The effects on expected inflation, by the purchasing of treasury
securities and MBS by the Federal Reserve.
Casimir G. Oostlander Thesis Economics Bsc Universiteit van Amsterdam
Roetersstraat 11 1018 WB Amsterdam
Februari 2013
Index
INTRODUCTION ... 3
RELATED LITERATURE ... 7
MECHANISM OF QUANTITATIVE EASING ... 7
CENTRAL BANKS ... 9
YIELDS ... 9
INFLATION ... 10
STATISTICAL MODEL ... 11
DATA AND TIMELINE ... 11
MODEL ... 12 CONTROL VARIABLES ... 13 RESULTS ... 14 MODEL ... 14 TREASURY SECURITIES ... 15 MBS ... 16 CONTROL VARIABLES ... 17 ROBUSTNESS CHECK ... 18 TREASURY SECURITIES ... 19 MBS ... 19 CONCLUSION ... 21 DISCUSSION ... 22 BIBLIOGRAPHY ... 23 2
Introduction
From December 2008 till August 2009 the Fed (Federal Reserve, the central bank of the United States) initiated the first QE (quantitative easing) program with the
purchase of treasury securities and MBS (mortgage backed securities). The second QE program was announced by the Fed in November 2010, it would purchase an additional 600 billion dollars of treasury securities by the end June 2011, according to the Fed the QE program will reduce long-term interest rates and thereby increase economic growth and inflation over time.1 By the end of the second QE program the Fed’s balance sheet had increased by 1.25 trillion dollars. In September of 2012, the Fed announced its third round of QE, this involves the Fed buying an additional 40 billion dollars of MBS and 45 billion dollars of treasury securities a month till the end of 2013, figure 1 illustrates the rapid growth of MBS and treasury securities held by the Fed on its balance sheet. In the beginning of 2014 the Fed began slowly tapering the QE programs.
Figure 1 Y-axe USD (millions), X-axe timeline. Source: Federal Reserve
1
http://www.federalreserve.gov/newsevents/press/monetary/20081125b.htm
3
The Fed had to present a new monetary policy because as shown in figure 2 presented below, the FFR (Federal Funds Rate) was close to the zero bound, the FFR is an important monetary policy measure, which can affect the economic growth and inflation over time. If the FFR is close to the zero lower bound the Fed is unable to influence economic growth or inflation by changing the FFR. The Fed had to introduced an unconventional monetary policy measure to counter the effects of the US financial crisis in the period 2007 - 2009.2 This policy was referred to as QE, which is an OMO (Open Market Operation) aiming at lowering long-term interest rates by buying long-term securities.3 Besides QE the Fed also provided unlimited liquidity supply to banks and investment banks. QE was first adopted in March 2001 by the BoJ (Central bank of Japan) to increase inflation.4
Figure 2 Y-axe Rate (percentage), X-axe timeline. Source: Fed of St. Louise
Former Fed chairman Ben Bernanke describes QE used in the United States as ‘credit easing’.5 This is different from QE that was implemented by the BoJ. Both the BoJ and the Fed expanded the reserves they hold on the liability side of the balance sheet, but the BoJ bought mostly treasury securities, which it holds on the asset’s side of the balance sheet. In contrast the Fed bought treasury securities and MBS that it holds on the assets side of the balance sheet, this differs from the BoJ
because the BoJ only purchased treasury securities. According to Fawley and Neely 2 http://www.nber.org/cycles.html 3 http://www.timeline.stlouisfed.org/index.cfm?p=faq#16 4 http://www.imes.boj.or.jp/english/publication/edps/2002/02-E-03.pdf 5 http://www.federalreserve.gov/newsevents/speech/bernanke20090113a.htm 4
(2013) the Fed was able to lower long-term interest rates through the use of the QE program this is elaborated in chapter ‘related literature’.
The Federal Reserve act states the main goals of the Fed: ‘to promote effectively the goals of maximum employment, stable prices, and moderate long-term interest rates.’6 With lending from the capital markets in the United States dysfunctional after the financial crisis of 2007 – 2009, the Fed had to improve the liquidity of the capital markets by lowering long-term interest rates to increase economic output, inflation and bank liquidity. To achieve this they had to buy long-term securities, which are traded on the capital markets, since conventional monetary policy measures could not be used anymore.
