1
UNIVERSITY OF AMSTERDAM
Faculty of Economics and Business
Bachelor Thesis:
Unconventional Monetary Policy at the Zero Lower Bound: Forward
Guidance and its Effect on Interest Rate Expectations
Matteo Ruozzo
BSc in Economics and Business
Supervisor: Christiaan van der Kwaak
25/10/2014, Amsterdam
2
Abstract:
Since the onset of the 2008 financial crisis, many central banks have been forced to keep target rates at their effective lower bound in an attempt to provide stimulus and foster recovery. As a consequence, monetary policy makers faced the loss of the ordinary tool of monetary policy transmission, manipulation of the aforementioned rates. This resulted in the necessity to adopt new, unconventional methods of monetary transmission to provide further stimulus. This paper focuses on one of these methods, namely forward guidance, a communication instrument used by central banks to disclose their stance on future monetary policy. By providing such indications, central banks hope to put downwards pressure on longer-term interest rates, hence lowering the cost of credit and improving financial conditions at the zero lower bound. Through this paper, we attempt to assess the effectiveness of forward guidance by analyzing whether this policy can effectively modify market expectations about the future path of interest rates. We examine the impact of forward guidance statements released under the zero-lower bound constraint both trough a literary review of previous studies and our own empirical research. Specifically, we try to verify whether there is a significant downward change in interest rates and yields for two types of financial assets, namely three-month futures on interbank rates and government bonds yields, after announcement releases. We rely on intraday data to conduct our analysis. We conclude that forward guidance can, and in several cases has been able to influence the expected path of future interest rates downwards. However, the policy has not always been successful and several factors can hamper its effectiveness.
3
Table of Contents
Abstract ... 2
1 Definition and Theoretical Framework ... 4
1.1 The Rationale Behind Forward Guidance ... 5
1.2 Theoretical Framework in the New Keynesian Model ... 5
1.2.I Optimal policy without the Zero Lower Bound... 7
1.2.II Optimal policy at the zero-lower bound ... 7
1.2.III The role of Forward Guidance at the Zero Lower Bound ... 8
1.3 The Time-Inconsistency Problem ... 9
2 Historical Remarks ... 10
3 Literary Review of Previous Empirical Findings ... 11
3.1 Words versus Actions, the Effect of Speech on Agents’ Expectations ... 11
3.2 The Effect of Forward Guidance on Interest Rates Volatility ... 15
3.3 The Reduced Sensitivity to Macroeconomic News ... 15
4 An Empirical Study of Central Bank Forward Guidance Statements at the Zero Lower
Bound. The Case for the FED, the ECB and the Bank of England ... 17
4.1 The Federal Reserve ... 18
4.2 The European Central Bank ... 21
4.3 The Bank of England ... 22
4
Introduction
A scenario where target interest rates hit the zero lower bound, so that the central bank can no longer stimulate the economy through their reduction, was widely regarded as a theoretical curiosity in the past. After the onset of the 2008 financial crisis, however, many central banks have been forced to keep target rates at their effective lower bound for a prolonged period, in an attempt to provide monetary stimulus and promote recovery. Once target rates reached the zero lower bound, the central bank lost its power to conduct monetary policy in the traditional way of altering their level through open market operations. Given that target rates cannot be lowered any further, a growing concern of monetary policy makers has become finding new, unconventional methods of policy accommodation.
Among the proposed methods, large-scale asset purchases, also known as quantitative easing, and forward guidance were the most widely adopted since the financial crisis. In this paper, we choose to focus on forward guidance, a
communication instrument used by central banks to disclose their stance on future monetary policy. At the zero lower bound, forward guidance statements aim clarify the future behavior of target rates set by the central bank. In many cases, central bank try to deliver the promise that target rates will be kept low in the future, even when economic conditions would dictate otherwise. If the statements are understood and believed by the public, expectations about future short- to medium-term rates, which are closely related to target rates, will be influenced downwards. Accordingly, and in line with the expectation hypothesis of the term structure of interest rates, current long-term rates will also decrease.
Thus, we decide to investigate the effect played by forward guidance on market expectations about the future path of interest rates. Our research question is the following: “Do forward guidance statements issued under the zero lower bound constraint successfully manage to influence the expected path of future interest rates downwards?” We attempt to find an answer both through a literature review of previous findings and through our own empirical research on the topic. The structure of the paper is as follows: In Section 1 we provide a definition of forward guidance and analyze its theoretical effectiveness in the New Keynesian model. We also turn to the time-inconsistency problem which may hinder the policy effectiveness. Section 2 looks at the history of forward guidance, both during ‘normal times’ as well as during periods of zero lower bound constraints. In Section 3 we conduct a literature review of previous studies. Lastly, in section 4 we perform an empirical study on forward guidance announcements made by the Federal Open Market Committee, the Bank of England and the European Central Bank, and compare our conclusions with those of previous findings. Section 5 concludes.
1 Definition and Theoretical Framework
This section provides a definition of forward guidance and outlines the theoretical rationale behind the adoption of this policy, building on Campbell’s (2012) primer based on the New Keynesian model. In section 3 and 4 we attempt to assess whether this framework is confirmed by empirical evidence.
Forward Guidance is a tool used by central bankers to communicate, through explicit public statements, the future course of monetary policy and of the central bank’s policy interest rate. At the zero lower bound, forward guidance turns into a promise to keep rates lower for an extended period. If the central bank can be trusted to deliver the promise to keep rates ‘lower for longer’, agents shift their expectations about future nominal interest rates and commercial banks’ willingness to lend at lower rates increases. Moreover, if lower rates are expected to bring inflation above target, a perceived reduction in real rates will follow accordingly (Williams, 2014). This overall reduction stimulates the economy by encouraging consumer spending rather than saving, particularly on expensive and durable goods, and by boosting business investment. Note that forward guidance is a policy that enhances transparency through better communication. As such, the policy is not limited to periods of
5
zero lower bound constraints on rates and was adopted by many central banks well before the beginning of the crisis, not specifically aiding at additional monetary stimulus.
Forward guidance announcement can consist in numerical guidelines about the forward path of interest rates, or less quantitative verbal statements (Den Haan, 2013). Moreover, in recent years, a distinction between two types of forward guidance-aimed statements has been introduced. Delphic forward guidance statements are free of any commitment and merely constitute a forecast of future macroeconomic performance as predicted by the central bank (Campbell et al., 2012). Such statements aim to lower the uncertainty of markets by feeding them information that is more accurate. To the extent that markets believe the central bank information to be superior, their expectations about future interest rates will be modified accordingly (Campbell et al., 2012). On the contrary, Odyssean forward guidance is the central bank’s public promise to engage in a future course of action, resulting into a binding commitment. Just like Odysseus tied to the mast to resist the sirens’ temptations, the central bank chooses to remain anchored to its own strategy by making a clear promise of future behavior, hence signaling a change in its reaction function. (Campbell et al., 2012). If the central bank promises to lift the commitment after a specific time interval, the statement is defined as time-contingent. If the commitment is linked to specific
macroeconomic goals that need to be achieved, the statement is of state-contingent type (den Haan, 2013). The concept of Central Bank reaction function, or reaction curve, will be widely used throughout this paper. This concept refers to a function that models the behavior of central banks in setting their target interest rates, considering the responsiveness of
unemployment and inflation to these rates, as modeled by Taylor’s Rule. Odyssean Forward Guidance has been playing an increasingly important role during times of zero lower bound constraints, mainly due to its ability to mitigate the time-inconsistency problem that we highlight in section 2.6.
