Home » Research Blog » Excess inflation and asset class returns

Excess inflation and asset class returns

Jupyter Notebook

Excess inflation means consumer price trends over and above the inflation target. In a credible inflation targeting regime, positive excess inflation skews the balance of risks of monetary policy towards tightening. An inflation shortfall tips the risk balance towards easing. Assuming that these shifting balances are not always fully priced by the market, excess inflation in a local currency area should negatively predict local rates market and equity market returns, and positively local-currency FX returns. Indeed, these hypotheses are strongly supported by empirical evidence for 10 developed markets since 2000. For fixed income and FX excess inflation has not just been a directional but also a relative cross-country trading signal. The deployment of excess inflation as a trading signal across asset classes has added notable economic value.

A Jupyter notebook for audit and replication of the research results can be downloaded here. The notebook operation requires access to J.P. Morgan DataQuery to download data from JPMaQS, a premium service of quantamental indicators. J.P. Morgan offers free trials for institutional clients.

This post ties in with this site’s summary of trading strategies based on macro trends.

Also, there is an academic research support program that sponsors data sets for relevant projects.

What is excess inflation?

For this post, excess inflation is defined as a real-time measured meaningful consumer price inflation trend over and above the formal or informal inflation target communicated by the public authorities of the currency area. Values can be positive or negative, indicating inflation excess and shortfall respectively.

In this post, we use as consumer inflation trends a seasonally and jump-adjusted change in core CPI, whereby “core” here means that volatile items, such as food and energy, are excluded, in accordance with local convention. Jump adjustment means that two types of outliers are adjusted for. First, large “spikes”, i.e. two subsequent large moves of the seasonal index in opposite directions are averaged. The causes for such spikes are often holiday shifts between two months, while the causes for jumps are mostly indirect taxes, administered prices, or FX turbulences. Second, large one-off jumps in the index are replaced by the local trend. The criteria for adjustment are statistical and based on pattern recognition. These are standard common-sense adjustments for backing out underlying inflationary pressure from headline reports.

For a meaningful analysis of the impact of economic trends on market returns, we use indicators of the J.P. Morgan Macrosynergy Quantamental System (“JPMaQS”). Quantamental indicators are real-time information states of the market and the public with respect to an economic concept and, hence, are suitable for testing relations with subsequent returns and backtesting related trading strategies.

The main quantamental metric used for this post is the real-time information state on core CPI, seasonally and jump-adjusted, % change of the last 6 months over the previous six months minus the official inflation target. Find documentation here. Most charts below are based on this series as a predictor or trading signal.

We also double-check all results by using alternative measures of excess inflation, such as

  • the same type of 6-month-over-6-month trend based on a headline CPI, and a consistent CPI, where the latter uses the same core definition across countries. Find documentation here.
  • conventional annual rates of headline, core, and consistent core CPI inflation, and
  • 1-2 year inflation expectations, which are based on a formulaic estimate explained here.

The measurement of excess inflation and all related analyses are applied to 10 developed market currency areas: Australia (AUD), Canada (CAD), Switzerland (CHF), the euro area (EUR), the United Kingdom (GBP), Japan (JPY), Norway (NOK). New Zealand (NZD), Sweden (SEK), and the United States (USD). The sample period for analyses is 2000 (January) to 2023 (February or March).

Why does excess inflation matter for market trends?

One of the principal functions of central banks in developed markets is the pursuit of price stability. The definition of price stability varies, but most banks abide by a commitment to keeping inflation close to their formal or informal target over the medium term. Hence, all other things equal, this means that they tend to set tighter monetary conditions if inflation threatens to exceed the target and easier conditions if inflation seems to fall short of the target.

Since future inflation is not easy to predict and central banks are regularly under scrutiny, meaningful recent inflation trends play a great role in shaping inflation target risk. Everything else equal, a recent inflation trend above the target increases the chances of a target miss on the high side. A recent inflation trend below the target increases the chances of a target miss on the low side. The balance of risk plausibly impacts subsequent central banks communication and policy actions. To the extent that this risk is not efficiently priced in the market, its direction is a valid predictor of the direction of verbal and policy interventions.

Monetary policy interventions should have pervasive effects across macro asset classes. By itself, a policy shift towards tightening should be negative for the performance of local-currency bonds (or interest rate swap fixed receivers) and equity markets. Moreover, if the shift towards tightening is relative to another currency area it should support the appreciation of the currency.

Inflation as a predictor of fixed income returns

Directional return prediction

First, we estimate the predictive power of excess inflation for two-year interest rate swap returns. The two-year tenor in the rates market should be most closely aligned with the monetary policy outlook. In particular, we investigate the relationship between the excess core inflation trend as available at the end of the month and the swap receiver return performance during the following month. Receiver positions have been calibrated to a 10% predicted volatility target in order to make return variation comparable across countries. We would expect a negative relation, i.e. high inflation giving rise to low or negative returns.

