Is there an optimal degree of central bank transparency?
Abstract
Is there an optimal degree of central bank transparency? Starting in the last decades, central banks are increasing their degrees of transparency substantially. This tendency led researchers to wonder whether high degrees can turn out to be detrimental for monetary policymaking. However, this empirical branch of research is still young and the findings of the studies are inconclusive. Since the appropriate methods to employ and the relevant set of control factors to use are at the center of this discussion, this thesis contributes to the debate by evaluating the inclusion of two potentially important factors. These factors are the degree of central bank independence and the depth of financial markets. Moreover, since datasets are limited, this thesis contributes to the debate by providing evidence from a substantially larger dataset.
In the regressions with the larger dataset, but without central bank independence and financial market depth, optimal degree results from previous studies are not robust. A closer look points out that the differences mainly stems from the larger time range utilized in this paper (it is extended with 2006-2010). Whereas controlling for possible explanations like the financial crisis does not change the results, including central bank independence and financial market depth does. They prove to be important for the estimations and their inclusion results in an optimal degree of central bank transparency for the larger dataset.
Master’s Thesis 2013-2014 Track: Monetary Policy and Banking Faculty of Economics and Business University of Amsterdam Arjen Goedmakers, 5876486 Supervisor: dhr. prof. dr. L. H. Hoogduin September 2014
Table of Contents
1. Introduction 3
Part I The theoretical framework for optimal central bank transparency 6 2. The initial preference for secrecy over transparency in the central banking literature 6
2.1. The FOMC versus Merrill case 6
2.2. Dynamic-inconsistency arguments in favor of secrecy 8
3. The tendency of increasing transparency 9
3.1. The increase in central bank independence 10
3.2. Efficient transmission of monetary policy 10
3.2.1. Anchoring inflation expectations 11
3.2.2. Adaptive learning 12
3.2.3. Deepening financial markets 12
3.3. The empirical preference for transparency: particular measures 13 3.4. The empirical preference for transparency: overall central bank transparency 15
3.4.1. Frameworks, indices and datasets 15
3.4.2. The effects of overall central bank transparency 17
4. Optimal central bank transparency 18
4.1. The case for an optimal degree of overall transparency 19 4.1.1. The tradeoff between policy-flexibility and managing inflation 19 4.1.2. Heterogeneous information and coordinated actions 20
4.1.3. Perceived deteriorating quality of information 22
4.2. Relating the degree of transparency to a measure for monetary policy 25 5. Empirical research concerning optimal central bank transparency 28
5.1. Extended datasets 29
5.2. Empirical studies: optimal transparency with inflation (persistence) 30 5.2.1. The endogenous relation between central bank transparency and inflation 30 5.2.2. The inflation persistence minimizing degree of central bank transparency 32 5.2.3. Potentially important factors: empirical evidence 33
6. Estimating the effects of overall transparency 35
6.1. Methods to estimate the effects of central bank transparency on inflation persistence 35 6.2. The controls: including central bank independence and financial market depth 37 6.2.1. Controls for inflation persistence and the level of inflation 38
6.2.2. Central bank independence 42
6.2.3. Financial market depth 47
Part II The empirical analysis 51
7. The method 51
8. The data 53
8.1. Central bank transparency data 53
8.2. Inflation data 54
8.3. Central bank independence data 56
8.4. Financial market depth data 57
9. Empirical analysis 57
9.1. Robustness of the optimal degree of central bank transparency to a larger dataset 58 9.2. Including central bank independence and financial market depth 61
9.2.1. Including central bank independence 61
9.2.2. Including financial market depth 63
9.2.3. Including central bank independence and financial market depth simultaneously 64 9.3. Robustness of the optimal degree of central bank transparency to multiple lags of inflation 67 9.4. High-independence central banks versus low-independence central banks 68
9.5. OECD economies versus Non-OECD economies 70
10. Conclusion 72
Appendices 75
References 103
1. Introduction
The view upon the economic costs and benefits of the secrecy and mystique that characterized the policymaking of monetary authorities for so long changed remarkably in the last two decades. Whereas society’s attitude towards secretive and paternalistically operating institutions already shifted in the early sixties, economic arguments regarding efficient monetary policymaking still favored a more secretive stance.
In the early post-World-Wars’ society, where a structural functionalistic paradigm was held by the mainstream of the social scientists in the Western world, a paternalistically operating government was seen as efficient from society’s perspective (Ritzer, 2013). Starting roughly in the sixties, a shift towards a more skeptical and distrustful attitude towards institutions became visible, resulting in an increasing demand for institutional accountability. New mainstream paradigms started putting more emphasis on dysfunctional properties of institutions, such as incentives to avoid accountability (Ritzer, 2013). The change in the attitude of the mainstream resulted in alterations to Western legal systems, which deteriorated the position of public institutions to operate secretively (Goodfriend, 1986). Karl Brunner (1981) translates the more distrustful paradigm in a highly critical paper, where he emphasizes that the secretively and mystically operating central bankers do this to shield themselves from accountability and because they believe in being all-knowing. This is representative of the critiques on secretively operating central banks. The arguments were mainly based on a demand for accountability to mitigate the distrust for paternalistically operating institutions.
In fact, in the dominant view, secrecy was seen as economically preferable on the assumptions that it provides central banks with tools to surprise the market and implies fewer distortions from markets anticipating on central bank interventions (Goodfriend, 1986). In the nineties, the economic arguments for operating transparently became dominant in the monetary literature, although the debate was still inconclusive. With increasing attention shifting to the benefits of central bank communication and the inflation-fighting potential of gaining credibility, high transparency became dominant to secrecy. In addition, empirical research studying the effects of transparency-increasing practices mainly found positive results (Van der Cruijsen and Eijffinger, 2007). This resulted in a tendency towards increasing transparency, which is shown in the figure in appendix A1.1. It shows the increases in the overall transparency dataset of Dincer and Eichengreen (2014), where the dots represent different central banks, between 1998 and 2010.
The tendency in figure A81.1, together with the inconclusiveness in the theoretical debate, led academics to wonder whether transparency can go too far (Mishkin, 2004). The theoretical
literature started favoring an optimal intermediate degree of overall transparency. At low degrees of transparency, more information can lead to better forecasts, but at high degrees of transparency, the public might be unable to process the information (Van der Cruijsen, Eijffinger and Hoogduin, 2010). Furthermore, public information can crowd out private information and increase the exposure of the private sector to risk (Morris and Shin, 2002). Whereas the empirical studies that investigate the effects of particular transparency-enhancing measures are highly valuable and informative, they do not shed much light on this discussion. It is clear that there are certain measures that improve the efficiency of monetary policymaking, no matter what the degree of overall transparency is, and it is essential to accumulate knowledge about the instruments to increase transparency. Nevertheless, it is important as well to gain insight in the effects of overall central bank transparency. Whereas the implementations of particular measures provide policymaking improvements, it is still possible that the implementation of a broader set of measures has negative effects as well.
