Macroeconomic Exposures
Measuring Portfolio Sensitivity to Macroeconomic Variables
Prevailing macroeconomic conditions are known to be important drivers of stock level returns ([1] [2]). Investors seeking a particular factor exposure are inadvertently exposed to various macroeconomic factors such as rising or falling interest rates or inflation. Managing these latent macroeconomic exposures is a notoriously difficult, yet important, challenge.
We measure macroeconomic exposures, making it possible to identify unintended or concentrated exposures to macroeconomic tilts. The macroeconomic factor profile displays a portfolio's exposures to breakeven inflation, implied volatility, term spread, credit spread, short-term rates and long-term rates. These macroeconomic variables are market-based and forward-looking in nature and thus integrate the market view on their future movements.
Methodology
For a robust estimation of macroeconomic exposures, we rely on a simple bivariate regression model. Our objective is to exploit differences in asset-level macro exposures without altering access to the market premium. Therefore, we regress an instrument’s returns on innovations in macroeconomic variables while controlling for the overall market exposure. That is, the return of an instrument ii at time tt is described using a time-series regression
where represents the overall market return, captures the innovations (returns) in some macroeconomic variable of interest (e.g., long-term rates or volatility), and and stand for the respective loadings. That is, our definition of macroeconomic sensitivity expresses how strongly a security’s nominal return covaries with the particular macro variable while controlling for the market return. Such an approach is similar to the one applied by Bekaert, Geert and Wang ([3]) who focus only on inflation. Macroeconomic betas are calculated over the five years prior to time using weekly returns.
Weighted least squares with decaying weights
The estimation of the loadings problems face a basic trade-off between sample size and reactivity to exposure changes. Using long historical tracks improves the number of points available for the estimation, which improves its statistical precision. However estimating sensitivities over large samples will give the same weight to past behaviour as recent one, whereas the most recent behaviour . On the other hand, relying on a short estimation window, such as only a recent year or two, will lead to imprecise estimates due to the small sample size. We suggest overcoming this problem by using weighted least squares (WLS) on the long-term history of returns. Specifically, we apply an exponentially decaying weighting scheme
where denotes the observation horizon and the factor specifies that as approaches infinity, half of the total weights are attributed to the first 260 weeks, i.e., five years. The objective of this method is to benefit from a large sample while also capturing the recent dynamics of macro exposures.
Choice of macroeconomic risk factors
Traditionally, policymakers focus on backward-looking measures of the state of the economy such as gross domestic product (GDP), consumer price index (CPI), economic growth, unemployment rate, or money supply to measure the macroeconomic environment. While these measures are important, they are reported with a lag and often revised ex-post, making them unsuitable for real-time allocation decisions. For these reasons, we choose macroeconomic variables that reflect immediate changes in the uncertainty of future investment returns. Each macroeconomic variable satisfies a set of well-established criteria from the empirical asset pricing literature [4][1]. First, all macroeconomic variables must be sufficiently reactive to changes in the expectations of investors. This rule automatically disqualifies macroeconomic variables that are not updated frequently enough or are at high risk of post-release data revisions. Next, all macroeconomic variables must be tied to aggregate economic conditions beyond a single asset class (e.g., equity). Finally, a conceptual and interpretable link between the macroeconomic variable and the factors established through academic literature and empirical studies is required.
As a result, six macroeconomic state variables have been chosen that satisfy the above criteria. These are linked to critical aspects of the economy such as monetary policy, business cycle, forward interest rates, risk compensation in corporate bonds and the aggregate level of implied risk. Such state variables are forward-looking, reflect investor expectations, come at arbitrarily low frequency and are well documented in quantitative finance literature.
Each macroeconomic state variable is described in detail in the following table:
Macroeconomic State Variables | Proxy | Definition |
|---|---|---|
Expected Inflation | US 10 Year Government Treasury (GT10:GOV) – US Government Treasury Inflation Indexed Bond (GTII10:GOV) |
|
Short-term rate | US 3 Month Government Bills (GB3:GOV) |
|
Long-term rate | US 10 Year Government Treasury (GT10:GOV) |
|
Term spread | US 10 Year Government Treasury (GT10:GOV) – US 12 Month Government Treasury (GB12:GOV) |
|
Expected Volatility | Chicago Board Options Exchange Volatility Index (VIX) |
|
Credit Spread | Baa Corporate Bond Yield (DBAA) - Aaa Corporate Bond Yield (DAAA) |
|
References
1 Noël Amenc et al. (2019). Macroeconomic risks in equity factor investing. Journal of Portfolio Management. 45(6): 39-60. 10.3905/jpm.2019.1.092.
2 Noël Amenc et al. (2019). A Framework for Assessing Macroeconomic Risks in Equity Factors. Scientific Beta Publication.
3 Geert Bekaert and Xiaozheng Wang. (2010). Inflation risk and the inflation risk premium. Economic Policy. 25(64). 10.1111/j.1468-0327.2010.00253.x.
4 Ralitsa Petkova. (2006). Do the Fama-French factors proxy for innovations in predictive variables?. Journal of Finance. 61(2): 581-612. 10.1111/j.1540-6261.2006.00849.x.