Macroeconomic variables like inflation, monetary policy, GDP growth, and commodity prices are important in explaining asset class performance and style premia — academic studies confirm it. The more directly the macroeconomic environment or specific economic variables affect a sector’s operating environment and financial results, the greater the impact on that sector’s performance. For example, oil prices directly affect the revenue, profitability, and asset value of Oil & Gas Explorers and Producers. And, interest rates movements are one of key drivers of Banks’ net interest margins.^{1}
The goal of our research here is to identify sectors and industries that have greater performance sensitivity to specific macroeconomic variables. This gives investors a starting point for the kind of topdown sector analysis that can help guide investment decisions.
Certain subsectors or industries may exhibit greater sensitivity to macroeconomic variables than the broad sector. Our analysis of the 11 GICS sectors and 18 industries (see Appendix I) identifies this specific opportunity set for investors.
We focused on a few key indicators broadly recognized to be influencers of asset class performance, as shown in Figure 1. Then, we selected sample time periods based on the availability of historical sector performance data (starting from July 2003) or the full cycle of yield changes.
Figure 1: Macroeconomic Variables
Macroeconomic Variables  Time Period 

10Year Treasury Yield (%, Level Change)  January 2018 – December 2022, which is a full cycle of 10year decline from 3% and rebound to 4% 
10Year Breakeven Rate (Proxy for Inflation Expectation, %, Level Change)  July 2003 – December 2022 
US Dollar Index (%, Price Change)  July 2003 – December 2022 
WTI Crude Oil Prices (%, Price Change)  July 2003 – December 2022 
Source: State Street Global Advisors, SPDR Americas Research, as of August 2023. Past performance is not a reliable indicator of future performance.
Figure 2: The Approach to Identify Sectors Highly Sensitive to Macro Variables
A simple linear regression model evaluates:
Focusing on relative returns removes effects of market beta on sector performance and transforms the time series to a stationary data set for regression analysis. We also tested the significance of coefficient (ttest), autocorrelation, and normality of the error distribution from the regression model to ensure coefficient estimates are reliable and statistically different from zero.
With so many unmeasured variables that can impact sector performance and all the noise in macroeconomic data, identifying a single macroeconomic variable that can explain even a small portion of sector performance gives investors valuable information.
To uncover that information edge, we set a minimum threshold of 0.08 for Rsquared to screen for sectors with a strong relationship to the variable. What that means is if the macroeconomic variable can explain more than 8% of the variance of sector returns, we consider the relationship between the variable and the sector to be strong.
The magnitude of the impact of macroeconomic variables, measured by beta, is also worth review. For example, both Metals & Mining and Capital Markets industries have exhibited strong negative correlation with the US dollar (USD). However, beta for the Metals & Mining is much greater than for the Capital Markets industries. In other words, a 1% depreciation of the USD may provide more tailwinds for Metals & Mining than for Capital Markets.
To evaluate Rsquared and beta altogether, we calculate weighted average zscore of Rsquared and beta for sectors that passed the previous screen under each macroeconomic variable. We gave Rsquared a greater weight (60%) in the zscore, since the strength of the relationship is a prerequisite to consider sectors for positioning against macroeconomic variables.
What is a zscore? Zscore measures how many standard deviations an element is above or below the population mean. A sector zscore can be calculated from the following formula. z = (X  μ)/σ, where X is the sector value of the metrics, μ is the mean of 11 sector values for a certain metric, and σ is the standard deviation of the value of 11 sectors.
There is one caveat to the sector beta estimates, which led us to group sectors that passed the initial screen into two tiers based on the weighted average zscore, instead of directly ranking them. While the beta for some sectors is statistically different from zero, their 95% confidence intervals — or the range of the beta that covers the true value 95% of the time — are quite wide and sometimes overlap with each other (see Appendix II). We placed sectors with a smaller weighted average zscore in the Tier 2 group.
See Appendix II for Rsquared, beta, and zscores of listed sectors. Sectors identified as having a strong relationship to macro variables are listed in the table below.
Figure 3: Sectors With Strong Relationships to Macro Variables
10Year Yield  10Year Breakeven  USD  Oil  

Positive  Tier 1  Insurance^ Financials Banks Regional Banks 
Oil & Gas Equipment & Services Metals & Mining 
Cons. Staples 
Oil & Gas Equipment & Services

