“big forces to worry about: growth and inflation. Each could either be rising or falling, so I saw that by finding four different investment strategies—each one of which would do well in a particular environment (rising growth with rising inflation, rising growth with falling inflation, and so on)—I could construct an asset-allocation mix that was balanced to do well over time while being protected against unacceptable losses. Since that strategy would never change, practically anyone could implement it.”
― Ray Dalio, Principles: Life and Work

Note: a big thank you to the always excellent Allocate Smartly for providing sector data for this post.

In the Business Cycle Sector Timing Model, I presented a simple sector rotation model that used the S&P500 as a proxy for expected economic growth and the business cycle to rotate between early and late stage economically sensitive sectors when the cycle is rising or expanding, and to defensive sectors when growth is falling or in a recession. Putting theory and backtests aside, there is empirical justification for using trends or momentum in the S&P500 for expected growth. Nicolas Rabener demonstrated that the stock market is a valid forward-looking economic indicator using long-term data:

“We extended our analysis back to 1900 using annual data from MacroHistory Lab. Since the stock market is forward-looking and tends to anticipate economic news flows, we instituted a one year lag. So for 2000, we compared that year’s GDP numbers with the performance of the S&P 500 in 1999……All of which suggests the S&P 500 was a good proxy for the US economy for much of the last 120 years.”. (Rabener, CFA Institute)

While this single factor model is useful it can be extended to a more robust two-factor model in order to account for inflation. The traditional Growth/Inflation matrix using real GDP and CPI was popularized by Ray Dalio from Bridgewater Capital to create an “All-Weather Portfolio”, and also referenced by Harry Browne (Permanent Portfolio), Geoffrey Moore at NBER, and Sam Stovall from Standard and Poor’s. The trends and/or momentum in these two variables using a two-state model (Up/Down or Rising/Falling) significantly explain cross-asset class returns (Brazen Capital). The Growth and Inflation Matrix is shown below along with four separate macroeconomic regimes:

“Goldilocks” refers to an economic regime characterized by low inflation and strong economic growth, with classic examples including the tech boom of the late 1990s and much of the 2010s. Reflation, on the other hand, is typically marked by an overheating economy with both high inflation and high growth, as seen in 2006-2007 during the housing boom when oil prices and other commodity prices began to rise significantly, and in 2009-2010 after the Global Financial Crisis in 2008. Stagflation, or stagnation, is characterized by high inflation and low growth, with the 1970s serving as a prime example. Deflation or disinflation, which is less common, involves falling growth and declining prices, with the most prominent examples being the Great Depression and the 2008 Global Financial Crisis. A good paper by Flexible Plan Investments called “The Role of Gold in Investment Portfolios” (I was a co-author in earlier versions) shows the regime performance for various asset classes using this Growth Inflation Matrix along with several other regime scenarios. Another good summary of using the Growth Inflation Matrix to invest in different asset classes was written by Resolve Asset Management.

The Role of Inflation

To understand how these macroeconomic regimes affect various sectors and the broader economy, it is essential to first examine the role of inflation, as it is a key driver behind many of these economic shifts. Inflation plays a central role in the financial system. Without it, there would be much less need to put your hard-earned savings at risk. Inflation affects everything from monetary policy to consumer purchasing power and the returns of all asset classes including stocks, bonds, commodities and currencies. The late Ronald Reagan once said, “Inflation is as violent as a mugger, as frightening as an armed robber, and as deadly as a hit man.” While this is true of hyperinflation, as seen in Weimar Germany, Venezuela, and Zimbabwe, or during the stagflation of the 1970s, moderate inflation still plays a crucial role in a healthy economy. When kept in check, it encourages spending and investment, and fuels economic growth even as it gradually erodes purchasing power.

The stock market tends to perform best when inflation is moderate, stable, and predictable. Equity returns tend to be negatively correlated with unexpected inflation as it increases the discount rate on future returns. At the business level, inflation increases costs and erodes corporate earnings and purchasing power. Uncertainty about future costs makes long-term business planning difficult. On the other hand, deflation—seen during the Great Depression and Japan’s stagnation in the 1990s—can be far more concerning, once triggered it creates a self-reinforcing downturn that is difficult for policymakers to reverse.

