How Does Monetary Policy Impact Market Performance?
In the United States, the US Federal Reserve has long been tasked with a dual mandate of low unemployment and moderate inflation. To achieve these demands, the Federal Open Market Committee (FOMC) targets an interest rate for Fed funds that is comparable to the price of money. With cheap or expensive money likely to accelerate one outcome at the expense of the other, however gradually, the Fed may cycle between periods of a falling or rising policy rate.
Although the Fed weighs policy actions against economic indicators that respond with a lag, financial markets may respond immediately. As an estimate of the market impact of policy regimes, investors can match US stock market returns with changes in the Fed’s policy rate to compare performance.
Note: Annualized at 252-business days/year. Source: Bloomberg, QES calculations.
Over the past half-century, the S&P 500 Index’s daily price changes average about 3 basis points (bps).1 Filtering by policy accommodation or normalization, average daily returns increase to 4 bps or decline to 2 bps.
Can investors exploit Fed policy actions by piling into or rushing out of stocks based on those actions?
Beyond the difference of averages, equity investors might consider how periods of accommodating or normalizing policy affect the distribution of returns. Higher variance of returns about the mean during policy accommodation provides a familiar and unambiguous warning of a wider dispersion of possible outcomes. The return distributions’ higher moments of skewness (“skew”) and excess kurtosis (“kurtosis”) provide more esoteric yet useful warnings.
- Relative to a normal or Gaussian distribution with identical mean and variance, positive or negative skew may appear as long right or left tails, reflecting a lower likelihood of modal results and a higher propensity for results right or left of the mode.2
- Kurtosis can also reflect a higher likelihood of extreme results (“fat tails”) but paired with a higher likelihood of modal results. Together, this thins the probability density of intermediate results between the mode and extremes.
These effects can be seen by comparing daily SPX returns since 1954 with a Gaussian distribution parameterized by the former’s mean and variance.
In this chart:
- Negative skewness reflects higher frequencies towards the left of the distribution: The most common SPX return bin is -0.50% to 0.0% versus 0.00% to 0.50% under Gaussian returns.
- Positive excess kurtosis captures higher frequencies near the mode and at extremes: The 23 SPX returns at or below -5% reflect 0.14% of days, yet would normally arise on 0.00% of days.
Whereas the graph above compares all sampled daily SPX returns as actually distributed or as if distributed normally, we can sort returns by rate regime to visualize their more-negative skew and higher excess kurtosis during periods of accommodative monetary policy:3
Historically, equity investors reap a higher expected return during periods of policy accommodation. Yet -5σ or worse events occur with twice the frequency, while +3σ or better events are barely as likely.
Here, higher moments reveal how equity investment based on policy regimes could be more risky than is apparent from lower moments alone. Skew and kurtosis can help investors uncover risks that may be hidden by return variance.
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
1. Notwithstanding the contribution of dividends to compound growth, ignoring the sensitivity of dividend yields to interest rates lets us focus on variation in price returns: The inclusion of distributions will tend to increase average and compound returns but contribute little to higher moments of periodic return distributions.
3. Whereas the full-sample distribution plots frequency by bins of absolute value (since the SPX distribution and its normally-parameterized equivalent have identical variance), the conditionally-sampled distribution plots frequency by bins of standard deviation, accounting for the higher and lower dispersion of either respective subsample. This avoids a potentially misleading plot in which the frequency distribution of subsampled returns by bins of absolute value makes the “Normalizing” series appear more, not less, peaked than the “Accommodating” series.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: ©iStockphoto.com/RiverNorthPhotography