The effect of quantitative easing on the term premium in the UK

The Effect of Quantitative Easing on

the Term Premium In the UK

Bachelor Thesis

Jules Bronk 10156690

Table of Contents

Introduction……….……3

Monetary Transmission Channels……….….5

Literature review………..7

Empirical model………11

Results……….17

Conclusion……….………18

Introduction

As the damage caused by the 2007-2008 financial crisis intensified, the monetary and fiscal authorities realised that conventional monetary policy alone could not prevent further economic downfall as the short-term interest rate had already reached its effective zero lower bound. In normal times, monetary policy is executed through setting the official short-term interest rate, or Bank Rate, at which the Bank of England (BoE) lends to financial institutions and thus influencing the cost of borrowing, stimulating aggregate demand through investments as a result (Joyce et al, 2012). The level of interest influences the interest rates set by commercial banks and other financial institutions for their depositors and borrowers. It also affects the exchange rate and asset prices, such as government bonds and shares. UK government bonds are referred to as gilts from now on.

The BoE’s Monetary Policy Committee (MPC) loosened monetary policy by cutting the Bank Rate down by 3 per cent at the end of 2008 and lowering it by another 1.5 per cent in the beginning of 2009 (Joyce et al, 2011). As the MPC was concerned that nominal spending would still be too low, the interest rate was lowered to its effective zero lower bound of 0.5 per cent and has it

maintained that level up to now (Joyce et al, 2011). Unfortunately, setting the Bank Rate at its zero lower bound did not boost nominal spending enough to meet the BoE’s 2 per cent CPI inflation target in the medium term (Joyce et al, 2011). With the interest rate at its zero lower bound, conventional monetary policy became ineffective and forced the BoE to introduce a number of unconventional measures (Joyce et al, 2012).

The main unconventional tool the BoE engaged in was a large-scale asset purchase programme, mostly consisting of UK government bonds, financed by money created electronically by the BoE, also known as central bank money (Kapetanios et al, 2012). This process is known as quantitative easing, or QE, and is executed under the ‘Asset Purchase Facility’ programme (Red Book, 2014). The aim of the QE programme was to inject a large amount of money into the economy to boost nominal spending, increase liquidity and prevent inflation from undershooting. This is achieved by buying an enormous amount of gilts,

which will bid up their price, causing an increase in the net wealth of the holders of those gilts. The first round of quantitative easing was implemented between March 2009 and January 2010, totalling for an asset purchase worth £199.1 billion, consisting for £198.3 billion out of mostly long-term gilts (Joyce et al, 2012). But as the financial situation failed to pick up enough, it became clear that the inflation target might not be reached and the BoE executed another round of QE purchasing £125 billion of gilts (Joyce et al, 2012). After a short pause in large scale asset purchases, the BoE purchased another £50 billion worth of gilts, completed at the end of 2012 gilts (Joyce et al, 2012). In total, the BoE has spent over £375 billion on assets purchases divided over the three rounds of QE between March 2009 and July 2012 (Joyce et al, 2012). The asset purchases, accounting for around 25% of annual GDP, mainly consist of gilts.

Due to the fact that the Bank Rate has been at its effective zero lower bound since March 2009, conventional monetary policy has lost its power as the BoE was no longer able to influence the economy by setting the Bank Rate. Through QE, the BoE raised the price of long-term assets and therefore lowered their yields, making current capital spending projects more attractive as the cost of borrowing decreases (Christensen & Rudebusch, 2012).

This paper tries to answer the question whether the asset purchases by the BoE have had a significant effect on lowering the term premium. According to the expectations hypothesis, the expected return from holding a long-term bond is equal to rolling over multiple short-term bonds with a total maturity equal to that of the long-term bond (Kim, 2007). However, the expectations hypothesis ignores the interest rate risk involved with holding a long-term bond. Investors may require compensation for bearing the risk of any deviation from the expectation hypothesis (Kim, 2007). This compensation is called the term premium.

First, I will discuss how QE is implemented by the BoE and its impact on the economy by reviewing the monetary transmission channels through which the asset purchases will have their effect. The following section provides a literature review of papers on quantitative easing. Next, an Ordinary Least Square (OLS) regression is performed to estimate the partial marginal effect of QE on the term premium. The OLS regression is similar to the regression Gagnon et al (2010) performed and includes variables like the unemployment gap, core

Figure 1

Source: Joyce et al (2011)

inflation, quantitative easing, foreign holding of gilts and the Consumer Confidence Indicator.

Monetary Transmission Channels

Most existing literature analyses the effect of quantitative easing on the financial markets through the monetary transmission channels. Joyce et al (2011) focuses on the first round of QE during the period from March 2009 to January 2010. They make a clear distinction between transmission channels working through asset prices and other channels (Joyce et al, 2011). The asset price channel is affected through policy signaling effects, portfolio balancing effects and liquidity premia effects. The other channels are affected by confidence and bank lending effects (see figure 1).

The BoE purchased assets mostly consisting out of gilts, worth a total of £375 billion pounds with the intention of improving the economic turmoil (Joyce et al, 2011). By purchasing financial assets, the prices of those assets are pushed upward, increasing the wealth of their holders. As the prices are pushed

upwards, yields are pushed downwards causing decreasing borrowing costs and thus stimulating expenditure (Joyce et al, 2012). When assets and money are not perfect substitutes, holders will buy other assets that are closer substitutes than money. This effect is known as the portfolio balancing effect and has been specified as the most important channel through which the BoE operates (Kapetanios et al, 2012). A higher degree of substitutability between assets increases the impact of this channel, as money will become less attractive due to

relatively higher transaction costs (Joyce et al, 2012). The BoE primarily purchased long-term assets with maturities ranging between 5 and 25 years, because these assets are often held by non-bank financial institutions like pension funds and insurance companies (Joyce et al, 2012). By increasing their wealth, the BoE stimulates these institutions to invest in riskier, short-term assets boosting the prices and lowering the yields of assets more broadly (Red Book, 2014). The portfolio balancing channel also decreases the duration risk of long-term assets as a smaller amount of long-term assets will be in the private market (Joyce et al, 2012). The duration risk consists of the term premium that is given to the holder of the asset for bearing the interest rate risk.

Due to the financial crisis investors have become more aware of the fact that, unlike gilts, some asset classes have become more difficult to sell, as investors have become more risk averse. Therefore, investors would normally demand a higher return on their assets to compensate for this risk (Joyce et al, 2012). By purchasing assets the BoE has increased the liquidity on the financial markets, actively stimulating trade and preventing the return on assets from increasing. This effect is known as the liquidity premia effect. Still, this effect is believed to have had only a minor impact, as this effect will only prevail while monetary policy is executed and liquidity premia are relatively low on gilts (Joyce et al, 2011).

Even though the asset purchase program is not actively set up to signal future policy, it indicates that the BoE is determined to boost demand and meet inflation targets even when the Bank Rate is at its zero lower bound (Joyce et al, 2011). This has improved the overall level of expectations within the economy and prevented asset prices from falling and risk premia from rising (Joyce et al, 2011). By giving the credible signal that the BoE is committed to the inflation target, inflation expectations have been anchored. According to the Fisher equation, the real interest rate is equal to the nominal interest rate minus the inflation rate. As inflation expectations have been anchored, a rise in real interest rates is prevented (Joyce et al, 2012).

The gilts obtained under the ‘Asset Purchase Facility’, are purchased on a competitive auction open to authorized financial institutions (Red Book, 2014). The prices of the gilts are determined at the end of the auction (Red Book, 2013). The gilts are ranked according to the spread of the yields and purchased by the

BoE until the desired amount has been reached. Even though the BoE purchases assets mostly from non-bank institutions, banks are still involved as they act as intermediaries in the process (Red Book, 2014). Once the central bank money is created, bank accounts are credited, who then purchase the assets from the previous holder (Red Book, 2014). The previous holders will either spent their money earned on goods or services, boosting overall expenditure, or buy other assets whose price will also rise. As mentioned before, the BoE also purchases private sector assets under the Asset Purchase Facility in order to improve liquidity and increase the flow of corporate credit (Red Book, 2014). These private sector assets however are negligible as they make up less than 1 per cent of the total assets purchased (Nelson, 2012).