Recently, several studies aim to estimate whether such an optimal degree of transparency exists, for which the effectiveness and efficiency of monetary policy transmission are maximized. However, this branch of research is still young, in both data gathering and empirical estimations, and the findings are not unanimous. Whereas Van der Cruijsen et al. (2010) find an optimal degree of central bank transparency, based on its effects on inflation persistence, Dincer and Eichengreen (2009) find no relation between an optimal degree and this variable. The difference between these results is remarkable, since the studies use very similar datasets. This means that the difference stems from the methods and control factors that are used in the estimations. So, it is necessary to gain more insight in the nature of the relationship between central bank transparency and efficient monetary policymaking, since the current debate is inconclusive. To do this, it is important to determine what the appropriate methods and relevant factors are. Moreover, because the datasets are limited, robustness checks with broader or other datasets are essential.
Since it is important to determine whether there is an optimal degree of transparency for the conduct of monetary policy, this paper aims at shedding more light on this topic. The main contributions to the discussion are the insight in relevant control variables for the discussion and the usage of a transparency dataset that has almost the double amount of observations compared to datasets used by previous studies, recently provided by Dincer and Eichengreen (2014). Although this study investigates the optimal degree of transparency, it is important to keep in mind that such a degree is dependent on a variety of factors, and as such differs across central banks. Nevertheless, when an optimal degree is found it provides evidence for a detrimental effect of (too) high levels of transparency, whereas it provides an indication for the range of optimal degrees. This results in the
following research question for this paper: Is there an optimal degree for overall central bank transparency, based on the efficient transmission of monetary policy?
The research method is twofold. The theoretical framework for investigating the optimal degree of central bank transparency is established first. The academic debate is discussed to determine what processes are at play. This results in inflation persistence as the measure used to analyze whether there exists an optimal degree of monetary policy. With this measure, the appropriate methods and factors that should be used for investigating the research question are determined. The second part of the research method answers the research question empirically. With a cross-country time-series transparency dataset, which is more extensive than datasets employed in previous studies, the research question is answered using empirical estimations. Because disturbances for the employed dataset might be cross-panel and/or cross-time heteroskedastic, and auto correlated, the empirical estimations control for the panel and time-series features and use panel corrected standard errors (PCSE).
This paper starts with Part I, which is the theoretical framework. This part starts with chapter 2, which describes the economic arguments used for preferring the secretively operating central bank. This is followed by chapter 3, which describes the theoretical and empirical arguments that were underlying the tendency towards increasing transparency. Chapters 4 and 5 respectively describe the theoretical and empirical work that is relevant for an optimal degree of transparency. Chapter 6 is devoted to determining the appropriate estimation method to employ and the right set of control factors to use. This is the last chapter of Part I. Part II provides the chapters that are devoted to empirically answering the research question. Chapter 7 is the first chapter of Part II, and describes the employed empirical method. Chapter 8 describes the data for the most important variables (information about the other variables can be found in the appendix). The results from the empirical estimations are described and analyzed in chapter 9. The final chapter of this paper is chapter 10, which provides conclusions, limitations, and recommendations for future research.
Part I The theoretical framework for optimal central bank transparency
In this part the gradual shift from favoring secrecy towards favoring transparency is discussed. The debate first developed on theoretical arguments, mainly favoring a low degree of transparency. From the nineties onwards, the theoretical arguments started favoring an increasing degree of transparency. Only at the turn of the millennium the empirical literature took off, made possible by the development of transparency indices and the collection of related data that started a few years earlier. The empirical literature mainly focused on the effects of particular transparency enhancing instruments and concluded overwhelmingly in favor of increasing transparency. Even more recently, academics started advocating some caution with increasing the degree of central bank transparency, arguing for an intermediate optimal degree of transparency. However, this line of research that aims to analyze the effects of varying overall degrees of transparency is still in its development stage and faces wide discussion about the methods to use and the factors to include. It is at the end of this part that these research methods are evaluated and where some potentially critical factors for the measurement of an optimal degree of transparency are analyzed. The next chapter starts with an explanation of the early preference for secrecy over transparency, based on economic arguments.
2. The initial preference for secrecy over transparency in the central banking literature
The shift in paradigm led to alterations in Western jurisdiction that weakened the position of central banks and other public institutions to operate secretively. The critical paper of Goodfriend (1986) discusses the most famous legal case regarding the degree of transparency in the central banking literature. The Fed Open Market Committee (FOMC) versus Merrill case had to be decided on economic arguments against and in favor of secretive central bank behavior. As such, this case makes an interesting starting point for the evaluation of transparency on economic grounds. The legal case moreover inspired a line of theoretical research that started analyzing the desirability of a low or high degree of transparency, beginning with the paper of Cukierman and Meltzer (1986).
2.1. The FOMC versus Merrill case
The FOMC came up with several arguments in favor of secrecy. The first argument is the potentially costly reaction of the market to the disclosure of information by the central bank (Goodfriend, 1986). When markets are unable to attach the appropriate weight to the received information, they might overreact. Market interventions that are at forehand beneficial might turn into costly
operations from societal perspective. This argument then hinges on the ability of the central bank to predict market behavior and on the abilities of the market to interpret disclosed information. For this argument the capabilities and the capacities of the markets are highly relevant. An economy with a large financial sector is more able to value provisioned information appropriately. Furthermore, the larger the size, the greater ability the sector has to process larger amounts of information. In addition, competition arguments can be used to argue that deeper financial markets benefit the private sector’s efficiency in processing information.
The second argument of the FOMC is that the provision of information can lead to undesirable pre-commitment (Goodfriend, 1986). For example, the publication of operating directives would mean limited flexibility in the possible policy reactions to economic shocks and fluctuations (Jensen, 2002). This argument can also point at the other direction. When the central bank is highly transparent, it can respond to short-term developments without facing a deterioration in its long-term operations perception. Thereby, the central bank can adjust short-term policies without losing credibility (Geraats, 2002).
Thirdly, the disclosure of information can lead to undesirable effects, such as speculation and higher volatility (Goodfriend, 1986). One line of reasoning that provides credibility to this argument comes from the literature inspired by Morris and Shin (2002). When information is heterogeneous and costly to acquire, the provision of free or cheaper public information leads to more uniform behavior of the market. The implications of the provision are the crowding out of market efforts to gather valuable information, making the markets more dependent on public information. When one assumes differing noise and risk in public and private information, the high dependency on public information would make the market highly vulnerable to the noise and risk coinciding with this information. For a central bank with a high degree of risk, it is clearly not optimal to provide a great deal of information. However, even a high degree of transparency by a central bank with intermediate or low risk can be harmful for economic stability, because of the overexposure to the noise. Although the probability of a bad forecast is small, the interconnectedness of the economy makes shocks highly disruptive because the reaction from the market is likely to be more uniform and coordinated when information is released.
