Contents

List of Figures vii List of Tables ix Notes on the Contributors xii Introductory Comments, De®nitions, and Research on Indexes of Monetary Services Michael T. Belongia 1

Part I New Results in Theory and Practice 1 Beyond the Risk-neutral Utility Function William A. Barnett and Yi Liu 11 2 Neural Networks with Divisia Money: Better Forecasts of Future In¯ation? Robert E. Dorsey 28

Part II Evidence from European Economies and the Planned EMU Area 3 Weighted Monetary Aggregates for the UK Leigh Drake, K. Alec Chrystal and Jane M. Binner 47 4 Weighted Monetary Aggregates for Germany Heinz Herrmann, Hans-Eggert Reimers and Karl-Heinz Toedter 79 5 Simple-sum versus Divisia Money in Switzerland: Some Empirical Results Robert Fluri and Erich Spoerndli 102 6 Weighted Dutch and German Monetary Aggregates: How Do They Perform as Monetary Indicators for the Netherlands? Norbert G. J. Janssen and Clemens J. M. Kool 120 7 Divisia Aggregates and the Demand for Money in Core EMU Martin M. G. Fase 138

Part III Evidence from the Paci®c Basin 8 Broad and Narrow Divisia Monetary Aggregates for Japan Kazuhiko Ishida and Koji Nakamura 173

v vi Contents

9 The Signals from Divisia Money in a Rapidly Growing Economy Jeong Ho Hahm and Jun Tae Kim 200 10 Divisia Monetary Aggregates for Taiwan Yen Chrystal Shih 227 11 Weighted Monetary Aggregates: Empirical Evidence for Australia G. C. Lim and Vance L. Martin 249

Part IV Evidence from North America 263 12 The Canadian Experience with Weighted Monetary Aggregates David Longworth and Joseph Atta-Mensah 265 13 Consequences of Money Stock Mismeasurement: Evidence from Three Countries Michael T. Belongia 292 Index 313 Introductory Comments, De®nitions, and Research on Indexes of Monetary Services Michael T. Belongia

With apologies to Mark Twain, reporting practices of modern central banks beg the expression, `Lies, damned lies, and monetarydata .' Although demonstrably wrong in their construction, simple-sum measures of the money stock continue to be the of®cial data published by central banks and are used to guide policy decisions if monetary quantity variables are part of that process. Moreover, whether by tradition or ease of access, academic research also persists in using simple-sum monetary aggregates to test hypotheses about the effects of money on economic activity. In fairness to all involved, however, the conventional wisdom changes slowly. Indeed, less than thirty years have passed since and Anna Schwartz, in MonetaryStatistics of the United States (1970) ended their discussion of the potential usefulness of weighted aggregates by concluding that, `So far there is only the barest beginning of an answer [of how to do it properly]' p. 152. Indeed, just prior to the publication of Friedman and Schwartz's book, two papers in the Federal Reserve Bank of St Louis Review were dramatic in that they called attention to money at all ± the now-famous study by Leonall Andersen and Jerry Jordan (1968) reporting a primary linkage between money and nominal spending (and the ineffective- ness of ®scal actions), and Karl Brunner's (1968) introduction of the term `monetarism', with a summary of its main principles. In some senses, these papers also marked the mid-point of a sweeping change in orthodox . Only nine years earlier, a paper by Brunner and Anatol B. Balbach presented evidence on money and economic activity that, in many ways, was more compelling than that of the Andersen±Jordan study. Disregarded by attendees of the Western Economic Association meetings of 1959, this paper is virtually unknown today. By 1979, however, arguments and evidence that could have been drawn directly from Brunner and Balbach were the basis of a fundamentally new focus for monetary policy: Most of the world's major central banks adopted monetary aggregate targeting in their efforts to control an accelerating rate of in¯ation. In view of this history, it is perhaps not

1 2 Introduction surprising that superlative indexes of the money stock are going through a typical gestation period prior to their broad acceptance. And, while new ideas are accepted slowly, it can also be said that no debate in economics is really `new'. From the ®rst discussions of the Equation of Exchange, knew that ®nding a measure for `M' was of central importance to empirical work. Schumpeter's (1954) Historyof Economic Analysis, for example, cites numerous views on the de®nition of money from both European and American perspectives over the period 1870±1911. Still, by the early 1930s, Lauchlin Currie was compelled to create his own money supply data to test the hypothesis that restrictive money growth caused the Great Depression. Why? As he put it at the time:

It is a rather startling conclusion that the growth of money under the Federal Reserve System has been largely a matter of accident or, at best, an incidental by-product of the system's other policies. In this connection it is highly signi®cant that while an enormous mass of statistical data is available on the composition of member bank assets, there does not exist in any of the system's publications, so far as I am aware, a series of money.1

It was not until 1948 however, that monetary data for the USA were published by the Federal Reserve Board. It also is notable that those initial data were not revised and improved at the Board, but rather were guided by work by William Abbott under the direction of Homer Jones at St Louis. To many, all this Sturm und Drang produced one of the greater ironies in the history of economic thought: after two decades of effort to get the data reported, to convince people that money mattered, and to push central banks towards the adoption of monetary aggregates as intermediate targets ± everything went wrong. Previously stable velocity functions shifted quickly and erratically. The strong connection between money growth and in¯ation all but disappeared. Wildly alarmist warnings of a resurgent in¯ation were embarrassingly wrong. The role of a quantity variable in implementing monetary policy was discredited. Some observers have interpreted the events since 1980 as a clear repudiation of using monetary quantity variables for the conduct of policy. Others have remained steadfast in their belief that money has potent effects on economic activity, but admit that the published quantity data are erroneous indicators of what the central bank is doing. Still another group has focused on the possibility that fundamental problems of measurement may be responsible for the recent monetary turmoil. This volume is directed to providing empirical evidence on this last proposition.