Literature Review

Joyce et al (2011) focuses on the announcement effects of QE, which will influence gilt yields as the first round of large-scale asset purchases almost entirely consisted of gilts. They estimated that gilt yields in total have fallen by 100 basis points. The biggest reaction was in March 2009, accounting for a fall of 75 basis points (Joyce et al, 2011). As risk-free interest rates are sensitive to macroeconomic news announcements and risk-free rates only dropped by less than 10 basis points due to the asset purchases, most of the fall in gilt yield can be assigned to portfolio rebalancing effect instead of announcement effects (Joyce et al, 2011). Later QE announcements effects have had a smaller effect on gilt yields. This may be caused by the fact that QE at some point was seen as conventional monetary policy (Joyce et al, 2012). The announcement reactions on the prices of assets other than gilts and corporate bonds have proven to be smaller and less obvious. Even though it is hard to make an accurate estimate of the QE announcements effects due to numerous other influences and lags involved, Joyce et al (2011) suggests that households’ net financial wealth has increased by approximately 16 per cent (Joyce et al, 2011).

In a subsequent paper, Joyce et al (2012) tries to identify the impact of QE again through announcement effects and investigates the direct effect of asset purchases on gilt yields and other asset yields during the first two rounds of QE. They account a large fall in gilt yields during the first round of QE mainly due to

the large reduction of liquidity premia, the increased scarcity and changes in the macroeconomic outlook (Joyce et al, 2012). During the second round of QE the announcement effects had a much smaller impact on gilt yields. In fact, it caused a slight rise in yield (Joyce et al, 2012). Still QE proved to be a strong instrument, as the direct asset purchase effect of QE2 did not have a smaller impact on gilt yields. Joyce et al (2012) acknowledges that it is complicated to separate the effect of QE on other asset prices from other factors driving asset prices. But the overall event-study results provide proof that QE also influences other asset prices. QE1 caused a fall of high-yield corporate bond yields of 150 basis points, while QE2 caused a fall of 40 to 50 basis points (Joyce et al, 2012). The

narrowing of the yield spreads on corporate bonds and long-term interest rates is about 75 basis points and is in line with QE reducing default risk for firms and longer-term interest rates (Joyce et al, 2012). The default risk of firms is reduced due to the fact that the fall of corporate bonds yields causes a rise of the price of those bonds and thus a net increase in wealth. The size of the overall reaction of QE2 is smaller due to the fact that this time the asset purchases were smaller, more anticipated and there were growing concerns about the euro area, which increased risk aversion (Joyce et al). The increased risk aversion will decrease the effect of the portfolio balance channel, as investors are less willing to

reinvest in riskier assets. Concluding, there are no precise results on the impact of QE1 and QE2 as it is impossible to separate the effects of the asset purchases from other policies conducted, but there is no reason to believe that the direct impact of QE on nominal demand is different between the two rounds of asset purchases (Joyce et al, 2012). There is a significant difference between the announcement effects of both rounds (Joyce et al, 2012).

Kapetanios et al (2012) focus on the macroeconomic impact of QE1 on output and inflation through a time-varying parameter vector auto regression (VAR), a change-point structural VAR and a large Bayesian VAR. Their approach includes a counterfactual analysis in order to compare the outcome of QE1, based on the findings of Joyce et al (2011), to the state of the economy if the BoE did not engage in the asset purchase program (Kapetanios et al, 2012). The counterfactual forecast is used as the baseline instead of the actual levels of GDP and inflation, as the actual levels might include the effects from other factors (Kapetanios et al, 2012). Also, it is assumed that the yield on long-term gilts

would have been 100 basis points higher without QE. The timeframe used for the BVAR model covers monthly data from April 1993 to September 2010 and

comprises 43 variables. The other two models include monthly and quarterly data from 1963 to 2011 (Kapetanios et al, 2012). Unfortunately, there are also some weaknesses to this approach. With the use of a counterfactual analysis comes uncertainty. It is also undefined through which of the monetary transmission channels QE is transmitted in these models (Kapetanios et al, 2012). All models indicate that without QE inflation would be well below the inflation target of 2 per cent or even negative. Output would have experienced a decline between 1.2 and 2.6 per cent (Kapetanios et al, 2012). The ranges of these estimates are affected by varying the different assumptions.

