How did Quantitative Easing affect
the term premium in the United
States during the credit crisis?
Floris van Eijck
10017623
Statement of originality
This document is written by student Floris van Eijck, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no other soureces other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
STATEMENT OF ORIGINALITY 2
TABLE OF CONTENT ERROR!BOOKMARK NOT DEFINED.
INTRODUCTION 4
IMPLEMENTATION OF QUANTITATIVE EASING BY THE FED 5
LITERATURE REVIEW 7 EMPIRICAL MODEL 9 RESULTS 12 CONCLUSION 15 APPENDIX 16 REFERENCES 19
Introduction
As the economic prospects worsened in 2007, the Federal Open Market Committee (FOMC) began cutting the overnight federal funds rate in order to ease the stance of monetary policy. The FOMC’s initial target was 5.25%, but late in 2008, they realized they had to lower the rate further and reached its effectively zero lower bound (Blinder, 2010). Even though it had reached its zero lower bound, the economic growth prospects still worsened and there was a threat of deflation. In order to counter this, the Fed started purchasing large amounts of medium and long-term assets to push down bond yield and provide more monetary policy stimulus to the economy (Christensen & Rudebusch, 2012). By doing these large-scale asset purchases (LSAPs), the Federal Reserve’s balance sheet expanded from about $850 billion to more than $4.4 trillion. The significant growth of the additional assets may remain in place for years to come. This process of these LSAPs is known as quantitative easing (QE) (Krishnamurty & Jorgensen, 2010 & Gagnon ét al, 2010).
The way LSAPs work is through the reduction of the amount of securities that are being held by the private sector, while simultaneously increasing the amount of risk-free bank reserves held by the private sector. The purchases bid up the price of the assets and by doing so, lowering the yield. In other words, the risk premium of an asset is being purchased by the Federal bank (Gagnon ét al, 2010).
The basic theory of the term structure of interest rates is the expectations hypothesis. According to this hypothesis, the expected return from holding a long bond until maturity is the same as the expected return from rolling over a series of short bonds with a total maturity equal to that of the long bond. That is, the long bond yield is the average of the expected short-term rates. Equivalently, the forward rate (the short-term rate at which investors agree now to borrow or lend in the future) is the expected future short-term rate. Though the expectations hypothesis provides a simple and intuitively appealing interpretation of the yield curve, it ignores interest rate risk. Except if
calculated until maturity, the nominal return on a long bond is uncertain, and investors may require compensation for this risk. The “term premium” refers to this compensation and any other sources of deviation from the expectations hypothesis (Kim, 2007). For Treasury securities
In this study, I will answer if QE that was implemented by the federal bank affected the term-premium. This is relevant because we need to learn how monetary policies affected al economic variables during the crisis in order to ensure better
responses in future recessions. To answer this question, I will first explain how QE was implemented. After that I will review existing literature about QE and the term-premium. Following the literature review, the model used by Gagnon et al (2010) will be discussed and eventually will be complemented by the variable quantitative easing in order to answer the question stated above. In the final two sections of this paper I will present the results, which will be concluded with by discussing them.
Implementation of quantitative easing by the Fed
Before discussing how the Fed implemented QE, it is important to know through which transmission channels QE works. There are multiple transmission channels that work through asset prices, the portfolio balance effect, expectations/signaling effect and liquidity premia effect. The other channels are confidence effects and the bank lending effect (Joyce et al, 2011). According to Kapetainos et al (2012) the most important channel is the portfolio balance effect. LSAPs purchase the risk premium of the assets that are being bought. It causes a reduction in the amount of the security that the private sector holds while in the meantime causing an increase in the amount of short-term, risk free bank reserves that he private sector holds. This increases the prices of securities and assets, while lowering their yield (Gagnon et al, 2010). Another way the portfolio balance effect reduces the yield is through the reduction of the duration of the risk of long-term assets, because there are less long-term assets in the private market. The main component of the duration risk is the term premium (Joyce et al, 2012 & Gagnon et al, 2010).Another way the LSAPs lowered the yields was through the signaling channel. The announcement of QE lowered the expected average short-term part of long-term rates (Bauer & Rudebusch, 2013). According to the Fisher theory, the real interest rate depends on the nominal interest rate minus the inflation rate. Because the Fed was committed to the inflation target and gave a credible signal about it, the expected inflation equaled the inflation. Protecting the real interest rate to rise during this time (Joyce et al, 2012).