Progress on monetary measurement since 1970

After Friedman and Schwartz concluded that weighted monetary aggregates had intuitive appeal but were, as yet, not well de®ned in theory, a solution to Michael T. Belongia 3 the problem developed rather quickly. Using Erwin Diewert's (1976, 1978) results on aggregation and index number theory, William Barnett (1978) derived a measure of `prices' (user costs) for the ¯ow of monetary services from a stock of monetary assets. With these prices and the readily-available quantity data for monetary asset stocks, Barnett (1980) then applied an index from Diewert's class of superlative indexes ± the Divisia ± to create weighted monetary aggregates for the USA.2 The principles behind his new measures and a step-by-step guide to their construction were discussed at greater length in Barnett (1982). Although alternative superlative indexes could serve the same purpose, Barnett's careful and thorough examination of the Divisia index led to its adoption in most applications of weighted monetary aggregates.3

De®nitions of common concepts

To avoid redundancies across papers in this book, some common de®nitions are provided here for general reference. Unless speci®cally noted otherwise in an individual chapter, the expressions below are those employed in each study. For greater detail on technical issues associated with adjusting data to match these general expressions, the reader is referred to Farr and Johnson (1985), Dietrich and Kliesen (1992), and the country-speci®c documentation in the individual chapters. A Divisia index of monetary service ¯ows is expressed as: Pk à à ÁlnDivMt ˆ 0:5 sit ‡ si;tÀ1† Álnqit , where sit is the share of total iˆ1 th expenditures on monetary services allocated to the i asset at time t and qit is the quantity of balances in the ith asset category. The expenditure shares are Pk à à de®ned as: sit ˆ uit qit = uit qit †, where uit is the user cost (rental price) of the iˆ1 th à i asset. Nominal user costs are determined as uit ˆ {(Rt À rit )} / (1 ‡ Rt )} Pt , where Rt is the rate of return on a benchmark asset, rit is the own-rate of return th to the i asset and Pt is a cost-of-living index. The marginal tax rate also is included in Barnett's (1978) original derivation but, in the absence of a consistent time series for this concept (at least in the USA), it has been omitted in most subsequent studies.4

With the rit and qit readily-observable from data and the use of the CPI or the GDP De¯ator as a proxy for Pt , treatment of Rt has been the major consideration of many researchers attempting to construct Divisia indexes.5 Indeed, of the three most frequent criticisms of Divisia monetary aggregates, discussions of the benchmark return have been among the most persistent.6 As the return on a completely illiquid asset (one that is incapable of producing any monetary services), Rt should be something akin to the rate of return on human capital in a world without slavery. Without a measure for this, the 4 Introduction search for an empirical proxy has focused on identifying some asset where

Rt  rit 8it . Early studies chose the return on B-grade bonds for this purpose until disin¯ation and an inverted yield curve produced negative user costs from this formulation. Some addressed this problem by adding an arbitrary constant to their chosen value for Rt such that user costs always would be positive. The practice was justi®ed by arguing that the constant captured the effects of service fees, minimum balance requirements and other factors not re¯ected explicitly in own-rates of return. But, while solving the practical problem, it was widely recognised for the arbitrary choice it was. A better solution to this problem has been offered, however, and it has become the predominant approach to measuring the benchmark return.

Viewing Rt as the maximum-available holding-period yield at each point in time and recognising that expenditures on monetary services are part of lifetime utility-maximization problem, consumers will adjust their portfolios of money and goods in every period as they face new vectors of relative prices. In this context, it is possible (if not likely) that different assets will occupy the role of the benchmark asset at different moments in time. This strategy is both consistent with the more general consumer choice problem behind Divisia aggregation, and solves the practical problem of assuring that all values for user costs are non-negative.7 Some chapters in this book have addressed the benchmark rate in different ways, however, and these exceptions are noted in the text.

Other issues The general expressions given above describe how a Divisia aggregate can be constructed and this book evaluates the empirical properties of such an index for a variety of countries. At least one other issue is worth noting, however, before examining this evidence. For the non-specialist, the expression for a Divisia aggregate merely notes summation across k monetary assets. In most applications, including many in this volume, Divisia aggregates are con- structed from the same asset collections used to determine the simple-sum aggregates reported by central banks. Thus, for example, Divisia M3 is evaluated for Germany and Divisia M2 ‡ CDs is evaluated for Japan. Although Barnett's (1982) survey paper expressly notes that a choice criterion should be applied to determine which asset bundles are candidates for aggregation, much work, including some of his own, relies on asset collections identi®ed by ad hoc criteria of central banks. The reason for doing so is related to the discussion in Note 5: with ®rst-order gains from the use of a superlative index, an error in the choice of the best asset collection has been regarded as one of second-order importance. This thinking is ampli®ed by the logic that, unlike simple-sum measures, Divisia aggregation will give a smaller weight to this type of mistake. The last chapter in this book, however, suggests that the consequences of such choices may be larger than has previously been believed. By testing the Michael T. Belongia 5 weak separability condition required for aggregation, Belongia ®nds that some of®cial aggregates fail this test, often because they include one or more time deposit categories. Some of the chapters in this book also test for weak separability as a preliminary step to aggregation. Others accept of®cial central bank de®nitions and construct Divisia measures of them. In either case, most of the authors indicate that the choice of an asset bundle is a question worth addressing, as results generally do vary across broad and narrow measures. An area of research in need of attention encompasses the development of better tests for weak separability and the application of them to monetary data for a variety of countries.

