On Inflation

How inflation drives asset prices and how to predict inflation

By: Igor Vasilachi

Inflation is the economy’s Jay Gatsby: talked about by many, understood by few. Inflation can have big impacts on markets, yet predicting inflation is notoriously difficult and most investors’ portfolios are not optimized for inflationary environments. Since the 1980s when inflation peaked in the United States, research on this important economic topic has all but completely dried up.  Below is a Google Ngram chart of the frequency of appearance of the word “inflation” in published sources.

Figure 1: Frequency of Appearance of the Word “Inflation” in Published Sources

Source: Google NGram

We have spent months trying to understand how changes in inflation impact asset prices and how to better predict it, in the context of investing through crises and the economic cycle more broadly. Our initial findings were unsurprising: inflation is hard to understand and even harder to predict. And there is no perfect way to protect your portfolio against inflation.

Today, we are sharing what we have learned so far about inflation: what works in inflationary environments, which indicators best predict inflation, and, most importantly, how this can generate returns in simple trading strategies.

Profiting in Inflationary Environments
TIPS seem to provide a simple answer to how to protect a portfolio during periods of rising inflation. Contractually linked to inflation, this is the closest investors can get to an “inflation insurance.” And, just as buyers of fire insurance don’t make a living as a result of a burning house, TIPS investors can’t expect to generate abnormal returns in inflationary environments, but to merely be compensated for it. This held true in practice, as short duration TIPS have averaged 0.1% in annualized real returns in periods of rising inflation. This was more than the -3.7% that 10-Year US Treasuries returned in such periods and slightly ahead of the S&P 500, which generated -0.4%. But this is hardly an impressive number, especially considering the kind of exposure to TIPS that is required to make a difference in a broader portfolio.

To profit from inflation, investors might instead consider buying assets that have a strong reaction to inflation, rising when inflation rises and falling when inflation falls. Borrowing a term widely used by others, we can think of this as “inflation beta.” We regress asset class returns against changes in the level of inflation to calculate inflation beta: a beta of 1.5 would suggest that a 1% increase in inflation would lead to a 1.5% increase in asset returns for the given period. Below, we show the inflation betas regressed on monthly returns for a subset of assets. In addition, we show real returns over the full period of our analysis compared to periods of rising inflation.

Figure 2: Inflation Beta and Average Annualized Real Returns by Asset, 1970–2020

Source: Bloomberg, FRED, Verdad. * REIT data available since 1990 (Dow Jones REIT Total Return Index); ** Short duration TIPS data available since 2005 (Barclays 1-5Y TIPS Total Return Index); *** Long duration TIPS data available since 2000 (Barclays 10Y+ TIPS Total Return Index).

Commodities and gold have the highest inflation beta in regressions on monthly data as shown above – and appear to be the only two assets whose real returns in periods of rising inflation were higher than over the full period. Conversely, 10Y treasuries have the most negative inflation beta. For investors looking to introduce assets to their portfolio that will do well when inflation is rising – and putting pressure on the stocks and bonds that make up most investors’ portfolios – gold and commodities seem to offer the most promise. However, commodities and gold are excessively volatile, have historically generated considerable drawdowns and are less correlated with inflation (only 25% of their long-term returns variability is explained by changes in inflation).

It’s important to note that the inflation betas presented in the table above are measured on monthly returns. In the case of TIPS, the above analysis does not fully capture the contract that compensates a TIPS holder for realized inflation upon maturity of these inflation-protected bonds. Since this inflation contract is tied to the principal of TIPS, which is paid out at maturity, we should also evaluate the inflation beta of TIPS on a duration-matched basis. For example, with TIPS that mature in 5 years, the inflation beta could be measured over 5-year horizons rather than monthly horizons. In a duration-matched analysis, we found that 5-year TIPS have an inflation beta of 2, which is similar to gold, but with a much higher correlation to inflation (more than 62% of the return variability of 5-year TIPS is explained by changes in inflation). In practice, some examples of duration-matching with TIPS include 529 college savings plans that hold TIPS with maturities over the four years in the future when a beneficiary is expected to attend college, and retirement plans that hold TIPS with maturities over an investor’s expected retirement years.

Predicting Inflation
There are two popular methodologies for predicting inflation. The first is to rely on the Survey of Professional Forecasters, the oldest US survey of macroeconomic forecasts by economists, and the second is to rely on the breakeven rate, the spread between the US Treasury yield and the yield for TIPS, as a measure of market expectations. Below we show the mean and standard deviation of realized versus expected inflation since 2003, the longest overlapping period between the three series.

Figure 3: Mean and Standard Deviation, %, 2003–2020

Source: FRED, Philadelphia Federal Reserve website. * 5-year breakeven rate, the shortest available horizon.

