The Risk Premium on the Global Financial Cycle
In the previous dispatch, we documented the predictive information contained in our measure of global liquidity. Specifically, we showed that the flux of cross-border funding strongly predicts 1-qtr forward changes in the level of the dollar. The level of the dollar itself, suitably detrended, contains information on prevailing financial conditions. These insights should allow us to price global assets. The risk posed by the tidal whipsaw of the global financial cycle to national stock markets should have a price — investors should be compensated for bearing that risk. Ie, stock markets that are more sensitive to the global financial cycle should sport higher expected excess returns. We can extract the risk premium associated with the global financial cycle from the variation in expected returns.
Before we can extract the price-of-risk, we must put the fifty odd stock markets on a uniform footing. The approach we shall take is informed by the macrofinance paradigm. We assume that the marginal investor in global assets is a Metropolitan market-based intermediary, like a global bank, reckoning her profits and losses in dollars. So we deflate national stock market indices by the exchange rate against the dollar prevailing in the quarter. Then we compute the dollar-based rate of return for national stock market i in quarter t+1 as:
$latex R_{i, t+1} := \log(M_{i, t+1} / X_{i, t+1}) - \log(M_{i, t} / X_{i, t})$,
where M is the level of the stock market index and X is the prevailing level of the exchange rate against the dollar. To really satisfy the assumptions of asset pricing, we also need the risk-free rate. Since our marginal investor is an American market-based intermediary, we use the risk-free rate prevailing in the United States. We obtain it from Kenneth French's website. We subtract the risk free rate from the dollar returns and, in an abuse of notation, denote the same by $latex R_{i, t+1}.$
Dynamic asset pricing requires a return forecasting or price-of-risk factor along with a risk factor. Our price-of-risk factor is 1-qtr lagged values of our stochastically-detrended global liquidity measure, as described in the previous dispatch. Our risk factor is contemporaneous AR(1) innovations in the same. That is, we assume that our feature has predictive information about future returns and that systematic risk is captured by contemporaneous shocks to our feature. In order to estimate our dynamic asset pricing model, we use a version of Adrian et al. (2015)'s 3-pass OLS estimator for the special case of constant betas and time-varying prices of risk. More precisely, we first regress excess returns in the time series against lagged values and contemporaneous innovations in our feature to obtain parameters stratified by country:
$latex R_{i, t+1} = \alpha_{i} + \gamma_{i} LIQUIDITY_{t} + \beta_{i} U_{t+1} + e_{i, t+1}, \qquad t=0, \dots, T,$
where U is the AR(1) innovation in LIQUIDITY for the quarter, and e is the end of quarter pricing error. In the second pass, the cross-sectional regression, we project "the constants," $latex \alpha_{i}$, and the slopes of last quarter's LIQUIDITY, $latex \gamma_{i}$, onto the betas $latex \beta_{i}$, the sensitivity of exposure to contemporaneous innovations in LIQUIDITY, in the cross-section, to obtain the price of risk parameters, $latex \lambda_{0}$ and $latex \Lambda$, respectively. That is, we run cross-sectional regressions of the form:
$latex \alpha_{i} = \bar\lambda_{0} + \lambda_{0}\beta_{i}, \qquad i=1, \dots, N,$
$latex \gamma_{i} = \bar\Lambda + \Lambda\beta_{i}, \qquad i=1, \dots, N.$
The time-varying risk premium associated with the global financial cycle is then given by,
$latex \Lambda_{t} = \lambda_{0} + \Lambda*LIQUIDITY_{t}.$
Thus, the price-of-risk is modeled as a linear function of our feature. Having spelled out the methodology, we are now ready to report our results. Table 1 displays the model estimates for the first "constants" cross-sectional regression, whose slope is $latex \lambda0$. We obtain a very good fit. Our feature betas explain two-thirds of the variation in the country fixed-effects, $latex \alpha_{i}.$ The slope parameter is large and extremely significant.
Table 2 displays the model estimates for the second "slopes" cross-sectional regression. We obtain an excellent fit. Sensitivity to contemporaneous shocks to global liquidity, as captured by the betas, explains 86 percent of the variation in the return predictability of national stock market returns, as captured by the slopes of lagged liquidity in the time-series regressions.
What these results show is that global stock markets are extremely sensitive to the global financial cycle. The next graph shows global liquidity betas. The sign of the betas is an artifact of the way the AR(1) innovations are constructed. From the previous dispatch, we know that positive shocks to global liquidity are positively associated with contemporaneous stock market returns. The betas are large and significant in 43 out of 46 markets for which we have data.
We also document the predictive slopes against one-quarter lagged global liquidity. Our measure forecasts stock market returns in 45 out of 46 countries for which we have data.
Finally, what is the risk premium associated with the global financial cycle — the excess compensation that investors can expect by exposing themselves to its tidal whipsaw? The next figure displays our time-varying price-of-risk parameter, computed as explained above. Exposure to the global financial cycle is rewarded with a handsome risk premium. On average, an additional unit of exposure to the global financial cycle is associated with an extra 13.3 percent quarterly return in excess of the risk-free rate. This is "too large" and very likely an artifact of the scaling of our feature. In any case, the risk premium is significant enough to warrant attention. There is also significant diachronic variation in the risk premium. The premium collapsed during the financial crisis and never fully recovered.
We have shown that the whipsaw of the global financial cycle is priced into the cross-section of expected excess returns in global stock markets. Exposure to the global financial cycle is associated with a significant risk premium. Global investors can systematically harvest this risk premium to earn high risk-adjusted returns. This is interesting and useful. I am glad I revisited the information content of the flux of global credit flows.