Skip to content

The Determinants of Inflation (Kinlaw, Kritzman et al.): summary and key takeaways

The paper splits US core inflation history into four regimes with a Hidden Markov Model. It then uses a Mahalanobis distance attribution to rank eight economic variables by their influence on regime likelihoods. For February 2022, federal spending ranked first by a wide margin.

Kinlaw, W.B.; Kritzman, M.; Metcalfe, M. et al. · 2022 · SSRNSummary updated 20 November 2026
Published2022Working paper
Asset focusMacroInflation
TypeResearchInflation & Macro
Reading time4 minthis summary

The Determinants of Inflation (Kinlaw, Kritzman et al.) at a glance

The question
Which economic variables were the most important drivers of changes in US inflation at different points in time, including the 2021-2022 spike?
Dataset & period
US core CPI (all urban consumers, not seasonally adjusted, excluding food and fuel) from the St. Louis Fed, monthly, January 1960 through February 2022. Eight monthly economic variables over the same span. Asset returns use the S&P 500, a long government bond index and cash.
Method
Inflation shift is the latest one-year inflation rate minus the average rate of the prior three years. A Hidden Markov Model fitted with the Baum-Welch algorithm finds four regimes: Steady, Rising Stable, Rising Volatile and Disinflation. For each month, the Mahalanobis distance of eight variables to each regime's average gives a regime likelihood. The derivative of those likelihoods with respect to each variable, scaled by full-sample standard deviations and normalised, gives its relative importance.

Key findings of The Determinants of Inflation (Kinlaw, Kritzman et al.)

  • All four regimes are highly persistent, and Rising Volatile is the most persistent. The authors read this as a poor sign for the recent inflation spike fading quickly.
  • Producer prices, wages and personal consumption grew fastest in the Rising Volatile regime and slowest in the Disinflation regime. Federal spending still grew at a fair pace in Disinflation, a regime that often overlaps recessions.
  • Equities and bonds fared worst when inflation was both rising and volatile, before and after adjusting for inflation. Cash gave only slight cover, and almost none in real terms.
  • In the late 1960s, federal spending was the top driver from the third quarter of 1966 to January 1968. Money supply growth led from April 1968 to May 1971, and inflation expectations dominated from June 1971 to mid-1978, apart from a few months.
  • In 2008-2009 the main drivers moved from inflation expectations to personal consumption and federal spending, then to wages and salaries. Toward the close of 2021, federal spending and inflation expectations were the leading forces.
  • As of February 2022, federal spending carried 41.6% of relative importance, well ahead of every other variable. The authors compare the setup to the late 1960s.

Robustness, as reported in the paper

  • The regime likelihoods built from the eight economic variables track the Hidden Markov Model regimes closely, which the authors take as evidence that the specification is sound.
  • Data are used as originally available at each date without adjusting for revisions, because the aim is explanation, not out-of-sample forecasting.

Caveats and limits

  • The paper explains inflation in sample and does not test forecasting, trading rules or out-of-sample performance.
  • Importance scores measure a variable's influence on regime likelihood, not its size or a causal effect. The authors state this explicitly.
  • Several inputs are spliced or proxied: headline PPI before core PPI exists, a CPI-based expectations proxy before survey data, and quarterly federal spending repeated across months.
  • The variable set is limited to eight series chosen by the authors, and results depend on that choice (inferred).

Items marked (inferred) are our reading of the design, not statements by the authors.

Why it matters

The attribution shows that the dominant inflation driver changes over time, so a single static factor view can mislead. For practitioners, it links regime identification to asset-class behaviour in real terms. It also offers a method for monitoring which macro variables currently explain inflation pressure.

Questions about this paper

What was the main driver of inflation in February 2022 according to the paper?

Federal government spending, with 41.6% of relative importance, far above the other seven variables.

How does the paper define an inflation regime?

A regime is based on the change in inflation, meaning the latest one-year rate minus the prior three-year rate. A Hidden Markov Model sorts months into four regimes.

How did assets behave in rising and volatile inflation?

Stocks and bonds did worst in this regime in nominal and real terms. Cash offered only slight protection.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.2139/ssrn.4137861. We link to the paper and do not host the PDF.

Follow new quant research

Download MacroFinance market data as CSV or Excel, or pull it by API, and test the ideas in these papers yourself. New summaries are added as we review them.

Create a free account
Summaries are written by MacroFinance from the published papers, in our own words, for information only. They are not investment advice and may omit detail. Papers are not hosted here: each page links to the original at its publisher.