SUMMARY
The review of each Korean livestock sector provided the background needed to develop an econometric version of the sectors. The first step in the process was the estimation of each of the identified behavioral equations. Chapter 3 provides the results of the estimations. Each equation has an identified R-square, Durbin-Watson statistic and a t-statistic for each parameter in the equation. A variable definition list can be found near the end of chapter 3.
The general approach used in each sector was to first identify the primary supply point. In the case of beef it is the number of female cows greater than two years of age. For pork, it is the number of sows in the pork herd while for chickens the number of chickens hatched drives the supply side.
Each of these equations has a similar specification. These equations contain a lagged dependent variable to help capture the dynamics of the supply portions of the sectors. A ratio of output to input prices is also contained in the equations that drive the response of these equations to changing economics. The reader can look in the equations sheet of the Excel model to find the short and long run elasticities of each of the equations.
At the other end of the spectrum, each of these sectors contains primary domestic consumption equations. The consumption of each meat product depends on its own price, the price of the other two meats in the system and income. Since these equations are estimated in double-logarithmic form the coefficients can be interpreted as elasticities. The demand side of this model remains estimated in a single equation approach rather than using a system approach as the lack of precision in gathering the data causes a system approach difficulties even though it allows for more demand theory to be imposed on the system.
The other important portion of the model is the interaction with world markets. Trade equations are estimated that include comparisons of world to domestic prices, exchange rates and other trade barriers that limit trade to something less than the free trade solution.
A beef import and chicken net trade equations are estimated in the system. No trade equations are estimated for pork since trade has been a very small portion of the Korean pork market.
The remaining equations provide two primary functions to the model. First, the remaining supply equations ensure that the biological constraints of each of the sectors are not violated. That is, animals or meat are not created or lost in the system. These equations will ensure that when a calf, pig or chick is born that the animal remains in the system until it is slaughtered, dies or goes back to the breeding herd.
The other set of equations provide the means of translating retail meat prices back to producers. Many of these equations can be classified as derived demand equations that show what an actor is willing to pay for an animal of a particular size and age to turn it into the next stage of the process. The most obvious case is what a meat processor is willing to pay for live animals to turn them into meat.
Although estimation of each of these behavioral equations is important in building the model, the larger objective is to have a system that performs satisfactorily. Tables on the first two pages of appendix 2 provide performance statistics of the entire system over the 1995 to 2002 period. These statistics are dynamic in that errors in one year are allowed to be carried into subsequent years. The far right column of these tables provides the annual percent root mean square error for all endogenous variables. In general, the beef model performs the most poorly of the sectors but that is expected given the number of lags that are contained in the model. Even though it performs the least satisfactorily, many of the errors are below 20 percent. The pork and chicken sectors perform very well over the period. Appendix 2 provides graphics of the actual versus predicted values from the dynamic simulation.