The last transmission channel that work through asset prices, is the liquidity premium channel. As the LSAPs caused an abundance in reserves compared to long-term bonds, the long-long-term rates are submitted to a downward pressure (d’Amico et al, 2012). Because of the reduced yields, investors needed a higher return on their assets. By implementing the LSAPs, the Fed prevented the increase of the return while also
stimulating trade and increasing the liquidity of financial markets. The effects of the liquidity premium channel is believed to be small and only exist during the LSAPs (Joyce et al, 2011).
According to Joyce et al (2011) the asset purchases boosted the consumer confidence and willingness to spend. They write that the monetary policy might have generated an improvement in economic outlook and that this improvement is reflected in higher asset prices, especially by reducing the risk premium (Joyce et al, 2011).
The last channel discussed is the bank lending effect. The LSAPs have increased the amount of reserves banks have stored at the central bank. These liquid assets could be a motivation for banks to extend more new loans than usually. However, because of the pressure on banks during the crisis, the impact of this effect is expected to be small, or even non-existent (Joyce et al, 2011).
Understanding the transmission channels, gives a better understanding of the events that happened during QE. The Fed engaged in three successive rounds of QE since 2008, but before these rounds, the Fed had already engaged in certain forms of QE. The first type was early in 2008 when the Fed started selling its holdings of
treasuries and buying less liquid assets. The intention was to provide more liquidity for markets that were in urgent need of them (Blinder, 2010). During the second
engagement, the Treasury started borrowing in advance of its needs while depositing the excess funds at the Central Bank (CB). This was an attempt to enable the Fed to purchase more securities without increasing bank reserves, but the Fed started lending the funds to try to rescue the Bear Stearns. But then Lehman Brothers failed and the monetary policy changed dramatically (Blinder, 2010).
The actual QE started in November 2008, when the announcement was made that the Federal Reserve would purchase up to $100 billion in agency debt and up to $500 billion in mortgage backed securities (MBS) (Gilchrist & Zakrajsek, 2013). The assets of the Federal Reserve rose from $907 billion on September 2008 to $2.214 trillion on November 12, 2008. Simultaneously to this growth in assets, the bank reserves grew from $11 billion to $860 billion. Since the capital barely changed, the Fed’s leverage ratio rose from 22:1 to 53:1 (Blinder, 2010). The first round of purchases was concluded early in 2010 and is labeled as QE1.
The second round of LSAPs (QE2) began after the announcement in August 2010. The FOMC wanted to re-invest the principal payments of the agency debt and mortgage backed securities into longer-term Treasury securities (Fratzscher & Straub, 2016). The goal of this re-investment was to stabilize the amount of securities held by the Federal Reserve. In November 2010 the LSAPs were committed to add $600 billion US treasuries to its balance sheet, making treasury purchases program the main tool of QE (Fratzscher & Straub, 2016).
The last round (QE3) started in September 2012, in which it yet again expanded the Federal Reserve’s balance sheet with purchases of mortgage-backed securities and
treasury bonds. This time, the Fed announced it would purchase $40 billion of mortage-backed securities and $45 billion in treasury bonds per month. Starting in 2013, the Fed started to reduce the amount by approximately $10 billion per month until it halted in 2014 (Fratzscher & Straub, 2016).
When the Fed started with quantitative easing, they used the LSAPs reactively. It was primarily used as a tool to rescue companies. However it soon became more
systematic (for example by setting fixed monthly repurchases) and became more focused on the market. The goal changed from rescuing faltering institutions to pushing down risk premiums, which had increased significantly due to the panic during the previous months (Blinder, 2010).
Literature review
Considerable research has been done to determine the impact of LSAPs on long-term yields. Most of these researched the announcement effect of Quantitative Easing, using high-frequency changes in bond yields. Another big part of the researches, analyzed the portfolio balance effect using lower-frequency data to test the implications. The main consensus is that the portfolio balance channel is the main transmission channel through which LSAPs work and that there is a positive relation between the maturity structure and the term premium. This translates in most studies into a lower term premium during the unconventional policies used by central banks during the crisis.