These two methodologies have tended to slightly underestimate inflation but have been quite close to the actual results. However, the forecasts have been far less volatile than actual inflation. And research from Bridgewater and Hedgeye suggests that markets react more to changes in the rate of inflation than the rate of inflation itself. They argue that instead of worrying about high or low inflation, investors should worry about rising or falling inflation – and expert forecasts and breakeven inflation have been good at the former but terrible at the latter.

We found two sets of indicators that are better at capturing the short-term volatility of inflation: business cycle indicators and commodity returns.

Business cycle indicators can provide meaningful information about future inflation, as credit expansion has tended to be inflationary and credit contraction has tended to be deflationary. The high-yield spread and the slope of the yield curve, two of the most widely known business cycle indicators in both practitioner and academic circles, signal the expansion or contraction of private and public credit and thus can be helpful in forecasting inflation.

Commodity prices can also help predict the direction of future inflation. Despite the transition to services, manufacturing is still a core component of our economy and commodities often represent a large portion of manufacturing costs, either as raw materials or sources of energy. The highly liquid commodities market ensures that any price increases are passed along easily to manufacturers. At the same time, consumer prices are sticky, which creates a lag between when additional costs are incurred by manufacturers and the moment these additional costs are passed through to consumers, thus translating into higher prices.

These indicators all provide meaningful information about future inflation and combining them can generate a stronger signal. Below we show the three-month forward excess accuracy of predicting the inflation direction for each of the above indicators, as well as expert forecasts and market expectations priced into TIPS. By excess accuracy, we mean the difference between the predicted accuracy and the historical probability of positive or negative inflation change.

Figure 4: 3M Forward Excess Accuracy of Predicted Inflation Direction by Indicator

Source: Verdad. *10-year minus 1-year treasury yields. **Junk- minus investment-grade bond yields.

Business cycle and commodity price indicators tend to predict the short-term direction of inflation better than experts or the market. Moreover, combining these indicators into a synthetic signal can further increase excess accuracy. It is worth noting, however, that these accuracy rates are still quite weak. Think of a 12% increase in excess accuracy as akin to moving from 50% accuracy to 62% accuracy – helpful, but still wrong a large percentage of the time. These signals are helpful but not dispositive.

Testing these Insights
The key test of our work on how different assets respond to inflation and how to predict inflation changes is whether these indicators can generate excess returns. We sought to test if our indicators can help generate excess returns through a simple two-asset strategy:

1. Long inflation-positive assets when indicators predict rising inflation;

2. Long 10-Year US Treasuries when indicators predict falling inflation.

We tested two alternatives of inflation-positive assets. We tested TIPS, given they protect investors against inflation contractually (i.e., the principal rises if inflation rises, and vice versa). We also tested gold, given its high inflation beta, lower volatility compared to the broader commodities class, and attractive historical performance in inflationary environments.

Below we show the annualized total period returns, information ratio, and max drawdowns for the TIPS strategy (i.e., long TIPS when indicators predict inflation) and compare this performance with a buy-and-hold of a 50/50 portfolio comprised of TIPS and 10Y treasuries.

Figure 5: Performance by TIPS1 Strategy and Indicator, 2005–2020

Source: Bloomberg, FRED, Verdad. 1) Barclays 1-5Y TIPS Total Return Index (available since 2005) used as proxy for TIPS; 2) 50% 10Y US Treasuries Total Return and 50% Barclays 1-5Y TIPS Total Return Index.

We found that most indicators, except for the slope of the yield curve, would have helped generate returns in excess of the 50/50 portfolio buy-and-hold strategy. However, the increased returns came at the cost of higher volatility and drawdowns and a reduced Sharpe Ratio, meaning investors willing to use leverage might prefer the 50/50 approach.

Replacing TIPS with gold as the inflation-positive asset would have doubled these returns, in line with gold’s higher inflation beta albeit at a cost to drawdowns and information ratio, as shown below.

Figure 6: Performance by Gold Strategy and Indicator, 2005–2020

Source: Bloomberg, FRED, Verdad. 1) 50% 10Y US Treasuries Total Return and 50% Gold Spot.

Similarly, most indicators, except for the slope of the yield curve, would have helped generate returns in excess of the 50/50 portfolio buy-and-hold strategy. The margin of outperformance on this strategy is greater than with TIPS, and this strategy also results in an information ratio comparable to a 50/50 strategy.

Conclusion
In summary, inflation is a poorly understood and difficult to predict economic indicator. Despite the recent inflation stability, experts and investors alike seem to underestimate its long-term level while consistently failing to capture its short-term changes. We found that using business cycle indicators and commodity prices can lead to more accurate forecasts of the direction of inflation. There is no perfect answer when investors are facing the trade-off between hedging and generating returns when inflation is rising. However, allocating to inflation-positive assets such as TIPS or gold when inflation is predicted to rise and allocating to 10-Year US Treasuries when inflation is predicted to fall might offer an alternative to holding a balanced 50/50 portfolio.