Time series analysis of monthly beef cattle prices with non-linear autoregressive models

This paper analyzes the dynamics underlying a time series of the monthly average beef cattle price received by producers in the State of São Paulo (Brazil). The time series under study records monthly prices since 1954. An exploratory analysis suggested that after a period of intense government intervention in the cattle and beef markets, the underlying dynamics seem to be settling to a pattern similar to the one observed prior to that period. In order to try to verify if the underlying dynamics after the interventionist phase are similar to those in former times, a forecasting procedure has been used based on non-linear autoregressive models. This type of models were used after the BDS test showed significant results which can be interpreted as non-linearities in the data. The results discussed in the paper seem to suggest that after a period of intense interventions that lasted over two decades, the current underlying dynamics are close (from a forecasting point of view) to those observed more than thirty years ago.