In 2008, aggregate demand in the United States collapsed and triggered deflation. The U.S. Federal Reserve, unable to lower short-term interest rates, turned to large-scale asset purchases (LSAP) to stimulate the economy. It purchased housing agency debt, mortgage-backed securities, and longer-term U.S. Treasury bonds. How did these actions affect financial markets and deflationary expectations?
Glick and Leduc (2012) collected 10 events, either statements by the Federal Open Market Committee (FOMC) or speeches by Fed Chairman Ben Bernanke, between November 2008 and November 2010. They included five events from the first round of asset purchases (QE1) and five from the second round (QE2). They reported that news of looser monetary policy caused the 10-year Treasury rate, the value of the dollar against several currencies, and the S&P Goldman Sachs Commodity Index to fall.
Kozicki, Santor and Suchanek (2015) culled 20 events related to unconventional Fed monetary policy. In addition to announcements during QE1 and QE2, they included events from the third round of LSAP that began in August 2012 (QE3). They found that news of LSAP lowered gold and silver prices during QE1 and QE2 and raised them during QE3.
Brainard (2017) observed that there is still much that policymakers do not understand about how LSAP affects the economy. In contrast, she noted that the effects of short-term interest rates have been extensively investigated.
In one study using short-term rates, Cook and Hahn (1989) found that increases in the federal funds rate target over the September 1974-September 1979 period raised Treasury rates of all maturities. Similarly, Kuttner (2001) reported that unexpected increases in the funds rate target over the June 1989-February 2000 period increased interest rates at all horizons.
Cook and Hahn's (1989) and Kuttner's (2001) results could imply that contractionary monetary policy raises longer-term real interest rates. The nominal interest rate equals the real interest rate plus the expected inflation rate. If contractionary monetary policy lowers expected inflation or leaves it unchanged, then evidence that it increases the nominal interest rate implies that it must be increasing the real interest rate also. Since contractionary monetary policy should raise short-term rates because of a liquidity effect and lower long-term rates because it reduces expected inflation, it is puzzling that increases in the Federal Reserve's target for the federal funds rate raised interest rates on long-term Treasury securities.
Romer and Romer (2000) offered an explanation for these findings. They demonstrated that the Fed has more information about future inflation than commercial forecasters do. They also found that Federal Reserve policy actions reveal some of this information (the inflation revelation channel). An increase in the funds rate target could thus raise interest rates by increasing expectations of future inflation.
One way to test whether how monetary policy surprises affected inflation expectations is to look at how they impacted daily traded commodity prices. Commodities such as gold and silver are hedges against inflation. Frankel (2008) and others demonstrated that if monetary policy actions are expected to increase real interest rates, they will lower commodity prices, and if they are expected to increase inflation, they will raise commodity prices. Thorbecke and Zhang (2009) reported that funds rate increases in the 1970s raised gold and silver prices and that increases in the later sample period lowered gold and silver prices. For the 1970s, these results suggest that R&R's explanation is correct. For more recent years, they indicate that funds rate increases affect real rates. These findings imply that the Fed's credibility was lower in the 1970s, when inflation was out of control, than in more recent years when inflation was quiescent (Note).
In a recent paper (Thorbecke, 2017), I investigated how news of unconventional monetary policy affected investors' inflationary expectations. If investors believe that LSAP would raise inflation, they would respond to news of LSAP by selling assets exposed to inflation and purchasing assets that hedge against inflation. This would lower the prices of assets that are harmed by inflation and raise the prices of assets that benefit from inflation, generating a positive relationship between inflation betas and asset returns. If investors instead interpret LSAP news to imply lower inflation, they would react in the opposite way and produce a negative relationship between inflation betas and asset returns.
To obtain inflation betas, I estimated a multi-factor model including returns on 60 assets on the left-hand side and inflation and other macroeconomic variables on the right-hand side. For LSAP announcement dates, I used what Roache and Rousset (2013) call the standard event dates for QE1, QE2, and QE3. I then regressed returns on the 60 assets over the 24 hours bracketing LSAP events on the assets' inflation betas.
Table 1 reports the results. Positive values indicate that investors expect more inflation and negative values imply the opposite. For the first seven events, the coefficients in Table 1 are always negative, indicating that news of LSAP cause investors to expect less inflation. Examining the first five events, Wright (2011) found that events 1, 2, 3, and 5 were episodes when policy was more expansionary than investors expected. The surprise expansionary components were especially strong for events 3 and 5. Swanson (2017) reported that the fifth event corresponded to a surprise 5.6 standard deviation expansionary shock.
These events influenced financial markets both by causing them to expect expansionary policies that might raise output and inflation and by indicating that the Fed was expecting lower inflation through the inflation revelation channel. The negative coefficients imply that markets did not expect LSAP to raise inflation. In the months when events 1, 2, 3, and 5 occurred, the consumer price index was experiencing deflation. The deflation rate when events 1 and 2 were announced was easily the highest the U.S. economy had witnessed over the last 60 years, and was almost six standard deviations away from zero. The combination of deflation and untried policy tools left investors unconvinced that the Fed could raise inflation.
For the fourth event, the coefficient in Table 1 equals -0.0078 and is significant at the 1% level. This coefficient implies that the assets with the largest inflation betas in the sample fell on average by 2.8%, and the assets with the most negative inflation betas rose on average by 2.8%. According to Wright's calculation, the fourth event was a contractionary surprise to markets. So this event both underwhelmed investors in terms of what the Fed was doing to fight deflation and conveyed news of low inflation through the inflation revelation channel.
The last event of QE1 occurred on November 4, 2009. The coefficient is positive and significant at the 1% level. The U.S. economy was recovering at this time and exiting from deflation, and the announcement caused investors to expect more inflation.
For QE2, the last two events in October and November 2010 caused returns on assets that hedge against inflation to fall. Both announcements were classified by Wright (2011) as events when monetary policy was more contractionary than expected. This contractionary policy news, combined with forecasts of lower inflation through the inflation revelation channel, caused market participants to revise their perceptions of inflation downwards.
QE3 began two years later, in August and September of 2012. The seasonally adjusted annual change in the price index for personal consumption expenditures excluding food and energy for the previous two quarters equaled 2.1% and 1.9% respectively, close to the Fed's target of 2%. Both of the QE3 announcements in the third quarter of 2012 caused investors to expect higher inflation.
These findings shed light on Glick and Leduc's (2012) and Kozicki et al.'s (2015) results that news of looser monetary policy lowered commodity prices during QE1 and QE2 and raised them during QE3. Glick and Leduc suggested that their findings implied that LSAP announcements during QE1 and QE2 signaled lower inflationary pressures. The negative coefficients in Table 1 on most announcements during QE1 and QE2 support their explanation, as they indicate that LSAP news lowered anticipated inflation during the first two rounds. Similarly, the positive coefficients in Table 1 on the announcements during QE3 are consistent with Kozicki et al.'s findings that LSAP news raised commodity prices during the third round.
These results indicate that, as actual inflation approached its target, the Fed was better able to influence inflationary expectations in the desired direction. The ability to affect expected inflation is important for monetary policy. At the zero lower bound, increases in expected inflation cause one-for-one decreases in the real interest rate. This provides needed stimulus when the economy faces deflationary risks. At higher interest rates, the ability to keep inflationary expectations anchored reduces the extra return that bondholders require to compensate for the risk of inflation. This keeps long-term interest rates from rising too high and choking economic activity. To influence inflation expectations, the Fed should remember the time-honored lesson that inflationary credibility increases as inflationary outcomes improve.
Table 1. Relationship between Assets' Returns and Their Inflation Betas on Days of Large-Scale Asset Purchase Announcements
||Coefficient on Inflation Beta
|Note: The table presents coefficients from a cross-sectional regression of returns on 60 assets over the 24-hour period bracketing announcements of large-scale asset purchases on inflation betas for the 60 assets. Inflation betas are obtained from iterated nonlinear seemingly unrelated regression estimation of a multi-factor model including returns on the 60 assets on the left-hand side and the Treasury bond/Treasury bill spread, the corporate bond/Treasury bond spread, the monthly growth rate in industrial production, the change in expected inflation, and unexpected inflation on the right-hand side. Unexpected inflation comes from the residuals of a regression of inflation on lagged inflation and current and lagged Treasury bill returns. QE1 refers to the first round of asset purchases, QE2 to the second round and QE3 to the third round.
*** (**) denotes significance at the 1% (5%) level.