Empirical evidence

The Fed’s balance sheet expansion officially ended with the completion of the tapering process in October 2014, but it will still take time until normalization is achieved and thus until a full evaluation of the impact of unconventional policy measures is possible. However, it is possible to assess which features of the QE program were effective and to what extent it helped to realize the monetary policy

31

goals (Rosengren, 2015). A number of studies have examined the effectiveness of large-scale asset purchases and forward guidance policies at the ZLB, which will be discussed in more detail in the following.

3.3.1.1. Transmission channels

QE in the US was believed to work through the portfolio balance channel and signaling channel. A number of empirical studies use lower-frequency data to investigate the portfolio balance effect by modeling the relationship between long-term Treasury yields, term premia, and the maturity structure of the public’s holding of Treasury debt (Gagnon et al., 2011; Hamilton & Wu, 2011;

Krishnamurthy & Vissing-Jorgensen, 2011). Gagnon et al. (2011), estimate that QE1 reduced the term premium by about 52 basis points and the 10-year Treasury yield by about 82 basis points. Hamilton and Wu (2011) estimate a three-factor term structure model to predict the effects of changes in the maturity structure of US debt from asset purchases and find similar but smaller effects than Gagnon et al. (2011). D’Amico and King (2013) find that QE1 and QE2 operated through scarcity and duration channels, but find no evidence of a signaling channel. Krishnamurthy and Vissing-Jorgensen (2011) on the other hand, find no evidence in support of the duration risk channel but conclude that QE effects on safe and risky assets were likely transmitted through the signaling channel. Also, Bauer and Rudebusch (2014) argue that previous studies wrongly attributed changes in term premia solely to portfolio balance effects and find that signaling effects for the LSAP’s were economically and statistically significant.

3.3.1.2. Forward guidance policy

The Fed used both calendar-based and threshold-based forward guidance in the past. Calendar-based forward guidance informs about the intended path of it policy rate over a certain period, while

threshold-based forward guidance is contingent on certain economic conditions and recognizes that the policy rate adjusts with changes in those factors (Campbell, Fisher, Evans, & Justiniano, 2012).

Williams (2012) argues, that the statement on August 9, 2011, where the FOMC announced it

“anticipates that economic conditions... are likely to warrant exceptionally low levels for the federal funds rate at least through mid-2013” successfully shifted market expectations. The introduction of

32

calendar-based forward guidance led to a fall in Treasury yields by one- and two-tenths of a percentage point, which was suggested a considerable drop (Williams, 2012).

A few studies have tried to assess the impact of forward guidance policies when the interest rate is at the effective lower bound. For instance, Campbell et al. (2012) show that surprises associated with FOMC policy announcement substantially influence Treasury bond rates, corporate borrowing rates, and private macroeconomic forecasts. Moessner (2013) studies the impact of explicit FOMC policy guidance on short- to medium-term interest rates and finds that forward guidance successfully flattened the yield curve for both Eurodollar futures and US Treasuries, by managing expectations about future monetary policy. While in theory, threshold-based forward guidance can, by providing more clarity about the conditionality, lead to increased credibility of the central bank, it is also more complex and therefore difficult to convey. The empirical evidence confirms that conditional forward guidance had less impact on futures rates and long-term bond yields (Filardo & Hofmann, 2014).

A recent study by Gavin, Keen, Richter, and Throckmorton (2015) examines the Fed’s forward guidance via news shocks to the monetary policy rule in a New Keynesian model at the ZLB. They conclude that forward guidance can stimulate economic activity but only if households believe the economy will recover; otherwise, interest rates will remain near zero. Thus, the current economic outlook likely influenced the effects of the FOMC communications.

The empirical results suggest that the Fed's forward guidance about future monetary policy

successfully reduced long-term interest rates. However, it remains unclear, to what extent the decline was due to the Fed’s communication and how much was due to simultaneous changes in economic conditions. Also, the magnitude of large-scale asset purchases potentially facilitated the credibility and thus the effectiveness of the Fed’s forward guidance. Since the policies were mostly conducted at the same time, it is difficult to isolate their market impacts (Engen, Laubach, & Reifschneider, 2015).