The chart below from O’Shaughnessy Asset Management below shows average equity market returns in different inflation regimes.

O’SHAUGHNESSY ASSET MANAGEMENT, L.L.C.

Clearly moderate inflation is the best regime for the equity market, while really high inflation is the worst. In addition, really low inflation is also undesirable as this is correlated with a stagnant or deflating economy. If this chart included the Great Depression, presumably this low inflation quintile might show even worse overall performance.

So how would inflation be used for sector rotation? Sectors are well-known to have distinct characteristics based on the long-term nature of their businesses, market dynamics, regulatory environment, and other factors. These unique traits can lead to predictable patterns in how sectors perform in both the short and long-term in response to changes in macroeconomic variables like growth and inflation. For example, Morningstar divides the major sectors into three separate super sectors – Defensive, Cyclical and Sensitive- based on their respective macro sensitivity. However this categorization is based primarily on the business cycle and the growth dimension but it fails to capture the inflation dimension directly.

Inflation can impact sectors in distinct ways, with some proving more resilient than others. Classic fundamental analysis of their business models and basic economics can reveal how they respond to inflation. In the short term, “bond-like” sectors with high operating leverage, such as Consumer Staples, Healthcare, and Utilities are more immediately impacted by inflation because they face delays in passing on price increases, which can lead to short-term margin compression. Their variable costs, such as energy, food, and labor, are highly exposed to inflation. Since these sectors have high dividend yields and low growth, they are also more sensitive to rising interest rates. However, in the long run, they tend to benefit from pricing power due to contracts that allow price hikes in line with rising prices and their role in providing essential goods and services. This makes the defensive sectors more resilient to sustained inflation. At the opposite end, consumers cut back on discretionary purchases such as automobiles and electronics during sustained inflation as wages fail to keep up with rising prices. A lack of pricing power, and falling demand among other factors can cause Consumer Discretionary and Technology companies to perform very poorly.

The chart below from Investors.com and Schroders Investment Management shows the historical performance of different stock sectors in sample (measuring contemporaneous returns during inflation) from 1973-2021 in excess of CPI when inflation is high and rising:

source: Schroders and Investors.com

As expected, Energy is a standout sector during rising inflation along with Precious Metals and Real Estate. Less intuitive is the below-average performance of the Materials sector, which can occur when energy costs rise faster than revenues that can be tied to other commodities in the long term. Energy is a key input for materials production, and if energy prices increase disproportionately to the prices of the materials themselves, it can lead to margin compression and weaker sector performance. It is also interesting to note that Utilities performs the best during inflation of the defensive sectors but this long-term result masks the sector’s poor performance during Stagflation in the 1970s. Besides high interest rates, what has changed in the Utility sector over time is that companies now have more flexibility to raise prices post deregulation and their power sources are less reliant on oil and gas given the push towards clean energy. Another predictable result we see in the table is that long duration/high growth assets like Technology and Consumer Discretionary are at the bottom of the list. These sectors suffer from a lack of pricing power in the long-term and falling demand as well as being sensitive to interest rates that can rise to combat inflation.

The real question is how do we measure inflation or better yet how do we find a leading indicator for inflation?

The problem with using CPI that not only is it slow and backward-looking but itis also unreliable due to restatements, methodology changes, and weighting issues. It is also a monthly indicator and cannot be used on a daily basis. This is why traders and economists often prefer market-based inflation measures (like breakevens or commodity trends) for real-time insights. The chart below from State Street- the issuer of Sector SPYDERs- shows the different primary sector’s forward-looking or expected inflation betas over the 10-year period from 2014-2024.

