Still confused what that fateful FOMC day just three weeks away from today may bring, in the aftermath of a Jackson Hole symposium which was mostly focused on the adverse side effects of Quantitative Easing and the proper sequencing of unwinding the Fed’s nearly $4 trillion balance sheet? Here is the explanation straight from the firm whose chief economist has dinners with none other than the Fed shadow Chairman, Bull Dudley, on a frequent basis. To wit: “First, we expect Fed officials to adjust the “mix of instruments” somewhat away from QE towards forward guidance at the September meeting, which appears to be an appropriate strategy in light of these results. Second, we expect that the FOMC will focus most if not all of the tapering on Treasury purchases rather than (current coupon) MBS purchases, consistent with the evidence that the latter are more effective in lowering mortgage rates and easing financial conditions.”
So: $10-15 billion reduction in TSY monetization announced in September, enacted in October, and a seismic shift in FOMC communication away from actual intervention to promises of such, aka forward guidance. Judging by the recent track record of “forward guidance” so far, the global market volatility exhibited so far may well be just a walk in the park compared to what is coming.
Full Goldman note:
While the Jackson Hole conference was less eventful than in previous years?in light of the lack of a keynote address from a member of the FOMC leadership?a number of interesting academic studies were presented. In particular, a paper by Arvind Krishnamurthy and Annette Vissing-Jorgensen (or “KVJ” for short)?presented by Krishnamurthy at the conference?argues that QE has more narrow effects on asset prices than Fed officials have typically proposed. First, KVJ present evidence that MBS purchases have been more effective than Treasury purchases in lowering mortgage rates and, thus, supporting the economy. The reason is that they find only limited evidence for a broad “duration removal effect,” through which taking out duration risk from the Treasury market boosts other, riskier, asset prices. Instead, they argue that Treasury purchases work mainly through a scarcity effect that pushes down Treasury yields but leave riskier asset prices largely unchanged. Purchases of MBS?which lower private borrowing costs more directly?are therefore a more direct and more effective way to boost the economy. Second, KVJ argue that within mortgages, purchases of newly issued (current coupon) MBS have been more effective in lowering mortgage rates than higher coupon MBS purchases. The basic intuition for this is that purchases of current coupon MBS are more effective in encouraging new MBS issuance. KVJ’s main vehicle for establishing these results is an event study that looks at the response of different asset prices to the Fed’s announcement of its unconventional policy steps.
The main shortcoming of such an event study?as stressed by the discussant of the KVJ paper at the Jackson Hole conference, Anil Kashyap of the University of Chicago?is that it is hard to control for the influence of other factors. In particular, the event studies do not control properly for the role of expectations. Also, the event studies focus on one-day changes, making it difficult to explore the persistence of the effects.
In the remainder of today’s comment we therefore provide additional evidence on this issue. Specifically, we extend our previous analysis and model jointly the effects of forward guidance, Treasury purchases and MBS purchases (both current and higher coupon) on mortgage rates, and financial conditions more broadly. We proceed in three steps.
First, we create a proxy of the FOMC’s forward guidance and the effects of its Treasury purchases on long-term rates. Following our previous work, we decompose the change in 10-year Treasury yields observed around FOMC decision announcements into the change in the rate expectations path (a proxy for forward guidance) and the change in the Treasury term premium (a proxy for the effects of Treasury purchases). We do so using the Board staff’s statistical approach (the so-called “Kim-Wright” model). To increase the chance that we only capture the effects of Fed policy on rates, we focus on the change in the Treasury term structure during the 1-hour window surrounding FOMC announcements. We then map the observed change in 2-year, 5-year and 10-year Treasury yields into a change in the term premium using a (daily) regression of the historical Kim-Wright term premium estimate on the structure of the yield curve. Our intra-day data goes back to 2001.
Second, we construct a proxy for the Fed’s mortgage purchases, distinguishing between current coupon and higher coupon MBS.
