Should the Fed Ignore the Most Important Bias in the CPI?

In the United States, there is an ongoing debate about the possible adoption of an explicit numerical inflation target for the Federal Reserve. An important facet of this debate regards what numerical value such a target should have. Bernanke et al. (1999, p. 28-30) discuss four reasons why an inflation targeting central bank should target a positive inflation rate: 1) CPI-based measures of inflation may be biased upward by 0.8 to 1.6 percentage points per year (Boskin et al., 1996). 2) Downward nominal wage rigidity may imply that a positive inflation rate could lower the “natural” rate of unemployment by facilitating real wage declines in declining industries (Akerlof et al., 1996). 3) Due to the zero lower bound on nominal interest rates, low inflation may constrain the central bank’s ability to stimulate the economy in the event of a recession (Summers, 1991). 4) Undershooting a zero inflation target may be more costly than overshooting such a target by the same amount due to liquidity and solvency problems in the financial system (Bernanke and James, 1991). The later three of these four possible reasons for a positive inflation target are hotly debated. However, there seems to be a widespread consensus that if measured inflation is biased upward the inflation target of the central bank should be shifted upward relative to what is decided to be an optimal level of “true” inflation by the amount of this bias. This consensus view was clearly stated by Ben Bernanke in his speech on November 14th 2007 that introduced changes to the communication strategy of the Federal Open Market Committee. In that speech Bernanke said: