This paper proposes a protocol for agricultural and food price analysis, which goes beyond the existing practices of focusing on the time series properties of the variables in question, such as the existence of a unit root and of a
cointegrating relationship among nonstationary prices, by entertaining structural breaks when conducting those tests. Specifically, this study applies Perron’s unit root test and Gregory and Hansen’s cointegration
test (both of which allow for structural change of unknown timing) to the U.S. beef/cattle prices
and pork/hog prices. To obtain consistent estimates of the break dates in the retail beef,
wholesale beef and farm cattle price linkage equation and those in the retail pork, wholesale pork
and farm hog price relationship, the study employs Bai and Perron’s and Kejriwal and Perron’s
global sum of squares minimization procedure to search for the optimal break dates, utilizing the
principle of dynamic programming to ease the computation burden. The empirical results strongly suggest that beef/cattle prices are nonstationary, while there is ample evidence to support the conclusion that pork/hog prices are stationary.