In the United States companies borrow through the capital markets instead of bank loans. When the demand of capital markets dropped for long-term securities a purchase in long-term securities is unlikely to have the same impact for banks as lending the same amount to banks, which banks can invest in long-term securities. According to Shleifer and Vishny (2009) a capital injection in a bank will not solve the liquidity problems because the capital will be used to acquire more distressed
securities, such as MBS, which have troubled collateral. A more mainstream view would argue that a combination of capital injection and the QE program has helped to reduce the capital problems of banks. As a consequence the Fed had to buy long-term securities to increase their prices through an OMO, so that the balance sheets of banks would increase and lending can restart again. An OMO can increase the money supply measure M2, which can increase inflation in the US according to the equation of exchange by Friedman (1987). With QE increasing the money supply measure M2 by large quantities to increase economic growth, there is a possibility of also a large increase in inflation.
6
http://www.federalreserve.gov/pf/pdf/pf_2.pdf
5
This thesis tries to find an effect by QE on expected inflation. Expected inflation is chosen as dependent variable and not inflation because it is the expected increase or decrease of overall prices within a future timespan and it enables central banks such as the Fed to plan future monetary policy measures. With the QE program the Fed tried to increase inflation over time and thereby prevent deflation, also lower long term interest rates enhances economic growth. To elaborate this a relationship needs to be found between QE program and expected inflation. The main objective of this thesis is to analyze the effects on expected inflation, by the purchasing of treasury securities and MBS by the Federal Reserve. The next chapter will discuss related literature further on a statistical model will be presented. The chapter ‘results’ will present the main statistical findings of this thesis. The last two chapters
‘conclusion’ and ‘discussion’ will summarize the main arguments and give recommendation for future research.
Related literature
As mentioned in the previous chapter, QE is a relative new subject within monetary policy measures, first in this chapter the supposed mechanism of QE will be
explained. To evaluate the subject further, this chapter will summarize the
developments in research regarding QE. Besides presenting the related literature this chapter will give arguments why this thesis gives a renewed insight in QE.
Mechanism of quantitative easing
The Fed adjusts the short-term interest rate by the FFR. The FFR is the rate that banks charge each other for short-term loans. The Fed uses an OMO to increase or decrease the money supply to the economy to maintain the federal funds target rate.7 A change in FFR is passed on to other short-term interest rates but also to long-term interest rates of assets like MBS and treasury securities. Long-term
interest rates, including those of MBS and treasury securities, generally followed the short-term interest rates. But in the financial crisis of 2007-2009 the economy began to contract, which led to a decrease in employment. The risk of deflation rose, so the Fed reacted by decreasing the short-term interest rates to enable companies to borrow cheaper and increase economic growth and inflation. By the end of 2008 short-term interest rates were close to zero bound.8 However long-term interest rates such as 30-year maturity treasury securities and 30-year conventional mortgaged rate did not follow the short-term interest rates such as 1-year maturity treasury securities, in the beginning of the financial crisis, see the figure 3 below.
Figure 3 Y-axe rates in percentage, X-axe timeline. Source: Fed of St. Louis
7 http://www.federalreserve.gov/monetarypolicy/openmarket.htm 8 http://www.federalreserve.gov/newsevents/press/monetary/20081125b.htm 7
The Fed could no longer influence long-term interest rates by FFR, with short-term rates near there zero bound level. This meant that the Fed could no longer prevent deflation or enhance economic growth. Fed had to devise an unconventional
monetary policy measure that could decrease long-term interest rates. The Fed used with QE an OMO to buy securities from the security holder. The Fed pays the sellers bank with newly created electronic money, the security is then transferred to the Fed. The security seller can retrieve the funds from the bank or leave it there. A large-scale security purchase such as QE can affect the price and interest rate of the security. To obtain more insight in how this works the relation between price and interest rate needs to elaborate further. If the demand for securities is high because the Fed buys large amounts the price will increase. Interest rate and the price of a bond have an inverse relationship if the price increases this will mean that the interest rate will decrease. According to Fawley and Neely (2013) the Fed was able push down long-term interest rates through the use of QE program. The long-term interest rates fell and businesses and households were able to borrow at lower rates and assets values increased.9 The United States economy slowly recovered from the financial crisis in August 2009 see figure 4 below.