Tier 2  Oil & Gas Exploration & Production Industrials^ 
Oil & Gas Exploration & Production* 
Metals & Mining  
Negative  Tier 1  Communication Services  Cons. Staples Health Care 
Metals & Mining Materials Oil & Gas Equipment & Services 
Health Care Cons. Staples Utilities*^ 
Tier 2  Tech^  Capital Markets* 
Source: State Street Global Advisors, SPDR Americas Research, as of December 2022. Past performance is not a reliable indicator of future performance. *Rsquared is greater than 0.08 and less than 0.1. ^Sectors that didn’t show strong correlation for the sample periods when the macroeconomic variable had a greater than one standard deviation move.
Our findings on how macro variables impact sectors are generally in line with economic intuition.
Figure 4: 5Year Rolling Correlation of Oil Monthly Return and 10Year Yield Changes
To find out whether relationships between sectors and macroeconomic variables would be strengthened or weakened by dramatic changes in variables, we divided the data sample into two groups based on the macroeconomic variable’s deviation from its historical average. If the variable is more than one standard deviation away from the average, we consider the change to be dramatic.
The table below shows the Rsquared of the simple linear regression between sectors and each macroeconomic variable in the two data groups. Most sectors show stronger correlation to the variable when there are more dramatic changes to the variable.
On the other hand, when changes are within one standard deviation, Rsquared of all sectors (except for Energy industries in relation to oil prices) declined below the minimum threshold of 0.08. This indicates an insignificant linear relationship when movements in macroeconomic variables are less significant.
Figure 5: Sector Sensitivity During Macroeconomic Shocks
Between 1stdv R Sqr 
Greater than 1 Stdv R Sqr 


10Year Breakeven Monthly Level Change Since 7/1/2003 
Oil & Gas Equipment & Services  0.054  0.285 
Metals and Mining  0.03  0.379  
Oil & Gas Exploration & Production  0.06  0.12  
Health Care  0.033  0.171  
Cons. Staples  0.018  0.214  
USD Weekly Return Since 1/1/2000 
Metals and Mining  0.047  0.496 
Cons. Staples  0.011  0.363  
Materials  0.028  0.283  
Oil & Gas Equipment & Services  0.052  0.343  
Capital Markets  0  0.393  
Oil Weekly Return Since 1/1/2000 
Oil & Gas Equipment & Services  0.196  0.447 
Oil & Gas Exploration & Production  0.184  0.459  
Energy  0.143  0.402  
Metals and Mining  0.047  0.273  
Health Care  0.028  0.109  
Cons. Staples  0.029  0.108  
Utilities  0.005  0.028  
10Year Treasury Yield Monthly Level Change Since 7/1/2003 
Insurance  0.001  0.001 
Oil & Gas Equipment & Services  0.024  0.19  
Financials  0.012  0.113  
Energy  0.019  0.123  
Regional Banks  0.03  0.108  
Bank  0.04  0.228  
Oil & Gas Exploration & Production  0.038  0.141  
Comm Svs.  0.041  0.117  
Industrials  0.003  0.05  
Tech.  0.002  0 
Source: State Street Global Advisors, SPDR Americas Research, as of December 2022. Rsquared greater than 0.08 is highlighted in green. Past performance is not a reliable indicator of future performance.
The slope of the yield curve has been closely watched by investors and monetary policymakers to project the future state of the economy. Monetary policy has a significant influence on the yield curve spread, economic activity, and shortterm equity market performance.
Expectations of future inflation and monetary policy contained in the yield curve spread also influence forecasts for economic growth, which in turn influence stock prices. The yield spread of 10 and 2year Treasurys is used as a proxy for the slope of the yield curve. Widening yield spreads indicate a steepening yield curve, while tightening spreads indicate a flattening yield curve.
We first conducted the Chi Square Test for Independence to determine if there is a significant relationship between the types of yield curve change (steepening or flattening) and sector performance (under/outperform the market). This helped us narrow our focus for further analyzing impact down to these nine sectors: Banks, Regional Banks, Capital Markets, Oil & Gas Equipment & Services, Software & Services, Consumer Staples, Financials, Real Estate, and Utilities.
We broke down the yield curve changes into six categories based on the direction and relative level of changes in 10year and 2year yields and created five dummy variables X1 ~ X5 = (0,1) to represent each type of yield curve in multiple linear regression analysis, as shown in Figure 6.
The intercept of the regression model (β_{0}) is interpreted as the average relative return when the yield curve is bear steepening. β_{0} + β_{1} , β_{0 }+ β_{2}, …… β_{0} + β_{5} are the mean estimate of relative returns given other five types of curve changes.
Figure 6: Yield Curve Multiple Regression Model and Types of Yield Curve Change
Sector Relative Return= β₀+ β₁×X₁+β₂×X₂+β₃×X₃+β₄×X₄+β₅×X₅
Yield Curve Change  Definition  Variables and Coefficient  Mean Estimate of Relative Return  No. of Months in the Data Sample (Since July 2003) 

