THE LONG MEMORY PROPERTY OF HUNGARIAN MARKET PIG PRICES: A COMPARISON OF THREE DIFFERENT METHODS

The present study investigates the long memory property of market pig prices. Simply knowing that these time series have long term dependence could have strong significance when forecasting prices. The presence of long memory is crucial information in making business decisions and creating portfolios. Long memory can be measured by calculating the so-called Hurst exponent. In our article, we studied and described three different methods (Rescaled range, Detrended Fluctuation Analysis, Autoregressive Fractionally Integrated Moving Average). Data consist of four time series (piglet, young pig, sow, slaughter pig) between 1991 and 2011. Before conducting the econometric analysis, all the series were seasonally adjusted using the TRAMO/SEATS method. Data preparation was followed by differencing the time series and testing their normality and stationarity. In the next step, we divided the analysed period into four parts and determined the Hurst exponent for each sub-period, using all three methods. In summary, results showed that slaughter pig prices are random; pig and piglet prices developed similarly and have long memory, while sow price changes definitely have short memory. Among the methods of pinpointing long term memory, ARFIMA was used for making the forecast. The forecasting ability of the method was compared to the traditional ARIMA model, with ARFIMA proving to be the better of the two.