Gilchrist & Zakrajsek (2013) analyzed the impact of changes in the benchmark market interest rates prompted by the LSAP announcements on the market-based indicators of corporate credit risk. Using identification through heteroscedasticity-based approach they found that the announcements led to a reduction in the cost of insuring against defaults for corporate credits. Their results support that the large-scale asset repurchases induced a significant easing in both household and business sectors. However they did not find any measurable effect on credit risk that is specific to the financial sector. They argue that this unexpected result might be because the LSAPs flattened the yield curve, resulting in lower future profitability of financial institutions. This effect could outweigh the improved economic expectations. Another reason they argue is that the large-scale asset repurchases have eradicate the risk associated with the credit crises.
In the paper of Hancock and Passmore (2011), they also found that the announcement of the Federal Reserve’s mortgage-backed securities program
significantly enhanced market functioning. They used an empirical pricing model for the yield of mortgage backed securities and mortgage rates to measure the relative
importance of the channel through which mortgage rates were affected. Their results also suggest that the LSAPs resulted in clearer government support and better anticipation of portfolio rebalancing. Not only did the LSAPs significantly improve the portfolio rebalancing during the program, they even found a substantial effect after the program ended because of the market conditions that were improved.
Christensen & Rudebusch (2012) analyze the announcement effect of the LSAPs announcement and the decrease in government bond yields that followed in the U.S. and the U.K.. In order to decompose the decline in government bond yield into changes in expectations about future monetary policies and future term premiums, they’ve used empirical dynamic term structure models. Their result suggest that the announcements had a different effect for the U.K. and the U.S.. They found that the main effect of the Fed’s LSAPs program was through the expectations channel, while in the U.K. yield declines would have been solely driven by reduction in the term premium. The reason for these differences might be the result of different policy communication and market structure. The Fed’s forward guidance was more guided toward providing monetary policy near to the zero bound, while the BoE’s forward guidance on interest rate was absent. However they do note that there is a lot of uncertainty regarding these conclusions.
Krishnamurty and Vissing-Jorgensen (2011) evaluated the effect of the Fed’s LSAPs during QE1 and QE2 on interest rates. They use an event-study methodology to study the effect on the nominal interest rates on long-term risk-free assets using daily and intra-day data. Their results found that QE1 and QE2 reduced the nominal interest rates significantly. The main reason found for the decrease was the increase in the safety price premium of risk-free assets. The yields were reduced by more than 100 bps as a result of QE1 and QE 2 reduced by nearly 20 bps. This indicates that the effects of QE on mortgage-backed securities works through an additional channel to affect the price of mortgage-specific risk. The last founding they note is that the expected inflation increased significantly during QE1 and less in QE2, resulting in bigger reductions of real-interest rates compared to the reductions in the nominal rates.
The first article of Joyce et al (2011), studies the impact of the announcement effects of the LSAPS made on financial markets and the economy in general. Using quarterly data and a small structural vector autoregression consisting out of variables such as policy rate, ten-year spot rate, real GDP growth and CPI inflation. They estimate that the total yields of securities have fallen by 150-300 basis points and that the biggest reason for this reduction is the portfolio balance effect. However they do stress that there is a lot of uncertainty regarding the magnitude of the impact.
Joyce et al. (2012) studies the impact of the announcement effects again, but this time by investigating the direct effect of the LSAPs on yields during QE1 and QE2. They found a contradiction when comparing QE1 and QE2. In QE1 there was a significant reduction of the yields that were mainly the result of the liquidity premium effect. However the effect in QE2 was a slight rise in yields. Due to these counterfactual findings it is hard to draw a strong conclusion about the impact of QE1 and QE2. A reason might be that investors became less willing to rebalance their portfolios with riskier assets in these uncertain times. They argue that this may increase the lag and that it will take longer for the portfolio balance effect of QE to pass through into asset prices. They conclude that the impact of QE is probably the same during the two
periods, but because of the lag different results were found. They keep stressing the difficulty in assessing the impact and an improvement of understanding is needed.
Gagnon et al (2010) discusses the channels through which LSAPs can affect the economy, by explaining the variation in the term premium during 1985 until before the announcement of QE1. In order to do so they use factors related to the business cycle, uncertainty factors about the economy and the net public-sector supply of long-term debt securities in an ordinary least squared regression to determine the effects on the
variation of the term premium. The variables relating to the business cycle and
uncertainty that they use are the unemployment gap, core CPI inflation, long-run inflation disagreement and the 6-month realized daily volatility of the on-the-run 10-year treasury yield as a proxy for interest rate uncertainty. The variables relating to the net public-sector supply of long-term debt securities they use is publicly-held Treasury securities, Treasury securities held in the Fed’s SOMA portfolio and debt securities held by foreign securities. According to the paper, the main transmission channel through which the LSAPs work is the portfolio balance channel, which is in line with findings in most studies. They conclude their paper with stating that the LSAP program implemented by the Fed was successful in reducing the net supply of long-term assets and that the magnitude of the reduction in the term premium is between 30 and 100 basis points.