Nelson (2012) analyses nine aspects of the recent large-scale asset purchases in the United Kingdom. Firstly the paper discusses the chronology of the QE program and acknowledges that buying longer-term assets, which the private sector will exchange for short-term assets as they are better substitutes than money, will put downward pressure on the term premium of longer-term assets. Next, Nelson analyses the assets purchased in the programs. The UK program distinguishes itself by purchasing mostly government bonds. The amount of corporate bonds purchased can be neglected, as the purchases

consisted for more than 99 per cent out of gilts. The scale of the large-scale asset programs is put in perspective by comparing it to the BoE’s total assets to get a better idea of the scale of the operation (Nelson, 2012). Nelson also mentions that the purchases were unsterilized in both countries, which means that monetary base is increased. The historical background and institutional

background may have motivated the recent policies conducted by the BoE. The view that long-term interest rates and term structure determination can be influenced date back to the immediate postwar period, where the UK authorities performed open market operations in the longer-term bond markets (Nelson, 2012). However, there is no clear evidence that the postwar open market operations have had an effect on the long-term interest rate as opposed to QE. This difference may be caused due to the lack of postwar data or the fact that at the time short-term interest rates were not at the effective zero lower bound which makes it harder to isolate the effect of the open market operations on the long-term interest rate (Nelson, 2012).

Christensen & Rudebusch (2012) try to separate the impact of the

signaling and portfolio balance effect in the UK through a series of dynamic term structure models (DTSM). For both countries, bond yields are primarily affected by the portfolio balance model as the central banks purchases bonds. The

purchases push up the prices of bonds and their substitutes, and lower the yield and term premiums. Part of the decline in long-term yields may also be assigned to the signaling channel. The relative impact of the signaling and portfolio

balance channel depends on the financial market structure and monetary policy conducted (Christensen & Rudebusch, 2012).By using a DTSM, change in gilt yields can be split up into term premium and short-term interest rates

components (Christensen & Rudebusch, 2012). The DTSM points out that the long-term yield component is responsible for about a quarter of the decline in gilt yields. The portfolio balance channel is thus responsible for three quarters of the decline in gilt yields and is the primary channel through which the QE

program works (Christensen & Rudebusch, 2012).

Next, Christensen & Rudebusch (2012) investigate cross-country yield responses between the US and the UK as US asset purchase announcements might have an impact on the UK gilt yields. The results point out that the UK transmission channels are not susceptible to US announcements (Christensen & Rudebusch, 2012). For symmetry, they also analyzed the response of US bond yields on UK announcements. The US interest rates also indicate little reaction to the UK QE announcements (Christensen & Rudebusch, 2012). This is in line with the finding that the gilt yields are mainly affected through the term premium as the signaling channel is of little importance.

Gagnon et al (2010) explains the historical variation in the term premium using an ordinary least square regression model using monthly American data over the period January 1985 to June 2008. It ends just before the first

announcements of the large-scale asset purchase (LSAP) program. The OLS regression is discussed and reproduced in the next section, only this time on data from the UK and including the large-scale asset purchase program. Before they present the model with the preliminary found results they explain the

mechanisms through which the asset purchases influence the economy. They find that the primary channel through which the LSAP’s work is the term premium, which is affected by the portfolio balance effect as explained in the

previous section (Gagnon et al, 2010). In order for investors to be willing to sell their assets and acquire new, short-term, risk-free bank reserves, the yield of their current holdings must drop. As the Federal Reserve purchases assets from the private sector, indeed their yield drops and price rises. The portfolio balance effect should not only cause a decrease in long-term yields, but also spill over into the yield on other assets (Gagnon et al, 2010). As the prospective on agency debt decreases, investors should bid up the prices and thus also lower the yield on corporate bonds (Gagnon et al, 2010). Concluding, Gagnon et al (2010) finds that reducing the net supply of long-term assets should successfully reduces the term premium somewhere between 30 and 100 basis points.

Empirical Model

Following Gagnon, Raskin, Remache and Sack (2010), I use a set of

macroeconomic variables to estimate the impact of the MPC’s large-scale asset purchase programme on the term premium of 10-year gilts. The term premium is a component the risk premium, which is most sensitive to asset purchases (Nelson, 2012). The term premium on 10-year gilts is calculated by subtracting monthly short-term interest rate from the monthly interest rate on 10-year gilts. The short-term interest rate used in this case equals the Bank Rate as the Bank Rate forms the expectations about the future course of short-term interest rates. In order to understand fluctuations of the long-term interest rate, it is necessary to investigate the fluctuations of the term premium. The BoE has removed an incredible number of long-term assets from the financial markets. With less duration risk left to bear, the term premium should decrease (Gagnon et al, 2010). I therefore expect a negative relationship between the term premium and quantitative easing.

The variables used in this paper are not identical to the ones used in the Gagnon model. The observable variables used by Gagnon et al (2010) are based on the business cycle, the net public-sector supply of longer-term securities and uncertainty about economic fundamentals (Gagnon et al, 2010). They perform an OLS regression to estimate the partial marginal effect of quantitative easing on the term premium of 10-year treasuries:

Where is the nominal 10-year yield term premium, which is calculated by subtracting the short-term yield1 _{from the yield on 10-year Treasury yields}2_.