Future research on superlative indexes of money In a tribute to Homer Jones on his retirement as Director of Research at the Federal Reserve Bank of St Louis, Milton Friedman, his former student and teacher, stated that regular publication of data on the money supply and the price level by the St Louis `Fed' probably contributed more than any other factor towards changing the profession's view of money. To repeat an earlier theme, however, it should be noted that twenty-two years (almost to the day!) separated Jones's appointment as Director of Research at St Louis and the Fed's adoption (at least in words) on 6 October 1979 of M1 as its primary intermediate target variable. The professional acceptance of Divisia aggregates appears to be moving at a parallel pace. But, if Friedman's interpretation of events still applies, measures of the money supply published by central banks will be changed when a suf®ciently large and persuasive body of empirical evidence requires that it is to be done. Unfortunately, like the physicist who cannot predict which additional grain of sand will cause the existing pile to cascade, it is not clear when superlative indexes of money will be used more widely. It is clear, however, that more and better evidence will be needed to force this change. It is a long leap from Francis Walker's (1878) homily, `Money is that money does', to the very hard work exempli®ed by the contributors to this volume. Fortunately, economic theory and index number theory have been linked to provide guideposts for careful measurement of the money stock. Theory also provides arguments against the use of simple-sum measures of money. When some perspective is given to the range of issues facing macroeconomists, it seems clear that more attention must be directed to basic measurement problems before the profession can be engaged in informed discussion of important questions. In fact, much of the evidence in this book suggests that several key `problems' in might well disappear if data measurement were to be grounded better in economic theory.

Notes 1. See Lauchlin Currie, The Supplyand Control of Moneyin the United States (1935), p. 54. This classic work, and the large body of his related writings in the Journal of Political 6 Introduction

Economy and the QuarterlyJournal of Economics in the early 1930s, are largely overlooked as a consequence of accusations in 1948 that he had supplied information to Soviet spies while working in the Roosevelt Administration. Although the FBI was never able to substantiate any of the charges brought against him, Currie stated that, `You never live down guilt by association'. He emigrated to Bogota, Colombia in 1950 and attained Colombian citizenship in 1958. He remained there until his death at the age of 91 on 23 December 1993. 2. Naturally, even these developments are not entirely new. The name for the Divisia index (and its capitalisation) comes from the index number work of French statistician FrancËois Divisia (1925). Moreover, the fundamental problems with simple-sum aggregation were discussed in detail by (1922). (1930) also recognised the problems of simple-sum indexes. 3. The primary alternative has been the Fisher Ideal index, which is the geometric average of the Paasche and Laspeyres indexes. It can be expressed as:

Xk Xk 0:5 FIt =FItÀ1†ˆ si;tÀ1 qit =qi;tÀ1 †g=f Sit qi;tÀ1 =qit †gŠ iˆ1 iˆ1

where quantities (qit ), user costs (uit ), and expenditure shares (sit ) are as de®ned in the text. Because this index is expressed in levels, it has often been used to connect periods when new assets are introduced to a Divisia aggregate. This need arises because a Divisia index, which is expressed in growth rates, is unde®ned at these points. Other work has simply constructed Fisher Ideal indexes and then taken their growth rates for empirical applications. 4. Hahm and Kim (Chapter 9 in this volume) address this issue by using after-tax own- rates of return. 5. This, however, has not been the only concern. As discussed by Longworth and Atta- Mensah (Chapter 12 in this volume), for example, legitimate questions can be raised about the effects of minimum balance requirements, service fees and cross- subsidisation of accounts by banks on the observed own-rates of return. Moreover, when aggregating across assets with different maturities, yield curve adjustments must be made so that all own-rates of return are expressed on a common basis. But while these are legitimate issues and should not be dismissed out of hand, it clearly is not correct to use them as a primary defence for the continued publication and use of simple-sum aggregates. The reason is simply one of order of magnitudes. While comparisons of Divisia and simple-sum aggregates show ®rst-order effects from the choice of alternative index numbers (see, for example, Belongia, 1996), the impact of the points noted above on own-rates of return (some of which are off-setting) occurs at a much smaller order of magnitude. For example, studies of the robustness of measurement to choice of alternative values for the benchmark return, Rt , all indicate that its effects are quite small. See, for example, Barnett (1991) and Hahm and Kim (Chapter 9 in this volume). 6. See, for example, comments by Issing (1992) and the response by Belongia (1995). The other two ongoing criticisms are, ®rst, that the empirical differences of Divisia and simple-sum aggregates of the same asset collections are small. The second is that, even if a Divisia aggregate were a better indicator variable, it could not be used as an intermediate target because endogenous interest rates in the Divisia expenditure share weights would eliminate the central bank's ability to control its behaviour. The ®rst objection is strongly rejected by the evidence in this volume and the growing empirical literature in the journals. The second, however, is merely an Michael T. Belongia 7

unsubstantiated assertion. In fact, in one of the few cases where this issue has been investigated, Belongia and Chalfant (1989) ®nd that Divisia aggregates are more controllable by the Fed than their simple-sum counterparts. This is seen in the data

by plotting the growth rate of a money multiplier derived as: Áln mt ˆ Áln DivMt À Áln AMBt , where AMB is the adjusted monetary base. One explanation for lower variance in a Divisia aggregate's multiplier would be a strong negative covariance between portfolio substitutions of the public and the banking system in response to a given change in interest rates. 7. See Barnett et al. (1992), p. 2108, for more discussion of this topic.