Indeed, the empirical evidence clearly confirms the negative correlation between excess inflation at the end of a month and next month’s returns for the overall panel. Moreover, to gauge the significance of the relationship we use the Macrosynergy panel test based on a methodology explained here. Simply looking at the significance of the correlation for the pooled data set (which is near 100%) can be misleading. This is because features and targets across currency areas are not independent and subject to common global market factors. Simply stacking data means “pseudo-replication” and surely overstates significance. Instead, the Macrosynergy panel tests check significance through panel regression models with period-specific random effects, adjusting the predictive regression for common (global) influences. Put simply, the method automatically accounts for the similarity of experiences across markets when assessing the significance.

Yet even accounting for communal global effects, the Macrosynergy panel test suggests that the probability of negative correlation based on the sample is 99.8%.


The negative relation has prevailed for 9 out of 10 countries since 2000. The only exception has been Japan, which is plausible since Japanese policy rates had been trapped near the zero lower bounds for most of the sample period and persistent deflation constrained the central bank’s influence on financial conditions. Switzerland faced similar issues. This suggests that a more advanced trading signal should also consider swap yield levels.

Other statistics confirm the negative predictive power of excess inflation for IRS returns. Thus, balanced accuracy, the average ratio of correctly predicted positive and negative returns, has been 52.4%. Moreover, all other alternative metrics of inflation trends posted a negative correlation with subsequent receiver returns and balanced accuracy ratios above 50%.

The excess inflation signal for IRS returns has been somewhat seasonal. Balanced accuracy of the prediction has been above 50% in two-thirds of all years since 2000. This is intuitive. Excess inflation can only be a meaningful signal for rates if it is meaningful in size or different across countries.

We calculate naïve PnLs based on the following assumptions. First, positions are taken based on panel z-scores of the excess inflation trends. The z-scores are winsorized at 2 standard deviations to mitigate data outliers and to avoid excessive risk-taking in any single market or period. Second, positions are rebalanced monthly with a one-day slippage for trading. And third, the long-term volatility of the PnL for positions across all currency areas has been set to 10%. These are standard procedures that we have been using in previously published proof-of-concept analyses. Note that this PnL is called “naïve” because it does not consider transaction costs and realistic risk management rules.

Value generation of the directional excess inflation signal has been respectable, albeit seasonal. Compared to a long-only portfolio, the excess inflation-based strategy underperformed in the 2000s but outperformed since 2000. The long-term naïve Sharpe ratio (prior to transaction costs) has been 0.79 with a slightly negative correlation to the S&P500 returns.

Relative cross-country return prediction

Relative here refers to the difference between excess inflation and return in one country and the basket of all 10 developed markets. We use relative excess inflation to predict relative vol-targeted 2-year IRS returns. We would expect countries with higher excess inflation to underperform in terms of their subsequent swap returns over the next month.

This underperformance hypothesis is borne out by empirical evidence. The relative relation is significant with the panel test-based probability of a positive correlation of 99.8%. The correlation of the signal with next month’s relative return has been positive for 8 out of 10 countries, with Australia and Switzerland being the exceptions.

Balanced accuracy of relative IRS return prediction is lower than for the directional case, but still 51% for the main core trend indicator and between 51% and 52.5% for all alternative excess inflation metrics. This is a frequent feature of macro trend factors: their relative accuracy is lower than directional accuracy, because of cross-country differences in the measurement and the importance of factors. Also, the relative excess inflation predictor has been quite seasonal with above-50% accuracy in only about half the years since 2000 and a positive correlation in 63% of the years.

Notwithstanding its seasonality, a trading signal based on relative excess core inflation would have produced significant value since 2000 with a naïve PnL producing a long-term Sharpe ratio of over 0.6 before transaction costs and near zero equity market correlation. Across a range of similar excess inflation signals Sharpe ratios would have been in a range of 0.38 and 0.66.

Inflation as a predictor of FX forward returns

Directional return prediction

If central banks across developed markets are assumed to be similarly committed to their inflation targets, monetary policy in areas with higher excess inflation should have a greater bias towards tightening and their currencies should tend to outperform. Put simply, relative excess inflation should predict currency outperformance. We test this proposition for eight “smaller” developed countries: Australia, Canada, Switzerland, the UK, Japan, Norway, New Zealand, and Sweden. For all these currencies we measure 1-month FX forward returns against their natural benchmarks, i.e. either the U.S. dollar (AUD, CAD, JPY, NZD), the euro (CHF, NOK, SEK), or both (GBP).

As a trading signal, we use the same core inflation metric as above, except that we focus on a consistent core CPI definition that excludes the same food and energy prices for all countries. This is because, for the FX market, it is the excess core inflation differential between the base and the main reference currency that matters, and fair comparison calls for similarity in concept. Our targets are 1-month FX forward returns, on positions that are based on 10% vol targets (similar to the IRS case).