1.1 The Rationale Behind Forward Guidance
Aside from additional monetary stimulus, authors highlight several reasons why it would be ideal for a central bank to announce its intentions about future policy rates, both during times of lower bound constraints as well as ‘normal’ times. The first of these reasons is the fact nobody, better than the central bank, should be capable of making predictions about the central bank’s future course of action. Thus, not doing so would represent a withholding of useful information that may result in
misalignments (Goodhart, 2013). A related reason is that forward guidance may facilitate the public in understanding the bank’s own strategy given the current economic outlook, known as the bank’s reaction function (den Haan, 2013). As den Haan (2013) points out, this is extremely relevant during unusual and volatile periods, when markets cannot base their predictions of central bank strategy on historical data. A third and final rationale that justifies the adoption of forward guidance is its success in reducing the sensitivity of interest rates to other news and data surprises that are not linked to the conditionality of the announcement (den Haan, 2013). This claim is empirically verified in Swanson and Williams (2012), a paper which we will review in the following section.
Forward guidance, however, does not come free of several drawbacks. Among these, Kool and Thornton (2012) highlight how forward guidance statements could disrupt financial markets when too much confidence is placed on the announced policy path, so that other relevant information is disregarded. This would result in a crowding out of private information that may otherwise be useful to policymakers for analyzing trends and making decisions. In addition, the tying of central bank’s hands to a specific commitment greatly reduces its flexibility and puts its credibility at stake, if the strategy is not pursued when better conditions dictate a different policy (Filardo and Hofmann, 2014).
1.2 Theoretical Framework in the New Keynesian Model
Having verified the existence of a sound rationale to adopt forward guidance both during ‘normal’ times as well as time of lower bound constraints, the theoretical background which underpins this instrument still needs to be clarified. This sub-section aims
6
to find an answer as to whether forward guidance is theoretically effective in providing additional monetary stimulus. For our analysis, we turn to the theoretical model outlined by Campbell (2013) and the definition of time inconsistency problem by den Haan (2013).
Campbell (2013) relies on the New Keynesian model, which summarizes the behavior of producers, households and the central bank. This is a non-stochastic model, meaning that it does not allow any room for random variation. The first equation is the New Keynesian Phillips curve, which links inflation to the current output gap and the expected inflation for the following year. This equation represents the ‘supply side’ of the economy, because it models the behavior of producers in setting their prices, as shown:
πt = κyt+ βπt+1 + mt (1)
As this equation shows, current inflation is determined by three main factors. The first factor, the dependent variable yt, is the
output gap, or the percentage deviation of actual output from its potential. We assume that a greater output gap correlates with a higher marginal cost of production, which is directly reflected on the price level. This influence is greater when prices are frequently adjusted, as reflected by the value of the coefficient κ (the Phillip’s curve slope). With perfectly rigid prices, an increase in marginal cost (caused by a change in the output gap) does not raise inflation, and κ is equal to zero. With perfectly flexible prices, an increase in marginal cost is immediately reflected on inflation, so that the coefficient κ becomes equal to infinity. The second factors that plays an influence on inflation is the expected inflation for the upcoming year (πt+1 discounted
by a factor β). The third factor are exogenous price shocks, unrelated to changes in marginal cost (the factor mt).
The second equation is an intertemporal substitution curve, which represents the demand side of the economy and shows a tradeoff between saving and consumption that is responsible for the output gap. The demand side is modeled by the following equation:
yt = – 1/σ (it – πt+1 – rnt) + yt+1 (2)
In this model, a household can choose to invest in a one year risk free bond at the nominal interest rate it and receive a return
of it – πt+1, which is the real interest rate. The term rnt is the natural rate of interest, the real interest rate that would place
economic activity at its potential and keep price levels invariable and stable, in the absence of any new shocks. Campbell (2013) assumes that total consumption is equal to total income, given that the economy has no other means of wealth accumulation. Hence, the current output gap is fully determined by the difference between current ex-ante level of real interest rates it – πt+1
and the natural rate rnt and the next period consumption gap. The parameter σ is called intertemporal elasticity of substitution
between consumption now and consumption tomorrow, it is a positive parameter so that increases in interest rates induce more savings and less consumption, resulting in a more negative output gap. Finally, a higher future consumption (positive future output gap) reduces the incentive to save today and has a direct effect on current consumption.
Lastly, we introduce a central bank loss function, which is minimized by keeping current and future deviations from zero inflation and a zero output gap to a minimum. The assumption of zero inflation is a simplification which we introduce for the sake of simplicity, given that many central banks target level of inflations which is slightly above zero (such a 2%). The loss function is the following:
L = ∑ βt(π2t + λy2t ) (3)
7
and future losses. Note that this loss function is compatible with the Federal Reserve double mandate of promoting the goal of maximum employment and stable prices. However, by setting λ equal to zero, the function models a central bank with the single mandate of inflation targeting, while setting λ equal to positive infinity models a central bank with a single mandate of stable growth. Finally, we decide to impose a restriction on interest rates, which must be strictly above zero. In fact, given that the nominal return of money is equal to zero, all bondholders would choose to substitute bonds for money if rates were to drop below zero. This restriction is represented by the following inequality:
it ≥ 0 (4)
We call this restriction the zero lower bound. The central bank controls the nominal rate of interest, which can be targeted to offset shocks to the natural rate rnt on output gap (through the IS curve) and inflation (through the Phillips curve).
1.2.I Optimal policy without the Zero Lower Bound
We can now graph the Phillips curve as a straight line crossing the y axis (inflation) at βπt+1 + mt and the x axis at the output gap
level, as shown in figure 2. The slope of the Phillips curve is given by κ. The figure shows that if βπt+1 + mt is different from zero,
it is not possible to achieve a zero output gap and zero inflation level, because the curve never touches the origin. In this case, there is a constant tradeoff between a higher inflation or a greater output gap (Campbell, 2013). The central bank loss function is drawn a set of indifference curves that show all combinations of inflation and output gap that yield a constant loss value, as shown:
Figure 2A: Optimal policy without the zero lower bound. The central bank is subject to the Phillips curve constraint and chooses an inflation output gap level which is tangent to the Phillips curve. Here, the IS curve does not play a role but it merely determine the level of interest which guides the tradeoff between Investment and Consumption. Note that in this figure, the interest rate is strictly above zero (ZLB constraint). Source: Campbell, 2013.
To achieve the lowest loss, the central bank chooses to position itself at the point of tangency between the Phillips curve and the indifference curve (Campbell, 2013).
1.2.II Optimal policy at the zero-lower bound
At the zero lower bound, the interest rate constraint plays an important role, because it restrains central bank’s actions by putting an upper limit on the output gap. To find this constraint, we rearrange equation 2 to isolate it on the left side.
8
yt = – 1/σ (it – πt+1 – rnt) + yt+1 (2)
implies that:
it = -σ(yt
-
yt+1) + πt+1 + rnt (5)We substitute this equation into the zero lower bound of equation 4, and rearrange for yt on the left side of the inequality, as
shown:
it ≥ 0 (4)
We plug equation 5 on the left side of this inequality:
-σ(yt - yt+1) + πt+1 + rnt ≥ 0 (6)
Solving for yt we obtain:
yt ≤ yt+1 + (πt+1 + rnt)/σ (7)
Given that the output gap is the difference between actual and potential output, a negative upper limit indicates a negative output gap, which is likely to result in a recession. In figure 2b, Campbell (2013) shows the effect of monetary policy at the zero lower bound constraint, which in this case is graphed as a negative constraint. The dotted line indicates the output gap constraint of equation 7, faced by the central bank.
Figure 2b: monetary policy decisions at the zero lower bound. The central bank cannot choose the point of tangency between its loss function and the Phillips curve constraint, because this would only be possible with negative interest rates. Thus, the central bank policy choice is suboptimal. Source: Campbell, 2013.
1.2.III The role of Forward Guidance at the Zero Lower Bound
As picture 2B shows, the upper limit on the output gap can be shifted to the right by increasing expected future inflation or decreasing next year’s negative output gap. By delivering the promise to keep rates lower for longer in the future, the central bank can boost spending and investment, thus increasing expected inflation and reducing the future output gap. Hence, the upper bound on output gap shifts to the right, allowing the central bank to achieve lower levels of loss, as shown in figure 2C. Note how, together with a rightward shift in the lower bound, we also obtain an upward shift in the Phillips curve, caused by
9
higher level of expected inflations (the Phillips curve intersects the Y axis at βπt+1 + mt ). Thus, there is always a trade-off
between the cost of additional current inflation and the benefit from the rightward shift in the output gap constraint. Whether this strategy causes an improvement in current outcomes depend on the initial levels of inflation and output gap (Campbell, 2013).
Figure 2C: the central bank makes a commitment to maintain low rates for an extended period of time. This accommodative policy causes an increase in inflation and a reduction in the negative level of output gap. The effect is twofold: the Phillips curve shifts upward due to higher expected inflation, and the restriction on output gap shifts to the right. Source: Campbell, 2013
However, Campbell (2013) highlights that if the central bank has the reputation of a successful inflation targeter, the cost of forward guidance will be small. Moreover, given that inflation levels during the financial crisis tended to remain below target, we may deem the policy as beneficial.
1.3 The Time-Inconsistency Problem
Having demonstrated how forward guidance can theoretically help the central bank in achieving a lower outcome of its loss function, we turn to the time-inconsistency problem highlighted by den Haan (2013). Time-inconsistency is a situation that arises when policymakers indicate in advance that they will commit to a certain policy in the future, but choose a different one when the time to implement arrives.In figure 3, den Haan (2013) shows the policy interest rate pattern for an economy that is gradually recovering. The policy interest rate is the rate that central banks aim to target directly. Examples of these rates are the Federal Funds Rate for the Federal Reserve or the Bank Rate for the Bank of England. Central banks control these rates through open market operations, until the desired level is reached. Upon this targeted rate depend all other market rates in the economy.
Figure 3: Forward path for the policy interest rate. Source: den Haan, 2013, p.10
10
target rate strictly above zero. On the other hand, the dotted line represents the promise to keep future rates ‘lower for longer’, even when times call for a tightening of monetary policy (den Haan, 2013). This promise is exactly what forward guidance statements aim to deliver. If properly announced and understood by the public, such a promise can lower the expected value short- and medium- term interest rates in the future, over the period covered by the announcement. This could effectively lower the level of long-term rates today, which depend current and future expected short-term rates as per the expectation hypothesis of the term structure of interest rates. A reduction in term nominal rates would result in a reduction in long-term real rates if we assume prices to be sticky in the short run. In turn, these low rates would stimulate purchase of durables and investment, given the lower cost of financing (den Haan, 2013).
At time T*, however, the central bank has a clear incentive to break its promise and reposition itself on the solid line because of the improved economic conditions. If the public foresees this inconsistency and does not believe what the central bank says at T0, public expectations will remain unchanged and long-term rates unaffected, making the policy ineffective (den
Haan, 2013).
Central banks, however, can mitigate this problem by making a commitment so clear and explicit, that breaking it would result in a loss of reputation. Hence, the importance of Odyssean statements highlighted at the beginning of this section emerges. The drawback is an inevitable loss of flexibility caused by a tying of the central bank’s hands (Filardo and Hofmann, 2014). In fact, Filardo and Hofmann (2014) point out that the design of forward guidance policies at the zero-lower bound is subject to a continuous trade-off between flexibility and effectiveness.
2 Historical Remarks
In section 1, we point out that forward guidance policies result in increased transparency and a better understating of the central bank’s reaction curve. As such, this policy need not be used only when the necessity for additional monetary stimulus arises. In this section, we provide a short overview of the history of forward guidance, starting from times where the adoption of this policy was not aimed at providing monetary stimulus, but rather higher central bank transparency.
The first adopter of forward guidance was the Bank of New Zealand, which began announcing the forward path of its policy interest rates in 1997, at a time when policy rates were well above the zero-lower bound (den Haan, 2013). Similar action was taken by the Norwegian Norges Bank in 2005, the Swedish Riksbank in 2007 and the Czech National Bank in 2008 (den Haan, 2013). In this context, the adoption of forward guidance policies constituted only a forecast of future action, rather than a commitment of any kind, and was therefore fully characterized by a Delphic component. Historically, the first bank to adopt forward guidance policies at the zero-lower bound was the Bank of Japan, which in 1999 stated that ‘the Bank will maintain the
zero interest rate policy until deflationary concerns are dispelled’ (Shiray, 2013). The Bank of Japan was in fact the first to find
itself in a position where policy rates could not be lowered any further, in what came to be defined as the first real liquidity trap since World War II (den Haan, 2013).
As of today, three major central banks have joined the Bank of Japan in adopting forward guidance at the zero lower bound, drawing attention to this unconventional measure of monetary policy and calling for further research. Namely, forward guidance measures at the zero lower bound were adopted by the Federal Reserve (December 2008), the European Central Bank (July 2013) and the Bank of England (August 2013). Without a doubt, the central bank that mostly experimented with forward guidance at the zero lower bound is the Federal Reserve, which began issuing statements of this kind in August 2008. Since 2008, the Federal Reserve’s forward guidance policies have evolved from an open-ended Delphic forecast, to an Odyssean time-contingent and finally state-time-contingent commitment. The European Central Bank, instead, issued only one statement in July 2013, announcing that rates were likely to remain low for an extended period of time. Free of any commitment, this statement constituted clearly a weak form of guidance, thus falling into the category of Delphic forward guidance. Finally, the Bank of England issued two statements since August 2013, under the new leadership of governor Carney. These statements were
state-11
contingent, as the decision to maintain low Bank Rates was conditional on the level of unemployment and inflation. However, given the statements did not underlie a shift in the reaction function, they constituted only a Delphic type of commitment.
This overview highlights how different central banks chose to adopt forward guidance policies in different ways. In fact, it appears evident that there exists no standard recipe for issuing such statements. Hence, through our literature review and empirical study, we seek elements that made a statement particularly successful in providing additional monetary stimulus.
3 Literary Review of Previous Empirical Findings
The model of section 1 explains the theory behind forward guidance. In the remainder of this paper, we analyze whether theory is matched by empirical evidence on the effectiveness of this policy in shifting interest rate expectations. This section reviews findings from previous studies. We choose to investigate the empirical effectiveness of forward guidance by seeking answer to three important questions:
1.Can we prove empirically that central bank speech is just as important as central bank actions, and that forward guidance is consequently able to shift expectations of future interest rates?
2.Can forward guidance effectively decrease the volatility of future expected rates, meaning that it provides more clarity about the central bank policy intentions?
3.Can forward guidance effectively decrease the sensitivity of interest rates to other type of macroeconomic shocks? One should not forget that the central bank decision to adopt forward guidance policies is a mean towards the end of fulfilling their mandate of price stability, or price stability and maximum employment in case of a dual mandate. In our literature review, we do not seek evidence that the central banks were successful in fulfilling their mandate during times of zero lower bound constraint. However, we believe that the ability to shift interest rate expectations is a valid starting point in achieving the central bank’s objective. This shift in expectations is in fact the mechanism underlying forward guidance policies at the zero lower bound which, if proven effective, may help spurring growth and curbing unemployment. On the contrary, as den Haag (2013) points out, if forward guidance were not able to affect the public’s expectation, then there would be little point in implementing it at all. Thus, we decide to focus on the relationship between forward guidance statements and the change in interest rate expectations, as stated in our research question.
3.1 Words versus Actions, the Effect of Speech on Agents’ Expectations
We begin by investigating the importance of central bank speech. Are central bank words equally or more powerful than central bank actions, or do they not play any role in affecting expectations? For an effective forward guidance, communication about central bank future intentions should be able to shift agent expectations just as much as a change in the current target
rate. To verify whether this is the case, we begin by reviewing an influential paper by Gürkaynack, Sack and Swanson (2004).
The authors build their research on previous approaches that link changes in asset prices to surprises in monetary policy actions, according to the following regression:
∆yt = + xt + t (8)
In this equation, the change in asset prices (∆yt) is caused by the surprise component of the change in the federal fund rate
target (xt). The surprise component consists of the realized value of the federal fund rate minus the financial markets’
expectation for that value, as indicated by federal fund futures rates. To estimate the surprise component of policy
announcements, Gürkaynack et al. rely on high frequency data using a ‘tight’ window (10 minutes before the announcement, 20 minutes after), a ‘wide window’ (20 minutes before and 45 minutes after) and a daily window (Gürkaynack et al., 2004, p. 8).
12
The importance of using high-frequency, intraday data stems from the fact using monthly or quarterly data could reflect nuances caused by other type of macroeconomic news, such as employment releases. The surprise component is calculated for a series of monetary policy announcements from January 1990 to May 2004. The monetary policy announcements taken under consideration are press releases where the Federal Open Market Committee (hereafter: FOMC) announced a change in the federal fund target rate which surprised markets, as reflected in the change in federal fund futures rates. Also taken under consideration are announcements were no change was made in the federal fund rate target, but this nonetheless surprised markets because they were expecting a change (Gürkaynacket al., 2004). After running the regression, the resulting coefficients imply that the surprise component of a change in the federal fund target causes significant changes on asset prices, specifically on Treasury yields and on the S&P 500 returns (Gürkaynacket al., 2004). As expected, stock returns are found to decrease and Treasury yields to increase following a monetary policy tightening (Gürkaynacket al., 2004). This regression, however, does not make a distinction between what part of the announcement causes the surprise component, and therefore the change in asset prices. In particular, it does not make a distinction between announced changes in the current federal fund rate target and changes in the future path of this target. Hence, Gürkaynack et al. (2004) choose to extend previous research through factor analysis. The statistical procedure used by Gürkaynack et al. to estimate the number of factors required for explaining movements in federal fund rates following the announcements will be briefly outlined.
Gürkaynack et al. (2004) reject the hypothesis that a single factor is responsible for causing movements in rates, but do not reject the hypothesis that two factors are responsible for the movement in rates. The authors rely on principal component analysis, a procedure that transforms a set of possibly correlated variables into uncorrelated variables using an orthogonal transformation, to isolate these factors. By employing this procedure, the authors isolate a first factor
corresponding to the change in forecast of current federal funds target and a second factor which involves other aspects of the announcement that moved near term rates (Gürkaynack et al, 2004). They call the first target factor and the second path factor. The significance of the path factor demonstrates the ability of policy statements to affect the economy by providing information about what will happen to interest rates in the future. Gurkaynack et al. (2004) find that the path factor accounted for more than three-fourth of the explainable variation in movement of Five and Ten year Treasury Yield around FOMC meetings. Accordingly, they conclude that central bank promises about future policy do in fact matter just as much as central bank actions. This conclusion is fundamental for our investigation on the importance of central bank words, given that many empirical studies on the effectiveness of forward guidance rely on this assumption.
Campbell et al. (2012) attempt to verify whether these findings continue to hold when the sample is extended to July 2007, before the outburst of the crisis, as well as during the financial crisis between August 2007 and December 2011. Strikingly, they find estimates that highly resemble those from the pre-crisis period, with both path and target factor having a significant influence on Treasury yields at 2, 5 and 10 year to maturity (Campbell et al., 2012). Moreover, they extend previous research by analyzing the response of corporate bonds’ yields with at least 20 years to maturity to the path and target factors (Campbell et al., 2012). These yields are of particular interest, given their direct relationship to firms’ investment decisions. Interestingly, they find that the target factor has no significant influence on bond yields, while a 100 basis points (hereafter: bp) positive path factor realization can raise these yields by 30 to 35 bp for the pre-crisis period and approximately 60pb for the crisis period (Campbell et al., 2012).
Building on Gurkaynack et al. findings, Woodford (2012) seeks further evidence on the effectiveness of central bank speech. The author conducts a study on the effect that central bank announcements play on overnight interest-rate swaps, hereafter OIS. In particular, Woodford (2012) focuses on the change in OIS rates immediately after announcement releases, using high-frequency data. OIS are financial instruments that involve the exchange of a fixed interest rate for the floating overnight interest rate. Woodford (2012) looks at three statements in particular, one issued by the Bank of Canada on April 21th, 2009, and two issued by the Federal Reserve on August 9th 2011 and January 25th, 2012. He points out that the Fed began
13
using forward guidance at the end of 2008, but earlier statements would be less suitable for an event study of this type as they incorporated other type of policy changes (such as quantitative easing) and did not constitute an Odyssean commitment (Woodford, 2012). Below, we show the change in Intraday US dollar OIS rates for the three events at different maturities, as found by Woodford (2012):
Figure 4: Intraday OIS rates in Canada, April 21st, 2009. The dotted line indicates the time of the announcement release. Source: Woodford (2012)
Figure 5 and 6: Intraday US dollar rates on, respectively, August 9th 2011 and January 25th 2012. The dotted line indicates the time of the announcement release. Source: Woodford (2012).
