The New Policy Framework of the Bank of Japan: Central Banking in an Uncertain World
Speech Prepared for a Macroeconomics Seminar at Uppsala University by Kiyohiko G. Nishimura, Member of the Policy Board, the Bank of Japan, on June 27, 2006
June 28, 2006
Bank of Japan
- I. Fundamental Uncertainty and Central Banking in a Transitional Phase
- A. Two-Perspective Strategy
- B. Collective Wisdom and a "Democratic Policy Board"
- 2. A New Policy Framework: Interpretation
It is my great pleasure to speak on the topic of present-day monetary policy in East Asia before an audience at the oldest university in Scandinavia. East Asia and Scandinavia may seem far apart geographically, but the global economy links them in a tight weave of business activity. Thus, though what I say today may sound a bit remote at first, it is not in reality: my subject concerns new challenges all monetary policy authorities face these days around the globe: how to communicate with financial markets and the general public, both home and abroad in an increasingly integrated global economy.
Three months ago, the Bank of Japan ended its unprecedented, unorthodox Quantitative Easing Policy five years after implementing it. The Quantitative Easing Policy was introduced on March 19, 2001, in response to a deterioration in economic activity. It consisted of two parts: the first was to make banks' current account balances at the central bank a policy target and induce them to hold far larger current account balances than required,1 and the second was a commitment to keep this policy until the CPI core inflation rate became stably zero or positive. On March 9, 2006, the Bank of Japan returned to a policy regime of targeting the uncollateralized overnight call rate, and the current target rate is around zero. At the time of this historic change in policy, the Bank also issued a document explaining its policy stance, entitled "The Introduction of a New Framework for the Conduct of Monetary Policy." (I hereafter call it the New Policy Framework.)
The policy change and its accompanying New Policy Framework stirred a lot of discussion, prompting both praise and criticism. In retrospect, the policy change itself was well received by financial markets and the general public alike, as suggested by the relatively smooth transition from the Quantitative Easing Regime that followed. The policy message expounded in the New Policy Framework also won a good reception in general. However, it is also fair to say that there emerged several questions and even criticisms about the policy stance of the Bank of Japan as summarized in the Framework.
I would like to use this special occasion given to me today, to delineate the New Policy Framework in as coherent a way as I can, and to answer some of the typical questions (and related criticisms) directed at it so far. However, it should be noted that the views expressed here are my own as a Policy Board member, and should in no way be interpreted as the agreed-on view of the Policy Board.
In the first part of this speech, I explain that the "fundamental uncertainty" we face in the present world is the most important background of the New Policy Framework. Heightening this fundamental uncertainty, Japan also faces extra uncertainty stemming from the long stagnation in the 1990s. The Framework has two basic principles that aim to meet these challenges: a Two-Perspective Strategy and the collective wisdom of "agreeing to disagree."
In the second part of this speech, I answer commonly asked questions (and criticisms) about the New Policy Framework. The questions are mostly concerned with the Policy Board members' understanding of price stability contained in the Framework. Some also relate to specific indexes that the Bank of Japan monitors.
1 The required balances at that time were around 6 trillion yen, and the target was the range between 30 and 35 trillion yen.
I. Fundamental Uncertainty and Central Banking in a Transitional Phase
We are in the midst of fundamental uncertainty. After a decade of low inflation and stable growth, it now seems that the global economy is proceeding to a different stage, although its destination is still uncertain.
Here we must distinguish two different kinds of uncertainty we now face. The first one, which is now familiar to all of us and often called risk, is formulated as a known probability distribution (with possibly unknown parameters learnable from past experience) of, say, stock prices and the GDP growth rate in the near future. The past decades have witnessed significant advances in dealing with this kind of uncertainty, especially in financial markets, and their application to economic policy analysis. Value at risk and option pricing, for example, have become part of our day-to-day conversation, at least in the financial quarters of the economy.
However, there is another kind of uncertainty, which deserves the name of true fundamental uncertainty. This is uncertainty that cannot be reduced to a known distribution, often called Knightian uncertainty in recognition of the writing of Frank Knight. Not only are we uncertain about the future value of stock prices and GDP growth in a known probability distribution, but also we do not have clear knowledge of their probability distribution itself. To put it differently, we do not have clear confidence in "stochastic models" (whether mathematical or nonmathematical) that we often use to describe economic activities in the real world.2 And it is not so uncommon for policymakers to make decisions in circumstances in which some unknown unknowns might occur. (This may recall the intriguing remarks of U.S. Secretary of Defense Rumsfeld about weapons of mass destruction in Iraq, which appear at the head of this section.)
This fundamental uncertainty is particularly keen in the case of Japan. The Japanese economy stagnated for almost twelve years after 1990, including a mini-crisis at the end of the 20th century. The period is often called Japan's "Lost Decade," and it is unprecedented compared with the rapid and then relatively stable growth that preceded it. There has been much discussion about its possible causes and remedies, but it is fair to say that the jury is still out.
The Japanese economy finally started recovering around 2003, after a long and painful economy-wide restructuring of business and employment practices. We are apparently in a transition period from the quagmire of nonperforming loans and other problems to a normal, growing state of the economy, and we are left with fundamental uncertainty about the working (or the "true stochastic model") of the Japanese economy as of today.
The way I see it, the Policy Board of the Bank of Japan has adopted two basic principles to deal with this fundamental uncertainty.
2 One way to formulate policy decisions under this fundamental uncertainty is a maximin criterion with multiple priors. See I. Gilboa, and D. Schmeidler, "Maximin Expected Utility with a Non-Unique Prior," Journal of Mathematical Economics, 18 (1989), pp. 141-153. It should be noted that, if one were sure about the relative probability of each prior, the problem would be no longer of fundamental uncertainty, and would be reduced to ordinary (though complicated) expected utility maximization. What makes fundamental uncertainty different from the ordinary type is that we do not have a clear idea of the relative "probability" assignable to each prior.
A. Two-Perspective Strategy
The first principle can be called a "Two-Perspective Strategy," which bears some resemblance to the Two-Pillar Strategy at the European Central Bank, though the former has a broader scope. This strategy involves a two-step policy consideration: policy should be guided by a most probable scenario of the future course of the economy, but at the same time it should incorporate "risk factors" (including "Knightian uncertainty factors") that affect the economy adversely if they happen to materialize, even though they may not be in the main scenario. This principle is stated in Section 1 (2) of the document issued on March 9, which is found in the Appendix of this text.
Firstly, the Policy Board will look at the most probable scenario of economic development in the immediate future ("one to two years"), and then assess the appropriateness of the current policy rate in achieving "sustainable growth under price stability."
Secondly, the Policy Board will also examine "in a longer term" "risk factors that will significantly impact economic activity and prices when they materialize although the probability is low."
Here the word "risk" is used rather vaguely as in everyday language, not in the context of the risk-uncertainty distinction of Frank Knight. In fact, "risk factors" here include Knightian uncertainty of unknown unknowns. We may be, say, 95% sure that the main scenario (the stochastic macro model) eventually will prevail, but with 5% certainty we fear some other unpleasant scenarios may instead materialize.3 We should take these possibilities, which may be remote but may have grave consequences if they materialize, properly into our policy consideration.4
3 This is one example of the epsilon contamination of confidence. The axiomatic foundation of the epsilon contamination is given in K. G. Nishimura, and H. Ozaki, "An Axiomatic Approach to Epsilon-Contamination," Economic Theory, 27 (2006), pp. 333-340. An example of the epsilon contamination and its impacts on economic agents' behavior is found in K. G. Nishimura and H. Ozaki, "Search and Knightian Uncertainty," Journal of Economic Theory, 119 (2004), pp. 299-333.
4 If the most probable scenario were some dynamic general equilibrium macroeconomic model, then we could set up an explicit model of the maximin optimization with epsilon-contamination (or perturbation) of the main model. However, it is still a long way to obtaining a reliable main scenario. Thus, actual implementation of the Two-Perspective Strategy is more "qualitative" than quantitative in its nature.
B. Collective Wisdom and a "Democratic Policy Board"
The second principle is collective wisdom that is flexible enough to adjust to the changing environment. The basic idea is quite simple: two heads are better than one, and four eyes see more than two. The Bank of Japan has nine Policy Board members of diverse background beyond financial markets. The Bank's staff are surely among the most competent researchers of various economic data and the most knowledgeable experts in forecasting future economic conditions. However, Policy Board members may have knowledge that differs from that of the Bank's staff, and this may contribute to prudent policy decisions. This is particularly important when we face the fundamental uncertainty I have described. Policy Board members may disagree about the model that underlies economic forecasts, but this diversity itself is a very important source of information about underlying uncertainty.
Thus, the collective wisdom principle allows different views of the economy within the Policy Board. No member knows nor pretends to know the "true model" of the economy. Experts' diverse views are better than one opinion committed to one particular model in the face of fundamental uncertainty. In essence, the collective wisdom here is to agree to disagree: different views about the economy become common knowledge among Policy Board members, and ultimately between the Bank of Japan and the general public. This is the essence of the "Democratic Policy Board" we now see in Japan.5
It should be noted that diverse views do not at all lead to diverse, conflicting policy subscriptions. Diverse views are in fact "multiple priors" alluded to before, and through the Two-Perspective Strategy (a sort of "maximin" criterion as I have already explained), we usually reach a unanimous policy decision although there are occasional dissenting votes.
Here arises an important communication policy issue. We should minimize possible confusion about future monetary policy stemming from possible "conflicting signals" by Policy Board members. To avoid this unpleasant consequence of the democratic Policy Board system, we should make it clear that only the Governor, who is the chairman of the Policy Board, represents it with respect to monetary policy. The basic message is clear: do not speculate about the Policy Board's policy direction based on one quote from one member. His/her vote is only one of nine, though admittedly it contains important information about the direction. It is informative but not decisive. The market should infer the future course of monetary policy mostly from the Governor's statements and the minutes of the Monetary Policy Meetings. The market should be accustomed to this working of the "democratic Policy Board." To put it differently, we have a dictum: Policy Board members avoid expressing their own, independent views about immediate policy rate changes beyond those stated in the minutes of the Monetary Policy Meetings and official reports of the Policy Board, as much as possible. However, they speak up in terms of their views on future economic developments. I know that it is a delicate balance, and both the Policy Board and the general public must learn this balance going forward.
5 This principle is also time-consistent. In contrast, the Policy Board's commitment in stipulated rules is not. Policy Board members are political appointees, and their composition changes over time. This means the Policy Board's explicit commitment to certain rules is not literally enforceable.
2. A New Policy Framework: Interpretation
I now turn to some commonly asked questions and/or criticisms about the New Policy Framework, especially those related to the understanding of price stability therein (i.e., Section 2, titled "The Bank's Thinking on Price Stability").
Now it is widely acknowledged that the Framework is not an inflation target that should be achieved in a certain period of time, nor the definition of price stability in that the Bank of Japan commits institutionally. Therefore, some may ask what role this understanding plays. The answer is now apparent in our discussion of the background of the Framework: it is a collection of Policy Board members' views (based on their own formal or informal stochastic models) about medium- to long-run price stability in the transitional phase of the Japanese economy as of today. Though I do not think these views change year by year, there is a possibility of change in the future when the economy is firmly in a normal state. Actual policy shall be determined by the Two-Perspective Strategy (a sort of a maximin criterion, as explained), based on this collection of views.6
That said, I give my own answers to some commonly asked questions about this framework in the following. However, as explained before, the "understanding of price stability" is a collection of Policy Board members' views (including my own). I am not in the formal position of representing the Policy Board when I answer questions. So the following answers should be interpreted as my interpretation of probable answers that the majority of Policy Board members might have in mind.
A typical question or criticism about the understanding of price stability in the framework is as follows.
Since this issue seems very important in understanding the Two-Perspective Strategy, I will raise this question/criticism first and give a rather detailed answer to it.
The bottom line of my answer is that there is a basic difference between Japan and other countries in inflation sensitivity. An inflation objective of 2% might possibly upset some Japanese firms and households, which are backward-looking and used to historical price stability of around 1%. Here I am not talking about the deflationary period of the late 1990s and the early 2000s, in which the average CPI inflation was in the negative range. I am referring to the period between 1985 and 1997, when the average CPI inflation was 1.2%.
For those firms/consumers with a lingering memory of the period of the "bubble economy," the inflation objective of 2% might indicate an inflationary bias, or rather, an asset inflationary bias of monetary policy. Thus, it might trigger some speculative, potentially distortional activities, especially in property markets. We should remind ourselves of the fact that the CPI inflation rate was just around 2% in 1989 and 1990, a final stage of the "bubble economy" where, for example, commercial property prices in Tokyo rose eight-fold between 1981 and 1990.7 About a decade later in 1999, Tokyo commercial property prices were one-eighth of the peak value, and such a precipitous decline in properties' collateral values no doubt strained the banking system.
From the standpoint of the Two-Perspective Strategy, or in other words, the so-called "risk management" in monetary policy, we should give proper attention to this possibility. However remote its possibility is, when it happens its impacts on economic activities are detrimental, as amply exemplified by the bursting of the "bubble economy" that afflicted the Japanese economy during the lost decade.8
The second group of often-asked questions is related to the first one and about what kind of consideration is given to possible upward biases in the CPI and the safety-margin against the zero nominal bound of interest rates.
6 This is also consistent with the fact that the Policy Board regularly publishes the collection of its members' views about one- and two-year-ahead forecasts of main economic variables, namely, GDP growth, the Corporate Goods Price Index, and the core CPI.
7 Transaction price data for Tokyo commercial properties as well as residential properties are found in K. G. Nishimura and C. Shimizu, "Biases in Appraisal Land Price Information: The Case of Japan," Journal of Property Investment and Finance, 24 (2006), pp. 150-175. The paper also shows large biases in official property price data.
8 This argument rests on the presumption that there may be a sizable number of backward-looking firms/households (and possibly, some politicians). There is in fact some evidence of this. One strand of evidence is that a sizable coefficient is found for a backward-looking inflation component in a hybrid Phillips curve containing both forward-looking and backward-looking inflation components, although there is little institutional wage rigidity as of today in Japan. Labor unions are weak, and nominal wages are not rigid. Prices are also relatively flexible and decline sharply in the markets of household appliances and ICT products. All in all, the hybrid Phillips curve result suggests that some firms/consumers are truly backward-looking, with decisions based on past experience or the status quo.
The first issue here is the so-called "safety-margin" over the zero inflation rate. The safety margin depends on various factors, including institutional rigidity, especially in labor markets. In addition, one may argue that the Japanese experience of the late 1990s shows the necessity of a wide safety margin if financial systems are in deep trouble due to extremely distressed property markets, in allocating funds efficiently. However, at this stage of recovery, Japanese financial systems are finally back on track, and there is no sign of institutional rigidity in labor markets. This implies a necessary safety margin is rather small compared with an economy of rigid wage and price setting. In that sense, the safety margin right now is not as large as a simple extrapolation of past experience into the future may suggest. Moreover, the economy is now recovering firmly, which implies a positive equilibrium real interest rate. Since the possibility of adverse negative shocks of the magnitude of the 1990s is now not as large as before, the welfare benefits of having large safety margin are likely to be much smaller than the welfare costs of pursuing higher inflation.
The second issue is the direction of bias in CPI indexes. In this respect, the Boskin Report of 1996 has had a tremendous influence in shaping what we think about CPI indexes. Based on studies in the 1980s and early 1990s, the report concluded that there was a 110-basis-point upward bias in U.S. CPI indexes, compared with so-called superlative indexes. A similar study carried out in Japan by a Bank of Japan staff person estimated a 90-basis-point upward bias based on the 1990-base-year CPI indexes. However, almost ten years have passed since then, and statistics have improved considerably in both countries, especially in Japan. Remaining upward biases seem small in Japan, at most in the range of 15-30 basis points. I suppose similar efforts by statistical authorities in other countries, and remaining biases are now considerably smaller than many casual observers may assume.9
In contrast, there emerges the possibility of a downward bias due to over-adjustment of quality in using a hedonic approach to PCs and related products. The Japanese Statistical Bureau has adopted a hedonic approach in estimating quality-adjusted prices of desktop and laptop PCs and printers. In doing so, it has reduced PC and printer quality-adjusted prices by more than 30% each year currently. Since actual prices of PCs and printers do not change so much, this is mostly due to quality improvement. That is, according to the Statistical Bureau, the rate of utility increase due to quality improvement is 30% per year. However, it is hard for laymen to believe that this year's PCs produce 30% higher utility than last year's PCs, since they carry out the same old tasks of word processing and spreadsheet calculation. This may suggest over-adjustment of quality in PCs and printers. The possibility is all the more important since the Japanese weight of PCs and printers is much higher than the U.S. counterpart.10
In this respect, it is not so surprising if some member(s) of the Policy Board see CPI biases going possibly in either directions, or view price stability as a zero-percent change in the CPI. However, as a reading of the document showed, some other members did not agree on this. This clearly shows that it is tricky to define price stability and to find a comfort zone including a safety margin. The Policy Board thus determined not to adopt one magic number or range that conceals inherent difficulty.
The third typical question is about food and energy in CPI indexes.
9 As a former member of the Statistical Council responsible for government statistics, I feel deep disappointment that few economists and financial journalists appreciate these changes, especially in Japan.
10 My own very rough estimate of this possible downward bias in the CPI is around 10 basis points as of December 2005. See K. G. Nishimura, "Upward Biases in CPI Are Now Substantially Reduced," Nihon Keizai Shimbun, December 6, 2005.
It is very important to distinguish between what kind of CPI indexes central banks should stabilize in the medium to long run, and what kind of CPI indexes they should monitor to achieve the goal of price stability. From the viewpoint of economics, it is clear that the CPI index to be stabilized should be one including all items, that is, the headline CPI. However, since some items like fresh food in the headline CPI fluctuate according to weather conditions that are beyond the reach of monetary policy, central banks usually monitor a subset of the headline CPI that tracks the movement of the headline CPI in the medium to long run very well. This subset of the CPI is often called core CPI.
It is clear now that price stability should be framed in the context of the headline CPI, in the manner followed by the Bank of Japan. In order to achieve long-run stability of the headline CPI, central banks should monitor the core CPI, which tracks the long-run movement of the headline CPI. This subset of the CPI, the core CPI, may differ from country to country. And in fact, that is the case. Recently, Shigenori Shiratsuka of the Bank of Japan carefully studied how various subsets of the CPI tracked the headline CPI in Japan and found that the CPI excluding fresh food (the current official Japanese core CPI) has a much better track record than the CPI excluding food and energy, a U.S./European mold of the core CPI.11 In fact, the latter showed the worst performance among four candidate subsets of the CPI he considered. This clearly warns of the pitfall of simplistic, blindfold application of other countries' best practice to a country with different preferences and institutions.
Finally, there is a commonly asked question about the future evolution of the framework.
11 S. Shiratsuka, "Core Indexes of CPI," Bank of Japan Review, 2006-J-7. Shiratsuka examined the current core CPI (CPI excluding fresh food), core CPI in the U.S./European model (CPI excluding food and energy), CPI excluding agricultural produce, energy and utilities, and a 10%-truncated average.
The answer to this question is "yes, but very carefully." We are in a transition from a very distressed, stagnant economy to a normal, growing one. The past negative legacy of the bubble economy and the lost decade that followed is still lingering around us. The Bank of Japan itself is in the process of rebuilding its credibility. This implies we should move forward very carefully.
The Policy Board made explicit this time-varying nature of the understanding of price stability in its announcement of a new policy framework. An annual reexamination of the understanding signifies the necessity of a regular review of its views about the economy.
One may argue that such a possibility of changes undermines financial markets' confidence about the future course of monetary policy implicit in the understanding of price stability and thus increases risk premiums reflecting uncertainty surrounding future policy. I am aware of this negative effect. However, in facing fundamental uncertainty about the future, I believe that the benefits of flexibility outweigh those of predictability.
I would like to close my speech with the following remarks. Carl Linnaeus of Uppsala University, or Carl von Linne, is a great figure of the 18th century who gave names to all living things, or more precisely speaking, started a system for doing so. In this way, Linne reduced fundamental uncertainty in identification surrounding natural sciences. Central banking, however, is not like the natural sciences, and we cannot reduce fundamental uncertainty about the future by simply "identifying" a magic number of 2% or so of core CPI inflation. What is more important than a quest for the magic rate to cure all problems is to face squarely the dynamic complexity of economic systems. I believe that the New Policy Framework of the Bank of Japan is the move in the right direction, though it is a modest start.