3.3.1.3. Impact on government bond rates

Several event studies have tried to evaluate the effects of the QE programs using a narrow time window around the policy announcements. There exists considerable evidence from QE1, suggesting that the program reduced 10-year government bond yields by about 100 basis points; Krishnamurthy

33

and Vissing-Jorgensen (2011) evaluate the effects of the Fed’s asset purchases on interest rates using daily and intra-day data and find that bond rates decreased by 107 basis points. Gagnon et al. (2011) show that the Fed’s first LSAP program successfully lowered the term premium, with a reduction of 91 basis points in Treasury yields. Glick and Leduc (2013) find similar results, reporting a reduction of 100 basis points while Meaning and Zhu (2011) suggest a smaller effect of around 80 basis points.

The evidence about QE2 announced on the 3rd of November 2010 indicates that there was no

substantial effect on Treasury yields. Krishnamurthy and Vissing-Jorgensen (2011) find a decline of 18 basis points in government bond rates using a 1-day window while the effect is larger for a 2-day window with a reduction by 30 basis points. Glick and Leduc (2013) on the other hand find no effect on the 10-year Treasury yields and a negligible decline in the 5-year rate. Meaning and Zhu (2011) study the impact of the Operation Twist announced on September 21, 2011, and find a small decrease in 10-year bond rates, but a rise in the 5-year rate.

Figure 3 QE announcements and 10-year Treasury yields (%)

Source: US. Bureau of Economic Analysis (2016) based on Rosengren (2015)

Figure 3 shows the 10-year Treasury yields around the time of the announcements of QE1, QE2, and QE3. The first program was announced on November 25, 2008, and in the FOMC statement on December 16, 2008, with additional purchases declared in the policy statement of March 18, 2009.

The 10-year Treasury yield fell 24 basis points on the day of the announcement in November,

First QE1 announcement

2.00 2.50 3.00 3.50 4.00 4.50

29-Oct-08 7-Jan-09 18-Mar-09

QE2 announcement

2.00 2.50 3.00 3.50 4.00 4.50

7-Oct-10 4-Nov-10 2-Dec-10

announcementQE3

1.00 1.30 1.60 1.90 2.20 2.50

9-Aug-12 8-Sep-12 8-Oct-12

34

followed by a further decline of 12 basis points on the next day. The FOMC statement in March led to another drop in the 10-year Treasury rate by 51 basis points. When QE2 was initiated on the 3rd of November 2010, the Treasury yields decreased by 14 basis points immediately the day after, followed by an increase over the next weeks. Treasury yields were already at historically low levels when QE3 was announced in the Fed’s policy statement of September 13, 2012. On the day after, the 10-year Treasury yields fell by 13 basis point and continued the downward trend over the next two weeks by 20 basis points. Overall, Treasury yields fell by approximately 200 basis points over the course of the three programs (Rosengren, 2015).

3.3.1.4. Impact on other assets

There is considerable evidence that the Fed’s LSAP programs affected corporate bond rates. For instance, Krishnamurthy and Vissing-Jorgensen (2011) find a strong impact of QE1 on corporate bond rates, reporting a reduction of 77 basis points for AAA-rated bonds, 93 basis points for A, 60 basis points for BA and 43 basis point for B bonds. Similarly, Neely (2013) suggest a decrease of 78 basis points in BAA-rated corporate bonds. Hancock and Passmore (2015) estimate substantial effects of the LSAPs on the secondary market yields on agency MBS and slightly smaller effects on mortgage rates. They suggest that the FOMC met its goal to put downward pressure on longer-term interest rates, and support mortgage markets to ease financial market conditions.