Empirical Model
As mentioned above, considerable research has been done to determine the impact of LSAPs on the term-premium. Most of them researched through which channels QE affected the term premium and looked at only one or two rounds of the unconventional monetary policy. However, I did not find any papers that study the effect on the term-premium over the course of the entire period during which QE was implemented, to try to answer if the results hold up when analyzing the entire period. In this paper I will try to answer that question by basing my model on one, which other economists have cited frequently and is generally accepted by the community, and adding the variable QE to the model.
The model used in this paper is based on the ordinary least squared (OLS) regression model used by Gagnon et al. (2010). The model explains the historical time variation in the term premium using a set of macroeconomic variables. The observable variables used by Gagnon et al (2010) are based on the business cycle, net public sector supply of long term securities and uncertainty about economic fundamentals. I use the same variables and use the OLS regression to estimate the partial marginal effect of Quantitative Easing on the term premium of 10-year securities. The model used by Gagnon et al (2010):
𝑡𝑝𝑡10=𝛽0+ 𝛽1∗ 𝑈𝑔𝑎𝑝+ 𝛽2∗ 𝐶𝑃𝐼 + 𝛽3𝐼𝑛𝑓𝑑𝑖𝑠 + 𝛽4∗ 𝑑𝑎𝑖𝑙𝑦𝑣𝑜𝑙 + 𝛽5∗ 𝑓𝑒𝑑 + 𝛽6∗ 𝑓𝑜𝑟𝑒𝑖𝑔𝑛 + 𝛽7∗ 𝑝𝑢𝑏𝑙𝑖𝑐
The model is estimated by obtaining monthly data over the period between January 2006 and December 2013. This timeframe is chosen, because it includes the
three successive rounds of QE as well as a period without QE. I expect that with the chosen period there will be enough observations to get reliable results to answer the research question. Before analyzing the model, I will first look at if the term-premium is reduced by comparing the mean before and after the first announcement made by the Fed.
The dependent variable is the nominal 10-year yield term-premium of Treasury yields. The data was obtained from Federal Reserve Bank of St. Louis. The value of the term-premium is expected to decline during the LSAPs that were implemented by the Fed.
As noted above, the variables are divided into three categories they are based on. The following variables are based on the business cycle:
𝑈𝑔𝑎𝑝 is the unemployment gap and is calculated by subtracting natural
unemployment rate1 from the actual unemployment rate2. The data from both the natural unemployment rate and the actual unemployment rate were obtained from the economic research website of the Federal Reserve Bank of St. Louis. For the natural
unemployment rate quarterly data was obtained, which can be used because it’s a slowly adjusting variable. Gagnon et al (2010) added the unemployment gap, because it’s a strong indicator for the economy. According to the Philips-curve, there is a trade-off between unemployment and inflation. If the actual unemployment is bigger than the natural rate, inflation will rise. So I expect a negative effect of the unemployment gap on the term premium, a lower unemployment gap will cause higher inflation. Which
increases the amount of compensation that is needed for carrying risk, this translates into a higher term premium.
The second variable used is the core CPI inflation3, which is also retrieved from the Federal Reserve Bank of St. Louis website. In order to reduce the volatility of this variable, the CPI rate is corrected for food and energy prices. I expect a positive effect of the core CPI rate on the term premium, because when interest rates rise, the cost of borrowing money increases. So when an economy is overheating, the CB will increase the interest rate to reduce the demand for money. A higher interest rate results in a higher amount of compensation needed for carrying risk, resulting in higher term premium.
1https://fred.stlouisfed.org/series/NROU#0 2https://fred.stlouisfed.org/series/NROU
3https://fred.stlouisfed.org/series/CPILFESL
As a proxy for interest rate uncertainty, the 6-month realized daily volatility of the 10-year treasury yields is used (infdis)4 and is obtained from the Federal Reserve Bank of St. Louis website. This variable was added, because the LSAPs reduced the
uncertainty about the inflation, because the Fed announced they committed to the lower-bound level for the interest rate. According to Hartman & Makin (1982), inflation
uncertainty has a negative effect on the nominal interest rate. So I expect that the certainty provided by the Fed will have a positive effect on the term premium.