References Abbott, William J. (1960) `A New Measure of the Money Supply', Federal Reserve Bulletin, October, pp. 1102±23. Abbott, William J. (1962) `Revision of Money Supply Series', Federal Reserve Bulletin August, pp. 941±51. Andersen, Leonall C. and Jerry L. Jordan (1968) `Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabilization', Federal Reserve Bank of St Louis Review, November, pp. 11±23. Barnett, William A. (1978) `The User Cost of Money', Economics Letters, vol. 1, no. 2, pp. 145±49. Barnett, William A. (1980) `Economic Monetary Aggregates: An Application of Index Number and Aggregation Theory', Journal of , September, pp. 11±48. Barnett, William A. (1982) `The Optimal Level of Monetary Aggregation', Journal of Money, Credit, and Banking, pt 2, November, pp. 687±710. Barnett, William A. (1991) `A Reply to Julio J. Rotemberg', in, M. T. Belongia (ed.), MonetaryPolicyon the 75th Anniversaryof the Federal Reserve System (Norwell, Mass: Kluwer). Barnett, William A., Douglas Fisher and Apostolos Serletis (1992) `Consumer Theory and the Demand for Money', Journal of Economic Literature, December, pp. 2086±119. Belongia, Michael T. (1995) `Weighted Monetary Aggregates: A Historical Survey', Journal of International and Comparative Economics, pp. 87±114. Belongia, Michael T. (1996) `Measurement Matters: Some Recent Results in Monetary Economics Re-examined', Journal of , October, pp. 1065±83. Belongia, Michael T. (1999) `Consequences of Money Stock Mismeasurement: Evidence from Three Countries', Ch. 13, this volume. Belongia, Michael T. and James A. Chalfant (1989) `The Changing Empirical De®nition of Money', Journal of Political Economy, April, pp. 387±98. Brunner, Karl (1968) `The Role of Money and Monetary Policy', Federal Reserve Bank of St Louis Review, vol. 50, no. 7, pp. 9±24. Brunner, Karl and Anatol B. Balbach (1959) `An Evaluation of Two Types of Monetary Theories' in Proceedings of the Thirty-Fourth Annual Conference of the Western Economics Association, September 2±4 (Santa Barbara, Calif.), pp. 78±84. Burns, Arthur and Wesley C. Mitchell (1946) Measuring Business Cycles, National Bureau of Economic Research, Studies in Business Cycles, no. 2 (New York: Columbia University Press) Currie, Lauchlin (1935) The Supplyand Control of Moneyin the United States (Cambridge, Mass.: Harvard University Press). Dietrich, Lynn D. and Kevin L. Kliesen (1992) `Appendix to Thornton and Yue', Federal Reserve Bank of St Louis Review, November/December, pp. 46±52. 8 Introduction

Diewert, W. Erwin (1976) `Exact and Superlative Index Numbers', Journal of Econometrics, May, pp. 115±45. Diewert, W. Erwin (1978) `Superlative Index Numbers and Consistency in Aggregation', Econometrica, July, pp. 883±900. Divisia, FrancËois (1925) `L'indice monetaire et la theorie de la monnaie', Revue d'economie politique, pp. 980±1008. Farr, Helen and D. Johnson (1985) `Revisions in the Monetary Services (Divisia) Indexes of Monetary Aggregates', Special Studies Paper No. 59 (Washington, DC: Board of Governors of the Federal Reserve System). Fisher, Irving (1922) The Making of Index Numbers (Cambridge, Mass.: Houghton Mif¯in). Friedman, Milton (1976) `Homer Jones: A Personal Reminiscence', Journal of Monetary Economics November, pp. 433±36. Friedman, Milton and Anna J. Schwartz (1970) MonetaryStatistics of the United States (New York: Columbia University Press). Issing, Otmar (1992) `Theoretical and Empirical Foundations of the Deutsche Bundes- bank's Monetary Targeting', Intereconomics, November/December, pp. 289±300. Keynes, J. M. (1930) A Treatise on Money, vol. 2 (London: Harcourt Brace & Company). Schumpeter, Joseph A. (1954) Historyof Economic Analysis (New York: Oxford University Press). Thornton, Daniel L. and Piyu Yue (1992) `An Extended Series of Divisia Monetary Aggregates', Federal Reserve of St Louis Review, November/December, pp. 35±46. Walker, Francis A. (1878) Money (New York: Henry Holt & Company). Index

A1 174, 177, 183±5, 196 Banking Law 1989 (Taiwan) 227 A2 174, 177, 186±7, 196 banks Abbey National Building Society 49 Korea 202±3 Abbott, W. 2 Taiwan 227 additive aggregates 82±3 UK 48±9 admissible asset groupings 207±12, 293±4 Barnett, W. A. 3, 15, 123, 124, 125, 139, Afriat, S. 294 252, 261 aggregation 206±7 asset collections 4 across countries 138±9 Barnett continued issues 293±5 benchmark rate 140, 207 problems of 82±3 generalised Divisia index 12, 19±23 simple-sum vs superlative 295 passim transaction costs approach 83±5 index number theory 265, 267 aggregation bias 59 user cost 17, 50 Akaike information criterion 64, 65, Barnett critique 12, 14±15, 23, 24 272±4 Barten, A. P. 209 Alston, J. M. 294, 298 Batten, D. S. 96 Andersen, L. 1, 96 Bayesian information criterion 64, 65 Arrow±Pratt measure of absolute risk Belgium 149±53, 155 aversion 20±1 Belongia, M. T. 28, 96, 115, 140, 174, 217 arti®cial neural networks (ANNs) 28±43 benchmark asset 4, 22, 124, 268±9 asset groupings 4±5, 292±312 benchmark rate of return 3±4, 50, 116, admissible 207±12, 293±4 124, 207, 271 evidence on 295±9, 309 risk aversion 21±2 asset price ¯uctuations 174, 181±2 Taiwan 231±3 augmented Dickey-Fuller (ADF) tests see Benelux 140±3, 145±53, 156±63 Dickey-Fuller tests Berk, J. M. 129, 130 Australia 249±62 Berndt, E. R. 209, 210 business cycle behaviour 259±61 bimodality 257±8 long-run demand for money 249±50, bivariate VAR tests 107±9 254±5, 261 bond yield 113, 114, 114±15 non-linear structures 255±8 Boughton, J. M. 143 statistical properties of the data 253±4 Bowden, R. J. 259±60 statistics 250±2 Bretton Woods system 122 weighted monetary aggregates 252±3 Brock, W. A. 255 Austria 130 Brunner, K. 1 automatic teller machines (ATMs) 104 building societies 48±9 autoregressive (AR) model for in¯ation Building Society Act 1986 (UK) 48 34±41 Bundesbank 129, 130 experience with monetary targeting Balbach, A. B. 1 80±2 Bank of Canada 266, 269 interest rate policy 96±7 bank debentures 231, 232 business cycle 49, 50 behaviour in Australia 259±61 Bank of Korea (BOK) 201±6 turning points 301±6

