Causes and Consequences of Bias in the Consumer Price Index as a Measure of the Cost of Living

The accurate measure of prices is fundamental to almost every important issue in economics, from measuring economic progress to the conduct of monetary policy to the indexation of private contracts and public programs and tax rules. This paper reviews the causes of bias in the United States Consumer Price Index (CPI), updates the estimate of such bias (now roughly 0.8 percent per annum) following several improvements by the Bureau of Labor Statistics (BLS), notes the likely far larger substitution bias than previously estimated and calls for a series of priority improvements. Particular attention is called to the over 40 basis point slower growth of the BLS' C-CPI-U compared to the CPI-U, more than double the early 1990s estimates, which highlights the importance of moving to a formula such as the chained Tornqvist C-CPI that corrects for traditional substitution bias. The implications for mismeasuring the growth of real wages, real median income, and real returns to stocks and bonds are developed, as are the budgetary implications of the overindexing of spending and tax brackets resulting from the overstatement of changes in the cost of living.