Asset pricing in created markets.

A further possibility is that quota prices adjust slowly in response to changes in profit conditions, and that the contemporaneous lease price is an insufficient indicator of expectations about future profits. This possibility could warrant the inclusion of multiple lagged values of the lease price in the estimation equation, as in the model of Burt (1986). We explored this by including the one-year and two-year lagged lease price in the fixed and random effects models, finding that these lagged prices were statistically insignificant and did not increase the total effect of lease prices on asset prices. In addition, we explored an adaptive expectations model (as described earlier) by including the lagged asset price as a regressor and estimating the model according to the approach of Arellano and Bond (1991) to account for the lagged dependent variable. The estimated coefficient on the lagged asset price was very small (0.08) and was statistically insignificant from zero, again suggesting that using the contemporaneous lease price in conjunction with the other variables affecting expectations is acceptable.

In general, the estimated coefficients on the other regressors are consistent with the predictions of the theory outlined earlier. Periods with higher interest rates have lower asset prices, ceteris paribus, as predicted by the basic present value model. As measured by higher mortality rates, stocks with more uncertain returns also tend to have lower asset prices, as is expected in the presence of risk-aversion.

With respect to the magnitude of these estimates, we refer back to the implications of the strict interpretation of the logarithmic approximation given by equation (6), which are [[beta].sub.2] [approximately equal to] - (1 + r)/(r + [theta] - g) and [[beta].sub.3] [approximately equal to] -[theta]/(r + [theta] - g), where each of the variables in these formulae is taken to be its mean value. For r = 6.4% and r + [theta] - g = 8.9% (based on the mean lease-to-asset price ratio), we would expect [[beta].sub.2] [approximately equal to] -12. Our estimate is [[??].sub.2] = -4, which is in the same realm, but somewhat muted relative to what the simple theory suggests. At the same time, an average risk premium of [theta] = 3.8% (based on [theta] = 8.9% - r + g and g = 1.3%) yields [[beta].sub.3] [approximately equal to] -0.4, which is similar to our estimate of [[??].sub.3] = -0.3. Note that although we do not have a direct measure of the risk premium, the mortality rate proxy we use should yield approximately the same estimated coefficient if it is directly proportional to the true measure of ln[theta].

Table 3 also reports evidence that stocks with faster-growing returns have higher asset prices, controlling for other factors. As noted earlier, growth in returns may be due to rising prices or falling costs. The former clearly has an important impact on quota asset prices, as growth in export prices is found to be strongly associated with asset prices in all specifications where this effect could be estimated. Regarding the magnitude of this effect, earlier we set out the hypothesis that [[beta].sub.4] [approximately equal to] (1 + g)/(r + [theta] - g), which yields [[beta].sub.4] [approximately equal to] 12, while our estimate is approximately [[??].sub.4] = 4. Interestingly, although the estimated coefficients on ln(1 + r) and ln(1 + g) are both lower than the theoretical expectation, they are approximately equal and opposite in sign, as suggested by the theory. One possible explanation is the presence of measurement error (the "errors-in-variables" problem), resulting in the usual bias toward zero.

Stocks where fishing costs are expected to fall over time are also found to have higher asset prices. This is seen in table 3 in two ways. First, recovering stocks tend to have higher asset prices, as expected. Contrary to our hypothesis, however, we find no evidence that this premium has dissipated over time, with a very small and statistically insignificant coefficient on the time trend found for recovering stocks. (18) One explanation for this finding may simply be that the expected future recovery of these stocks has yet to be fully realized, due to the life-cycle characteristics of the fish populations and/or ocean environmental conditions.

Second, we find that high levels of trade in the quota asset market are associated with higher asset prices across stocks, after controlling for other effects, but that this effect is statistically insignificant. The positive point estimate is consistent with the notion that stocks experiencing a high degree of rationalization after the introduction of the quota system feature decreasing fishing cost and thus become increasingly valuable over time.

Finally, the shellfish dummy (i.e., for rock lobster, abalone, and scallops) is found to enter specifications (i), (iii), and (iv) with a highly significant positive coefficient. This suggests that shellfish stocks tend to have higher asset prices than other stocks, ceteris paribus. One possible explanation for this additional effect of shellfish stocks is that the biomass of these species is typically estimated with more precision, and, hence, their catch rates are more certain. There is also anecdotal evidence that these stocks tend to have more effective cooperative management institutions (Yandle 2003).

Conclusion

When there are competitive fishing quota markets, rational asset pricing theory suggests that the price of quota should reflect the expected present value of future profits in the fishery. Evidence of economically rational asset prices implies that the market is conveying appropriate incentives to quota owners. Unless the TACs are set to achieve the optimal stock levels, however, quota prices are unlikely to internalize the full stock externality. Nevertheless, the incentives will be more closely aligned with economic optimality than they would be under traditional fishery management methods.

Random effects and other panel data models revealed that quota asset prices were related to contemporaneous lease prices in the expected manner in the New Zealand ITQ market. We also find that asset prices are higher when interest rates are low and for stocks that experience less biological fluctuation. Furthermore, stocks with higher growth rates of fish output prices tend to have higher quota asset prices. We also find that stocks thought to have experienced reductions in costs since the introduction of the ITQ market have higher asset prices, ceteris paribus, although these effects did not diminish over time as expected.

We conclude, therefore, that the New Zealand quota system as a whole has functioned reasonably well and the prices at which quota have sold appear to reflect expectations about future returns on specific fish stocks. The U.S. government's ocean action plan and recent legislative proposals encourage the regional fishery management councils to adopt market-based systems for fisheries management. For skeptics of these plans, and for fishery managers currently designing ITQ programs in the Gulf of Alaska, Gulf of Mexico, and along the west coast of the United States, our results provide additional statistical evidence that real world ITQ programs are transmitting the correct incentives to quota owners to address the common pool problem in ocean fisheries.

More generally, the relationships between the assets and dividends are further empirical support for the ability of tradable rights systems to lead to a more efficient utilization of resources.

We are grateful to Suzi Kerr, the research assistants at Motu Economic and Public Policy Research (New Zealand) and Resources for the Future, and the New Zealand Ministry of Fisheries for the provision of confidential trading data. We also thank Resources for the Future and the New Zealand Ministry of Fisheries for providing funding for this research.

[Received June 2005; accepted May 2006.]

References

Alston, J.M. 1986. "An Analysis of Growth of U.S. Farmland Prices, 1963-82." American Journal of Agricultural Economics 68:1-9.

Annala, J.H. 1996. "New Zealand's ITQ System: Have the First Eight Years Been a Success or a Failure?" Reviews in Fish Biology and Fisheries 6:43-62.

Annala, J.H., K.J. Sullivan, and C.J. O'Brien. 2000. Report from the Fishery Assessment Plenary: Stock Assessments and Yield Estimates. Wellington: New Zealand Ministry of Fisheries.

Arnason, R. 1990. "Minimum Information Management in Fisheries." Canadian Journal of Economics 23:630-53.

Arellano, M., and S. Bond. 1991. "Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations." Review of Economic Studies 58:277-97.

Balestra, P., and J. Varadharajan-Krishnakumar. 1987. "Full Information Estimations of a System of Simultaneous Equations with Error Component Structure." Econometric Theory 3:223-46.

Baltagi, B.H. 2001. Econometric Analysis of Panel Data, 2nd ed. New York: John Wiley and Sons.

Banerjee, A. 1999. "Panel Data Unit Roots and Cointegration: An Overview." Oxford Bulletin of Economics" and Statistics' Special Issue:607-29.

Barichello, R. 1996. "Capitalizing Government Program Benefits: Evidence of the Risk Associated with Holding Farm Quotas." In J.M. Antle and D.A. Sumner, eds. The Economics of Agriculture: Papers in Honor of D. Gale Johnson, Volume 2. Chicago: University of Chicago Press, pp. 283-99.

Barry, P.J. 1980. "Capital Asset Pricing and Farm Real Estate." American Journal of Agricultural Economics 62:549-53.