Bear Steepen  10year yield increase > 2year yield increase  Intercept, β_{0}  β_{0}  52 
Bear Flatten  10year yield increase < 2year yield increase  X_{1, }β_{1}  β_{ 0 }+ β_{1}  49 
Bull Steepen  10year yield decrease < 2year yield decrease  X_{2, }β_{2}  β_{ 0 }+ β_{2}  22 
Bull Flatten  10year yield decrease > 2year yield decrease  X_{3, }β_{3}  β_{ 0 }+ β_{3}  72 
Twist Flatten  10year yield decrease, 2year yield increase  X_{4, }β_{4}  β_{ 0 }+ β_{4}  20 
Twist Steepen  10year yield increase, 2year yield decrease  X_{5, }β_{5}  β_{ 0 }+ β_{5}  19 
Source: State Street Global Advisors, SPDR Americas Research, as of December 2022. Past performance is not a reliable indicator of future performance.
Linear regression models for Banks, Regional Banks, Real Estate, and Utilities show an adjusted Rsquared greater than 0.08, indicating yield curve movements have significant explanation power for these sectors’ returns.
The table below shows the mean estimate of relative returns for various yield curve changes. The estimates that passed the significance test of coefficient (ttest) are highlighted in green. However, estimates for Bull Steepen, Twist Flatten, and Twist Steepen types of the yield curve should be taken with a grain of salt, since there are only about 20 observations under each of those scenarios in our data sample.
Figure 7: Estimated Mean of Relative Sector Returns (%)
Bear Flatten  Bear Steepen  Bull Flatten  Bull Steepen  Twist Flatten  Twist Steepen  

Banks  0.404  1.846  1.914  0.686  0.574  1.244 
Regional Banks  0.385  1.955  1.785  0.465  0.235  1.125 
Real Estate  0.463  2.09  1.783  0.036  1.871  0.369 
Utilities  0.125  2.2  1.278  0.158  1.597  0.417 
Source: State Street Global Advisors, SPDR Americas Research, as of December 2022. Past performance is not a reliable indicator of future performance. Green shades highlight the estimates that passed the significance test of coefficient (ttest).
This analysis of the yield curve’s impact on sectors is consistent with expectations:
Figure 8: Summary of Yield Curve’s Impact on Sectors
Bear Flatten (10year yield increase < 2year yield increase)  Bear Steepen (10year yield increase > 2year yield increase)  Bull Flatten (10year yield decrease > 2year yield decrease)  
Positive  Bank; Regional Bank  Real Estate; Utilities  
Negative  Bank; Regional Bank; Real Estate; Utilities  Real Estate; Utilities  Bank; Regional Bank 
Source: State Street Global Advisors, SPDR Americas Research, as of December 2022. Past performance is not a reliable indicator of future performance.
While this research focused on the impacts of a short list of macroeconomic variables, we acknowledge that sector performance is influenced by many variables beyond the ones analyzed. This includes other economic variables, industryspecific secular trends, valuations, monetary policy, and shortterm market sentiment.
Given the complexity and interactive nature of economic variables, it’s difficult both to anticipate which variables will drive sector returns and also to judge whether the macro expectations are priced in. Rather than predict sector performance using these variables, this research helps investors understand which sector relationships with macroeconomic variables appear most meaningful over a nearly 20year period.
Due to the limitation of linear regression models, this research identifies only sectors that have strong linear relationships with the macro variables. Sectors may have more complicated relationships that require a nonlinear model to formulate. Of course, a more complicated nonlinear approach would come at the expense of an easily understood and interpretable model.
Rather than use this research alone to predict sector performance or provide sector rotation trading signals, investors should use it together with sector fundamental analysis and our sector business cycle framework, to evaluate the merits of investing in certain sectors under specific economic conditions.