As expected, a positive relative excess inflation has been associated with higher or positive FX returns. The Macrosynergy panel test of predictive relation suggests that the relationship has been highly significant with a probability 99.8%.

Positive correlation prevailed for 7 out of 8 currencies, with the UK being the only exception. The monthly balanced accuracy of relative excess inflation-based predictions of monthly FX returns has been more than 53% for the panel. Positive balanced accuracy prevailed in 70% of all years. Across alternative versions of the relative excess inflation signal accuracy (balanced) has been between 50.3% and 53.1%.

Judging from the naïve PnL simulation, value generation has been reasonably consistent across time. The Sharpe ratio of the strategy since 2000 has been 0.42 with a 12% correlation with the S&P500. Across the range of similar excess inflation signals Sharpe ratios have been in a range of 0.17 to 0.64. In fact, headline inflation metrics would have produced higher value than core inflation signals.

Relative return prediction

We can also test a “double relative” excess inflation signal for FX trading. The hypothesis is that among the eight smaller DM currencies those with higher relative excess inflation versus their base currencies outperform those with lower or negative relative excess inflation.

As in the previous cases, the positive correlation between double-relative excess inflation and relative FX returns is confirmed by the evidence. The Macrosynergy panel test assigns a 99.9% probability to the significance of the predictive relation over the past 23 years since 2000. The positive relation prevails for 7 of 8 small developed market currencies.

The balanced accuracy of monthly return predictions has been 53.8% and was above 50% in three-quarters of all years since 2000. Across the range of similar signals, balanced accuracy has been between 52.4% and 54.7%. Across time, accuracy was on the low side during the low inflation period between 2012 and 2020.

Naïve PnL generation has recorded a long-term Sharpe ratio of 0.5 with a 22% S&P500 correlation but also has been extremely uneven across time. Most trading profits were produced from 2000 to 2007 and strategy returns flatlined from 2014. Across similar “double relative” excess inflation signals, naïve PnL Sharpe ratios have been between 0.35 and 0.67.

A “double relative” signal typically needs more careful calibration to make it comparable. Since inflation cycles are similar across countries, the direction and size of positions depend on the differences in measure and impact. For example, excess inflation has less impact on monetary policy if the central bank is more dovish.

Inflation as a predictor of equity returns

Directional return prediction

If the risk of monetary policy is tilted towards tightening (and not fully priced) it translates into upside risks for the stochastic discount factor of the dividend discount model. This implies downside risk for equity returns. From this, we derive the hypothesis that excess inflation predicts subsequent equity returns negatively. We investigate this negative relation for eight liquid developed market country equity index futures. The targets are volatility-targeted futures returns, analogously to the IRS and FX cases.

As for interest rate swaps, the negative relation is borne out by the empirical evidence since 2000. However, in the wide panel of developed equity markets, the probability of a significant relationship is low (57%), according to the Macrosynergy panel test. This seems odd at first glance but reflects an essential structural feature of equity markets: monthly returns across countries are dominated by global factors and the U.S. index (S&P500) plays a far bigger role in capturing these and the smaller country indices. For the data set, this has two consequences:

  • Monthly equity returns are highly correlated across developed markets and the consideration of non-U.S. data does not add much independent local return variation. This dependence is picked up by the panel test. Indeed, the stronger these global effects, the greater the importance of deviations from the period-mean in the panel regression that underlies the test. This “relative” effect is evidently much weaker than the outright directional effect of inflation.
  • The non-U.S. excess inflation rates have a lot less influence that the U.S. data because they have only modest importance of global equity trends. A panel analysis, however, checks if local inflation is a significant predictor for local returns across all countries. And this general hypothesis, which treats all countries’ inflation rates as equally important, fails.

By contrast, if a panel test is performed on the U.S. alone the negative correlation is found to be significant with a probability of 96%. Thus, even though the sample is much smaller, the hypothesis that U.S. inflation is a major concern of equity performance is more sensible and backed by the data.

Across countries, the negative relation prevails in 7 of 8 markets, with the U.K. being the only exception. The reliance on U.S. inflation does not prevent the accuracy of panel predictions for subsequent returns from being fairly high, with balanced accuracy of over 54% and simple accuracy near 59%.

Value generation has been strong but due partly to the long bias of the inflation signal (77% since 2000). The long-term Sharpe ratio has been 0.75 with a nearly 30% correlation with the S&P500. As for the panel relation analysis, history has shown the inflation signal to be valuable, but not pervasively enough to attest to high significance.

Relative return prediction

Relative excess inflation rates failed to show significant predictive power and investment value, in line with the reasoning that U.S. inflation is decisive for equity performance rather than local trends.


Related articles