This paper will use the yield on 10-year gilts instead. is the set of observable variables that will be discussed below and is an error term. The timespan for this model is from January 2005 to December 2013, which contains substantial periods in which QE was both active and not active. All three rounds of

quantitative easing are thus included in the model.

1. UGAP = Ut - UNAIRU

Gagnon et al (2010) included the unemployment gap in the model since it’s a strong economic indicator, strongly correlated with the business cycle, which therefore also has a big impact on the interest rate. Higher unemployment means more slack in the economy, which causes a lower price pressure. As the price pressure is low, there is a smaller fluctuation of the real interest rate and thus a lower term premium. On 12 February 2014 the MPC declared that the Bank Rate remains at its current level of 0.5 per cent to stimulate the economy until the unemployment rate has fallen below 7 per cent (Bank of England, 2014). is the monthly unemployment rate3 _{and is the yearly Non Accelerating}

Inflation Rate of Unemployment4_{. The yearly value of the NAIRU can be taken}

since it is a slowly adjusting variable. When unemployment is at NAIRU, there is just enough unemployment in the economy to keep the inflation from deviating from the target inflation. The NAIRU depicts both the trade off between

unemployment and inflation and acknowledges that there will always be

1_Source: http://www.bankofengland.co.uk/statistics/Documents/rates/baserate.pdf 2_Source: http://www.ecb.europa.eu/stats/money/long/html/index.en.html 3_Source: http://www.ons.gov.uk/ons/datasets-and-tables/dataselector.html?cdid=LF2Q&dataset=lms&tableid=9 4_{Source: http://stats.oecd.org/Index.aspx?QueryId=36376#}

structural unemployment. The trade off between unemployment and inflation is summarized by the Phillips Curve, which can be written as

Which shows that inflation will stay at the target level as long as cyclical unemployment is equal to zero and there are no supply shocks. I expect a

negative relation between unemployment and the term premium due to the fact that higher unemployment will cause lower inflation. Lower inflation results in a decrease of future compensation for bearing risk and thus a lower term

premium.

2.

The core inflation rate5 _{can be calculated by taking the CPI rate corrected for}

food and energy prices. Food and energy are excluded since these goods tend to be very volatile, causing excess movement in the inflation rate. Core inflation is included in the model since it is strongly correlated with the interest rate. As interest rates are increased, consumers tend to have less money to spend as borrowing becomes more expensive. With lower spending, the economy slows down, resulting in a lower inflation. With economic growth, the demand for money will increase. In order to prevent an economy from growing too fast, causing even higher inflation, the government will increase the interest rate. I therefore expect a positive relationship between the level of inflation and the term premium.

3. Consumer Confidence Indicator6

According to Gagnon et al, one of the factors that influence the historical

variation of the term premium is the business cycle (Gagnon et al, 2010). Apart from the unemployment gap, which is strongly correlated with the business

5_{Source: http://www.ons.gov.uk/ons/taxonomy/index.html?nscl=Price}

cycle, I choose to include the Consumer Confidence Indicator (CCI) as well since consumer confidence will have an impact on the overall level of spending. The Consumer Confidence indicator is based on the following questions7_:

- How does the financial situation of your household compare with what it was 12 months ago?

- How do you think the financial position of your household will change over the next 12 months?

- How do you think the general economic situation in this country has changed over the last 12 months?

- How do you think the general economic situation in this country will develop over the next 12 months?

- Do you think there are any benefits in people making major purchases at present time?

4. & 5. Gilts purchased by foreign official agencies8 _{and held by the Bank of}

England9

As Gagnon et al (2010) points out, the portfolio balance effect is not affected by who makes the asset sales or purchases. The size of the purchases however, does matter. The response of private investors should not differ, when the purchases are executed by the BoE or any foreign agency (Gagnon et al, 2010). The holding of gilts by the BoE in this case does not include the gilts purchased under the Asset Purchase Facility program. If these variables are not included in the model, the partial marginal effect of QE by the BoE on the term premium will be too high. For the BoE’s holding of gilts, all monthly gilt holdings are added and normalised by annual nominal GDP10_{. The time series on gilts purchased by}

foreign officials is quarterly, so all values are divided by three and also 7_{Source: http://www.bloomberg.com/quote/UKCCI:IND} 8_Source: http://www.dmo.gov.uk/reportView.aspx?rptCode=D5N&rptName=9c60062a-08b9-4170-9eda-485d6d240cd6%7C%7CGILT%20MARKET%20(3)&reportpage=Gilts/Overseas_Holdings 9_Source: http://www.dmo.gov.uk/reportView.aspx?rptCode=D4L&rptName=2010e6cb-09d3-4e90-88e6-a8485546d5d9%7C%7CGILT%20MARKET%20(9)&reportpage=Gilts/Government_holdings

normalised by nominal GDP.