A related reason to refrain from providing more transparency is the increasing possibility of bank runs when information about the fragility of financial institutions is made public (Cukierman, 2009). In a classical Diamond and Dybvig (1983) model, the release of information about the health of the financial system might trigger bank runs. Cukierman (2009) discusses the undesirable coincidence of too much transparency in such a setting, where the central bank releases private information about segments or institutions of the financial system. The increasing vulnerability to
bank runs harms investment and the risk sharing facilities of the financial system. Dincer and Eichengreen (2014) apply this line of reasoning to the recent financial crisis and look at the short term reaction of central banks on world-scale. They compare the pre- and post-crisis transparency indices and see no reversal in the overall trend of increasing transparency1. Geraats (2010) moreover
argues that the provision of information about financial sector and institution health creates an incentive effect to apply more prudent policies. This would create financial stability over the longer horizon.
2.2. Dynamic-inconsistency arguments in favor of secrecy
The dynamic-inconsistency research of Kydland and Prescott (1977), Calvo (1978) and Barro and Gordon (1983) put central bank credibility at the center of economists’ attention. The paper of Cukierman and Meltzer (1986) extended this field to the theoretical analysis of central bank transparency and led to the dominant position of a low degree of transparency versus a high degree until the mid-nineties (Faust and Svensson, 2001). The most important arguments from this line of research are the loss of policy flexibility, related to the second FOMC argument, and the costs of short-term focused politicians.
Cukierman and Meltzer (1986) address the question of whether a central bank should operate transparently or opaquely by analyzing an optimal policy model where the information between the central bank and the public is asymmetrical. The expectations of the public are influenced by the behavior and the information provision of the central bank. The central bank can vary its degree of transparency2, by having leverage over being secretive and creating noise. They
assume a central bank with an objective function that attaches weight to price stability and closing of the output gap. When the degree of transparency is low, the central bank can use unanticipated money creation as a tool to close the output gap at the cost of inflation. Vital to their analysis are the assumptions that only unanticipated money matters3 and a by the public unknown objective
function of the central bank. Although Blinder (2009) argues that the last assumption is outdated, a lot of central banks have low scores regarding this factor of transparency in the index of Dincer and Eichengreen (2014).
1 Although they see some central banks decrease their degree of transparency for the first time, only six central banks decrease transparency, while 18 central banks increase transparency. Moreover, the adoption of monetary policy tools such as forward guidance indicate no reversal in the increasing-transparency tendency of central banks.
2 Transparency implies the provision of consistent information
3 This assumption receives a lot of criticism nowadays. For example see Blinder (2009) or Walsh (2007). 8
In the economic line of reasoning from Cukierman and Meltzer (1986), a transparent central bank buys credibility at the cost of losing the surprise inflation policy tool. When the lower-inflation gains from more credibility are relatively low to the losses in flexibility to respond to supply shocks, secrecy may be beneficial (Van der Cruijsen and Eijffinger, 2007). Furthermore, a central bank with a high weight attached to mitigating the output gap in its objective function gains more by being opaque, because it is able to obscure its intentions with noise, such that expected inflation is lower.
Another important argument stemming from the time-inconsistency literature is related to the electoral pressures on politicians. Since most politicians have a time horizon that is not optimal for monetary policymaking, a structure with high influence of the government on monetary policymaking makes the case for secrecy (Mishkin, 2004). In (democratic) society, policymakers are eager to accept a somewhat higher inflation in order to provide their population with higher employment and output. When central bank independence is low, business cycles might cause the electorate and the government to put pressure on the central bank to operate with a too expansionary monetary policy in order to boost short term output (Mishkin, 2004). When central banks for example operate with transparent objectives, the pressures can cause the failure to meet these objectives. This might not only be suboptimal for the economy, but also deteriorates the perceived quality of the central bank because it fails to achieve the targets that it more or less pre-committed itself to. When a central bank is secretive about its operations, the electorate and government are less able to hold the central bank accountable for not surrendering to the pressure. So, the central bank is able to operate under less conflict with politicians and can carry out a more optimal policy from society’s perspective (Mishkin, 2007).
Until the nineties, secrecy was favored compared to transparency. The most appealing economic arguments were the potentially costly market reactions based on markets that can misinterpret the provisioned information, higher volatility stemming from more coordinated market reactions, the loss of policy flexibility of the monetary authority due to pre-commitment and credibility pressures, and the time-inconsistency problems related to democratically elected politicians. In the nineties, the attitude towards the benefits of secretively behaving central banks changed remarkably. Chapter 3 provides an overview of the most important literature that deals with this shift.
3. The tendency of increasing transparency
In the last two decades, we have experienced a remarkable turn in the view on secretive behavior of monetary authorities. The economic literature started to advocate high degrees of transparency on
theoretical arguments. Later on, made possible by the construction of datasets inspired by these theoretical arguments, economists began to test the effects and relations of transparency with econometric means.
The theoretical arguments for favoring increasing transparency are provided in the first sections of this chapter. These can be broadly put in the following categories, respectively that the increasing independency of central banks implies the need for accountability, and that high transparency facilitates a more efficient transmission of monetary policy. This chapter ends with a discussion of the developing empirical central bank transparency literature and its initial preference for high degrees of transparency.
3.1. The increase in central banks independence
The benefits of delegating monetary policy to an independently operating monetary authority are widely accepted. They do not suffer from the political pressures like their dependent colleagues, which enables them to conduct more optimal monetary policy in the long-term. The independence results in a reduction of the inflationary bias of the central bank. But independence irrevocably leads to the need for accountability. Highly dependent central banks have no need for political accountability. However, the increasing independence of central banks, which started a few decades ago and is still going on, changes this radically (Dincer and Eichengreen, 2009). The controllability of the monetary authorities is reduced because the ability to punish bad performance through democratic elections decreases. In order to avoid pressures to curb the independence again, central banks should be able to prove that they operate in society’s interests and avoid suspicion about their motives. Being transparent is one of the most powerful and efficient means to accomplish this democratic accountability (Dincer and Eichengreen, 2009).
3.2. Efficient transmission of monetary policy
A high degree of transparency can be a useful tool to gain credibility and build up a good reputation. A central bank that is transparent about its objectives and operations is able to gain credibility by committing to them. This enhances the central bank’s abilities to anchor inflation expectations, improve market planning and increase its policy flexibility. So, another argument for a high degree of transparency is that it improves the efficient transmission of monetary policy. This section provides the theoretical arguments related to the transmission of monetary policy. The overview of the empirical literature follows in the next sections.
3.2.1. Anchoring inflation expectations
Several lines of research attribute the property of an improved ability with anchoring inflation expectations to increasing central bank transparency. In the economic literature based on the work of Barro and Gordon (1983), such as Christiano and Gust (2000), disruptive inflation expectations can be tamed by transparency. When a country finds itself in an inflation expectations trap, where unions demand higher wages because of the expected inflation and these wage demands result in higher prices, a transparent regime can be useful in offsetting this self-fulfilling spiral. A central bank that attaches a high weight to price stability and that can credibly communicate this objective, can lower inflation expectations and as such the costs (for example losses of competitiveness and wage negotiating costs) that coincide with them (Christiano and Gust, 2000). In addition, a highly transparent central bank can more efficiently manage inflation expectations and avoid expectation traps.
The literature that deals with the commitment versus discretion discussion argues in favor of high transparency to anchor inflation expectations too. Commitment means that the central bank engages in a strategic long-term policy to influence the expectations of the public optimally. Discretion means that the central bank reacts to the economy without strategically taking the long-term private sector expectations into account. In order to commit, a central bank must be transparent. The long-term commitment strategy leads to lower output and inflation volatility than the discretion strategy, where the central bank does not invest in credibility-enhancing practices like transparency (Walsh, 2010)4. The private sector has a better understanding of the ideology behind
the actions of the central bank than under discretion and as such is better able to adjust its expectations and planning. Commitment moreover reduces the inflationary bias that a secretive central banker has, leading to lower inflation expectations by the public (Geraats, 2009).
However, Geraats (2009) acknowledges that transparency can go too far. She classifies the effects of transparency into two categories, namely information and incentive effects. Besides its benefits, the information effect can be negative, for example when the target for the output gap is disclosed. With this disclosure, it will be included in the inflation expectations of the public. When the output gap target increases in weight, the interest rate rises (depressing the gap). This results for the central bank in a hard time with closing the output gap. On the other hand, information disclosure enhances risk-management and improves the reactions of the market to policy actions. Furthermore, there are positive and negative incentive effects. Disclosure refrains central banks
4 Although endogenous inflation persistence means that the central bank faces a dynamic optimization problem under discretion as well, leading to a reduction of the inflationary bias (Walsh, 2010).
11
from the inflationary bias of a discretionary regime, but it could lead to coordination problems as is explained in section 4.1.2.
3.2.2. Adaptive learning
In the adaptive learning literature, the public uses an adaptive strategy to form expectations based on the perceived actions of the central bank, when it views the provided information as too low. When a central bank operates with a high degree of transparency, the public has no need to use the adaptive learning strategy. The credible communication forms the base of the expectations of the market, meaning that the central bank is able to anchor inflation expectations to a high extent. Hence, the predictability of monetary policy is high. This implies that interest rate risk and volatility go down. The transparent strategy moreover increases the flexibility of the central bank to react to rate changes (Bernanke, 2004). The improved predictability provides the central bank with more influence on consumption and investment decisions of the private sector. So it is enhancing the smoothing of economic cycles (Issing, 2005).
A low degree of transparency means that the public adapts its expectations to the perceived actions of the central bank. This implies that expectations are more persistent, since they are to a large extent based on past actions of the central bank. Economic cycles that attract monetary policy reactions persist in the expectations of the market. This means inefficient inflation expectations, consumption patterns and investment decisions (Blinder, Goodhart, Hildebrand, Lipton and Wyplosz, 2001; Mishkin, 2004). It must be noted that when the abilities of the market to produce economic forecasts increase, the need for transparency is lower. A large financial sector indicates a more information-producing and as such better forecasting private sector. Increasing transparency could then be harmful because it crowds out valuable private information.
3.2.3. Deepening financial markets
The liberalization, the growth and the interconnectedness of financial markets increased the importance of economic stability and the predictability of monetary policy actions. The increasing size of the financial markets and the transactions in these markets made them more important for achieving the central bank’s objectives, while the growth of the financial markets made the playing field less directly controllable and increasingly disruptive (Dincer and Eichengreen, 2007). For the transmission of monetary policy, this means that managing inflation expectations is gaining relatively more importance for the central banker (Winkler, 2000). The larger size of financial markets
increases the market’s reaction to interest rate changes. This means that these changes can be more disruptive, creating a more risky and volatile environment. Since transparency can be a tool to mitigate the risk and volatility of the interest rate, the increasing depth of financial markets provides the base for increasing transparency.
These were the most appealing theoretical arguments in favor of increasing the degree of transparency. The next section is devoted to the empirical research that developed, inspired by these arguments.
3.3. The empirical preference for transparency: particular measures
The increasing attention towards central bank transparency rested on more theoretical arguments at first. These arguments came from the same corner as the arguments from the literature to favor central bank independence. However, only at the end of the nineties economists started constructing transparency indices and datasets to study its effects empirically estimation. Because the empirical research that studies the effects of central bank transparency evolved so recently, it is still in its infancy. Most studies take the form of individual central bank assessments. These studies investigate the effects of particular transparency-enhancing measures and try to estimate the effects on key economic variables (Dincer and Eichengreen, 2014). They are important for the construction of central bank transparency research and provide insight in the effects of particular transparency instruments. They are however hard to generalize given their limited sample sizes and time ranges. The first attempts to study the effects of the overall degree of transparency cope with the same limitations. The datasets are small and the central banks are dominantly situated in advanced countries. It is thus important that more research is done in these fields, in order to establish a better understanding of the nature of the findings. Important factors to know are amongst others, whether the results are robust to more extensive time ranges and to more extensive panels.
This section discusses the literature regarding particular transparency enhancing measures, supported by some examples. The attention shifts in the next section towards the first research that desires to study overall central bank transparency.
The results from research that studies the effects of implementing transparency enhancing measures are typically positive. Van der Cruijsen and Eijffinger (2007) discuss relevant studies that focus on particular transparency-increasing elements and studies that analyze the effects of increasing general transparency. The popular view stemming from this literature is that transparency-increasing measures lead to an improvement in the anticipation of monetary policy. This leads to an improved predictability for the public, implying an improvement in the transmission
of monetary policy. Nevertheless, not all research finds positive anticipation effects. Especially the papers that investigate transparency-increasing measures that deal with more advanced and more diversified information coincide with lower results (Van der Cruijsen and Eijffinger, 2007). This implies that capacity and capability constraints can lead to negative effects of providing difficult but also large amounts of information. This is highly relevant for the next chapter, which discusses the case for an optimal degree of transparency.
Dincer and Eichengreen (2007) list a variety of papers that estimate the effects of the degree of transparency or the implementation of particular transparency-increasing measures on the level of inflation, inflation persistence and variability of inflation. Although some papers have their drawbacks (reversed causality, advanced country bias and small data periods), the results generally are that expected inflation is lower and that the predictability of monetary policy goes up.
Several authors investigate the impact of the more transparent regimes of the inflation-targeting central banks on variables like long-run inflation expectations, inflation forecast errors and the level of inflation. We have seen a remarkable expansion in the degree of central bank transparency (Figure A1 pictures the increase between 1998 and 2010). The most dramatic increases are implemented by central banks that adopted inflation-targeting regimes (Geraats, 2009). However, all types of central banks experienced increases, whereas the monetary base and exchange rate-targeting central banks experienced the smallest expansion (Geraats, 2009). So, especially central banks that are eager to improve their credibility and start to attach more weight to lowering inflation experience a shift towards transparency5.
Blinder (2009) summarizes research papers that look at these effects. These papers however cope with limited sample sizes, restricted time ranges, advanced country biases, and causality problems. Causality can run in both directions since either the more transparent policy of inflation targeting reduces inflation, or the willingness to reduce inflation leads central banks to adopt inflation targeting. Overall, the relations found in the literature provide evidence in favor of better anchoring of inflation expectations and a lower level of inflation, although the results are mixed and highly dependent on the control group that is chosen. An example is Gürkaynak, Levin and Swanson (2006) who look at the U.S, the UK and Sweden and address the question whether a high degree of transparency is related to lower inflation persistence. They find that in the U.S., where the inflation target is not provided to the public, the persistence of inflation is higher. So, their results argue in favor of the adaptive learning view that high transparency enhances monetary transmission efficiency.
5 It is important to keep in mind that inflation targeting is more extensive than increasing transparency alone. Objective transparency is integral in inflation targeting, but inflation targeting is more than just the provision of information (Dincer and Eichengreen, 2014).
14
Kuttner (2001) investigates the market response of policy board meeting-to-meeting announcements for the U.S. and finds that such announcements decrease interest rate volatility and increase monetary policy predictability, thus finding positive effects of procedural transparency. Central banks with more transparency about their objectives experience lower inflation persistence (Kuttner and Posen, 1999). By building up trust through being open about long-term objectives, the bank is able to respond more actively without losing credibility about the commitment to those long-term objectives. Contradictory to the literature that found negative effects of transparency on flexibility, King (1997) and Kuttner and Posen (1999) argue that a central bank that puts more effort in communicating about its long-term objectives and assumptions gains more flexibility in responding to shocks in the short run.
The studies that investigate the impact of implementing transparency-enhancing measures typically find positive results. However, important drawbacks are the limited datasets and the sampling bias towards central banks that come from advanced countries. This leads to problems with generalizing the results. In the next section, the research that first attempts to study the effects of increasing overall transparency is discussed. It becomes clear that these studies cope with similar difficulties.
3.4. The empirical preference for transparency: overall central bank transparency
Now the first set of studies that started constructing central-bank-transparency datasets and indices are discussed. These efforts finally enabled economists to estimate the impact of overall transparency on the performance of macroeconomic variables. The first subsection discusses the construction of the transparency indices and datasets. It is followed by an overview of the results from studies that were made possible by these datasets. These studies mainly find positive results of increasing the degree of transparency. Only at the end of the previous decade, the studies became more tentative in their attitude towards increasing the degree of transparency. This latter line of research is described in chapter 5, which deals with the empirical literature about an optimal intermediate degree of overall central bank transparency.
3.4.1. Frameworks, indices and datasets
Around the early 2000s, several central bank transparency frameworks and indices were constructed. Geraats (2000) proposes a framework for transparency, consisting of five categories of transparency. These categories are procedural, economic, political, policy and operational
transparency, and each category has three subcategories. Eijffinger and Geraats (2002) use this framework to construct a transparency index (Appendix A9 provides this index). They gather data for their index on nine central banks in 2001. Eijffinger and Geraats (2006) extend the time range to 1998-2002 for the same central banks. Another index is constructed by Bini-Smaghi and Gros (2001). They adopt 15 criteria as well and look at six central banks. They are however more focused on frequency and diversity of information disclosures, whereas Eijffinger and Geraats (2002; 2006) put more emphasis on the content. De Haan, Amtenbrink and Waller (2004) apply a similar index to the Bini-Smaghi and Gros (2001) index, for six central banks as well.
The first extensive study on the effects of overall transparency is carried out by Fry, Julius, Mahadeva, Roger and Sterne (2000). They constructed a transparency index with a database of 94 central banks in 1998. There are however some weaknesses. The most important ones are the limited time range of only one time point and a framework that does not capture the full concept of transparency very well.
One should be cautious with the Fry et al. (2000) dataset. Given the recent changes in the degrees of transparency, implementing their dataset might lead to unreliable results (the figure in Appendix A1 shows the increase of transparency for a sample of 120 central banks between 1998 and 2010). Moreover, since the more transparent central banks were largely situated in countries with relatively well-developed institutional frameworks, their results might rather account for this difference than for the effect of transparency (Van der Cruijsen and Eijffinger, 2007). A time-series dataset would mitigate this problem.
Another drawback of the Fry et al. (2000) dataset is that it uses a more restrictive framework for transparency than the one constructed by Geraats (2000). Their framework is criticized for not capturing all of the components of transparency (Van der Cruijsen and Eijffinger, 2007) and is questioned for being biased in favor of finding positive results of transparency (Dincer and Eichengreen, 2014). In their framework, they capture transparency as an average of three elements, which are the openness about policy decisions, the frequency and form of forecasting analysis, and the frequency of other forms of communication like speeches.
So, several frameworks and indices were constructed to measure the overall degree of central bank transparency. They enabled the construction of datasets which were typically restricted in their sample size and/or time range. Next, the research that studies the effects of the overall degree of transparency, with these constrained datasets, is discussed. The more extensive databases, that recently became available, are described in chapter 5 (about empirically estimating an optimal degree of transparency). They enable an increasingly reliable analysis of the topic.
3.4.2. The effects of overall central bank transparency
The developing frameworks, indices and datasets enable the empirical estimation of the effects of increasing overall central bank transparency. Chortareas, Stasavage and Sterne (2002) use the Fry et al. (2000) index and find that procedural and economic transparency both imply lower inflation variability, although this comes at the expense of higher output variability. The release of voting records and forecasts leads to a better predictability of monetary policy by the public. Cecchetti and Krause (2002) use the index as well and find that higher transparency is correlated with lower inflation- and output volatility, although the results are not significant at the usual significance levels.
Other studies (for example Demertzis and Hallett, 2007; Dincer and Eichengreen, 2007; 2009) use the index constructed by Eijffinger and Geraats (2002; 2006). This index measures transparency based on the five elements of central bank transparency that are provided by Geraats (2000). The Eijffinger and Geraats (2006) index encompasses more elements of transparency than the Fry et al. (2000) index, implying less suspicion of ideological biasedness and an increase in the fit of overall transparency. As such, it is this index that is used later on in the estimation sections of this paper.
Demertzis and Hallett (2007) apply the Eijffinger and Geraats (2002) index. They analyze the same nine central banks, but their time period ranges from 1990 to 2001. They look at the correlations between transparency degrees, and the level and variability of output and inflation. They find a statistical relation between transparency and the variability of output and inflation, but not with their average levels. The correlation between transparency and inflation variability is negative and the correlation between transparency and output variability is positive. When they utilize other transparency indices, such as the Bini-Smaghi and Gros (2001) index, or the De Haan et al. (2004) index, their results appear robust.
Demertzis and Hallett (2007) are, however, amongst the first to raise question marks about these findings. They argue that their result for transparency on output variability might be attributable to the effect of central bank independence. When there is a correlation between transparency and central bank independence, for example because of the accountability argument, one needs to control for the influence of this independence. In a crude example, imagine that independent central banks are less inclined to react to output shocks at the cost of price stability. The more independent central bank’s economy faces higher output variability and lower inflation variability. So, (part of) the relations that Demertzis and Hallett (2007) observed, might be due to a lack of controls.
Van der Cruijsen and Demertzis (2007) look at the impact of central bank transparency on inflation expectations. They use the data of Eijffinger and Geraats (2006) with a time period that ranges from 1989 to 2004 for nine central banks. They divide their sample into two categories, high and low transparency, and estimate the effect of the two degrees on inflation persistence. They find a positive link between the low-transparency category and inflation persistence, and less inflation persistence for the more high-transparency category.
Since the early nineties, most research regards increasing transparency as beneficial. The theoretical reasons are the need for democratic accountability due to growing central bank independence and the view that monetary policy transmission is enhanced by transparency. The empirical literature that supports the theoretical monetary policy transmission argument is still young, but accelerates rapidly. Most of this research takes the form of studies that look at particular transparency-enhancing measures, but attempts to estimate the effects of overall transparency are also undertaken. The hosanna in the enhancing-transparency literature however led other authors to question these ongoing benefits. Moreover, the development of overall-transparency indices makes it feasible to test these objections. The next chapters provide the theoretical and empirical arguments in favor of the view that central bank transparency can go too far.
4. Optimal central bank transparency
After having seen the most important arguments against and in favor of increasing transparency, this chapter discusses the literature that argues for an intermediate degree of transparency on a theoretical base. This literature inspired empirical research that studies the effects of increasing overall transparency, while taking into account the possibility of an optimal intermediate degree for central bank transparency. This empirical research is discussed in the next two chapters. It is important to keep in mind that although the theoretical arguments laid the base for the empirical research, this still is a dual process in which the insights from the empirical research also inspire new paths of argumentation.
In this chapter the main theoretical arguments in favor of an intermediate degree are discussed first, consisting of adapted versions of arguments from the sections above, as well as arguments that were not yet part of the discussion. Then, when there is a theoretical case for an optimal degree of transparency, the next part of this chapter is devoted to discussing how one is able to measure the optimal intermediate degree of transparency.
4.1. The case for an optimal degree of overall transparency
The arguments in favor of an optimal intermediate degree of transparency are summarized in this section. The first argument is an existent tradeoff between policy flexibility and the ability of the central bank to anchor inflation. Overexposure to noise and risk, stemming from public information, is the second one. Furthermore, the transmission of monetary policy can be disrupted by an overload of information and a deteriorating reputation of the central bank.
4.1.1. The tradeoff between policy flexibility and managing inflation
One of the main arguments for favoring secrecy is that transparency implies a loss of policy flexibility (Cukierman and Meltzer, 1986). Faust and Svensson (2001) extend the work of Cukierman and Meltzer (1986), by emphasizing the importance of a good reputation for the central bank. The central bank still sets a degree of transparency in which it has to balance between managing inflation expectations and output objectives. Since central banks already experienced a significant increase in their degree of independence, investing in a good reputation became increasingly important. Whereas a high degree of opaqueness was favored in the early literature, following Cukierman and Meltzer (1986), the reputation argument shifts the optimum more towards some intermediate degree of transparency. Jensen (2002) continues on this line of research by assuming a forward-looking private sector and a central bank that decides on a degree of information disclosure. When the central bank is more transparent, this improves the public’s ability to forecast future policy. This increases the sensitivity of inflation expectations to monetary policy, which forces the central banker to become increasingly concerned with inflation. Transparency-increasing central banks that have low inflation-fighting credibility gain inflation-fighting credibility because the disclosure implies commitment and accountability. However, the banks with higher credibility see their output shock stabilization costs rising, without offsetting credibility-benefits, because the marginal revenue of increased credibility is relatively small.
Gosselin, Lotz and Wyplosz (2009) use a model with a central bank that chooses the degree of transparency about its operations. The information the central bank gathers is subject to risk, ranging from low to high. The central bank uses the interest rate to strategically influence market expectations. They find that a central bank that is able to gather information with a medium to high degree of precision optimizes by setting an intermediate degree of transparency. The reason is that the strategic interest rate signaling device loses its power in the higher transparency setting when
risk is low. Full transparency is desirable, on the other side, when risk is high. Full transparency then makes the public aware of the flaws that plague central bank information.
Demertzis and Hallett (2007) study the effects of transparency on inflation and output. They find statistical evidence that transparency influences inflation and output variability, but not their average levels. Their results indicate that more transparency leads to reduced inflation variability. However, output variability goes up when transparency increases. This result provides econometric evidence for a tradeoff between the benefits of anchoring inflation and the loss of policy flexibility to respond to output fluctuations.
4.1.2. Heterogeneous information and coordinated actions
Another line of research that argues for intermediate transparency uses a heterogeneous information structure. Morris and Shin (2002) use a coordination setup where individual welfare is not only dependent on the state of the world, but also on the actions of other actors. In their model, actors have two types of information, namely public and private. The public uses both types of information, which are representing forecasts of fundamentals, to determine its actions. Both types of information contain a certain degree of noise regarding these fundamentals.
In the model of Morris and Shin (2002), the private sector uses public information as a coordination device. So, the public gains utility from acting uniformly. Public information might then be harmful, because the overall weight attached to public information is out of balance with its precision. This causes damage resulting from society’s overexposure to noise of the public information. Moreover, private information is crowded out when there is a lot of public information. Hence, the economy becomes even more vulnerable to the noise of public information.
Svensson (2006), however, argues that under reasonable assumptions regarding the parameters of the model, more public information provision is always beneficial. Svensson (2006) uses the original model of Morris and Shin (2002). In order to have a decrease in expected social welfare from increasing transparency, the private sector should attach more weight to the coordination-device component in its utility function than to the fundamentals-related component. When this is the case, the precision of public information should be very low relative to private information. Equation 1 shows the argument in summary. Here 𝑟 is the weight attached to coordination and (1 − 𝑟) the weight attached to acting in line with fundamentals. Since 𝑓(𝑟) must be positive, the weight attached to coordination should be higher than the weight attached to fundamentals, but it cannot be one.
