Structural Changes in the Time Series of Food Prices and Volatility Measurement

Volatility in product prices is of considerable interest in the food and agricultural policy arena. Standard deviation (SD) type of measures, including variance and the coefficient of variation, have been commonly used for estimating realized volatility. These methods are relatively simple to calculate and focus on the width of the data. Despite these merits, they have a shortcoming in that the statistics can be amplified when the time series is nonstationary. This study suggests a new way to measure price variation. The structural breaks in the unconditional mean of a time series are determined, and then the conventional SD type of measures for each regime are calculated. This method addresses the weak point of the SD type of measure and is a competitive alternative to the conditional variance type or the trend deviation type of measures when the time series at comparison have notably different data-generating processes or have nonstationarity. Copyright 2012, Oxford University Press.