313 314 Index business demand deposits (DDB) 202, cointegration tests 30±1, 126±7 206, 207 Australia 254±6 Butter, F. A. G. den 143 Canada 279±83, 288 core EMU 152±3, 168 Canada 265±91 Germany 91±3 empirical evidence for Canadian Korea 218±19 monetary aggregates 269±71 Switzerland 106, 107, 110±12, 113, empirical evidence on long-run demand 114 for money equations 279±84, Taiwan 238±40, 241 288; cointegrating vectors UK 54±60, 66±8 279±83; dynamic money demand Competition and Credit Control 48 equations 283; stability of the constant weighting 83, 85 money demand functions 283, consumer 285, 286 decision 15±16 new empirical evidence on short-run demand for monetary assets 15±19 indicator models 271±9; data existence of monetary aggregates for and summary statistics 272; 16±17 de®nitions 271; information consumption-based beta 23 content 272±5; multi-period control error 95±7 forecasting 278±9; one-quarter- controllability 196 ahead forecasting 275±7 Taiwan 243±5, 246 short-run causality tests 284±7 corporate sector 59 capital asset pricing model (CAPM) 12, Corset 48 19±23 CPI 272±88 passim causality tests CPI excluding food and energy 272±88 Canada 270±1, 284±7 passim German and Dutch aggregates credibility, public 28 125±32 Creedy, J. C. 255, 257 Switzerland 106, 107±10 crossover operation 36 UK 65±70 currency in circulation see also Granger causality tests Germany 80, 83±9 central bank money stock at constant Taiwan 228, 232 reserve ratios 80 currency equivalent aggregate (CE) 83, central banks 23±4 298±9, 300, 301, 302 policy rules and 13±14 Currie, L. 2, 5±6 see also under individual names CUSUMSQ test 61, 63 certi®cates of deposit 251, 253 negotiable 230, 232 Davidson, R. 274, 275 Chalfant, J. A. 217, 294, 298 Davidson and MacKinnon J-test see J-test cheque accounts 228, 232 decision, consumer's 15±16 Chong, Y. 279 demand for money 208±9 Chow tests 93 consumer demand for monetary assets rolling Chow tests 283, 285, 286 15±19 stepwise Chow tests 188±9, 192±3 core EMU 138±69; Divisia Christensen, L. R. 210 measures 147±53; simple-sum Chrystal, K. A. 14, 59, 140, 274±5, 276 aggregates 143±7 benchmark rate of return 50 functions for Japan 187±95 summation aggregates compared with instabilities 120±1 Divisia aggregates 270±1 long run in Australia 249±50, 254±5, Cockerline, J. P. 265±6, 268±9, 274±6 261 Index 315

long run in Canada 279±84, 288; error-correction models (ECM) 127 dynamic equations 283, 284; Canada 283, 284 stability of money demand dynamic for UK 60±4 functions 283, 285, 286 forecasting in¯ation in Korea 219±22 long run in Germany 91±3 Germany 91±3 portfolio demand for money 180±1, Switzerland 110, 112 182, 194, 196 errors stability for Taiwan 238±40, 242, control 95±7 243 forecasting 31±2, 36±9, 95±7, 133±4 Denny, M. 210 Euler equations 11±12, 13±14, 17, 18 deposit money banks 202±3 Euro-deposits 81±2 deposit shifts European Monetary System (EMS) 122, Germany 81 130 Switzerland 103±4 European Monetary Union (EMU) Taiwan 227, 234, 245±7 138±69 deregulation 227 aggregation method 138±9 time-deposit interest rates 174, Divisia aggregates 139±40, 141 181±2 money demand with Divisia development institutions 203 measures 147±53; analysis of diagnostic tests 255, 256 results 149±52; estimation results Dickey, D. A. 54, 126 147±9, 153±63; statistical properties Dickey±Fuller (DF) tests 30±1, 93, 126 152±3 Korea 217±18, 218±19 money demand and simple-sum Taiwan 238±40 aggregates 143±7; estimation UK 54, 55 results 145±7, 153±63; general Diewert, W. E. 3, 125, 206, 223 theoretical framework 143±5 direct calculation 139±40, 141 preliminary analysis 140±3 Divisia, F. 139 Exchange Rate Mechanism (ERM) 49 Divisia index passim exchange rate targeting 49 Divisia aggregates vs simple-sum exchange rates aggregates 295, 299±309; ®xed 47±8 comparison of long-run trends monetary aggregates for core EMU 306±9; differences at turning based on 140±63 passim, 164 points 301±6 Netherlands 122±3 extended 12, 18±23 expenditure shares 3, 87, 89 tracking property and risk aversion exports 131 11±12 extended Divisia index 12, 19±23 Dorsey, R. E. 34, 35, 36 initial extension 18±19 Drake, L. 50, 59 user cost of money under risk aversion Dutch central bank (DNB) 122±3 20±3 dynamic error-correction models 60±4 extrapolation tests 193±5 dynamic money demand equations Canada 283, 284 F-test for the exclusion of money from the Germany 92±3 St Louis equation 64, 65 Farr, H. T. 207 encompassing J-tests 279, 280 Fase, M. M. G. 122, 138, 139, 143 Engle, R. F. 30, 107 Faulkner, W. 47 Engle-Granger cointegration tests 107, Federal Reserve 2, 5, 31±2 113, 114, 240, 241 ®nancial innovations 28±9, 227, 267, envelope rate of return 21±2 272 316 Index

Fisher, D. 268, 269, 281 money stock measurement 292; asset Fisher Ideal index 6, 124±5, 206, 267 groupings 296±7, 299, 300; Canadian aggregates 269±70, 271±88 simple-sum vs Divisia ®xed exchange rates 47±8 measures 301±9 Fluri, R. 103, 109 problems of monetary aggregation forecast errors 31±2, 36±9, 95±7, 133±4 82±3 forecasting transaction costs approach 83±5 Canada 275±9, 287±8 GNP velocities 236±8 in¯ation forecasting with neural government bond yield 113±5 networks 28±43; Granger, C. W. J. 30, 66, 107 methodology 34±6; past forecast Engle-Granger cointegration tests 107, record 31±2; results 36±9 113±14, 240±1 M4 61, 62 Granger causality tests 30, 69±70, 127, foreign currency deposits 230, 232 261 France 142, 145, 147, 149±53, 155±6, in ®rst-differenced VARs 106; 160±3 Switzerland 107±10 Friedman, M. 1, 5 Granger representation theorem 66 Fuller, W. A. 54, 126 graphical analysis 51±4 Funahashi, K. 33 growth Fuss, M. 210 Granger causality tests for Switzerland 109±10 GARP (generalized axiom of revealed money growth and in¯ation 28±9, preference) test 294, 295±9, 300 29±31; Taiwan 240±3, 244 Japanese aggregates 174, 183, 186, growth rates of monetary aggregates 196±8, 299, 300 Australia 253±4 GDP Canada 272, 273 de¯ator 73, 74; Canada 272±88 Germany 88±9, 90 passim Japan 174±6, 178±80, 182, 183, 198 nominal 272±9, 280, 287±8 Korea 212±17 real see real GDP Netherlands 125 generalised Divisia index 12, 19±23 simple-sum vs Divisia aggregates generalised normal distribution 256±7 299±309; long-run trends 306±9; genetic algorithm 35±9 turning points 301±6 Germany 79±101, 120, 174 Taiwan 234, 235 control error and projection error 95±7 UK 52, 53; and in¯ation 66±8, 69 core EMU 138, 140±2, 145±53, 153±4, 156±63 Hall, S. G. 55, 56 data and construction of aggregates Hallmann, J. J. 93 85±91 Hendry, D. 279 experience with monetary targeting Herrmann, H. 131 80±2 HL (highly liquid assets) 203, 206 forecasting in¯ation with neural hoarding and dishoarding 143 networks 34, 36±9 Holland, J. 35 German monetary aggregates and the homothetic translog indirect utility Netherlands 121, 129±35; impact function 208±9 on the Dutch economy 129±32 Hornik, K. 33 link between money and prices 93±5 Hostland, D. 265±6, 269±70, 274±5, 276, money demand and its long-run 277 dynamics 91±3 household demand deposits (DDH) 202, money growth and in¯ation 29±30 206, 207 Index 317

Hylleberg, S. 152±3 Korea 202±5 hypothesis testing 210±12 Netherlands 122, 123, 125±9, 132, 134±5 impulse response functions 70±6 Switzerland 106±14 passim income investment institutions 203 cointegration analysis for UK 54, 56, Irie, B. 33 57, 58 Ishida, K. 173, 181 elasticities in Switzerland 111±12 Issing, O. 80 long-run dynamics in the Netherlands 125±9 and money in Australia 249±62 J-test real income see real income Canada 274, 275; encompassing income velocity 113±15, 236±8 J-tests 279, 280 index number, choice of 292±312 Taiwan 240, 244 simple-sum vs Divisia measures 295, UK 64, 65 299±309 Janssen, N. G. J. 129 indicator models 271±9, 287±8 Japan 173±99 indirect calculation 140, 141 analysis of money demand indirect utility function 208±9 functions 187±95; M1 188, 189, in¯ation 28±43 190; M2 ‡ CDs 188±95 control policies 120 construction of Divisia core EMU 143±53 aggregates 174±5 error correction forecasts for developments in Divisia aggregates Korea 219±22 175±87; A1 177, 183±5; A2 177, forecasting with neural networks 186±7; L 177, 182±3, 184; 32±41; methodology 34±6; results M1 175±8; M2 ‡ CDs 178±82 36±9 GARP test 174, 183, 186, 196±8, 299, Germany: control error and projection 300 error 95±7; link between money money stock measurement 292; asset and prices 93±5 groupings 297, 299; simple-sum money growth and 28±9, 29±31; vs Divisia measures 299±309 Taiwan 240±3, 244 Jensen, M. 12, 19, 20, 21, 23, 24 Netherlands 125±35; forecasts and real Johansen, S. 254, 270, 279 income growth 133±4 Johansen±Juselius cointegration past forecast record 31±2 test 126±7, 270, 279±81 Switzerland 106±14 passim Johansen maximum likelihood targeting in Canada 266 cointegration technique UK 48±9; causality tests 66±70; Switzerland 106, 110±12 targeting 49 UK 54±60 see also prices Johansen multivariate cointegration information content procedures 254±5, 256 Canada 272±5 Johnson, D. 207 Switzerland 106, 110±12, 114±15 Jones, H. 2, 5 insurance institutions 204 Jordan, J. 1 interest rates Judd, J. P. 82 core EMU 140, 143±53 Juselius, K. 270, 279 deregulation of time-deposit rates 174, 181±2 Germany 87, 88; interest rate Karnosky, D. S. 96 policy 96±7 King, M. 270, 274±5, 276 318 Index

