Aggregation versus Disaggregation - What can we learn from it?

This paper identifies a possible bias in the relevance of certain variables explaining CPI inflation when estimated on an aggregate basis. Since the bias can lead to an unjustified exclusion of variables from the CPI inflation regression, I propose the alternative technique in order to better identify the relevance of economic variables in explaining CPI inflation. This paper also shows that the approach of modeling inflation through the disaggregation of its components is superior to the aggregated approach, because disaggregation increases both the accuracy of inflation forecasts and the efficiency of monetary policy management. I compare the performance of the aggregated and disaggregated monthly structural models for inflation processes in Israel, whereby the central bank interest rate and other economic variables are determined simultaneously with the inflation rate. Within these two different structural models, I also examine the implications on the central bank loss function of incomplete information regarding certain unobserved economic variables, and misidentification of the true inflation process in the economy. It was found that the disaggregated model generates smaller loss to the economy then does the aggregated model.