6. One-year ahead inflation expectation11

One of the goals of QE has been to anchor the expected inflation to the inflation target (Kapetanios et al, 2012). If expected inflation falls below the target, real interest rate rises causing a higher real cost of borrowing and vice versa (Joyce et al, 2012). This variable is comparable to the long-run inflation disagreement used in Gagnon et al (2010), except that the range of expected inflation differs. The expectation is based on the question of how the public expects prices to change over the next 12 months. This survey is carried out on behalf of the BoE.

7. QE12

For this variable, the time series of the gilt purchases under the Asset Purchase Facility is used. The time series starts at the 11th _{of March 2009. This is the}

variable we are most interested in and will give an estimate whether QE has had a significant partial effect on the term premium. All daily purchases within a month are added in order to retrieve the monthly amount spent on gilts. The total amount spent by the Asset Purchase Facility at the end of 2013 is divided over 26 months and is approximately £382 billion in total. The monthly amount spent on gilts is, just like all other debt variables, normalised by nominal GDP. The purchases of corporate bonds by the BoE are negligible, as the large-scale asset purchases by the BoE consist for more than 99 per cent out of gilts (Nelson, 2012). I expect a negative relationship between QE and the term premium as its aim is to place downward pressure on longer-term interest rates (Nelson, 2012).

8. 6-month realized daily volatility of the 10-year gilt yields13

Gagnon included this variable to capture the interest rate uncertainty. To

11_{Source: http://www.bankofengland.co.uk/publications/Pages/other/nop.aspx} 12_{Source: http://www.bankofengland.co.uk/markets/Pages/apf/gilts/results.aspx}

calculate the 6-month realized daily volatility, the standard deviation of the daily gilt yields of the previous 5 months plus the daily yields of that current month is calculated. Interest rate uncertainty is included in the model due to the fact that a higher interest rate uncertainty will cause a higher compensation for bearing this risk and this a higher term premium.

9. Public debt14

The general government can be divided into four subsectors: social security funds, local government, state government and central government. The public debt is the amount owed by the government, which has built up to now and is excluding the temporary effects of financial interventions. This variable is also normalised by nominal GDP.

Putting all the variables together will give the following regression function:

14_Source:

http://www.ons.gov.uk/ons/datasets-and- tables/index.html?pageSize=50&sortBy=none&sortDirection=none&newquery=national+debt&content-type=Reference+table&content-type=Dataset

Results

Regression January 2005 - December 2013

Regression January 2005 - December 2013, robust

Regression January 2007 – December 2013, robust

Conclusion

The coefficient on the UGAP variable turned out to be positive, which is

contradictionary to the findings in Gagnon et al (2010). The positive coefficient indicates that a higher level of unemployment is associated with a higher term premium, while higher unemployment naturally causes lower inflation and thus a lower term premium as the inflation risk decreases. One reason for this

unexpected outcome is that we have already corrected for inflation in this model. Another possible reason for this positive correlation is the negative term

premium around 2005/2006. A negative term premium might sound illogical as the term premium compensates the risk for holding longer-term assets instead of short-term assets. But holders of longer-term assets could accept a lower long-term yield than the short-long-term interest rate to avoid short-long-term interest

volatility. Independent strategic consultancy firm Oxford Analytica (2005) wrote that prospects were that housing prices were expected to fall, the dollar was weakening and an economic stagnation of the euro-area was lurking, which should lead to an expected increase in short-term interest rate risks. It is likely that the long-term interest rate rises relatively less due to the fact that it is expected that the economy will eventually pick up again. This could possibly explain the negative risk premium. Under results there are two regressions excluding 2005 and 2006. The coefficient on the UGAP variable does become weaker, but is still very strong and positive. By increasing the timespan of the regression, the negative term premium of 2005 and 2006 will have a relative

smaller impact on the unemployment gap which might lead to a negative relationship between the two variables as expected.