The third variable included by Gagnon is the inflation disagreement and looks at how the public expects prices to change over the next 12 months. If the public
expectations are bigger than the actual price changes, the real interest rate will decline which increases inflation. I expect that the inflation disagreement will have a negative effect on the term premium. However due to the public announcements the Fed made I don’t expect this effect will be significant.
The next variables were added to capture the effects of changes in the net public sector supply of longer-term debt. All three variables are expressed as a percentage of nominal GDP:
Publicly held Treasury securities5 are debt that is owed by the government to the public. This variable is added to include the amount of asset purchases that are not obtained due to the unconventional policies implemented during the crisis. The data was retrieved from the website of the U.S. department of treasury normalized by the annual nominal GDP. The data of the nominal GDP was retrieved from the Federal Reserve Bank of St. Louis website
-Treasury securities held in the Federal Reserve’s SOMA portfolio6 is added to the model to prevent the effect of quantitative easing on the term premium to be overstated. The data was retrieved from the Federal Reserve Bank of St. Louis website and
normalized by the annual nominal GDP. Treasury securities held in the Federal
Reserve’s SOMA portfolio and the U.S. debt securities held by foreign official agencies, could have been combined together as Gagnon et al (2010) notes that the portfolio balance channel is only affected by the size of the securities held. It does not matter who holds it. Yet, I have chosen to not combine them in order to conclude if there aren’t any
4https://fred.stlouisfed.org/series/TYCSD678FRBCLE#0
5https://www.treasury.gov/resource-center/data-chart-center/tic/Pages/ticsec2.aspx 6https://fred.stlouisfed.org/series/TREAST
other factor might play a role when a different a different agent holds the securities. However, taking the fact that there should not be a difference, I expect both variables to have a similar effect on the term-premium.
U.S. debt securities held by foreign official agencies7 are obtained from Federal Reserve Bank of St. Louis website. And as noted above, this variable is added to the model to prevent the effect of quantitative easing on the term premium to be overstated
In order to answer the research question, quantitative easing was added to this model as a variable. The data used for this variable was retrieved by using the
announcements made by the Fed and is normalized as percentage of the nominal GDP, because the LSAPs have increased the price and by doing so will also lower the
compensation investors need. Another reason for this expectation is that LSAPs reduce the maturity of Treasury securities and the term premium should be reduced when maturity is reduced (Gagnon et al, 2010). Adding QE results in the model:
𝑡𝑝𝑡10=𝛽0+ 𝛽1∗ 𝑈𝑔𝑎𝑝+ 𝛽2∗ 𝐶𝑃𝐼 + 𝛽3𝐼𝑛𝑓𝑑𝑖𝑠 + 𝛽4∗ 𝑑𝑎𝑖𝑙𝑦𝑣𝑜𝑙 + 𝛽5∗ 𝑓𝑒𝑑 + 𝛽6∗ 𝑓𝑜𝑟𝑒𝑖𝑔𝑛 + 𝛽7∗ 𝑝𝑢𝑏𝑙𝑖𝑐 + 𝛽8∗ 𝑄𝐸
Results
In chart 1 can be seen that the term premium has indeed decreased (by average) during the three stages of QE. This was already known, however I decided to analyze if it actually did, decrease as to confirm the correct data is used. Compared to the period before the first announcement of QE1, the term-premium has dropped significantly due to the negative term-premium between 2011 and 2013. This might seem unnatural, as the term-premium is a compensation for the risk that comes with bearing long-term over short-term assets. Yet this could be explained by the fact that investors might rather accept a lower long-term yield in order to reduce the effect of short-term volatility.
7https://fred.stlouisfed.org/series/FYGFDPUN
Chart 1: Term-premium during the time period with means before and after the first announcement of QE.
In table 1 below, the results of the regression can be found
In contrast to the results found by Gagnon et al (2010), the unemployment gap I found are positive and significant at the one% level. This could be the result of the negative term-premium during 2011 and 2013. If the timespan used in this regression would have been longer, the relative impact of the negative term-premium might have less effect on the unemployment gap. The findings are quite illogical because in normal circumstances, a higher unemployment gap will cause a lower inflation (or even
deflation) resulting in a reduction of the term-premium.