Figure 4 Y-axe GDP percentage change, X-axe timeline. Source: Fed of St. Louis
9
http://www.federalreserve.gov/newsevents/press/monetary/20090128a.htm
8
Central banks
Research has also been conducted about QE programs at other central banks for example Yoshiyuki and Kozo (2013) revealed a relationship between interest rates and expected inflation, influenced by QE implemented by the BoJ. The authors give a recommendation for future research: it is necessary to pay more attention to the central bank’s balance sheet. According to Joyce, Mclaren and Young (2012) QE by the Bank of England (BoE) is more similar to that of the Fed, the BoE also focused on buying large quantities of mainly long-term government bonds and related assets. Joyce, Mclaren and Young (2012) found that the purchase of securities decreased long-term interest rates by around 1 percent in England.
Yields
To further understand the QE program set up in the United States research has to be conducted for not only QE implemented by the BoJ and BoE, but also for the Fed. Fawley and Neely (2013) show a relationship between QE and yields. This thesis tries to find a relation between QE and expected inflation. To make a statement about expected inflation, a relationship between QE and yields needs to be identified. If the purchase of treasury securities and MBS has an effect on yields, then they can also affect expected inflation by reducing borrowing costs and enhancing economic growth. Different papers show a reduction in yields by QE as anticipated by the Fed. Joyce, Miles, Scott and Vananos (2012) find a reduction of 1 percent in the 10-year treasury securities yields, Meier (2009) argues that there is an effect on treasury securities yields by QE of 0.60 percent and Meaning and Zhu (2011) find smaller effect of 0.50 percent. It is clear that there is an effect on yields by QE, in the next paragraph the effect of QE on inflation and expected inflation will be addressed.
Inflation
Joyce, Tong and Woods (2011) present a model, which can be used to estimate the effects of QE on inflation. The authors conclude that the first QE program by the BoE increased inflation by between 0.75 and 1.5 percent. This thesis is focused on the effects on expected inflation. The paper of Krishnamurthy and Vissing (2011) describes evidence for effects of QE on market expected inflation. They do so by evaluating the increase or decrease in yields that could be caused by QE and compare these yield changes to the market expected inflation. They use the difference between treasury securities yield and TIPS (treasury inflation protected securities) yield as dependent variable, to estimate the effects of QE on market expected inflation. Chapter ‘empirical model’ will further explain TIPS. Krishnamurthy and Vissing (2011) find an increase of 0.96 percent of market expected inflation by QE and is statistically significant with a p-value lower than 5 percent. The data of TIPS spread suggest that purchasing program QE of the Fed in long-term assets had a large and positive effect on market expected inflation.
This thesis will try to give a renewed sight on QE by evaluating the effects on expected inflation. The related literature mentioned above describes the effects of QE on interest rates, yields and market expected inflation. For this thesis expected inflation is chosen because central banks such as the Fed use it to evaluate the use of future monetary policies measures, which have many implications on the
economy. Also another new aspect is that this thesis uses the change of treasury securities and MBS on the balance sheet of the Fed, this increase is easier to measure than for example interest rate changes by QE announcements.
Statistical model
This chapter will mention the data and timeline that is used. Also the statistical model is presented, which could support an effect of QE or to be more specific treasury securities and MBS on expected inflation.
Data and timeline
The relevant data was found with DataStream and at the Fed database all data was selected in the timespan January 2003 till July 2013. This was done, because Fed only supplies treasury security and MBS data from January 2003 and when writing this thesis data earlier than December 2013 was not available yet. The data between July 2013 and December 2013 could not be used as data points because the GDP data from DataStream was available till July 2013. MBS and treasury securities are denoted as weekly data, but both real and nominal GDP is denoted in quarterly data. All data had to be recalculated to monthly data. First for MBS and treasury securities the monthly average was calculated by summing up the four weekly data points and divided them through number of weeks that is four to calculate the monthly average. Second the quarterly real and nominal GDP data was recalculated to monthly data points by using linear interpolation, by dividing the difference between quarterly data points through number of months that is three this will give a monthly average. The monthly average is then added to the first quarterly value for the first month, the second month will have twice the monthly average plus the first quarter and the third month will be the second quarter. For real GDP the growth was needed, the
difference between the first month and the second month was divided through the first month and then multiply with 100, which gave the growth of real GDP in percentages. Converting quarterly and weekly data to monthly data was chosen because the dependent variable was given in monthly data points.
