For all the younger traders in our audience, we would like to inform you that maybe not now, but once upon a time, markets actually used to respond to economic data. That includes both stocks as well as the market that has been historically considered far "smarter" than equities, the Treasury market. Sadly, as central banks took over, the significance of economic data released declined until recently it has virtually stopped mattering, something we predicted would happen back in 2009 when we warned that soon the only financial report that matters is the Fed's weekly H.4.1 statement.
Today, some six years later, Goldman picks up where we left off nearly a decade ago, and asks "Does the Treasury Market Still Care about Economic Data?"
What it finds is simple (and something even the most lay of market observers these days could have told them): no.
As Goldman's Elad Pashtan writes, "the sensitivity of US Treasury yields to economic data surprises has declined to near record-lows over the last two years. We find that the pattern of reactions to data surprises across the yield curve matches pre-crisis norms—with higher sensitivity for short-term rates than longer-term rates—but the average reactions are much lower; for breakeven inflation reactions to growth data are not discernible from zero."
So if it is not the economy, then what does the "market" respond to? Take a wild guess:
In contrast, Treasury yields have reacted more strongly to Fed communication, at least according to one measure of policy surprises, and the sensitivity of exchange rates to activity news has increased.
Here are the details:
Economic data “surprises”— the difference between reported values for major economic indicators and consensus forecasts—have had a limited impact on US Treasury yields lately. Typically, Treasury yields rise on news of stronger-than-expected economic growth as investors anticipate either higher inflation and/or tighter monetary policy, and fall on news of weaker growth as markets discount lower inflation and/or easier monetary policy. In recent months, yields have had a much smaller reaction than normal to these types of data surprises. In Exhibit 1, we show the estimated impact of a 10-point surprise in our MAP index—a scaled measure of US growth surprises—on Treasury yields by year, controlling for changes in both risk sentiment (using the VIX index) and oil prices. The impact on 2-year yields has fallen to the lowest level since 2012, and the impact on 10-year yields has fallen to the lowest level since our dataset begins.
Treasury Rates Becoming Less Responsive to Data Surprises
Here Goldman expresses its confusion: "The limited impact of data surprises on rates is surprising given that the funds rate is no longer pinned at zero, and the Federal Reserve is actively considering further rate increases. When the funds rate was at the zero lower bound (ZLB) and the Fed was easing policy through forward guidance and QE, investors rightly saw little prospect of near-term rate hikes, even if the economy firmed meaningfully. The responsiveness of short-term Treasury yields to data surprises picked up as the Fed approached liftoff last year, but has since retreated back to ZLB levels."
Actually, the reason for the collapse in the market's response is precisely because the same "market" no longer believes the Fed, or its reaction function, and as a result is no longer as concerned about rate hikes, as it was for example in 2015. Where things get even more confusing is that over the past several years, Fed policy has been largely driven by the market itself, which however no longer responds to the data but merely to the Fed, creating the most diabolical and reflexive "circular reference" in capital markets existence.
That particular discussion is the topic of a separate post, however. For now we are more interested by Goldman's "amazement" at something that had been largely obvious to most non-academics. Here is Goldman's attempt to "explain" this phenomenon.
One possible explanation for this phenomenon is that investors are now more focused on Fed communications, rather than to economic data releases—perhaps due to uncertainty about the central bank’s reaction function. And we do see some evidence along these lines. For example, we can use the same regression framework, but replace the MAP score variable with a measure of monetary policy surprises. We quantify monetary policy surprises using the correlation in daily returns across asset classes. Unexpected monetary policy changes create a particular pattern in market returns, which allow us to isolate them from growth and inflation shocks, and estimate their relative magnitudes over time (for further details see here). We calculate policy shocks using average correlations from 2000 through 2016 (i.e. the principal component loadings are fixed over this time period), so the regression results can be thought of as how the reaction in rates differs from the sample average. When we apply this regression to nominal forward rates on FOMC meetings and minutes release days, we see that interest rates have indeed become more sensitive to monetary policy events—both today and during the crisis—than they were during the pre-crisis era (Exhibit 3).
Treasury Reactions Similar but Larger to Monetary Policy Surprises
While most of Goldman's analysis is commonsensical, they do find an interesting tangent, namely that as the "sensitivity of the Treasury curve to data surprises has declined, the sensitivity of the dollar has increased." So are we all now just one big FX-trading family? Here's Goldman
Why are investors no longer reassessing their inflationary outlook in response to economic data? One possible explanation relates to investor perceptions about divergent global monetary policy regimes. While the sensitivity of the Treasury curve to data surprises has declined, the sensitivity of the dollar has increased: since mid-2014 the dollar has been roughly twice as sensitive to data surprises compared to pre-crisis levels (Exhibit 4, right panel). This result hints that investor may be focused on the effects of dollar pass-through to domestic prices, such that breakeven inflation remains stable even as activity data surprises markets (though we would note that the implied effects are larger than our normal pass-through estimates would suggest, and much more persistent as well).
Breakevens no Longer Sensitive to Data, but Dollar Sensitivity Higher
Summarizing Goldman's findings:
we find that that the sensitivity of nominal Treasury yields to US economic data surprises is currently very low, despite the fact that the FOMC has hiked once and is considering further increases. The reaction of breakeven inflation in particular is not discernible from zero. At the same time, Treasury markets appear more sensitive to Fed communication—at least according to one measure of policy surprises—and the dollar is reacting more strongly to activity data. Although it is difficult to draw strong conclusions, there could be a few explanations behind these disparate facts, including (1) higher uncertainty about the Fed’s reaction function, (2) investor focus on exchange rate appreciation and pass-through to domestic prices, and (3) low confidence that cyclical forces will lift domestic inflation.
While we are genuinely surprised at Goldman's surprise to the bond market's lack of a reaction to surprises, we would add a (4): the market, in its conventional role of a discounting mechanism which is constantly calibrated by processing an near infinite amount of information about the future, no longer does that and is simply responding to the latest statement or act by the Fed which - paradoxically - is reflexively responding to the market (especially if the market is selling off). Which is why on occasion you will find us writing it as "market", because thanks to the Fed, it no longer exists.