The core CPI rate is positive as expected, however there is no evidence that that the core CPI rate explains any variation in the term-premium. This result suggests that it is highly likely that there was a misspecification of the variable or that the data wasn’t obtained properly. It is highly unlikely that the core CPI has no effect on the
term-premium. If inflation increases investors should be compensated for bearing the risk of a lower real interest rate, which in turn will translate into a higher term-premium.
In line with precious studies and the expectations is the realized volatility, which has a significant negative effect on the term-premium. Also the inflation disagreement is exactly how I thought it would be, negative yet not significant. These results were part of the goal of the unconventional tools and it seems that the Fed has succeeded. However there should be a more detailed research to find how big the impact these variables were. -1.3000 -0.8000 -0.3000 0.2000 0.7000 1.2000 28/5/05 10/10/06 22/2/08 6/7/09 18/11/10 1/4/12 14/8/13 27/12/14 N omin al 10 -yea r yield ter m p remiu m Date
Coefficient Standard Error Constant 0,874 (0,000)* 0,265 Unemployment gap 6,846 (0,041)* 3,293 Core CPI 0,033 (0,890) 0,241 Inflation disagreement -0,028 (0,445) 0,036 Realized volatility -0,391 (0,000)* 0,033
Publicly held Treasury 0,275 (0,831)
1,287
Fed’s SOMA portfolio 3,430 (0,078)*
1,926
Foreign held debt -9,153 (0,000)* 2,017 Quantitative Easing -0,351 (0,004)* 0,120 # observations 96 Adjusted R-squared 0,9038 Standard Error regression 0,159
Table 1: OLS on the term-premium during entire period.
Only two of the variables that were added to capture the effects of changes in the net public sector supply of longer-term debt seem to be significant in explaining the variance of the term-premium being the Treasury securities held in the Federal
Reserve’s SOMA portfolio and the U.S. debt securities held by foreign official agencies. The U.S. debt securities held by foreign official agencies have a significant positive effect on the term-premium as expected. This expectation is based on the fact that for the portfolio balance channel it does not matter whom purchases/sales assets, but the amount does. Publicly held Treasury securities also has a positive effect on the term-premium, however there is no evidence to believe this effect is significant. The
explanation for this might be that the amount of publicly held Treasury securities is small in comparison to the LSAPs in the timespan that is researched. Using a longer period could see the effect of publicly held Treasury securities becoming significant as was expected. The Federal Reserve’s SOMA portfolio was expected to be positive but the
results suggest that it had a significant negative effect on the variance of the term-premium during the timespan that was researched. This might be the result of a misspecification of the variable used in this model.
In this model Quantitative Easing had a significant negative effect on the term premium as expected. Purchases made during QE have increased the price and by doing so lowered the compensation investors need.
The overall model explains 90,38% of the time variation of the term premium during the researched timespan. However a lot of correlation between the explanatory variables was present (see appendix table 1). Also a lot of the variables that other papers found to have a significant effect, did not seem to be significant in this model. Regrettably I think I’ve misspecified a couple of variables during my research and in hindsight, the timespan should have been longer in order to make definitive conclusions about the effects. The short time period became especially when the entire timespan was divided into two different periods: a pre-QE period (January 2006-November 2008) and a period during and after QE (November 2008- December 2013). In the appendix can be seen that these regressions give roughly the same answers as the original regression, with the exception of more variables not contributing significantly to the model.
Conclusion
In this paper I have tried to analyze the effect on the time variation in the term-premium using a set of macroeconomic variables in order to learn from the financial crisis in case another one happens. My analysis suggests that implementing QE during the crisis was an effective policy implemented by the fed during the financial crisis. By reducing the net supply of assets with long duration, the Federal Reserve’s LSAP
programs appear to have been successful in reducing the term-premium and stimulating economic activity. However, even without the results found in this study, the way the Fed implemented QE wasn’t the most efficient way. Especially QE1 was very reactive and did not target improving the long-term economic prospects, but was used in order to save failing companies. The Fed learned from these times and it became more
systematic and focused on the market. This is a valuable lesson for all the central banks across the globe and should not be underestimated.
In my opinion my model failed to properly estimate the effects of the used variables due to a couple of misspecifications and choosing a timeframe that was to short. However it does suggest that the monetary policy was successful in improving the economy through the term-premium. It is hard to determine the exact effect due to the periods in which QE was implemented where very short.