Source: State Street Advisors

What this table is showing is whether certain sectors have a positive or negative or neutral relationship with inflation expectations as proxied by breakeven rates. When inflation expectations are rising Energy, Industrials, Financials and Materials tend to rise. Real Estate, Consumer Discretionary and Technology are relatively neutral. Defensive Sectors such as Utilities, Health Care and Staples tend to fall as inflation expectations rise. The unusual disparity between long-term performance of sectors during rising inflation and the initial response to expected inflation (Financials and Materials for example) can be explained in the table below:

Constructing an Expected Inflation Indicator

Using this information about how sectors react to inflation expectations we can create an expected inflation indicator by tracking trends in various sectors. There is a benefit to this approach vs using breakeven rates-equity markets can potentially react faster than TIPS breakevens since equity markets are much larger and more liquid and hence can easily absorb changes in expectations. In contrast breakevens can be more noisy because they are more affected by shifts in liquidity, monetary policy, and stale pricing. This expected inflation indicator can be superior to CPI due to forward-looking pricing, while CPI is lagged and subject to revisions. We can create this indicator by first constructing portfolios with positive and negative expected inflation betas. The relative strength or trend between these portfolios acts as a market-implied inflation signal, potentially leading traditional measures like CPI and breakevens in real-time.

The goal here is to employ a simple approach that generally captures trends in expected inflation that can easily be used in trading systems. The relative strength between two portfolios is an easy way to accomplish this goal. To do that we need to create portfolios with positive and negative betas to expected inflation. To do that we will reference the previous table, and create simple portfolios with common sense rather than optimization. For the positive expected inflation beta portfolio we will overweight Energy which has a high positive expected inflation beta and equally weight Industrials, Financials and Materials (1/6 in each) which all have positive and highly similar expected inflation betas.

Next we will create a negative expected inflation beta portfolio by equally weighting the defensive sectors (Communications was omitted due to a lack of long-term data) since their betas are all negative and fairly similar.

To create this relative time series we will simply divide the Positive by the Negative Expected Inflation Beta Portfolio. This simple ratio-based indicator is shown below:

In this chart using the sector implied expected inflation indicator we can clearly see that the tech boom in the 1990’s was a stable inflation regime which is best for equities historically. After the bear market in the early 2000s, the indicator begins to climb during the bull market that follows and peaks in late 2007 during the Housing Boom before tanking during the Credit Crisis in 2008. In 2009 to 2010 the indicator climbs substantially as reflation follows the previous deflationary bear market. From late 2010 onward the indicator falls substantially and bottoms out during the COVID bear market in 2020 before rallying until the end of 2024 as the market reflated. All of these movements align with what we would expect to see given the major changes that have occured with inflation over time. Since this measure is constructed using data from 2014 onwards, it is interesting that it generalizes well going backward to 1990. This is because sectors have a fairly stable response to inflation expectations given their business models.

To test how well this tracks performance during periods with high CPI we can backtest how well different sectors perform when sector implied expected inflation is above the median. We can then then compare how well this aligns with long-term in-sample performance during inflation going back to 1973. For sectors we will use the Sector SPYDRs from State Street and the Vanguard REIT ETF (VNQ).

Next we will backtest how each sector performed when the indicator was above the 200-day median going back to 1990. The assumption is that even a lagged value should be a leading indicator of CPI and inflation. If that is the case then we should see a strong agreement between backtested sector rankings and long-term sector performance rankings when CPI was high.

The rank agreement between historical performance of sectors during high inflation and actual performance during predicted high inflation is very high correlation of .92 which is quite remarkable. This means that using the sector implied expected inflation is a good real-time indicator for forecasting trends in inflation. The biggest relative disparity is for the Materials sector which performed better during inflation in the past then the testing period. Deeper investigation reveals that as the composition of the sector has shifted over time from precious metals and mining toward industrial chemicals, gases, and specialty materials, reducing the weight of traditional miners due to drift as well as GICs changes making precious metals a smaller sub-industry. Another disparity was that Utilities performed better during expected inflation than in the past. Perhaps the simplest explanation is that the missing third variable here is interest rates. The rate on a 10-year bond was over 8% at the beginning of the testing period vs 4% at the end of the testing period. Since Utilities are highly interest sensitive this was a positive tailwind which exaggerated results. As we mentioned earlier there were also structural changes in the industry such as deregulation which allows them to raise prices in accordance with inflation while simultaneously having less exposure to high input costs from oil and gas and more clean energy. All other ranks were either the same or off by one.