Glick and Leduc (2013) consider the effect of QE on the value of the dollar using high-frequency intraday data. They document that the US dollar depreciated significantly for Q1, with a decline of 0.5 percentage points following a surprise announcement equivalent to a one percentage point decrease in the federal fund rate. Results for QE2 and QE3 were smaller. Neely (2013) reports a jump

depreciation of the US dollar during the announcement window from 3.5% to 7.8% following QE1 surprises. In a recent study of the effects of unconventional monetary policy on international risk premia Rogers, Scotti, and Wright (2015) find that carry trade portfolios from foreign countries are significantly related to the US monetary policy surprises. They then identify monetary policy shocks using a dynamic, structural VAR incorporating US and foreign interest rates, and exchange rates, which allows them to measure the effects on expectations and risk premia. The results suggest that

35

unconventional policy surprises lowered domestic and foreign risk premia, and led to a depreciation of the dollar.

The impact on other financial assets is rather small. Rosa (2012) estimates the impact of changes in Treasury term premia due to LSAPs on stock markets and finds only moderate effects, reporting that a 20 basis point decrease in 10-year Treasury rates lead to an increase in stock prices of only 1%.

However, other event studies of stock market responses in narrow time windows suggest somewhat larger reactions to the policy news about QE1 and QE2 of about two basis points for a 20 basis point decrease in the 10-year Treasuries. Nevertheless, it is argued that stock market reactions largely depend on the prevailing market conditions and event studies do not capture the full effect of the monetary policy actions by the Fed (Engen et al., 2015).

Figure 4 illustrates the increase in the S&P500 stock index over the course of QE1. There was no large momentum in stock prices from QE2, but there was a noticeable upswing following Operation Twist with a continuing upward trend over QE3. Following the Liftoff, in December 2015 the S&P500 saw a large drop which was largely due to a fall in oil prices and global growth concerns but markets started to recover at the beginning of 2016. The pattern confirms that a considerable part of the stock market response to the unconventional monetary shocks happened outside of the narrow response windows, but QE programs facilitated an improvement in equity prices over time.

Figure 4 S&P 500 composite stock price index

Source: US. Bureau of Economic Analysis (2016) QE1 starts

QE1 ends

QE2 starts QE2 ends

Operation Twist

QE3 starts QE3 ends

Fed Liftoff

500 700 900 1100 1300 1500 1700 1900 2100 2300

Nov-08 Dec-09 Jan-11 Feb-12 Mar-13 Apr-14 Apr-15 May-16

36 3.3.1.5. Impact on the real economy

Since the ultimate goal of the unconventional monetary policies is to impact the real economy, recent studies have attempted to capture the macroeconomic effects. While the aim of QE1 was mainly to stabilize financial markets and mitigate the ramifications of the crisis on output, the latter policy measures were meant to stimulate the economy.

Since event studies are not feasible to measure the impact of QE on GDP and inflation, most evidence stems from econometric studies using data from before the financial crisis. For instance, Baumeister and Benati (2013) evaluate the macroeconomic effects of a compression in long-term bond yield spread by estimating a time-varying parameter structural VAR (TVP-VAR) model for the US. They show that the US unconventional monetary measures prevented risks of a significant deflation and a

collapse in GDP comparable to those during the Great Depression. Further, without QE the US inflation rate was estimated to be minus one percent, output would have fallen by 10%, and unemployment would have been 10.6% in 2009.

Chung, Laforte, Reifschneider, and Williams (2012) conduct their analysis using a set of structural and time series statistical models rather than a single structural model. Their model simulations indicate the Fed’s balance sheet expansion prevented deflation and labor market conditions from being much worse than without the asset purchases. They further report that QE1 increased GDP growth by almost 200 basis points, while QE2 boosted GDP by around 100 basis points. Similar conclusions about QE2 are also reached by Chen, Cúrdia, and Ferrero (2011). They use a medium-sized DSGE model that explicitly accounts for the ZLB constraint and suggest that the program increased GDP growth by less than half a percentage point. Although they find only marginal effects on inflation, they predict that LSAPs combined with a commitment to keep interest rates low for a considerable time would double the impact on GDP growth and inflation.