The first announcement, made by the Bank of Canada, implied an immediate reduction in its target rate from 0.5 to 0.25 percentage points, together with a promise to keep the target rate at this level until the end of the second quarter of 2010, conditional on the outlook for inflation. The fact that this announcement included both a reduction in current rate as well as a promise to keep this rate unchanged for months in the future means that the immediate fall in OIS could be attributable to both factors. Nevertheless, Woodford highlights that in this instance, rates at longer maturities fell more than rates at shorter maturities. This implies a flattening in the OIS yield curve, which could not be explained by a simple reduction in the current rate (Woodford, 2012). Hence, the promise to maintain low rates for an extended period of time in the future seems to have played a prominent role in affecting expectations (Woodford, 2012). Similar results can be deducted from evidence on the two announcements by the FOMC. Note that these two statements did not contain any cut in the current target rate, which remained unaffected at range of 0 to 0.25 bpt. Instead, only a promise to keep rates unchanged for a long time was made (Woodford, 2012). However, in figure 5 and 6 we notice a fall in OIS rates together with a flattening of the OIS yield curve immediately afterthe announcement. While short-term rates remained almost unchanged (they were already at low levels due to the effect of previous announcements), rates at longer maturity fell substantially. Woodford’s evidence seems to strengthen Gurkaynack et al. findings on the importance of the path factor in explaining the variation in rate movements at medium-long-term maturities.
In order to extend our empirical assessment to other central banks who chose to employ forward guidance at the zero lower bound, we turn to the analysis conducted by Filardo and Hofmann (2014). In their paper, the authors focus on the major four players of central bank unconventional policies at the zero lower bound, namely the Federal Reserve, the ECB, the Bank of
14
Japan and the Bank of England. We find their assessment particularly useful, due to its extensive reach on several central banks. The authors focus on the change in three-month interbank future rates at different maturities, as well as the yield on the ten year treasury bond following various announcements (Filiardo and Hofmann, 2014). This change corresponds to the end-of-day value on announcement date minus the end-of-day value before the announcement. As Filardo and Hofmann (2014) point out, this method is somewhat imprecise, because it could reflect nuances from other macroeconomic announcements occurred on the same day, such as assets purchases. We report their findings in figure 7.
Figure 7: announcement effect on three-month interbank futures and bond yields, end-of-day value on announcement date minus end-of-day value on the day before theannouncement. Source: Filardo and Hofmann, 2013
As the figure shows, it is possible to identify an overall pattern, corresponding to a drop in interest rates following the announcement. This is particularly evident for early statements released by the FOMC and shown in the left panel of figure 7. However, this has not always been the case. For example, the effect of statements issued by the FOMC has faded over time, to the point that the last two statements caused an increase in interest rates at long-term maturities. Moreover, the effectiveness of forward guidance statement is almost inexistent in the case of the Bank of England and, for the Bank of Japan, the
announcement effects on future rates are rather limited, perhaps signaling the fact that they did not constitute much of a surprise given that low rates were already expected (Filardo and Hofmann, 2014). Given these findings, the authors conclude that in general, the overall trend is an immediate impact of forward guidance on the level of expected future rates, but with a diminishing effect over time (Filardo and Hofmann, 2014).
Finally, Swanson and Williams (2012) attempt to assess the ability of the Federal Reserve to affect public expectations by looking at the effect of statements on Treasury yields. In particular, they focus on two statements, the first one
communicating a tightening of monetary policy after a long period of accommodation in January 2004, and the second one promising exceptionally low levels of the Federal Fund rate for an extended period, in August 2011. Below, we report the change that these statements caused on Treasury yields as shown in Swanson and Williams (2012), using intraday data.
15
Figure 8: Daily change in Treasury yields at different maturities, following two announcements made by the FOMC. Source: Swanson and Williams (2012).
Despite the fact that both announcements did not imply a change in the current target rate target, the change in Treasury yields was substantial and significant. In particular, Swanson et al. point out that the January 2004 announcement caused an increase of 10 to 16 bp at maturities from 1 to 10 years. To cause a decline of equal size, a cut in federal fund rate of approximately 100 bp would be needed (Swanson and Williams, 2012).
At this point, we are able to provide an answer to our first question, according to our analysis of existing literature. Was forward guidance able to shift expectations about future path of interest rates? While the importance of the path factor both with and without the zero lower bound constraint seem to signal that this is indeed the case, empirical analysis provides mixed result. In the case of the Federal Reserve, evidence seem to suggest that the statements released in August 2011 and January 2012 managed to cause a negative drop in rates. These statements constitute the purest test of forward guidance, given that they were free of quantitative easing announcements and that were not accompanied by a drop in the current target (Woodford, 2012). However, the ineffectiveness of later statements and of statements issued by the Bank of England lead us to believe that some conditions need to be met for this policy to work as expected.
3.2 The Effect of Forward Guidance on Interest Rates Volatility
As Filardo and Hofmann (2014) point out, by providing greater clarity about the future path of policy interest rates, a central bank should be able to reduce the volatility of market expectation of future policy rates. The authors test this hypothesis by measuring the realized volatilities of three-month interbank futures during periods of forward guidance against periods without forward guidance (Filardo and Hofmann, 2014). This study is conducted on data for the Federal Reserve, the ECB, the Bank of England and the Bank of Japan. The result is a realized volatility which tends to be lower during forward guidance periods than during non-forward guidance periods (Filiardo and Hofmann, 2014). This relationship, which is strong at one-year maturity, appears to be weaker at longer horizons (Filiardo and Hofmann, 2014).While we could interpret this result as additional evidence that forward guidance did successfully manage to provide more clarity to markets, we note that forward guidance might not be the sole reason for a decrease in interest-rates volatility. In fact, we point out that a lower interest rate volatility might even be detrimental to monetary policy, if caused by exogenous reasons. For example, if we were to attribute a lower interest rate volatility to the fact that markets expect the recession to last for a very long time, this decrease might signify that the central bank has lost its ability to affect asset prices even at longer maturities. This would further hinder the central bank ability to accommodate the economy. Given that the authors do not specify the cause for this decrease in volatility, caution is advised when interpreting these results.
