Why doesn´t the Fed look at its own measures of price pressure?

From most accounts, the Fed is at a loss to understand why inflation is low. Some say their model is broken. Others that changes like globalization have messed their understanding of the inflation process.

A few years ago, the St Louis Fed developed an indicator to measure price pressure. According to them:

Policymakers usually want to know—to the extent possible—the probability that inflation over the next four or eight quarters will exceed the inflation target. Or, if inflation is very low, they may also want to know the probability that inflation will fall below zero (deflation).

Specifically, if Fed policymakers perceive a relatively high probability that inflation will rise above the 2 percent target rate over the next year, then the probability that the FOMC will raise the federal funds target rate likely exceeds the probability that the FOMC will reduce the federal funds target rate. Thus, assessing inflation’s likely path over some horizon matters to policymakers and those in financial markets.

To help policymakers, financial market participants, and others who have an interest in assessing future inflation probabilities, the Federal Reserve Bank of St. Louis has developed an index called the price pressures measure (PPM).

The PPM measures the probability that the expected inflation rate (12-month percent changes) over the next 12 months will fall in different ranges (0-1.5, 1.5-2.5 and above 2.5 percent.)

The chart below shows the probability of inflation in different ranges over the next 12-months during the recovery from the Great Recession.

For the past two or three years, the highest probability has been for inflation in the 0%-1.5% range. In late 2016-early 2017, the probability that inflation would be “on target” rose significantly, but has since recoiled strongly.

Despite those indications, the Fed keeps insisting that low inflation is due to temporary or transitory factors, and should soon climb towards the 2% target.

One can only wonder if this indicator will have the same fate as the “Change in Labor Market Conditions” index which was recently discontinued because “it wasn´t useful”, meaning it contradicted the Fed´s views on the labor market.



  1. //

    “We take a different approach in our analysis. First, our framework uses a pure time-series
    model to forecast inflation. Simple time-series models have been shown to be as accurate as
    larger, more complex structural models—and the resource demands on the forecaster are significantly
    smaller. Our model is a Bayesian vector autoregressive (VAR) model augmented
    with a set of factors that summarize disaggregated price, employment, and interest rate data.
    The set of factors is derived from approximately 100 economic and financial data series, including
    well-known measures of inflation expectations. We find, consistent with the NKPC, that
    inflation expectations matter. We use standard forecast accuracy tests to test whether our
    dynamic factor model produces a more accurate forecast than a simple, naive forecasting model
    (random walk) and a benchmark time-series model that forecasts future inflation based solely
    on lags of previous inflation. Finally, we use our dynamic model to produce forecast probabilities.
    For example, policymakers usually want to know whether the probability that inflation
    over the next four or eight quarters will exceed the Fed’s 2 percent inflation target is greater or
    less than the probability that it will fall short of 2 percent.”


  2. //

    An intuitive result. The TIPS spread time series we use in our NGDP forecast is probably one of the top three most predictive market inputs we use. If you figure the traders in the TIPS market are mostly trying to forecast CPI inflation, then over time at least some of them should develop robust and sophisticated methods for prediction, information which is baked into the TIPS spread quickly. In general, the actual policy makers at the Fed are far behind the consensus in academic Macro. A good case has been made since at least the 80s that inflation is largely set by expectations thereof, of course NGDP is the big story.

Leave a Reply