Engen et al. (2015) use survey data from the Blue Chip Economic Indicators to assess changes in private-sector expectations of the Fed policy rule and estimates from other studies about QE effects on term premia to evaluate the economic stimulus of the program. They find that the impact on real activity and inflation was limited in the first two years after the financial crisis since changes in policy

37

expectations were moderate. However, Engen et al. (2015) acknowledge that the Fed’s QE facilitated the economic recovery from 2011 onwards, once the private sector gained confidence in the Fed’s commitment to ease markets. They estimate that forward guidance and asset purchases may have lowered the unemployment rate by about 1.2% and increased the level of inflation by 50 basis points.

The inflation rate is illustrated in Figure 5 and represents the second major goal as part of the Fed’s dual mandate. The Core PCE inflation index has been persistently below the 2% goal, despite a peak after the Operation Twist. Other real variables however, such as housing prices and auto sales, showed a clear upward trend after the implementation of Operation Twist and QE3, while unemployment rates even fell below the desired 6% mark (Rosengren, 2015).

Figure 5 US Inflation Rate: Core personal consumption expenditure price index (annual changes %)

Source: US. Bureau of Economic Analysis (2016)

3.3.1.6. International spillover effects

It has been argued that monetary policies of developed economies not only impact the targeted domestic markets but also pose various spillover effects. Via the portfolio balance channel, QE might increase demand for longer duration and risky domestic assets, but may also trigger a portfolio rebalancing towards foreign assets (Fratzscher, Duca, & Straub, 2014). Through the liquidity channel, the decline in liquidity premia and thus borrowing costs increases overall bank lending activity, which can spillover to developing countries. Finally, signaling about future economic conditions and policy rates can affect the confidence channel about deflation and volatility, and are likely to increase

QE1 starts QE1 ends

QE2 starts QE2 ends Operation

Twist QE3 starts

QE3 ends

Fed Liftoff

0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3 2.5

Nov-08 Nov-09 Oct-10 Sep-11 Aug-12 Jul-13 Jun-14 Jun-15 May-16

38

investment activity. Thus, via those transmissions channels, QE has significant implications for

economic conditions and financial markets elsewhere in the world (Lim, Mohapatra, & Stocker, 2014).

Chen, Filardo, He, and Zhu (2014) examine domestic and spillover effects of the Fed’s QE on both advanced and emerging market economies (EMEs) using a global VAR model. They state that global portfolio rebalancing, signaling or confidence channels may have played a critical role in lowering emerging market bond yields, inflating equity prices, and causing upward pressure on exchange rates against the dollar. Overall, the impact of QE on emerging markets is stronger than in the US and other advanced economies and has contributed to an overheating in EMEs. Bowman, Londono, and Sapriza (2015) study the effects of QE on sovereign yields, foreign exchange rates, and stock prices in EMEs and find that EME assets prices experienced significant fluctuations around the Fed’s monetary easing announcements. The study suggests that the vulnerability of a country to changes in the US monetary policy is larger for those with high-interest rates, inflation rates, CDS spreads, current-account deficits and weak banking systems. Furthermore, their findings indicate that while QE affects EMEs, the impact has not been very different from past changes in US interest rates.

The benefits and costs of QE policies and the resulting spillover effects remain debatable. On the one hand, stronger domestic growth due to unconventional policy measures stabilizes the economic and financial conditions and increases demand for exports of foreign countries. In fact, it has been noted that the US QE measures have prevented a prolonged recession and severe deflation in the most advanced economies. However, especially EMEs hold the view that such measures led to an inflation of interest rate differentials and domestic currency depreciation, thereby triggering currency wars. All in all, while QE mostly benefited advanced economies, the detriments for the emerging economies depend partly on how the economies reacted and on their existing financial structure (Chen et al., 2014).

In document The Quantitative Easing Experience (Sider 31-39)