Abstract
This article has two purposes. The first is to evaluate multi-functionality of horse production. Although the multi-functionality of agriculture has been widely recognized, little is said about that of horse production. We evaluated the multi-functionality of horse production in the Hidaka district, where more than 80% of domestic blood horses are produced. The second is to assess the effects of truncation, endogenous stratification, and overdispersion on the estimation of count data travel cost models using on-site samples. There have been almost no considerations of those effects in previous studies in Japan.
Our main findings are as follows: 1) The Poisson estimators are biased and inconsistent in the presence of overdispersion; therefore the negative binomial models are more preferable. Applying the Poisson re-gression models to our data results in overestimation of the consumers' surplus. 2) With on-site samples, accounting for truncation of the dependent variable makes a substantial difference in the coefficient estimates. However, accounting for endogenous stratification does not make a marked difference in this particular case. 3) The multi-functionality of horse production in the Hidaka area is estimated at about 11.0 hundred million yen per year by the negative binomial model corrected for both endogenous stratification and truncation, and at about 11.3 hundred million yen per year by the truncated negative binomial model respectively. These results indicate horse production provides visitors large amount of benefit through multi-functionality.