Model
The recalculated monthly data points were then implemented in the statistical model with the control variables and dummy variable, the statistical model and variable information is represented below:
𝑦𝑦𝑡𝑡= 𝑦𝑦𝑡𝑡−1+ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑣𝑣𝑣𝑣𝐶𝐶𝑣𝑣𝑣𝑣𝑣𝑣𝐶𝐶𝑣𝑣𝑣𝑣𝑡𝑡 + 𝛽𝛽6�𝑁𝑁𝑔𝑔𝑁𝑁 𝐺𝐺𝐺𝐺𝐺𝐺𝑔𝑔𝑔𝑔𝑔𝑔 �𝑡𝑡+ 𝛽𝛽7�𝑁𝑁𝑔𝑔𝑁𝑁 𝐺𝐺𝐺𝐺𝐺𝐺𝑁𝑁𝑚𝑚𝑚𝑚 �𝑡𝑡+ 𝛽𝛽8(𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑦𝑦 )𝑡𝑡×
�𝑁𝑁𝑔𝑔𝑁𝑁 𝐺𝐺𝐺𝐺𝐺𝐺𝑔𝑔𝑔𝑔𝑔𝑔 �
𝑡𝑡+∈𝑡𝑡
• Y: expected inflation rate (monthly) (percentage) Michigan consumer index
• MBS: mortgage backed securities held outright by the Fed (monthly) Federal Reserve • Gov: U.S. treasury securities: notes and bonds, held outright by the Fed (monthly) Federal
Reserve
• Nom GDP: nominal gross domestic product (monthly) U.S. Bureau of Economic Analysis • QEDummy:
1 = Quantitative easing (1,2,3) 0 = no quantitative easing
The first variable (𝑦𝑦𝑡𝑡) is the dependent variable of the model, which denotes the expected inflation rate in percentages. The data reflects a survey of the monthly expected inflation under United States citizens conducted by the Michigan University research department. The second variable (𝑦𝑦𝑡𝑡−1) is a lag term because inflation expectation highly depends on the previous period inflation expectation. The control variables will be addressed at the end of this chapter. The variable for treasury securities is � 𝑔𝑔𝑔𝑔𝑔𝑔
𝑁𝑁𝑔𝑔𝑁𝑁 𝐺𝐺𝐺𝐺𝐺𝐺�𝑡𝑡, treasury securities are the sum of US bills, notes and bonds
bought by the Fed. Both treasury securities and MBS needed to be normalized because the nominal value is non-stationary. Non-stationary variables makes forecasting unreliable, in order to receive consistent, reliable results, the
non-stationary data needs to be transformed into non-stationary data.10 A stationary variable is when the probability distribution of that variable does not change overtime,
treasury securities and MBS are made stationary by divided them through nominal GDP because nominal GDP is a stationary variable.
10
Stock, & Watson. (2012). Introduction to Econometrics. In Stock, & Watson, Introduction to
Econometrics (p. 577 - 590). England: Pearson.
12
The variable for MBS is denoted by � 𝑁𝑁𝑚𝑚𝑚𝑚
𝑁𝑁𝑔𝑔𝑁𝑁 𝐺𝐺𝐺𝐺𝐺𝐺�𝑡𝑡in the model. (𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑦𝑦 )𝑡𝑡×
�𝑁𝑁𝑔𝑔𝑁𝑁 𝐺𝐺𝐺𝐺𝐺𝐺𝑔𝑔𝑔𝑔𝑔𝑔 �
𝑡𝑡 Denotes the variable as earlier described treasury securities but now a
QE dummy is added, which is one for the time period of December 2008 till July 2013 in which QE was implemented. For MBS a QE dummy was not added because the Fed did not buy MBS before the QE program was implemented, which was in December 2008.
Control variables
The control variables are added to the model because they adjust for business cycles and are a possible influence on expected inflation. Different economic factor such as real GDP growth and FFR can be used as control variables. For example the FFR affects economic growth and expected inflation when the FFR is lowered, borrowing becomes cheaper and tends to increase borrowing, which increase
demand. A higher demand with a stable supply will increase prices economic agents anticipate this higher demand, which leads to an increase in expected inflation. The same logic holds for a change in consumer confidence and real GDP. For the control variable unemployment the Philips curve could be used because it is a tradeoff between unemployment and inflation, a lower unemployment, higher income, higher demand, upwards price pressure leads to a higher rate of inflation and expected inflation.
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑣𝑣𝑣𝑣𝐶𝐶𝑣𝑣𝑣𝑣𝑣𝑣𝐶𝐶𝑣𝑣𝑣𝑣𝑡𝑡= 𝛽𝛽1𝐹𝐹𝐹𝐹𝐹𝐹𝑡𝑡+ 𝛽𝛽2𝑈𝑈𝐹𝐹𝑡𝑡+ 𝛽𝛽3𝐼𝐼𝑡𝑡+ 𝛽𝛽4𝐶𝐶𝐶𝐶𝑡𝑡+ 𝛽𝛽5𝐹𝐹𝑣𝑣𝑣𝑣𝐶𝐶 𝐺𝐺𝑄𝑄𝐺𝐺 𝑡𝑡 • FFR: effective federal funds rate (monthly) (percentage) Federal Reserve
• UR: unemployment rate (monthly) (percentage) U.S. Bureau of Labor Statistics • I: consumer price index (monthly) (percentage) U.S. Bureau of Labor Statistics
• CC: consumer confidence (monthly) (index with 1985=100) The Conference Board, Inc • Real GDP: gross domestic product (monthly) (growth rate) U.S. Bureau of Economic Analysis
Results
In this chapter, the results that were found for the model and its variables will be discussed also the possible implications of these results will be explained. All the p-values and coefficients that are referred to can be found in table 1. The p-value is the smallest significance level at which the null hypothesis can be rejected, in this thesis a boundary of 0.05 is used. The null hypothesis states that the variable has no effect on the dependent variable.11 The coefficient indicates the effect of the independent variable on the dependent variable.
Table 1. OLS regression for expected inflation.
Model
To indicate if the statistical model is able to find an effect between treasury
securities, MBS and expected inflation several tests for the model such as 𝐹𝐹2 and the F-test can be used. The 𝐹𝐹2 is 0.741, which indicate that 74.1% of the variation in expected inflation is explained by the independent variables of the model. The p-value of F-test the statistical model has a p-value of 0.00, which indicates that the independent variables of the model have an effect on expected inflation and that the model has explanatory powers. Next the variables of interest, treasury securities and MBS will be discussed.
11
Stock, & Watson. (2012). Introduction to Econometrics. In Stock, & Watson, Introduction to
Econometrics (p. 113). England: Pearson.
_cons 4.391278 1.444623 3.04 0.003 1.53002 7.252535 Gmult .025479 .0092634 2.75 0.007 .0071317 .0438263 Mcor .0034715 .0051619 0.67 0.503 -.0067522 .0136952 Gcor -.033426 .0126404 -2.64 0.009 -.0584618 -.0083901 RP .0885248 .0852657 1.04 0.301 -.0803547 .2574043 C .0004982 .0051033 0.10 0.922 -.0096096 .010606 F -.0177233 .0459588 -0.39 0.700 -.1087506 .0733039 U -.3115883 .1398061 -2.23 0.028 -.5884919 -.0346847 I .1944532 .0565083 3.44 0.001 .0825314 .306375 Y1 .5736195 .086637 6.62 0.000 .4020241 .7452149 Y Coef. Std. Err. t P>|t| [95% Conf. Interval] Robust Root MSE = .43325 R-squared = 0.7406 Prob > F = 0.0000 F( 9, 116) = 33.03 Linear regression Number of obs = 126
14
Treasury securities
For the variable treasury securities the following coefficient and p-value was found a coefficient of -0.033 and a p-value of 0.009, these results indicate that the purchase of treasury securities by the Fed has an effect on the expected inflation. Normally if the demand for treasury securities increases the US government can borrow at a lower rate, also consumers and banks will benefit from the increasing treasury securities prices, because price and interest rates have an inverse relationship. The increases in demand for treasury securities will eventually lead to economic growth and expected inflation to change. In this thesis a negative coefficient was found but this effect is very small. A possible explanation could be that the Fed decreased the purchase of treasury securities before the financial crisis, this could cause the
negative coefficient in this model. The Fed bought less treasury securities to stabilize inflation because the economy was overheating. The decrease was when the
financial crisis hit the US economy indicated by the gray area in figure 5 below.