Kobayakawa, S. 174, 183, 196±8, 198 aggregates 175±8, 195; GARP Kolmogorov, A. N. 33 test 197 Kool, C. J. M. 121, 123, 129, 130 Korea 202, 206 Korea 200±26 Switzerland 103, 104, 105 admissible aggregates 207±12; model M1A 105 speci®cation and data 208±10; asset groupings 296±7, 298 weak separability tests 210±12 Taiwan 228±9, 234, 235, 236±47 passim Divisia monetary aggregates 201±7; M1B 105 aggregation 206±7; benchmark Taiwan 227, 228±9, 234, 235, 236±47 rate 207; own rate of return 207 passim performances of admissible M2 aggregates 212±22; cointegration asset groupings 296±7, 298±9 tests 218±19; empirical Canada 270, 271±88 performance 218±22; error Divisia vs simple-sum measures 301, correction forecasts of in¯ation 302, 306, 308±9 219±22; historical behaviour forecasting in¯ation with neural 212±18 networks 34±9, 40, 41 Kremers, J. J. M. 93 Korea 200±1, 203, 206, 211; performance 212±22 passim L 270 Switzerland 103±4, 105, 108, 110 Japan 174, 177, 182±3, 184, 196, 197 Taiwan 227, 228±31, 234, 235, 236±47 Lagrange multiplier test 60, 61 passim Latan, H. A. 145 M2A 202, 206, 211, 212±22 passim Lawson, N. 49 M2B 206, 211, 212±22 passim Lee, T. H. 255 M2‡ 270, 271±88 Lim, G. C. 257 M2‡CDs 173, 174, 177, 196, 198, 299 Liu, Y. 12, 19, 20, 21, 23, 24 analysis of money demand Ljung±Box (LB) tests 152, 167 functions 188±95 LL 271±88 Divisia vs simple-sum aggregates LL‡ 271±88 178±82, 195±6, 301, 304, 308 long run GARP test 197, 198 demand for money: Australia 249±50, M3 299 254±5, 261; equations for Canada 270, 271±88 Canada 279±84 forecasting in¯ation with neural Divisia vs simple-sum measures 306±9 networks 34±9 dynamics in the Netherlands 125±9 Germany 80±2, 85±99 Switzerland 106, 110±12 Korea 206; performance 212±22 LS (long-term time and savings deposits) passim 202, 206 simple-sum vs Divisia aggregates 301, Lucas critique 13±14, 23±4 303, 306, 308 Lye, J. N. 255 Switzerland 103, 105 UK 48±9 M1 306 M3‡ 270, 271±88 asset groupings 296±7, 298 M4 49±76 Canada 270±1, 271±88 MacDonald, R. 14, 270±1, 274±5, 276 forecasting in¯ation with neural MacKinnon, J. 93, 94, 113, 274, 275 networks 34±9, 40 Major, J. 49 Japan 173, 174; analysis of money Mandel, M. J. 11 demand functions 188, 189, 190; Martin, V. L. 255, 257, 259±60 Divisia and simple-sum maximal Eigenvalue test 255, 256 Index 319

