Fed Chair nominee Warsh points to a lower measure of inflation to argue for rate cuts

Kevin Warsh has fleshed out his argument for lower interest rates by pointing to a less commonly-used measure of inflation that is close to the Fed's 2% target.  However, this measure may be giving the wrong signal and convincing his colleagues is likely to prove difficult.  Also, publicly advocating a lower measure after five years of missing the 2% target would dent the Fed's credibility.    

This month marks the start of a new era for the Federal Reserve, with former Fed governor Kevin Warsh on track to take over as Fed Chair from Jerome Powell in a couple of weeks.

(Powell had planned to resign from the Fed, but said this week that he would stay on as a “low-profile” governor until the “[government] investigation [into the Fed] is well and truly over with transparency and finality”, a decision supported by Senator Thom Tillis, a key Republican decision-maker.)

Warsh’s views might change once he is in the role, but he testified in his confirmation hearing that he wanted to make major changes to the way the Federal Reserve operates, including overhauling the policy framework, forecasting approach, communication strategy and balance sheet.

At that hearing, he was questioned about whether he would cut interest rates to satisfy President Trump – or as Republican Senator John Kennedy colourfully asked, would he be “the president's human sock puppet” – something he forcefully rejected, saying, “The president never asked me to predetermine, commit, fix, decide on any interest rate decision in any of our discussions, nor would I ever agree to do so”, stressing that he would be “an independent actor if confirmed as chairman”.

Nonetheless, Warsh’s testimony fleshed out his argument for lower interest rates by pointing to a less commonly-used measure of core inflation, namely the trimmed mean PCE deflator, which is currently close to the Fed’s 2% target.

Central bank targets are framed in terms of headline inflation, but operationally policymakers have always tried to ignore volatility in prices that is unrelated to the economic cycle by focusing on the trend in inflation that can be influenced by interest rates.

Unfortunately, like most things in economics, there is no unique way of measuring the trend in inflation and different approaches can yield different results, although they tend to be correlated over time.

1. The traditional approach is the exclusion method, where volatile prices – nearly always food and energy, but occasionally “non-market” or government-influenced goods and services – are excluded from the calculation of core inflation.
2. Another way of estimating trend inflation involves taking a purely statistical approach, simply excluding volatile prices by relying on either the ranked distribution of price changes or a statistical model.
3. A third approach that is much less common involves estimating the trend in inflation using either a broader macroeconomic model or a statistical model that incorporates other economic indicators.

In the US, the Federal Reserve has long relied on the first approach, focusing on the core personal consumption expenditure (PCE) deflator, which excludes food and energy.

However, Warsh has dismissed the core PCE deflator, testifying in favour of the trimmed mean PCE deflator, which measures core inflation by excluding large price movements from the distribution of price changes.

“[At the Fed] we used … [the] core PCE [measure of inflation], so we'd exclude food and energy, because it was sort of a rough swag [i.e., ‘scientific wild-ass guess’] as to what was going on. … The measures I prefer are … trimmed averages, where we take out all of the tail risks, all of the one-off items … [These trimmed mean measures are currently] not where they should be [relative to the Fed’s 2% target], but I think that the trend is quite favourable”.

In the US, the Federal Reserve Bank of Dallas has long published a measure of the trimmed mean PCE deflator, which excludes large price falls in the bottom and price rises in the top of the distribution of all price changes.*

Warsh will presumably direct his staff to construct their own trimmed mean measure, but using the Dallas Fed series as a guide, the Dallas Fed measure shows annual inflation running at 2.4%, well below the core PCE inflation rate of 3.2% and close to the Fed’s 2% target.

Divergences between the core PCE and trimmed mean inflation rates are common, but this is an unusually large positive gap. Excluding the pandemic, it is the largest positive gap since 2010 and before that, the early 1980s.

The large gap means that a relatively small number of goods and services are underpinning the higher core PCE inflation rate. This likely reflects the influence of tariffs boosting goods prices, where core goods prices have risen by 2.8% over the past year after briefly resuming their pre-COVID declines following the pandemic.

However, there is the clear risk that there is more to rising goods prices than the impact of tariffs. Some businesses might have become accustomed to raising prices and this risk could become more important now that the Iran war has caused major spikes in transport costs and key industrial and agricultural commodity prices, as well as disruptions to supply chains.

There is also a lingering concern over whether the Dallas Fed series excludes too many prices to be representative of the trend in inflation, given it only covers goods and services whose combined weights account for less than 50% of the PCE deflator.*

For its part, the Dallas Fed has acknowledged that the trimmed mean measure may be giving the wrong signal at the moment, noting that, "the currently lower trimmed mean PCE inflation rate relative to the core PCE inflation rate may not be strong evidence of disinflation ahead if price changes continue to exhibit a positive skew and the trimmed mean PCE cuts too many of these changes".

Perhaps, though, the more pressing problem for Warsh will be to convince his colleagues on the FOMC to switch their focus from the core PCE deflator to a trimmed mean alternative.

In this respect, the experience of New York Fed President John Williams, who will be Warsh’s second-in-command on the FOMC, serving as Vice Chair, could give Warsh pause for thought. 

Williams has for some time recommended to other FOMC members that they should use his staff’s model-based measure of core inflation – viz, Multivariate Core Trend inflation – without any success.^

Interestingly, though, the New York Fed series currently provides an almost identical signal to the core PCE measure, with annual inflation rates of 3.1% and 3.2%, respectively, such that the lower trimmed mean measure is the odd one out among the three measures.  

The final more practical problem in Warsh promoting a lower measure of underlying inflation to build a case for lower interest rates is it would dent the Fed's credibility.  As the Brookings  Institute correctly argued, "Switching to a new metric when inflation has been above the Federal Reserve’s 2% PCE target for five years could be seen as moving the goal posts, potentially threatening the central bank’s credibility."  

Notes: 
 * As best explained by the RBA, trimmed mean inflation is calculated by ranking price changes from lowest to highest and then excluding the largest price falls in the bottom and price rises in the top of the distribution, such that trimmed mean inflation equals the average inflation rate of the remaining set of price changes. Different central banks make different decisions on how much of the distribution of price changes to “trim” based on their own analysis. 

• The Bank of Canada focuses on a 40% trim for the CPI, comprising 20% off the top and 20% off the bottom. 
• The RBA focuses on the trimmed mean CPI that lops off 30% of the distribution, comprising 15% off the top and 15% off the bottom. 
• The RBNZ relies on a trimmed mean CPI that excludes 10% of the distribution, comprising 5% off the top and 5% off the bottom.

Unusually, the Dallas Fed uses a large asymmetric trim totalling 55%, comprising 31% off the top and 24% off the bottom of the distribution. 

^The New York Fed’s Multivariate Core Trend model aims to measure persistence of inflation by calculating the common trend using a dynamic factor model. The European Central Bank also calculates a measure of core inflation using a dynamic factor model, but most central banks have not publicly adopted this approach.