More Problems With “Inflation” – Coping With Volatility

For traders, volatility is breakfast, lunch and dinner – market turbulence is good for making money.

But for policy-makers — like the leaders of the Federal Reserve, the European Central Bank, and the like – volatility in macro-economic indicators is highly indigestible. This is especially a problem for inflation indices, which track prices of goods and services to provide guidance for monetary policy. Some of the leading metrics suffer from excessive volatility in some of the key components. For example, the extreme volatility of energy prices is well-recognized.  

Trying to manage monetary policy with gauges that fluctuate wildly would be like driving a car where the speedometer was constantly fluttering between 30 mph and 60 mph. Taking a long-term average may reduce the effect — but only for looking at the past history. Policy-makers are forward-focused. They need guidance on where the inflation trend is headed. High volatility obscures that trend. 

Statistical engineers at the Bureau of Labor Statistics (BLS), the Bureau of Economic Analysis (BEA), and various branches of the Federal Reserve have tried to address this problem by creating alternative metrics which eliminate or segregate the most volatile components. The concepts for reducing volatility fall into three categories: 

  1. “Core” Indices: these eliminate specific components “known” to be volatile (e.g. energy)
  2. Trimmed Indices: these remove the statistical outliers – the components showing the largest increases and decreases each period
  3. Sticky Price Indices: these exclude the components whose prices change most frequently

MORE FOR YOU

Core Indices

The earliest approaches addressed the problem by eliminating whole categories of goods that have historically exhibited excessive volatility. The inspiration for this approach was the OPEC oil embargo, which produced gasoline shortages and severe energy price rises in the 1970s. This led to the creation by the BLS of a modified version of the standard Consumer Price Index (CPI), which deleted the volatile energy and food components – which became known as the “Core CPI.” The BEA introduced a “core” version of its preferred metric, the Personal Consumption Expenditure Index (PCE).

Aside from the practical motivation (a less volatile index), the deletion of energy and food components is justified (by the BLS) “under the belief that food and energy prices are volatile and are subject to price shocks that cannot be damped through monetary policy.” 

Trimmed Indices

The Trimmed Mean PCE and the 16% Trimmed CPI – promoted by economists at several branches of the Federal Reserve – “solve” the volatility problem by eliminating the statistical outliers. They rely on well-understood principles of statistical regularization, developed to control the distorting effect of large outliers in a statistical series. Trimmed indices cut off the “tails” of the distribution that include the components which experienced the largest increases or decreases each period. The trimmed components are not always the same. In a given month, “energy” may be trimmed if it is undergoing a big swing from the rest of the index – or to may be retained if it is not showing such a deviation. 

An even simpler approach somewhat in the same vein is the Median PCE, which simply records the inflation for the single component “whose expenditure weight is in the 50th percentile of price changes.” (This is so stripped-down that it may best be seen as a heuristic device, rather than a proper index.) 

Sticky Price Indices

The Sticky Price Index, developed by the Atlanta Fed, ranks the components by the frequency with which their price changes. The concept of price stickiness plays a central role in economics and in the rationale for monetary policy. The current view seems to be that the average price “spell” (between important or structural price changes) is 1 to 1½ years, although the average frequency for all changes (which includes “temporary” changes such as sales) is a little more than 4 months. Some prices change much more frequently (like gasoline), and some change much less often. The designers of the Atlanta “sticky price index” draw the cut-off line at the 4 month mark.

  • “We thought the average frequency of price change was a natural separating point. If price changes for a particular CPI component occur less often, on average, than every 4.3 months, we called that component a “sticky-price” good. Goods that change prices more frequently than this we labeled “flexible-price” goods… About 70 percent of [the components are] sticky-price goods and 30 percent [are] flexible-price goods.

  How successful are these approaches? What light do they shed on the current “surge” in inflation? 

The Adjusted Measures Do Reduce Volatility 

The concept of a “Core” index — eliminating the volatile categories of Food and Energy altogether – does reduce the variability of the inflation metric (as measured by the standard deviation of the series for the last 10 years). 

The Sticky index uses an alternative method of volatility reduction – eliminating the categories that frequently change prices. The components that are deleted are similar to the Core model – food and energy are prominent – but the “flexible-price” outtakes are more comprehensive. 

  • “About half of the flexible-price CPI comprises food and energy goods, the remainder being largely autos, apparel, and lodging away from home.”

This reduces volatility by half compared to “pure” Core model. (The Core filter is also applied to the Sticky index itself, but there is essentially no further improvement in volatility reduction, showing that price-change frequency captures all the vol-reduction available from the hard-deletions of the Core approach.) 

The Trimmed Mean provides the strongest reduction in volatility. It is the most stable of any of the Inflation metrics used by the Federal Reserve. 

Vol-Adjusted Measures Run Higher Than Unadjusted Measures 

Surprisingly, the volatility-adjusted measures run higher than the “raw” CPI and other unadjusted indices under “normal” macro-economic conditions, such as during the period from 2012-2019. 

This is significant. The debate over the validity of various inflation metrics often questions the motives for selecting one or another index. These results show that measures taken to control volatility do not add a downward bias to the reported figures. Reducing volatility – during “normal” periods – slightly elevates the measured inflation.

The Volatility Message: The Outliers Are Driving “Inflation” Today

The inflation regime changed dramatically when the pandemic hit. Starting in January 2021 the average of the “raw” CPI rises rapidly above the volatility-controlled measures.

Increases were not driven by the “Core” components, but by the Outliers – the known high-volatility categories – like food, energy, used cars, hotels. 

The gap between the “raw” CPI and the Core CPI rose to its highest level in a decade. 

The Policy Implications of Volatility-Driven Inflation 

This may seem all bit overly technical. But it has important iplciatns, both technical and substantive. 

Technically speaking, managing volatility in an index like the CPI and its cousins and derivatives is an important statistical challenge. Too much volatility in the index vitiates its usefulness. The unadjusted inflation measures that capture the attention of the media, the politicians and the public are too volatile, too “raw” to serve as reliable guides to policy. The surge in prices is not as generalized as the “raw” figures make it seem. “Inflation” is highly concentrated in just a few highly volatile components, which are obviously affected by supply constraints, rather than excess demand. (The previous column has analyzed the “Used Cars” example in detail.)

Substantively, volatility turns out to be an information-rich signal for policy-makers, though it is not interpreted correctly in many cases. The meaning of the signal is this: Prices that are overly volatile are not amenable to traditional monetary policy countermeasures

To repeat, in bold, the cautionary comment by the Bureau of Labor Statistics cited above: 

  • These prices are volatile and are subject to price shocks that cannot be damped through monetary policy. 

This should be inscribed in every economist’s crib-notes when it comes to talking about what to do about “inflation.” Raising interest rates will not bring down the price of gasoline. “Tapering” or “quantitative tightening” (i.e., having the Federal Reserve sell off its bond holdings) – while these measures may make sense for other reasons – will not reduce the cost of a used car. There may be fiscal measures that can help – a gas tax holiday, or a temporary tax credit for used car buyers. (That might sound strange – and I am not really proposing it – yet consider that the electric vehicle market has been the recipient of a similar stimulus for years.) But the point is that the Fed is helpless against supply-shock driven price rises, no matter how much they ballyhoo the power of the yield curve.

Comments are closed.