3.3 The Reduced Sensitivity to Macroeconomic News
As a final point, we seek evidence on forward guidance’s ability to lower the sensitivity of interest rates to other types of macroeconomic news, as long as the conditionality of the announcement does not depend on this news (for example, if a forward guidance announcement is linked to a certain minimum level of unemployment to be attained, changes in
unemployment levels would constitute news that is related to the conditionality of the announcement). In order to verify this claim, Swanson and Williams (2012) estimate the time-changing sensitivity of yields to macroeconomic announcements. They investigate whether Treasury yields are more constrained, that is, non-sensitive to macroeconomic news, during periods of zero
16
lower bound than before (Swanson and Williams, 2012). They base their investigation on the regression of equation 8, where
Xt is a vector of surprise components of macroeconomic data releases on a certain day and ∆yt is the one-day change in
Treasury yield over the same day (2012, p.13). For convenience, we report equation 8 below:
∆yt = + xt + t (8)
Using this method, the authors compare the average sensitivity of Treasury yields under the zero lower bound constraint, up to the end of 2012, to that of a benchmark sample spanning between 1990 and 2000 (Swanson and Williams, 2012). Their findings show that the decreased sensitivity of Treasury yields to macroeconomic news is particularly prominent at the shortest end of the yield curve. In fact, 3-month and 6-month yields appear to be partially or completely constrained to macroeconomic news since spring 2009 (Swanson and Williams, 2012). The sensitivity of intermediate-maturity yields to macroeconomic news, however, appears to be less attenuated. In fact, 1-year and 2-year yields are found to be partially and fully responsive to news until late 2011. Only then this yields became unresponsive (Swanson and Williams, 2012). Finally, 5-year and 10-5-year yields are found to be essentially unconstrained throughout the length of the sample, up to the end of 2012 (Swanson and Williams, 2012). These findings are remarkable for our investigation on the effectiveness of forward guidance. As the authors highlight, the sensitivity of medium- and longer-term Treasury yields to macroeconomic news is closely related to the length of time that the federal fund rate is expected to remain at the zero lower bound (Swanson and Williams, 2012). Can we deem the reduced sensitivity of Treasury yields to macroeconomic news as positive phenomenon? Not necessarily so. Swanson and Williams (2012) point out that in order to answer this question it is necessary to identify what caused of this decreased sensitivity of yields to macroeconomic news. If we assume that, by issuing forward guidance statements, the central bank succeeded in manipulating expectations about the future level of the federal fund rate, then longer-term yields that are constrained to macroeconomic news would prove the effectiveness of this policy, because they would signify that markets are focusing on central bank words rather than on other type of information. Furthermore, they would not signify a loss in monetary policy effectiveness given that the commitment could be revoked at later times (Swanson and Williams, 2012). However, if constrained long-term yields were caused by exogenous reasons which are out of the central bank control and not dependent on forward guidance, then the constraint yields would indicate a loss in monetary policy effectiveness not only in the present but also in the future, given that yields are unresponsive (Swanson and Williams, 2012). Clearly, it is not easy to discern whether the reduced sensitivity of yields to macroeconomic news was caused by forward guidance statements or by exogenous reasons. In this regard, further research is advised. However, for the purpose of our investigation on forward guidance effectiveness, we may conclude that central bank statements were effective in lowering the sensitivity of
intermediate-rates only after 2011 and unable to decrease the sensitivity of long-term rates. Hence, we may conclude that, while forward guidance was partially effective in reducing the sensitivity of interest rates to other macroeconomic
announcements, substantial room was left for policymakers to affect the economy through unconventional monetary policy during the length of this sample.
17
4 An Empirical Study of Central Bank Forward Guidance Statements at the Zero Lower Bound. The
Case for the FED, the ECB and the Bank of England
In this section, we conduct our own empirical study to verify the effectiveness of central bank forward guidance statements in shifting interest rates expectations. The following methodology is employed: using intraday data, we look atthe change in interest rates and yields for two types of financial assets, namely three-month futures on interbank rates and government bonds, after announcement releases. We calculate daily rates for a time interval that spans from ten days before to ten days after the release of the announcements, at different maturities. This specific time interval is chosen because it allows us to investigate whether a specific trend is sustained in time. On the contrary, previous research tend to focus extensively on intraday data, failing to capture the sustainability of the changes in rates after the one-day interval. Moreover, this specific length of time results in a sample that is large enough to perform meaningful statistical testing. After obtaining the data, we proceed to calculate the average rate for the ten-days interval both before and after the announcement and investigate whether the two averages differ significantly. In order to investigate the difference in ten-days averages, a student t-test for equal mean with significance level of 0.05 (5%) is performed. This is a two-sided test where the null hypothesis of equal average interest rates before and after announcement releases is tested. Specifically, the two-sided version of the test is chosen because the average rate after the announcement could be either lower or higher than before its release. Moreover, as we explained in section 3.3, announcements may cause a reduction in the sensitivity of rates to other macroeconomic news. In this case it would be reasonable to expect a significantly lower variance of interest rates after announcement releases. While we do not test directly for this decrease in variance, we perform an F-test for equality of variances on our sample before and after announcement releases, with significance level of 0.05. If we reject the null hypothesis that variance of interest rates before and after the announcement is the same, our t-test is adjusted accordingly to account for unequal variance. To conclude, if average rates on the ten days following the announcement are significantly lower than average rates before, we infer that the statement has been successful in lowering the future path of interest rate expectations, which will in turn provide additional monetary stimulus.
Our research differs from existing one because of its extensive focus on all the major adopters of forward guidance at the zero lower bound. In fact, despite the abundance of empirical and econometrical research during periods of non-zero lower bound constraint, very few papers focus solely on forward guidance effectiveness at the zero lower bound, and research is particularly lacking for the ECB and the Bank of England. This scarcity of research is partially attributable to the lack of a sufficient amount of data. In fact, only few statements were released after the onset of the financial crisis, impeding the development of econometrical models. The significance of our research, and of investigating unconventional methods of monetary policy in general, may be understood by looking at figures 9, 10 and 11. These figures show the evolution of the three central banks’ target rates over time, since January 2000. For the three central banks that we investigate, target rates have now reached their lowest feasible level, implying that ordinary transmission mechanisms of monetary policy have become no longer an option.
18
Figure 9, 10, 11: the evolution of the central bank targeted rate over time since 01/01/2000: For the Fed, this rate is the Federal Fund Rate, for the Bank of England the Bank Rate and for the ECB the main refinancing rate. Source: Routers, ECB.
4.1 The Federal Reserve
We begin our case study by focusing on Federal Reserve statements. As previously specified, the FOMC, the branch of the Federal Reserve that decides on the formulation and conduct of monetary policy in the US, is the one that most experimented with forward guidance measures since 2008. Hence, focusing on the Fed may help us understanding what type of forward guidance is the most effective.