The coefficients on inflation and the Consumer Confidence Indicator validate the expectations and literature reviewed that both have a positive relationship with the term premium. Increased inflation will lead to an increased inflation premium to compensate for bearing additional risk of lower real

interest rates and thus a higher term premium. A higher CCI is likely to increase consumer expenditure, which will lead to higher inflation and a higher term premium.

For the foreign holdings of gilts, the BoE’s holdings of gilts and the QE variable it was expected that all negative coefficients were expected as the portfolio balance effect is not affected by who makes the asset sales or purchases (Gagnon et al, 2010). The strength of the coefficients should differ as the MPC purchased a considerable larger amount of gilts under the Asset Purchase Facility program. However, the coefficient on foreign holdings turned out to be very strong and negative. It is so strong due to the fact that asset purchases from non-resident investors result in lower yields and thus a higher price. As the price increases relatively more for non-resident investors, their net wealth will

increase more too, which results in a stronger portfolio balance model effect and a lower term premium. This outcome is contradictionary to the literature

reviewed. The coefficient on QE is very small and not significant, which means that in this model QE has little or no effect on the term premium component of the long-term interest rate. It could still have an effect on the long-term interest rate through the average expected level of short-term interest rates over the maturity of the gilts. The coefficient on the holdings of the BoE is positive instead of negative. Apparently, the regular purchases of gilts by the BoE affect the term premium through inflation. The regular holdings of gilts are used to control the money supply and thus the inflation. As the holdings of gilts increases, the money supply also increases, causing a rise in inflation and of the term premium as the theory predicts.

The expected inflation is an important variable which turned out to be very small and insignificant for both the normal and the robust regression. For the expected inflation data, a survey where they investigated one-year ahead inflation expectations of consumers performed on behalf of the BoE, was used.

As the term premium is determined by the markets, the expected inflation of the financial markets would have been a better option to prevent this variable from being insignificant.

The coefficient on the 6-month realized daily volatility of the 10-year gilt yields turned out to be positive as expected, but insignificant. According to this model, the volatility of the 10-year gilts does not play an important role.

The coefficient on the public debt is strong and positive. As the government issues mostly long-term gilts to finance excess government expenditure, the price of those long-term gilts will decrease and increase their yield more than the yield of short-term assets. As the long-term interest rises, the term premium will also rise.

References

Bank of England. (2014). The Bank of England’s Sterling Monetary Framework,

the ‘Red Book’.

http://www.bankofengland.co.uk/markets/Documents/money/publications/redb ook.pdf

Bank of England. (2014). Inflation report February.

http://www.bankofengland.co.uk/publications/Documents/inflationreport/201 4/ir14febeconrec.pdf

Christensen, J, Rudebusch, G, (2012), ‘The response of interest rate to US and UK quantititive easing’, The Economic Journal, no. 122 (November).

Gagnon, J, Raskin, M, Remache, J, Sack, B, (2010), ‘Large scale asset purchases by the Federal Reserve: Did they work?’, Staff Report, no. 441.

Joyce, M, McLaren, N, Young, C, (2012), ‘Quantitative easing in the United Kingdom: evidence from financial markets on QE1 and QE2’, Oxford Review of

Economic Policy, vol 28(4), pp. 671-701.

Joyce, M, Miles, D, Scott, A, Vayanos, D, (2012), ‘Quantitative easing and unconventional monetary policy – an introduction’, The Economic Journal no. 2012-122.

Joyce, M, Tong, M, Woods, R, (2011), ‘The United Kingdom’s quantitative easing policy: design, operation and impact’, Bank of England Quaterly Bulletin 2011 Q3, vol 51(3), pp. 200-12.

Kapetanios, G, Mumtaz, H, Stevens, I, Theodoridis, K, (2012), ‘Assessing the economy-wide effects of quantitative easing’, The Economic Journal, no. 2012-122.

Kim, D, (2007), ‘The bond market term premium: what is it, and how can we measure it?’, BIS Quarterly Review, June 2007.

Krishnamurthy, A, Vissing-Jorgensen, A, (2011), ‘The effects of quantitative easing on interest rates: channels and implications for policy’, National Bureau of

Economic Research, working paper No. 17555.

Nelson, E, (2012), ‘Key aspects of longer-term asset purchases programs in UK and US monetary policy’, Oxford Economic Papers, 2013-65.