The Effects of Temporal Data Aggregation on Price Transmission Analysis

Applied price transmission analysis is often carried out with temporally aggregated data. However, the effects of temporal aggregation on estimation and interpretation are largely ignored in the price transmission literature. Following Marcelino (1999) we show how temporal aggregation affects the parameters of the vector error correction models (VECMs) that are commonly used in price transmission analysis. Temporal aggregation does not affect the parameters of the long-run equilibrium relationships between prices, but it does affect the adjustment and autoregressive parameters that describe the short-run dynamics of price adjustment. Temporal aggregation also introduces moving average components into the error terms of VECMs. We present a Monte Carlo experiment and an application to wheat prices in the EU and the US to illustrate how failure to account for these moving average effects further distorts estimates of the dynamics of price transmission. These results have important implications. First, estimation with different temporal aggregates of the same price data will lead to differing and sometimes contradictory conclusions regarding the speed of price transmission, impulse-response patterns and price discovery. Second, it is not possible to derive conclusions regarding the dynamics of price adjustment at a higher frequency from estimates generated with lower-frequency data. In general, reliable insights into the dynamics of price transmission can only be generated if the frequency of the data employed in estimation coincides with the time decision intervals of the agents whose transactions determine prices.