Between December 2008 and December 2013, a total of seven forward guidance statements were issued by the FOMC. These statements, ranged from open-ended, to time-contingent and finally state-contingent. The first version of Fed’s forward guidance was arguably vague, stating that rates would remain low for “an extended period of time” (Filardo and Hofmann, 2014) . Instead, later statements became clearer and better defined, linking lower rates to a well-specified time interval and, in its final version, to a specific macroeconomic outlook. In table 1, we show the effect that the statements played on Eurodollar future rates at different maturities, by taking the average rate ten days before and ten days after the
announcement. Given the high availability of data, we chose to use Eurodollar future rates as a proxy for the US Interbank rates. As Swanson and Williams (2012) point out, Eurodollar futures rates are the most heavily traded in the world and closely resemble market expectations of the federal fund rate from three to six months ahead. Moreover, they have been widely adopted in previous studies on forward guidance (see for example, Woodford, 2012 and Swanson and Williams, 2012). The results of our study on three-month Eurodollar future rates are diverse and somewhat inconsistent. As the table shows, the first four statements caused a significant decrease in interest rates in almost all cases, and for all medium-term maturities. However, the time-contingent statement of September 2012 acted in the opposite way, causing an increase in interest rates at 1-year and 2-year maturities after release.
19
The same holds true for later statements issued on December 2012 and December 2013, where interest rates at short-medium maturities are found to increase significantly in the 10 days following the announcement release.
Table 1: the effect of forward guidance statements released by the FOMC on US Interbank rates, as reflected by futures on the 3-month Eurodollar rate, widely used in previous literature. Source: Datastream
We turn now to the second type of financial instrument under scrutiny, namely Treasury yields. Investors tend to look at the yield curve to predict future economic conditions. Thus, studying its evolution may provide a good indicator of investment decisions. What we seek is a reduction in Treasury yields after statement releases, especially at longer maturities, implying a flattening of the curve. The results we obtain, shown in table 2, are very similar to those found for Eurodollars future rates. In particular, we note that earlier statements were able to reduce significantly the 10-days average rate at all maturities, with greater reductions in medium-term rates that caused a flattening of the yield curve. However, this effect began to decrease since the January 2012 statement, which caused a very mild reduction in 2-year and 10-year rates, while successive statements resulted in rate increases at all maturities. We find two possible explanations for this progressively inverted trend. In the first case, we may assume that the central bank was no longer successful in explaining its intentions about the federal fund rate’s future path and its message went ignored. This may be due to a loss of credibility or bad communication. Alternatively, we may assume that markets challenged the FOMC’s words and did not believe that the bank would deliver on its promises and that interest rates would increase earlier than anticipated. We find the second assumption more probable. Given that the last two statements tied a low level of the federal fund rate to a 6.5% or higher level of unemployment, it is likely that markets that this threshold would be reached faster than what the central bank forecasted. The fact that the Federal Reserve announced a decrease in quantitative easing measures on June 19th 2013, conditional on the economic outlook, may signify that markets were indeed right about their belief.
20
Table 2: the effect of forward guidance statements released by the FOMC on US Treasury yields, average 10 days before and after the announcement. Source: Datastream
In relations to this fact, figure 12 shows the evolution of Treasury yields since 2008 at various maturities. We note a trend that consisted in a steady decrease, especially at medium-long term maturity, which lasted approximately until the tapering of bond purchases in June 2013. After that date, highlighted by the red vertical line, rates began increasing significantly.
Figure 12: the evolution of Treasury yields since December 2008. The vertical red line shows the date when the FOMC announced a progressive winding down ofquantitative easing measures. Source: Datastream
Moreover, figure 13 shows the evolution of unemployment rate expectations in the United States since 2008, as calculated by Reuters Poll. As the figure highlights, these expectations decreased constantly over time and, in 2013, closely approached the 6.5% bound that would have triggered a liftoff of central bank commitment.
21
Figure 13: The evolution of unemployment rate expectations since 2008. The red line constitute the 6.5% unemployment bound specified by the FOMC for the conditionality of its state-contingent announcements of Dec. 2012 and Dec. 2013. Source: Datastream
Accordingly, the results of our study on FOMC announcements are indicative of the fact that forward guidance in the US constituted a weak form of accommodative measure. While its ability to lower future rate expectations was strong in the beginning, it did not prevent markets from revising these expectations upwards once economic conditions started to improve. As we showed in figure 12, as soon as a decrease in quantitative easing measure was announced by the central bank (indicated by the red vertical line), Treasury yields began increasing steadily, despite the fact that forward guidance was still in place. Hence, we may conclude that forward guidance announcements, when not reinforced by quantitative easing, are not able to influence expectations fully. In fact, the decision to taper quantitative easing measures provided in itself anindication of economic recovery that voided the effect of the statements.
4.2 The European Central Bank
We now turn to forward guidance measures adopted by the European Central Bank under zero lower bound constraints. As we did for the FOMC, we assess the effect of these statements on the average level of interbank rates and government bond yields before and after the release.
The ECB adopted forward guidance through the release of only one statement, made on July 2013, which outlined how the Governing Council “expects the key ECB policy rates to remain at current or even lower levels for an extended period of
time”, given the subdued outlook for inflation (Filardo and Hoffmann, 2014). This constituted an open-ended time of
commitment, which left the ECB with enough flexibility to change its policy decisions at any point in time. As executive board member Praet points out, this statement was issued with the purpose of correctly instructing the public about the right shape of the ECB’s reaction function, implying that there was a previous misalignment of public expectations about the shape of this function (2013, p.4). To the extent that the statement correctly managed to provide more clarity about the central bank reaction function and assessment of macroeconomic outlook, we expect a decrease in interest rates immediately after the announcement. This is indeed what our findings highlight, as reported in table 3.
22
table 3: the effect of ECB forward guidance statements on interbank Euribor Rates and Government Bond Yields, 10 day average before and after the announcement. Source: Datastream
In particular, we note a small but significant decrease in Euribor rates (the interbank lending rate in Europe) at all maturities, with the highest drop, 21bp, at the one-year maturity. Further, by investigating a government bond index of AAA-rated Euro countries, we find similar drops in bond yields, although even more moderate. In particular, we find the highest drop, 16bp, at the two-year maturity. Despite the fact that these decreases were only mild (21bp at the maximum), they still resulted significant and considerable, given the law rates environment that already persisted. Accordingly, we can conclude that ECB forward guidance was weak but successful in lowering expectations about future interest rates.
4.3 The Bank of England
To conclude our empirical study, we turn to the Bank of England. Forward guidance in the United Kingdom began only in August 2013, when Mark Carney was elected as the bank’s governor. TheBank of England’s forward guidance consisted in two statements, of a state-contingent nature, which linked a low level of the base rate of interest (the Bank Rate) to an
unemployment threshold of 7% and several other “knockout” conditions, such as the level of inflation and financial stability. As in previous cases, we study the effect that these announcements played on interbank rates and government yield curves at various maturity. For interbank rates, we were unable to obtain data on Libor futures, so we focus on the commercial bank liability curve published on the bank of England website. From the same website, we also obtain the UK Gilts yield curve. The results of our study are shown in table 4. As the table highlights, forward guidance statements issued by the Bank of England did not cause a significant downwards shift in interest rate expectations at any maturity. Rather, we note mild but significant increases in these rates following most statements. Given that previous findings show that central bank statements can effectively shift interest rate expectations, we wonder what went differently in the case of the Bank of England. There are several possible reasons that might explain why the BoE forward guidance statements did not cause the drop in expected future rates that was hoped for. To begin, we note that both statements were linked to a 7% level of unemployment, in a period where unemployment rates were in fact very close to that specific threshold, as highlighted in figure 14:
23
Figure 14: Actual Unemployment Rates in the United Kingdom since 2008, according to the UK Office for National Statistics July 2014 revision. The red line shows the 7% announcement threshold, and the dotted lines correspond to the announcement dates. Source: Office for National Statistic
In this figure, the red horizontal line correspond to the 7% announced unemployment liftoff threshold, while the two vertical dotted lines correspond to the announcement dates. While a survey of expected unemployment rates, rather than realized figures, would be more appropriate to describe market expectations, we were not able to retrieve this information. However, this figure accurately conveys the underlying idea. Further, we believe that the inclusion of several
“knockout” conditions, that left the BoE with enough flexibility to lift its commitment at any point in time, deterred markets from believing in the bank’s promises. In other words, it is possible that the choice to avoid a strong commitment on the bank’s behalf created that time-inconsistency problem that we highlighted in section 2, especially during a period when the economy was showing strong signs of recovery. As for Federal Reserve, the results of our study on Bank ofEngland’s statements lead us to believe that forward guidance can be ineffective in certain instances.