Figure 5 Y-axe Treasury securities in USD (millions), X-axe timeline source: Fed
To look at the effect of QE, a dummy variable is added to the variable treasury securities. The QE dummy is one if the Fed implements QE if not it is zero. The QE dummy has the value of one from December 2008, which is the start of the QE program by the Fed. The variable for treasury securities purchased under QE, has a coefficient of 0.025 and p-value of 0.007, which indicates an effect of treasury
securities purchased under QE on expected inflation. It supports the effect of
treasury securities purchases by the Fed on expected inflation when the QE program was implemented. The demand for treasury securities bought in the QE program is beneficial for the US government and bondholders, because of increasing treasury securities prices. The increase in these prices will lead to more income and a increasing demand, which eventual leads to economic growth and higher expected inflation.
MBS
The variable MBS has a coefficient of 0.003 and is not statistically significant with a p-value of 0.503, which is an unexpected outcome. This indicates that MBS does not have an effect on expected inflation with a significant level of 0.05. It could be that consumers consider the effect of MBS to be smaller on expected inflation than treasury securities and thereby do not consider an effect of MBS.
Control variables
The control variables are added to the model because they adjust for business cycles. The control variables itself do not necessarily need to have an effect on expected inflation.12 The control variables of the model are further discussed, and possible effects will be elaborated. Lagged expected inflation is statistically
significant by a p-value 0.000 and has a coefficient of 0.574 indicating a positive effect. For inflation a p-value of 0.000 is found which indicates that inflation has an effect on expected inflation, this effect can be explained by high inflation today being related with high expected inflation tomorrow. Unemployment has a p-value of 0.028 indicating that it is statistically significant, unemployment has a negative effect on the expected inflation with a coefficient of -0.312, which indicates that the increase of unemployment could decrease expected inflation. This effect can be explained by the Philips curve, which implies that lower unemployment will lead to higher income, higher demand, upwards price pressure and will lead to a higher rate of inflation and expected inflation. Control variable consumer confidence has a p-value of 0.922 and consumer confidence has a small positive effect on expected inflation, with a
coefficient of 0.0005. Consumer confidence being not statistically significant is interesting because consumer confidence and expected inflation both are measured upon consumer emotion, if consumers feel that the economy is likely to grow
consumer confidence and expected inflation could both increase. The FFR is not a statistical significant variable with a p-value of 0.700 and has a coefficient of -0.177. The negative coefficient can be explained by a lower FFR implies a decrease in savings, increase in demand which will lead to higher inflation. Real GDP has a p-value of 0.301 and a coefficient of 0.088, given the p-p-value has no statistical significant effect on expected inflation.
For the variables of interest an effect by treasury securities and treasury securities purchased under QE was found on expected inflation, but MBS did not have an effect on expected inflation. Based on these findings the QE program did effect expected inflation, but only trough buying treasury securities.
12
Stock, & Watson. (2012). Introduction to Econometrics. In Stock, & Watson, Introduction to
Econometrics (p. 271 - 276). England: Pearson.
17
Robustness check
In this chapter a new dependent variable is introduced, which can indicate if the conclusions are robust. All the p-values and coefficients that are referred to can be found in table 2. If some parts of the model are changed but the core is left intact and if the model still presents a relationship between expected inflation, MBS and
treasury securities with supporting coefficients and p-values, the model is robust. The newly introduced dependent variable is the yield spread between TIPS and treasury securities. TIPS are a treasury security that is corrected for indexed inflation. Holders of TIPS do not suffer when inflation increases and an increase in inflation does not affect their real return on the investment. TIPS can reflect market expected inflation. To do so yields of TIPS are subtracted from the yields of treasury securities with the same maturity.13
𝑣𝑣𝑠𝑠𝐶𝐶𝑣𝑣𝑣𝑣𝑠𝑠𝑡𝑡= 𝑆𝑆𝑠𝑠𝐶𝐶𝑣𝑣𝑣𝑣𝑠𝑠𝑡𝑡−1+ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑣𝑣𝑣𝑣𝐶𝐶𝑣𝑣𝑣𝑣𝑣𝑣𝐶𝐶𝑣𝑣𝑣𝑣𝑡𝑡 + 𝛽𝛽6�𝑁𝑁𝐶𝐶𝑄𝑄 𝐺𝐺𝑄𝑄𝐺𝐺�𝑔𝑔𝐶𝐶𝑣𝑣
𝑡𝑡+ 𝛽𝛽7�
𝑄𝑄𝑣𝑣𝑣𝑣 𝑁𝑁𝐶𝐶𝑄𝑄 𝐺𝐺𝑄𝑄𝐺𝐺�𝑡𝑡
+𝛽𝛽8(𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑄𝑦𝑦 )𝑡𝑡× �𝑁𝑁𝑔𝑔𝑁𝑁 𝐺𝐺𝐺𝐺𝐺𝐺𝑔𝑔𝑔𝑔𝑔𝑔 �𝑡𝑡+∈𝑡𝑡
• Spread = (yield treasury securities –yield TIPS)
• TIPS: 10-Year Treasury Inflation-Indexed Security, Constant Maturity (monthly) (Percentage) Federal reserve
• Treasury securities: 10-Year Treasury Constant Maturity Rate (monthly) (percentage) Federal reserve
Table 2, OLS regression with market expected inflation.
13
Carlstrom, & Fuerst. (2004). Expected inflation and TIPS. Federal Reserve Bank of Cleveland.
_cons .3326245 .4172983 0.80 0.427 -.4938874 1.159136 Gmult .0084087 .002748 3.06 0.003 .002966 .0138514 Mcor -.0031074 .001959 -1.59 0.115 -.0069874 .0007727 Gcor -.01064 .0037615 -2.83 0.006 -.0180901 -.0031899 RP .0648315 .0237008 2.74 0.007 .0178891 .1117739 C .005552 .001864 2.98 0.004 .0018602 .0092438 F -.0135339 .0173317 -0.78 0.436 -.0478616 .0207938 U -.0092805 .0422797 -0.22 0.827 -.0930208 .0744597 I .0316209 .0160431 1.97 0.051 -.0001544 .0633962 TIPS1 .8522884 .0647943 13.15 0.000 .7239551 .9806217 TIPS Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 21.4834825 125 .17186786 Root MSE = .14737 Adj R-squared = 0.8736 Residual 2.51930436 116 .021718141 R-squared = 0.8827 Model 18.9641782 9 2.10713091 Prob > F = 0.0000 F( 9, 116) = 97.02 Source SS df MS Number of obs = 126
18
Treasury securities
Table2 shows an OLS with the above-defined spread, the coefficients and p-values of the control variables will not be reviewed, because to have a robust model the effects of the MBS, treasury securities and treasury securities purchased under QE on the dependent variable must not change. The variable treasury securities had a coefficient of -0.011 and a p-value of 0.006, with a 0.05 significant level it would indicate an effect of treasury securities on market expected inflation.
The variable treasury securities purchased under QE has a coefficient of 0.008 and a p-value of 0.003, which does support an effect of treasury securities purchased under QE with a significant level of 0.05. This result is not different from the result found in the model with expected inflation. Adding a QE dummy to the variable treasury securities does not make a difference in terms of statistically significant effects for expected inflation or market expected inflation. The implementation of QE by the Fed for financial markets is evenly important as the general purchase of treasury securities by the Fed. The finding of treasury securities purchased under QE do corresponded with the effect of QE found by Krishnamurthy and Vissing (2011) on market expected inflation.
MBS
The variable MBS has a coefficient of -0.003 and a p-value of 0.115, this shows that the MBS does not have an effect on market expected inflation with a significant level of 0.05. This result does not differ from the result found for MBS with expected inflation, financial market and consumers have the same expectations for inflation regarding the purchases of MBS by the Fed in the QE program.
With a robustness check the following results were found, treasury securities and treasury securities purchased under the QE program did have an effect but MBS did not have an effect on market expected inflation. In table 3 below the results of
market expected inflation are compared with the results of the model with expected inflation, the results do not differ from the results found for expected inflation this implies that the model is robust for treasury securities, treasury securities purchased under QE and MBS.
Expected inflation Market expected inflation
Treasury securities Effect Effect
Treasury securities purchased under QE Effect Effect
MBS No effect No effect
Table 3
Conclusion
Chapter ‘introduction’ stated the main objective of this thesis, which is to analyze the effects on expected inflation by the purchasing of treasury securities and MBS by the Federal Reserve. The chapter ‘conclusion’ will summarize the most important
findings of this thesis to answer the main objectives as stated above.
The thesis introduced the subject of QE as an unconventional monetary policy measure of the Fed that was an OMO focused on buying treasury securities and MBS to reduce long-term interest rates, and thereby increasing economic growth and inflation over time. The papers from Joyce, Miles, Scott and Vananos (2012) and the paper of Meier (2009) have shown that the Fed has effectively lowered the long-term interest rates with the QE program also Krishnamurthy and Vissing (2011) found an effect of QE on market expected inflation. This thesis sets itself apart from previous research by researching the effect on expected inflation by using
macroeconomic variables and using the stock of Fed assets as independent variable.With the QE program the Fed tried to increase inflation over time and thereby prevent deflation, also lower long-term interest rates enhances economic growth. A statistical model was set up and tested to find a relationship between found QE program and expected inflation. To find a relation between expected inflation, treasury securities and MBS a statistical model was created, testing this model resulted in the following results. For the variables treasury securities and treasury securities purchased under QE a statistically significant effect was found but for MBS no statistically significant effect was found for expected inflation. Based on these findings the QE program did affect expected inflation, but only trough buying treasury securities. This could be explained by a purchase of treasury securities having a larger effect on the economic growth and expected inflation than a
purchase of MBS. With a robustness check the following results were found, treasury securities and treasury securities purchased under QE did have an effect but MBS did not have an effect on market expected inflation. The results do not differ from the results found for expected inflation this implies that the model is the robust for
treasury securities, treasury securities purchased under QE and MBS. The Fed is able to affect expected inflation and market expected inflation by buying treasury securities when QE was implemented, this way the Fed can control both expected inflations and achieve their goals of stable prices.
Discussion
This last chapter presents possible improvements for the model and a
recommendation for further research regarding the discussion surrounding QE. With the Fed tapering its QE program it could be interesting to investigated if expected inflations also decreases. Also the robustness check could be improved by
identifying the liquidity risk of the TIPS. According to Carlstrom and Fuerst (2004) TIPS would overstate market expected inflation by 0.5 to 1.0 percent, because treasury securities must compensate investors for inflation and liquidity risk. In this thesis the TIPS were used for estimating market expected inflation for robustness purposes to calculate the liquidity risk was too complex. In this thesis a statistically significant effect for expected inflation by purchasing MBS was not found, future research could focus on identifying the effect of MBS on other macroeconomic variables, also future research could look at the effect of the QE program on economic growth in the US.
Bibliography
Fawley, and Neely. (2013). Four Stories of Quantitative Easing. Federal Reserve
Bank of St. Louis Review .
Friedman, M. (1987). quantity theory of money. The New Palgrave: A Dictionary of
Economics .
Glick and Leduc. (2011). Central bank announcements of asset purchases and the impact on global financial and commodity markets. Journal of International Money
and Finance .
Joyce, Mclaren and Young. (2012). Quantitative easing in the united kingdom:
evidence from financial markets on QE1 and QE2. Oxford review of economic policy Joyce, Miles, Scott and Vananos. (2012). Quantitative easing and unconventional
monetary policy - an introduction. The economic journal.
Joyce, Tong and Woods. (2011). The United Kingdom's quantitative easing policy : design, operation and impact. Quarterly Bulletin Bank of England .
Keller, G. (2011). Managerial Statistics.
Krishnamurthy and Vissing. (2010). The Aggregate Demand For Treasury Debt.
Journal of political economy .
Krishnamurthy and Vissing. (2011). The Effects of Quantitative Easing on Interest Rates:Channels and Implications for Policy. National Bureau of Economic Research Martin and Milas. (2012). Quantitative easing: a skeptical survey. Oxford review of economic policy.
Meaning and Zhu. (2011). The impact of recent central bank asset purchase programmes. Bank of International Settlements Quarterly Review .
Meier. (2009). Panacea, Curse, or Nonevent? Unconventional Monetary Policy in the United Kingdom. IMF Working Papers .
Pilbeam, K. (2006). International finance. In K. Pilbeam, nternational finance (pp. 409-447). US: Palgave.
Shleifer and Vishny. (2009). Banking and Securitization. American Economic Review Stein, J. (2012). Monetary policy as financial stability regulation. The Quarterly
Journal of Economics.
Yoshiyuki and Kozo. (2013). Policy commitment and market expectations: Lessons learned from survey based evidence under Japan's quantitative easing policy. Japan
and the World Economy .
