Abstract: Precisely characterizing the dynamic behavior feature of hog industry chain is the basis of predicting the hog price trend, and the important basis for the government to formulate feasible policies to stabilize the hog price as well. However, from the perspective of relevant studies at home and abroad, no matter it is a linear model such as SVAR model, error correction model, GARCH model family, or a time series decomposition method such as HP, BP filtering method, EMD decomposition and B-N decomposition method, it only analyzes the external fluctuation appearance of hog market price time series,which makes it impossible to reveal the structural change characteristics and internal change laws of the hog market price time series behavior.This is also an important reason for the paradox that a series of policies to stabilize the hog price have been introduced intensely since 2006 in China, but the price appears abnormal fluctuation instead.
Therefore, based on the monthly data of hog price and pork price in China from June 1994 to June 2018, the hog price and pork price sequences are divided by using the multi-regime smooth transition autoregressive model to analyze the regime system transition characteristics. Then, the nonlinear dynamic adjustment behavior of hog price and pork price is further studied via the analysis of the characteristic root of the multi-regime smooth transition autoregressive estimation equation and nonlinear impulse response. It systematically describes the regime transition and nonlinear dynamic adjustment behavior of the hog price industry chain.
The results show that, the hog price follows a smooth transition process of three-regime, while the pork price is a smooth transition process of four-regime.The hog price has long duration only in "low-price regime" and "high-price regime", while the pork price has long duration in "depressed price regime","price firm regime" and "price recovery regime".From the perspective of regime transition path, the hog price can easily fall rapidly from the middle price to the low price, and the pork price tends to transfer from "price slump regime"to "depressed price regime"and from "price recovery regime" to "price firm regime".The impact of random shock on pork price is persistent and has a weak "self-regulation mechanism", and the impact on hog price is persistent but lacks "self-regulation mechanism". In the short term, random shock has a positive effect on the price fluctuation of hog market. In the long term, random shock has a negative effect on hog price and a positive effect on pork price.These conclusions provide important decision-making basis for the government to further optimize the policy of hog price fluctuation.

Key words: hog price, pork price, regime transition, multi-regime smooth transition autoregressive model, nonlinear impulse responsefunction