Table 4: the effect of BoE forward guidance statements on interbank rates and UK Gilts yields, 10-day average, before and after announcement dates. For interbank rates, we use the commercial bank liability curve published by the Bank of England. Source: Bank of England
24
5 Conclusion
Through this research, we aimed to analyze forward guidance at the zero lower bound and its effectiveness in lowering expectations of future interest rates. We believed that, to the extent that forward guidance is able to decrease interest rate expectations about short- and medium-term rates, long-term rates should also decrease. Accordingly, we attempted to verify whether interest rate expectations at various maturities changed significantly after announcement releases, which was the research question of this paper.
We began by proving the effectiveness of forward guidance in theory, following the primer developed by Campbell according to the New Keynesian model. We also pointed out the time-inconsistency problem, which may render the policy ineffective but could be mitigated by a strong Odyssean commitment. To verify the empirical effectiveness of forward guidance, we
conducted a literature review of previous findings and our own research focusing on the Federal Reserve, the European Central Bank and the Bank of England. From our literature review, we were able to draw three important conclusions:
1. That what central banks communicate about their future intentions is just as important as what they do about the current level of the targeted rate, especially at longer maturities.
2. That interest rates on interbank futures had a lower realized volatility during times of zero lower bound forward
guidance. While no evidence proves that the lower volatility was caused by the release of the statements, we may infer that the statements provided more clarity about central bank policy intentions if this were the case. Hence, we advise further research to verify whether a relation exists between forward guidance statements and lower interest rates volatility. 3. That the sensitivity of interest rates to other type of macroeconomic news partially decreased once forward guidance was introduced, even under zero lower bound constraints.
While this finding plays in favor of the policy effectiveness, the crowding out of private information may be harmful for central bankers who wish to make predictions. Moreover, this decrease in sensitivity was absent for longer-term rates and only partially constrained 1-year and 2-year yields, indicating that unconventional monetary policy was fully effective only at shorter maturities.
Our empirical review on central bank statements yielded mixed results. While earlier statements released by the FOMC and the statement issued by the ECB managed to reduce future rates expectations, later statements did not yield the same results. The same holds true for Bank of England’s forward guidance, which did not effectively manage to shift rate expectations for interbank futures and Treasury gilts. To answer our research question, we conclude that forward guidance can, and in several cases has been able to influence the expected path of future interest rates downwards. However, the policy has not always been successful and several factors can hamper its effectiveness. The Federal Reserve’s decision to wind down quantitative easing measures on June 2013 resulted in a steady increase in Treasury yields, despite the fact that forward guidance statement continued to hold. Hence, the policy seems to be most effective when reinforced by
quantitative easing, another unconventional method of policy accommodation. Moreover, the statement should come free of “knockouts conditions” which may be perceived as escape routes that the central bank could use to shy away on its commitment, reinforcing the time-inconsistency problem. Certainly, a central bank might have very good reasons to rely on these knockout conditions, which would allow it to abandon a low interest rate regime without losing credibility. Loss of credibility is indeed one of the fundamental weaknesses of this policy, as our paper highlights. Nonetheless, we continue to believe in the policy effectiveness and recommend further research on what conditions need to be met for a statement to be successful.
25
Web Links to Data:
Figure 9: ECB Main Refinancing rate, European Central Bank. Web: https://www.ecb.europa.eu/stats/monetary/rates/html/index.en.html
Figure 14: Actual Unemployment Rates in the United Kingdom since 2008, UK Office for National Statistics July 2014 revision. Web: http://www.ons.gov.uk/ons/taxonomy/index.html?nscl=Unemployment#tab-data-tables
Table 4: Data on UK Gilts Yields and Commercial Liability Curve, Bank of England. Web: http://www.bankofengland.co.uk/statistics/pages/yieldcurve/default.aspx
26
References
Campbell, J.R. (2013). Odyssean forward guidance in monetary policy: A primer. Forward Guidance: Perspectives from
Central Bankers, Scholars and Market Participants, Vox eBook, 87-105.
Campbell, J. R., Evans, C. L., Fisher, J. D., Justiniano, A., Calomiris, C. W. & Woodford, M. (2012). Macroeconomic Effects of Federal Reserve Forward Guidance. Brooking Papers on Economic Activity, 1-80.
den Haan, W. (2013). Forward Guidance: Perspectives from Central Bankers, Scholars and Market Participants. Center for
Economic Policy Research, Vox eBook, 1-25.
Filardo, A., Hofmann, B. (2014), Forward Guidance at the Zero Lower Bound, BIS Quarterly Review.
Goodhart, C. (2013). Debating the Merits of Forward Guidance. Forward Guidance: Perspectives from Central Bankers,
Scholars and Market Participants. Vox eBook, 151-157.
Gurkaynak, R. S., Sack, B. & Swanson, E. T. (2005). Do actions speak louder than words? The response of asset prices to monetary policy actions and statements. International Journal of Central Banking, 1, 55-93.
Kool, C. J., & Thornton, D. L. (2012). How effective is central bank forward guidance?. Federal Reserve Bank of St. Louis
Working Paper Series, 2012-063A.
Praet, P. (2013). Forward Guidance and the ECB. Forward Guidance: Perspectives from Central Bankers, Scholars and
Market Participants. Vox eBook, 25-35.
Swanson, E. T., & Williams, J. C. (2012). Measuring the Effect of the Zero Lower Bound on Medium-and Longer-Term Interest Rates. Federal Reserve Bank of San Francisco Working Paper, 2.
Shirai, S. (2013). Monetary Policy and Forward Guidance in Japan. Speech at the International Monetary Fund,
Washington DC, 19.
Williams, J.C. (2014). Monetary Policy at the Zero Lower Bound, Putting Theory Into Practice. Hutchins Center
on Fiscal and Monetary Policy, Brookings Institution.
Woodford, M. (2012). Methods of Policy Accommodation at the Interest-Rate Lower Bound. The Changing Policy