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 𝑜𝑓 𝑝𝑢𝑏𝑙𝑖𝑐 𝑖𝑛𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛
𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 𝑜𝑓 𝑝𝑟𝑖𝑣𝑎𝑡𝑒 𝑖𝑛𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛< 𝑓(𝑟) ≡ (2𝑟 − 1)(1 − 𝑟) 𝑤ℎ𝑒𝑟𝑒 𝑟 ∈ [
1
2, 1] Equation 1.
From equation 1 it becomes clear that the maximum of the right-hand side is 8 1 when 𝑟 =34. This means that for the original model of Morris and Shin (2002), Svensson (2006) found that the precision of public information should at least be eight times lower than that of private information to have decreasing welfare from increasing transparency.
When it is costly for the private sector to gather and process information, as in Demertzis and Hoeberichts (2007), the optimality of transparency is limited again. Since the costs of gaining public information are lower, private agents tend to overuse this kind of information. This leads to the return of overexposure to noise from public information. Moreover, Morris and Shin (2005) argue that a high degree of public information makes socially optimal forecasting more difficult. This stems from private sector actions that are largely based on the public information provisioned. The high dependence on public information causes prices to rather contain public information than information about fundamentals, for example. This means that the noise of public information increases compared to the setting where this is not taken into account.
Walsh (2007) and Kool, Middeldorp and Rosenkranz (2011) underpin the case for an intermediate degree of transparency, based on overexposure to public information. Walsh (2007) uses a New Keynesian model with economic transparency, with an inflation-targeting central bank. The Calvo imperfect price-flexibility assumption leads firms to adjust their prices before shocks realize. A considerably imprecise central bank, that provides public information, causes inefficient adjustments and is thus costly for the economy. Kool et al. (2011) show with an asset market model that increasing central bank transparency crowds out private sector investment in information gathering. This in turn has detrimental effects on the ability of society to predict monetary policy and make correct forecasts about the economy. Moreover, overestimation of the precision of the central bank decreases the desirability of information disclosure. It crowds out more private information than it should be credited for and thus increases the forecast error variance.
Ehrmann, Eijffinger and Fratzscher (2012) add to the discussion by empirically investigating to what extent increasing transparency functions as a coordination device. They confirm a sizeable reduction in forecaster disagreement resulting from increasing the release of public information. This seems natural since more uniform information is available, but also underpins the case of Morris and Shin (2002), where increased transparency (public information) leads to a higher exposure to public information noise.
4.1.3. Perceived deteriorating quality of information
A line of research related to Morris and Shin (2002) is based on information processing constraints. Van der Cruijsen, Eijffinger and Hoogduin (2010) advocate an intermediate degree of transparency, based on limited information processing abilities by the public that lead to inefficient valuation of information and misinterpretation of information. The limitations lead to errors in private sector expectations, although the disclosed information is additional to the baseline scenario (so, the private sector is able to form the exact same forecast as in the baseline when the additional information is neglected). When the public perceives that the available information leads to inefficient forecasts, it starts relying more on other devices for decision-making, such as past performances of economic variables. Limitations that lead to this perceived deterioration of information are basically twofold. Firstly discussed, is that the increase in information leads to confusion because the public faces binding information processing constraints. Furthermore, the misinterpretation of the complexity and conditionality of information could undermine the reputation of the central bank. Because the problems are stemming from high degrees of transparency, an intermediate degree might be more optimal.
Information overload and confusion
The information overload and confusion argument hinges on the assumption that markets and people are constrained in their capabilities and capacities to process information. Information provided at low levels of transparency leads to improved private sector forecasts, higher predictability of monetary policy and reduced inflation persistence (Gürkaynak, Levin and Swanson, 2006). Intuitively, increasing transparency when transparency is low benefits the perceived quality of information of the public. Besides, transparency-increasing practices undertaken at low levels of transparency are the relatively easy accessible and understandable measures (Van der Cruijsen et al., 2010). This means that the private sector can process them at low costs.
At higher levels of transparency, however, the information becomes increasingly complex. Therefore, it seems natural that the public starts facing problems with understanding the information, with attaching the correct weight to the information, and with focusing on the relevant information. This leads to a less effective transmission of monetary policy. So, inflation expectations become more difficult to influence, private sector forecasts are less predictable, and the forecasts incorporate larger errors.
Another channel, through which a high degree of transparency can cause problems, is the confirmation-bias that is natural to human beings when they process information. People tend to
attach more weight to information that confirms their current knowledge and expectations (Van der Cruijsen et al., 2010). The availability of more diverse information increases the likelihood that certain information can function as a confirmation of current knowledge. This induces people to focus on this information and neglect conflicting information. An example is that information about bandwidths results in people choosing the value in the bandwidth that matches their expectations, instead of the value that is most likely (Van der Cruijsen et al., 2010). More evidence is provided by Van der Cruijsen and Eijffinger (2010). They conducted a survey with a sample of 1800 Dutch households, where they asked them to report inflation expectations. About 80 percent of the households reported highly unlikely inflation expectations. This provides evidence for binding information processing constraints, since the information is freely available. Possible explanations for these constraints are, amongst others, limited understanding ability and public disinterest but also the confirmation-bias with their processing of information about inflation.
The information constraints can make it inefficient to aim at full transparency. It is proven that people make more use of heuristic methods when they are performing more complex tasks, while these methods are accompanied with more misjudgments than other methods. Too much information, that is increasingly complex, makes people rely more on heuristics. This makes the public more eligible to make mistakes (Tversky and Kahneman, 1974). Van der Cruijsen et al. (2010) argue that this could slow down the adaptive learning speed of agents, resulting in higher forecasting errors. Gigerenzer and Gaissmaier (2011) confirm these expectations in their analysis of heuristic decision-making. Providing an intermediate degree of transparency is thus potentially more efficient and effective than a degree that is high. Issing (2005) adds to this discussion by showing that people process certain types of information better than others and are selective in the information they process. This makes it more efficient to provide certain information instead of as much as possible.
Other reasons why too much transparency leads to persistent expectations are the existence of re-optimization costs, the existence of information processing/gathering costs, and rule-of-thumb behavior by part of the public (Van der Cruijsen et al., 2010). These all lead to persistent inflation expectations, either because the public is unconsciously reluctant to accept differing views, or because it is constrained by resources and understanding abilities.
So, an overload of information leads to confusion by the public, since it is unsure about the weight to attach to information, how to interpret the information and how to use the information. When the public gets confused and there is much and complex information available, the negative effects of touching the boundaries of the constraint becomes apparent. Whereas the central bank can steer inflation expectations efficiently when it can transmit its intentions and projections clearly,
the overload and confusion cause noise in this transmission. The confusion leads people to make decisions using heuristic devices, for example. This causes the inflation expectations to be more independent of information provided by the central bank. So, the efficient transmission of monetary policy is distorted.