Mayer, W. J. 36 one-quarter-ahead forecasting 275±7 McCallum, B. T. 31, 91 order of integration tests 54, 55 McNees, S. K. 32 Osterwald-Lenum, M. 127 mean absolute forecast errors 31±2, 97 output 106±14 measurement GDP see GDP issues 4±5, 292±312; asset groupings predicting real output growth 109±10 293±4, 295±9, 309; simple-sum vs Divisia measures 295, 299±309 P-star approach 93±5 progress since 1970 2±3 partial adjustment models 187±8 Meltzer, A. H. 31±2 passbook deposits 228±9, 232 Miyake, S. 133 passbook savings deposits 229, 232 monetary aggregation see aggregation Patterson, K. D. 50 monetary assets payments system, technology in 31 consumer demand for 15±19 perfect certainty 11, 17, 19 taxonomy for Korea 201±6 Perron, P. 54 monetary policy personal sector 59 Canada 266 Phillips, P. C. B. 54 Germany 95±7 Plosser, C. I. 30 Netherlands 122±3 Poole, W. 298 Taiwan 227±8, 234±6 portfolio adjustment costs 268 UK 47±9 portfolio demand for money 180±1, 182, monetary targeting 1, 120 194, 196 Australia 249 postal savings re-deposits 229±30, 232 Germany 80±2 Poterba, J. M. 13, 14, 15, 19 Korea 200 prices Taiwan 227, 236 Canada 278±88 passim money demand see demand for money core EMU 143±53 money-market funds 81±2 Korea 218±22 money shocks 73±5 link between money and prices 93±5 multi-period forecasting 278±9 Netherlands 125±9 multiplicative aggregates 82±3 Switzerland 106±14; price with constant weights 83 elasticity 111±12; `structural' Murray, J. 265±6, 268±9, 274±5, 276 price-change equation 112±13 MZM 298±9, 300, 306 see also in¯ation private sector 13±14 negotiable certi®cates of deposit consumer demand for monetary (NCDs) 230, 232 assets 15±19 Netherlands 120±37, 149±53, 154±5 projection error 31±2, 36±9, 95±7, causality tests and long-run dynamics 133±4 125±9 purchasing power parity (PPP) 140±63 data description 123±5 passim 164 forecasts of in¯ation and real income growth 133±4 quantitative ceilings 48 impact of German monetary aggregates quantity theory of money 125 on the Dutch economy 129±32 monetary policy 122±3 Rasche, R. H. 28 neural networks 28±43 rates of return nominal GDP 272±9, 280, 287±8 benchmark see benchmark rate of return non-linearities 255±8 Korea 207 non-parametric tests 54, 55, 223, 294 Taiwanese assets 228±31, 232, 233 320 Index real GDP Sims, C. A. 73 Canada 272±9, 280, 287±8 smoothed Divisia monetary aggregate UK 70, 73, 74 (SM) 85, 87±99 real income speci®cation analysis 152±3, 165±6 core EMU 143±53 spectral analysis 259±61 growth in the Netherlands 129±35 Spencer, P. 85, 268 recursive estimated coef®cients of real Spindt, P. A. 207 GNP 240, 243 SS (short-term time and savings deposits) Reimers, H. -E. 94 202, 206 rental price St Louis equations 64±5 negative 268 stability of money demand functions UK 54, 56, 57, 58 Canada 283, 285, 286 Reserve Bank of Australia 249 Taiwan 238±40, 242, 243 residual autocorrelation 152, 167 stability of velocities reuni®cation 91±3 Taiwan 236±8 risk aversion 11±27 tests for Korea 217±18 Barnett critique 14±15 standard error of residuals 149, 152 consumer demand for monetary assets stationarity 90±1 15±19 stepwise Chow test 188, 189, 192±3 extended Divisia index 18±23 stock market speculation 228±9, Lucas critique 13±14 245±7 user cost of money under 20±3 `structural' price-change equation 106, rolling Chow tests 283, 285, 286 112±13 root mean square forecast error 97, superlative monetary aggregates 275±7 drawbacks 268±9 Rotemberg, J. J. 13, 14, 15, 19, 83, 298 future research 5 Runkle, D. E. 75 vs simple-sum aggregates 295, 299±309 Savin, N. E. 209 theoretical construction 267 savings bonds 231, 232 Swiss National Bank (SNB) 102, 103±4, savings deposits 81, 83±9 105, 115 savings institutions 203±4 Switzerland 102±19, 120 Scadding, J. L. 82 calculation of Divisia monetary Schumpeter, J. A. 2 aggregates 115±16 Schwartz, A. 1 Divisia aggregates used by the SNB Schwarz's Bayesian information criterion 103±4 64, 65 empirical comparisons of simple-sum Schwert, W. G. 30 and Divisia aggregates 106±13; separability tests see weak separability tests Granger causality tests 107±10; Serletis, A. 270, 274±5, 276 income velocity and bond yield shocks, money 73±5 113, 114; information content short run 269 110±12; `structural' price-change causality tests for Canada 284±7 equation 112±13; test set-up and indicator models for Canada 271±9 methods employed 106±7 sight deposits 83±9 revision of monetary statistics 104±5 simple-sum aggregates 82, 266±7 vs Divisia aggregates 295, 299±309; Taiwan 227±48 comparisons of long-run trends construction of Divisia 306±9; differences at turning aggregates 228±34 points 301±6 controlability 243±5, 246 Index 321

linkage between aggregates and vector autoregressive modelling and economic activity 236±43; impulse response functions 70±6 in¯ation and money growth United States (USA) 29, 120, 174 240±3, 244; stability of money forecasting in¯ation with neural demand equations 238±40, 242, networks 34, 36±9 243; stability of velocities 236±8 money stock measurement 292; asset Tatom, J. A. 121, 129, 130 groupings 296, 298±9, 300; technology in payments system 31 simple-sum vs Divisia Teo, L. E. 257 measures 301±9 time deposits 299 user costs 19 deregulation of interest rates in Japan under risk aversion 20±3 174, 181±2 Germany 81, 83±9 variable weighting 83, 85 Switzerland 103±4 Varian, H. R. 294 Taiwan 229, 232 vector autoregressive (VAR) Toedter, K. -H. 83, 87, 94 modelling 65±6 trace test 255, 256 Granger causality tests in ®rst- transaction accounts 104±5, 115±16 differenced VARs 106, 107±10 transactions-orientated monetary and impulse response functions 70±6 aggregate (TM) 83±5, 87±99 lag speci®cation and cointegration 55±6 Treasury Bill rate 73, 75 vector error-correction models (VECMs) Treasury bills (B) 230, 232 270 turning points 301±6 Canada 284±7, 288 information content in Switzerland unit root tests 106, 110±12, 114±15 growth rates of simple-sum vs Divisia velocities 309 measures 308±9 Japanese aggregates: A1 184, 185; income velocity and bond yield in A2 186, 187; L 183, 184; Switzerland 107, 113, 114 M1 175, 178; M2‡CDs 180±2 Korea 217±18 Korea 215±16, 217±18 money demand equations for stability in Taiwan 236±8 Taiwan 238±40 UK 52±5 United Kingdom (UK) 47±78 causality tests of Divisia and simple-sum weak separability tests 4±5 M4 65±70 asset groupings 295±9, 300 cointegration analysis 54±60, 66, 67 Korea 206±7, 210±12 dynamic error-correction models wealth 58±9 60±4 variable in Japanese money demand graphical analysis 51±4 functions 193±4, 195 institutional and policy weighting schemes 82±3, 85 environment 47±9 Winder, C. C. A. 129, 130, 139 monetary data 49±51 St Louis equations 64±5 Yue, P. 103