What does the Consumer Price Index Measure? Bias Estimates for Poland

Considering the wide range of CPI applications in the economy, the accuracy of estimating changes in an average price of a consumer basket of goods attracts the attention of many economists. How the index of consumer prices is calculated is not, as it might seem, a merely statistical problem. The role of the CPI in monetary policy is particularly important in the countries whose central banks rely on direct inflation targeting strategy. The paper addresses the problem of bias in the measurement of inflation as provided by the price index of consumer goods and services (CPI) in Poland. We estimated the size of the bias resulting from two sources: substitution effect (Boskin et al. 1996) and the use of plutocratic weights in index calculation (Prais, 1959). The study involved a comparison of the official CPI index in Poland with superlative indexes (Diewert, 1978) and the democratic index in 2005-2009 period. The survey did not identify an upward CPI bias and the findings indicate a slight understatement of the CPI stemming from both sources (respectively: 0.1 and 0.3 pp. per annum). A downward bias due to substitution effect is rather unusual. A deeper analysis pointed to two possible explanations to this phenomenon. On the one hand, overstatement may be absent due to frequent adjustments in the weights used for CPI calculation, which results in a better match between the index and the changes occurring in the consumption structure. On the other hand, it was proved that in the period analysed, there was a faster-than-CPI rise in the prices of those goods and services the demand for which is relatively inelastic, and a positive growth of households’ real income has been observed over the recent decade. When looking into the “plutocratic gap”, it was found that the CPI (plutocratic) index for Poland was lower than the democratic index. Such a result of the “plutocratic gap” survey is in line with the research conducted for other countries.