Perceived deteriorating quality of the central bank
Another argument is that the credibility and the reputation of a central bank might suffer from too much transparency. They are highly dependent on whether the bank achieves its objectives and whether its forecasts are close to the truth. Therefore, the importance to be clear about the conditionality and risk regarding projections is straightforward. The perceived competence of the central banker is likely to deteriorate when it misses its targets on a regular base (Dincer and Eichengreen, 2009). Cukierman (2009) argues, furthermore, that the central bank should only provide information that is within its feasibility and desirability range. The central bank should abstain from providing information that could lead to misinterpretation and might be valued incorrectly. The provision of information should not lead to a misunderstanding about what the central bank knows and tends to achieve. The release of output gap estimates, which are impaired with high measuring difficulties and disputed identification strategies, should for example not be provided in detail since they increase the vulnerability of the central bank’s reputation (Cukierman, 2009). A deteriorating reputation could in turn lead to political pressures such as press statements or even political intervention. The central bank should thus optimize between providing clear information and communicating the conditionality of this information, given the information-processing constraints of the public (Van der Cruijsen et al., 2010).
When the public starts to put too much emphasis on the conditionality and the complexity of monetary policymaking because of the large amount of information provisioned, the perceived quality and usefulness of the information are deteriorating. This in return affects the reputation of the central bank (Issing, 2005; Van der Cruijsen et al., 2010). Meade and Stasavage (2008) empirically find that increasing transparency could incur costs at the expense of the quality of the economic debate. An example is that information about dissent during meetings decreases the effectiveness of policy measures because the central bank staff is perceived as less capable6.
Furthermore, committee members are more likely to behave in a strategic fashion which might be detrimental to the development of optimal policy measures. Van der Cruijsen et al. (2010) provide
6 The relevance of this argument became visible quite recently when ECB president Mario Draghi confirmed that the ECB needed unanimity for its decision to provide the Eurozone economy with an impulse to avoid being trapped by deflation. He argued that dispute about the measures would undermine the confidence and the power of the measures taken (Financieel Dagblad, 2014).
24
the example of the Bank of England which releases inflation fan charts. The information about the complexity and bandwidth might lead the public to perceive the central bank as less capable, because the information about the complexity and bandwidth are interpreted as a central bank that is not able to generate more precise information. This means a deterioration of the monetary transmission efficiency while the underlying quality of the central bank is unchanged.
The extents to which information processing constraints are binding affect the transmission of monetary policy. An overload of information might confuse the public, leading to disrupted expectations and forecasts. Furthermore, the reputation of the central bank might deteriorate when the degree of transparency is too high. When the reputation deteriorates, the perceived quality of (the information provided by) the central bank also deteriorates. People are then induced to use other devices for decision-making. This leads to a deterioration of the monetary transmission mechanism. In the next section, the method of Van der Cruijsen et al. (2010) to link transparency to optimal policy is discussed. With this method, the arguments mentioned in this chapter result in a verifiable hypothesis.
4.2. Relating the degree of transparency to a measure for monetary policy
Now that there is enough reason to believe in an optimal degree of transparency, that is neither secrecy nor full transparency, this section provides the theory to test this hypothesis. A way to test for the optimal degree of transparency is provided by Van der Cruijsen et al. (2010). As described above, too much or too little central bank transparency can mean that the private sector views the quality of its inflation forecasts as too low. This induces the private sector to form its forecasts on other sources like past inflation. Hence, the central bank is less able to influence inflation expectations. Thus, in the ideal case, the optimal degree of transparency is estimated on the perceived quality of inflation forecasts. However, because of the difficulty and subjectivity that come with the construction of such a variable, it is better to use a more feasible variable that approximates the perceived quality of inflation forecasts. Inflation persistence is more attainable to measure than the perceived quality of inflation forecasts. Furthermore, it is a widely used factor in monetary policy analysis. Now the case for using inflation persistence as the approximation for the perceived quality of inflation forecasts will be made.
The last section provided several arguments through which transparency relates to the perceived quality of private sector inflation expectations. An overload of information can cause confusion, making it more difficult for the private sector to form efficient inflation expectations. This
leads the public to perceive a deterioration of the quality of its forecasts. Furthermore, too much information about the conditionality and the risk that coincide with monetary policy making can cause the perceived quality of the central bank to deteriorate. Because the public perceives the information as unreliable or as too conditional, it starts to neglect the information when making its forecasts.
The Morris and Shin (2002) coordination structure leads inflation expectations to be more persistent when there is a lot of (complex) information available. Because the private sector values coordinated action, the uncertainty about which information is used by other market players leads the players to form more backward-looking expectations.
So, if the private sector has a hard time forecasting inflation and perceives its forecasts as of low quality, other factors such as past inflation start to play more important roles in its price-setting behavior. So, a deterioration of the perceived quality of inflation forecasts leads to a higher degree of backward-lookingness.
Increasing backward-lookingness enhances inflation persistence. Van der Cruijsen et al. (2010) and Walsh (2010) illustrate with the following New Keynesian model how backward-lookingness leads to inflation persistence. Here the model is provided, so that one can see this linkage:
𝜋𝑡 = (1 − 𝜙)𝛽𝐸𝑡𝜋𝑡+1+ 𝜙𝜋𝑡−1+ 𝜅𝑥𝑡+ 𝑒𝑡 𝑤𝑖𝑡ℎ 𝑒𝑡~𝑖𝑑𝑑(0, 𝜎2); 0 ≤ 𝜙 ≤ 1 Equation 2.
𝑥𝑡= 𝐸𝑡𝑥𝑡+1− 𝛼(𝑟𝑡− 𝐸𝑡𝜋𝑡+1)) 𝑤𝑖𝑡ℎ 𝛼 > 0 Equation 3.
Equation 2 is the New Keynesian Phillips curve with endogenous inflation persistence. Equation 3 is the forward-looking IS-curve. In this model, 𝜙 measures the degree of backward-looking behavior exhibited by inflation, 𝑥 is the output gap, 𝜋 is inflation, 𝑟 is the nominal interest rate, 𝐸 stands for expectations and 𝑡 indicates the time period, 𝑒 is an exogenous shock to inflation and 𝛽 is the discount factor. When this shock is autoregressive, inflation is not only endogenously persistent, but also exogenously persistent. Still, with or without exogenous persistence, a higher value of backward-lookingness (𝜙 is higher) leads to higher inflation persistence.
Then assume that the central bank minimizes a loss function that takes the form of equation 4, to determine its monetary policy. The expected loss is 𝐿 and 𝜆 is the central bank’s relative preference for output stabilization to price stability. One can see that a lower value for 𝜙, means a lower expected loss for the central bank. This is because inflation adjusts more rapidly towards its
