Analysis Of Pork Price Breakpoint Regression Based On Nonparametric Method

The hog industry is pillar industry of animal husbandry in our country. And pork takes an important part in our dietary pattern. The development of pork industry will affect peasants’ income and employment, the relevant industries’ development and the rural economy. For the past few years,because under the market mechanism function, the fluctuation range of pork price grew gradually. The big swells made swine breeder hard to judge, which led to the shortage or surplus of pork in the market. And that in turn can lead to heightened volatility. So the core problem of the hog industry is the fluctuation of pork price.Supply and demand are determinant factors of pork price. The influence factors of pork supply are:the income level, substitute products price, residents’ consumption habit, emergency and so on. And there are many the influence factors of pork demand, every link of hog production, circulation and sell will affect it. This paper selects CPI, urban residents disposable income, corn price and beef price as the influence factors of pork price.This paper chooses the pork price from January 2000 to March 2016 as the object of study. During the period, there exists some circles and the trend is up. The period of wave lasts 3 to 4 years. And according to the trend chart of pork price, we can divide the period into 4 parts. So we can suppose there exists 3 change point. And after detecting the position of each change point, we can take advantage of the change point to implement a regression discontinuity(RD) design. This paper assume the variance may be different, detecting the positions of 3 change points base on strong consistency theory of change point detecting. Then I implemented a nonparametric regression design to the pork price.Because in nonparametric regression, when the number of independent variables grows, the sample size should increase exponentially, or there will be not enough sample point in local space. As the sample size is 195, it is impossible to use more than 2 independent variables in nonparametric regression. So I chose CPI and corn price as the independent variables. This paper compares the result of nonparametric regression using the whole period’s data and the result of regression discontinuity. And the conclusion is that, the mechanism of different stage differs from each other. If we use the whole data to implement nonparametric regression, the trend of estimated value is opposite to the true value in the last stage. And the residual sum of squares is bigger. So the regression discontinuity design is necessary. And use this method, we can draw conclusions to serve the policies.