all of the impact on the time charter rates comes from the lagged dependent variable. Hence, the Koyck Lag Hypothesis is rejected. In the Rational Expectation case, the two tests conflicted. Based on the fact that the Rational Expectation Hypothesis includes the lagged dependent variable and that the Johansen test has been found to be a more robust test than the ADF test, the Rational Expectation Hypothesis is rejected. The fifth hypothesis is a reflection of the general bulk industry perception that the time charter rate is impacted by changes in the comparable spot rate and not much by the spot rate levels. In this case both the ADF and the Johansen test accepted the hypothesis for all markets. Thus, the paper concludes that the conventional market explanation of the time charter freight rate setting process is essentially correct-spot rate changes matter spot rate levels do not.