Now lets revisit the Growth and Inflation Matrix and see how each sector performs during periods where expected economic growth is high or low vs when expected inflation is high or low. To do that we will use the 200-day simple moving average for the S&P500 (SPY) to capture the expected growth dimension (above trend/sma is UP, Growth is UP, below is DOWN), and as before we will use the 200-day median of the expected inflation time series so that the periodicity of each indicator is aligned for the sake of consistency. First let’s look at broad stock market performance and median sector performance in each macro regime:

The broad stock market using the S&P500 as a proxy performs best in the “Goldilocks” regime which is characterized by expected growth being above trend (rising) and expected inflation being below trend (falling). However when we look at the median sector, performance is best in the Reflation regime which is characterized by expected growth and inflation both being above trend. What this means is that there is likely a larger disparity in sector performance in the Goldilocks regime where there are some sectors that do much better than others. In contrast, during Reflation most sectors perform well as this regime typically occurs coming out of corrections or bear markets or when the market overheats at the end of a bull market. In Deflation which is characterized by both expected growth and inflation being below trend, the broad stock market and median sector perform poorly and almost exactly the same. This indicates that there is less disparity in sector performance during this regime. This result is predictable as deflation or disinflationary periods tend to occur during bear markets or corrections. However the worst regime was Stagflation where expected growth was below trend while expected inflation was above trend. This result was surprising since Deflation is historically the worst regime for equities but the difference in performance is small. Bear in mind that we are using expected growth and inflation and expectations can change frequently. This means that expected deflation might contain multiple false signals vs looking at in-sample periods where we know exactly what happened historically.

Overall we can conclude that the expected growth component is the most important dimension for market and sector performance- equities perform better in bull vs bear markets regardless of inflation expectations. The picture below shows how expected growth and inflation macro regimes have evolved over time:

Notice that most of the 1990s and 2010s were Goldilocks conditions. The major Bear Markets such as 2001-2002 and 2008 and 2022 showed a combination of Deflation and Stagflation. Typically Stagflation deteriorated into Deflation in Bear markets . Most corrections in contrast were associated with Deflation. This helps to reconcile why Stagflation showed lower returns than Deflation.

Now lets get more granular and take a look at sector performance by macro regime:

As we might expect, the Technology sector performs best during Goldilocks conditions followed by Consumer Discretionary. Both are growth oriented sectors that perform well when consumers have more purchasing power During Relationary periods as expected the Energy sector is the best performer followed by Real Estate, this again aligns with expectations. During Stagflationary periods the defensive but less inflation-sensitive Health care and Utilities perform the best, this also makes logical sense. Curiously during Deflationary periods Materials and Consumer Discretionary perform the best with Consumer Staples showing solid performance as well. Why would Materials perform the best? Perhaps it is because input costs from inflation will fall faster than their revenues which are tied to other commodities and have longer-term contracts. This may also capture a mean-reversion effect as Materials is also a good early cycle performer. As for Consumer Discretionary this might also be due to a mean-reversion effect where the market expects a return to consumer spending (Consumer Discretionary is historically a very good early cycle performer). In either case the excess performance of Materials and Consumer Discretionary are not signficantly different from Consumer Staple performance which historically has been the best sector to hold in market downturns and makes more logical sense as a safe haven.