Specifically, we use the daily change in option adjusted spreads (OAS) on conventional 30-year MBS on FOMC decision announcement days. We define the current coupon spread as the OAS for a coupon that prices at par (i.e., at $100) and the higher coupon spread as the OAS for a coupon that prices at a 4% premium (i.e., at $104). While using daily, rather than intra-day, data might lead us to include influences other than the FOMC meeting, we believe that this is a reasonable simplification because we observe that MBS price changes around FOMC announcements drive most of the variation in daily changes in MBS spreads on those days over the last couple of years (when data are available).
Third, we use our proxies for forward guidance, Treasury purchases and MBS to explore their effects on the primary mortgage rate and financial conditions (measured by the GSFCI). In particular, we run regressions of changes in the primary mortgage rate/the GSFCI on our measures of guidance, Treasury purchases and MBS purchases. We use daily data back to 2001 (when the intra-day Treasury data are first available) and run separate regressions for different time horizons (after one to twenty days after the policy event).
Exhibit 1 shows our results for the primary mortgage rate. Specifically, the exhibit shows the cumulative effect of a 100 basis point increase in each of our Fed policy proxies on the primary mortgage rate after 1 to 15 days. First, we find that forward guidance (proxied with changes in rate expectations) has larger and more persistent effects on the primary mortgage rate than Treasury purchases (proxied with changes in the Treasury term premium). Second, changes in current coupon mortgage spreads (our proxy for current coupon MBS purchases) have larger effects on the primary mortgage rate than changes in the Treasury term premium (Treasury QE). Finally, our proxy for higher coupon MBS purchases (changes in higher coupon OAS) has negligible effects on the primary mortgage rate. Statistical tests (not shown) indicate that the effects of guidance and current coupon MBS purchases remain statistically significant even after 15 days, while the effects of changes in the Treasury term premium become insignificant after 10 days, and the effects of changes to higher coupon mortgage spreads are statistically never distinguishable from zero.
Exhibit 1: The Effect of Forward Guidance and Purchases on Primary Mortgage Rates…
Exhibit 2 shows that a broadly similar pattern is observed for our financial conditions index (GSFCI). Guidance, in particular, is highly effective in affecting financial conditions, followed by current coupon MBS purchases and Treasury purchases.
Exhibit 2: …and Financial Conditions
Our analysis is subject to a number of important caveats. First, it has been challenged whether the Kim-Wright model can distinguish accurately between changes in the rate expectations path and term premium at the zero lower bound. Attributing too much of the change in yields to the term premium?and too little to changes in rate expectations?could bias our results toward finding that signaling (per basis point) is more effective in easing financial conditions. Likewise, our analysis does not allow for an interaction between guidance and the effects of purchases, a channel we have found to be empirically relevant in past research. Second, our mortgage OAS variables are subject to a number of issues, including OAS model dependency and an arbitrary choice of what constitutes current coupon and higher coupon securities. Finally, all of our variables are proxies for the Fed’s actions, not direct measures. Ideally, we would have liked to explore directly the effects of the per-dollar purchases of Treasury and MBS purchases on private borrowing rates. This approach, however, is practically infeasible due to data limitations (asset purchase announcements were not broken down by current/higher coupon MBS) and expectational effects (asset purchases were typically anticipated prior to announcement).
Despite these caveats, our results are consistent with KVJ’s findings?and Fed officials’ evolving views over the last couple of years?that assign a more important role to forward guidance and (current coupon) MBS purchases than Treasury purchases. Moreover, these findings are consistent with our expectations for Fed policy in coming months. First, we expect Fed officials to adjust the “mix of instruments” somewhat away from QE towards forward guidance at the September meeting, which appears to be an appropriate strategy in light of these results. Second, we expect that the FOMC will focus most if not all of the tapering on Treasury purchases rather than (current coupon) MBS purchases, consistent with the evidence that the latter are more effective in lowering mortgage rates and easing financial conditions.
Please check back for new articles and updates at Commoditytrademantra.com