Appendix
Unemployment gap Core CPI Inflation disagreement Realized volatility Publicly held Treasury Fed’s SOMA portfolio Foreign held debt Quantitative Easing Unemployment gap 1.000 Core CPI -0,030 1.000 Inflation disagreement -0,315 0,2158 1.000 Realized volatility -0,850 0,328 0,084 1.000 Publicly held Treasury 0,712 -0,184 -0,173 -0,525 1.000 Fed’s SOMA portfolio 0,2430 0,075 -0,004 -0,078 0,815 1.000 Foreign held debt 0,699 -0,184 -0,141 -0,502 0,991 0,804 1.000 Quantitative Easing 0,799 -0,332 -0,480 -0,634 0,398 -0,063 0,411 1.000Coefficient Standard Error Constant -2,101 (0,340) 2,164 Unemployment gap 33,074 (0,051)* 16,133 Core CPI 0,079 (0,787) 0,290 Inflation disagreement 0,256 (0,013)* 0,096 Realized volatility -0,080 (0,548) 0,131
Publicly held Treasury 0,444 (0,895)
3,323
Fed’s SOMA portfolio 32,107 (0,022)*
13,208
Foreign held debt 0,552
(0,944)* 7,815 # observations 34 Adjusted R-squared 0,5484 Standard error of regression 0,120
Coefficient Standard Error Constant 2,935 (0,000)* 0,444 Unemployment gap 0,175 (0,956) 3,126 Core CPI -0,035 (0,899) 0,277 Inflation disagreement 0,069 (0,178) 0,051 Realized volatility -0,518 (0,000)* 0,045
Publicly held Treasury -2,191 (0,128)
1,415
Fed’s SOMA portfolio -3,084
(0,151)
2,115
Foreign held debt -1,628
(0,463) 2,203 Quantitative Easing -0,588 (0,002)* 0,182 # observations 62 Adjusted R-squared 0,9471
Standard Error regression 0,138
References
Blinder, A. S. (2010). Quantitative Easing: Entrance and Exit Strategies Federal Reserve Bank of St. Louis Review, 92(6), 465-479.
Christensen, J. H., & Rudebusch, G. D. (2012). The response of interest rates to US and UK quantitative easing. The Economic Journal, 122(564), F385-F414.
Kim, D. H., & Orphanides, A. (2007). The bond market term premium: what is it, and how can we measure it?. BIS Quarterly Review, June.
Gagnon, J., Raskin, M., Remache, J., & Sack, B. P. (2010). Large-scale asset purchases by the Federal Reserve: did they work?. FRB of New York Staff
Report, (441).
Krishnamurthy, A., & Vissing-Jorgensen, A. (2010). The Effects of Quantitative Easing on Long-term Interest Rates. Kellogg School of Management working paper. Rudebusch, G. D., Sack, B. P., & Swanson, E. T. (2006). Macroeconomic implications of
changes in the term premium. FRB of San Francisco Working Paper, (2006-46).
Fratzscher, M., Lo Duca, M., & Straub, R. (2016). On the international spillovers of US quantitative easing. The Economic Journal.
Gilchrist, S., & Zakrajšek, E. (2013). The Impact of the Federal Reserve's Large‐Scale Asset Purchase Programs on Corporate Credit Risk. Journal of Money, Credit
and Banking, 45(s2), 29-57.
Bauer, M., & Rudebusch, G. (2011). Signals from unconventional monetary policy. FRBSF Economic Letter, 36.
Bauer, Michael D. and Rudebusch, Glenn D., The Signaling Channel for Federal Reserve Bond Purchases (April 3, 2013). International Journal of Central
d’Amico, S., English, W., López‐Salido, D., & Nelson, E. (2012). The Federal Reserve's Large‐scale Asset Purchase Programmes: Rationale and Effects. The Economic Journal, 122(564).
Hancock, D., & Passmore, W. (2011). Did the Federal Reserve's MBS purchase
program lower mortgage rates?. Journal of Monetary Economics, 58(5), 498-514. Hartman, R., & Makin, J. H. (1982). Inflation uncertainty and interest rates: Theory and empirical
tests.
Joyce, M. A., McLaren, N., & Young, C. (2012). Quantitative easing in the United Kingdom: evidence from financial markets on QE1 and QE2. Oxford Review of Economic Policy, 28(4), 671-701.