The table below shows a more detailed breakdown of sector performance by regimes and also by sub-regimes along with S&P500 performance for comparison:

As we might expect, the Technology sector is the most sensitive to growth expectations and performs the best and the worst when expected growth is up or down. Materials surprisingly shows the least sensitivity to growth expectations, perhaps reflecting the shift in composition to becoming more like a defensive industrial sector. As expected, the Consumer Discretionary is the most sensitive to inflation expectations and performs the best and the worst when expected inflation is up or down. The Technology is not far behind and is also highly sensitive to expected inflation. Consumer Staples shows the least sensitivity overall to expected inflation given its stable demand and pricing power. When expected inflation is up we see that Energy and Real Estate both do very well. But the standout performer when expected inflation is up is Utilities which is even more unusual because it is in the Negative Portfolio Beta Portfolio- meaning it is likely falling relative to the Positive Portfolio Beta for which Energy is the largest component. Many of the reasons for this we have already discussed. That said historically Utilities have done fairly well during inflation relative to other sectors (4th best historically) but this outperformance is likely anomalous.

Creating a Simple Growth and Inflation Sector Timing Model

Now we are going to create a simple strategy to capture performance across regimes. Using a combination of empirical data and economic logic I chose the following sectors for each regime:

The first two regimes (the green regimes) are the most obvious because they align with both data and logic. They are the major return generators in the strategy: Technology is an overwhelming statistical favorite duing Goldilocks conditions and logically it makes sense to perform best in that regime. Energy is clearly the odds on favorite when inflation is expected to rise and especially when the market is rising during Reflation. The defensive regimes (red and orange) are meant to reduce risk and preserve capital as returns are difficult to generate when the broad market or expected growth is down. The choices here are less critical because there is much less disparity in sector performance. Holding Consumer Staples in both regimes (Stagflation and Deflation) is a simple solution and doesn’t detract much from model performance historically. During Deflation from a logic perspective it is most prudent to choose Consumer Staples (3rd best) even if it wasn’t the best performer in that regime. We know from the previous post on business cycles that it is the best performer during either slowdowns or recessions. Furthermore, the sector shows stable demand for its products with good pricing power and has the lowest sensitivity to inflation. For these reasons it is a much easier choice than Materials or Consumer Discretionary which performed better but do not have the same defensive characterisitics nor do they have strong logic for inclusion. In Stagflation we will choose the Healthcare sector because it is historically a top performer in that regime and also during expected inflation. From a logic perspective, Healthcare offers essential products and services that are often paid for by insurers. Furthermore the Healthcare sector has historically the second best overall performer in slowdowns and recessions and has lower interest rate sensitivity than Utilities which was also a top performer. A long-term study by MSCI using global sectors from 1975-2021 concluded: “Among sectors, health care has done favorably in stagflation.”

Now that we have our choices for each regime we can backtest the historical performance of using this simple sector timing model:

The strategy substantially outperforms the S&P500 with slightly higher volatility but lower drawdown risk. The returns of the strategy are much more asymmetric- higher relatively in bull markets and lower relatively in bear markets. Surprisingly this simple strategy isn’t that sensitive to parameters (it isn’t optimal either), but for a production strategy more work still needs to be done to make it robust and to reduce taxes and trading costs.

Overall the strategy produces very high returns especially with equity sectors that have historically been quite noisy. After employing a two factor growth/inflation matrix and seeing the sector performance disparity by regime, it is easier now to understand why simple relative strength strategies do not perform very well with sectors. Macroeconomic regimes change in a nonlinear way- they do not occur in a set or highly predictable order and they can change very quickly. Using a static time-based lookback will gather data that often straddle multiple regimes at once and can only perform well if a regime lasts a long time.

There is a good case for using the Growth and Inflation Sector Timing Model even if it has a higher risk than most are comfortable with: the strategy has the advantage of always being in the market, and therefore faces less timing risk than a tactical model. Even if you choose to get defensive when the market goes up, the defensive sectors will still likely outperform a cash or bond allocation. Overall the strategy can be considered a return enhancer as part of investor portfolios without creating the typical drag of risk management strategies. That said we have had a really good 15-year run in equity markets leaving the strategy vulnerable to market downturns and hence more muted expected performance. Furthermore there there are much better assets to hold in various macro regimes if we expand the opportunity set.