Rational Expectations and Inflation (Third Edition) - Hardcover

Sargent, Thomas J.

 
9780691158709: Rational Expectations and Inflation (Third Edition)
Hardcover
ISBN 10: 0691158703 ISBN 13: 9780691158709
Publisher: Princeton University Press, 2013

Synopsis

This collection of essays uses the lens of rational expectations theory to examine how governments anticipate and plan for inflation, and provides insight into the pioneering research for which Thomas Sargent was awarded the 2011 Nobel Prize in economics. Rational expectations theory is based on the simple premise that people will use all the information available to them in making economic decisions, yet applying the theory to macroeconomics and econometrics is technically demanding. Here, Sargent engages with practical problems in economics in a less formal, noneconometric way, demonstrating how rational expectations can satisfactorily interpret a range of historical and contemporary events. He focuses on periods of actual or threatened depreciation in the value of a nation's currency. Drawing on historical attempts to counter inflation, from the French Revolution and the aftermath of World War I to the economic policies of Margaret Thatcher and Ronald Reagan, Sargent finds that there is no purely monetary cure for inflation; rather, monetary and fiscal policies must be coordinated. This fully expanded edition of Rational Expectations and Inflation includes Sargent's 2011 Nobel lecture, "United States Then, Europe Now." It also features new articles on the macroeconomics of the French Revolution and government budget deficits.

"synopsis" may belong to another edition of this title.

About the Author

Thomas J. Sargent is professor of economics at New York University. His books include Robustness and The Big Problem of Small Change (both Princeton). He was awarded the 2011 Nobel Prize in economics.

From the Back Cover

"Sargent's work on inflation remains central to cutting-edge research in economics as well as to current and momentous policy decisions. Are the United States and Europe headed toward inflation with our large and intractable deficits? Will the European currency union survive? The breakthrough theoretical insights and brilliant case studies in this book are still the foundations that anyone thinking about these questions needs to read, and then to read again."--John H. Cochrane, author of Asset Pricing

"Rational Expectations and Inflation is a collection of classic articles on the subject, several of which were explicitly cited in the scientific background to Sargent's Nobel Prize. The contribution of this book is great."--Marco Bassetto, Federal Reserve Bank of Chicago

From the Inside Flap

"Sargent's work on inflation remains central to cutting-edge research in economics as well as to current and momentous policy decisions. Are the United States and Europe headed toward inflation with our large and intractable deficits? Will the European currency union survive? The breakthrough theoretical insights and brilliant case studies in this book are still the foundations that anyone thinking about these questions needs to read, and then to read again."--John H. Cochrane, author ofAsset Pricing

"Rational Expectations and Inflation is a collection of classic articles on the subject, several of which were explicitly cited in the scientific background to Sargent's Nobel Prize. The contribution of this book is great."--Marco Bassetto, Federal Reserve Bank of Chicago

Excerpt. © Reprinted by permission. All rights reserved.

Rational Expectations and Inflation

By Thomas J. Sargent

PRINCETON UNIVERSITY PRESS

Copyright © 2013 Princeton University Press
All rights reserved.
ISBN: 978-0-691-15870-9

Contents

List of Figures............................................................xi
List of Tables.............................................................xiii
Acknowledgements...........................................................xv
Preface to the Third Edition...............................................xvii
Preface to the Second Edition..............................................xix
Preface to the First Edition...............................................xxi
1. Rational Expectations and the Reconstruction of Macroeconomics..........1
2. Reaganomics and Credibility.............................................17
3. The Ends of Four Big Inflations.........................................38
4. Stopping Moderate Inflations: The Methods of Poincaré and Thatcher......111
5. Some Unpleasant Monetarist Arithmetic...................................162
6. Interpreting the Reagan Deficits........................................197
7. Speculations about the Speculation Against yhe Hong Kong Dollar.........211
8. Six Essays in Persuasion................................................228
9. Macroeconomic Features of the French Revolution.........................248
10. United States Then, Europe Now.........................................297
References.................................................................339
Author Index...............................................................357
Subject Index..............................................................361

Excerpt

<h2>CHAPTER 1</h2><p><i>Rational Expectations and theReconstruction of Macroeconomics</i></p><p>The government has strategies. The people havecounterstrategies.</p><p>Ancient Chinese proverb</p><br><p><i>Behavior Changes with the Rules of the Game</i></p><p>In order to provide quantitative advice about the effects of alternativeeconomic policies, economists have constructed collections ofequations known as <i>econometric models</i>. For the most part thesemodels consist of equations that attempt to describe the behavior ofeconomic agents—firms, consumers, and governments—in termsof variables that are assumed to be closely related to their situations.Such equations are often called <i>decision rules</i> because theydescribe the decisions people make about things like consumptionrates, investment rates, and portfolios as functions of variables thatsummarize the information people use to make those decisions.For all of their mathematical sophistication, econometric modelsamount to statistical devices for organizing and detecting patternsin the past behavior of people's decision making, patterns that canthen be used as a basis for predicting their future behavior.</p><p>As devices for extrapolating future behavior from the past undera given set of rules of the game, or government policies, thesemodels appear to have performed well. They have not performedwell, however, when the rules changed. In formulating advice forpolicymakers, economists have routinely used these models to predictthe consequences of historically unprecedented, hypotheticalgovernment interventions that can only be described as changesin the rules of the game. In effect, the models have been manipulatedin a way that amounts to assuming that people's patterns ofbehavior do not depend on those properties of the environmentthat government interventions would change. The assumption hasbeen that people will act under the new rules just as they haveunder the old, so that even under new rules, past behavior is stilla reliable guide to future behavior. Econometric models used inthis way have not been able to predict accurately the consequencesof historically unparalleled interventions. To take one recent example,standard Keynesian and monetarist econometric modelsbuilt in the last 1960s failed to predict the effects on output, employment,and prices that were associated with the unprecedentedlarge deficits and rates of money creation of the 1970s.</p><p>Recent research has been directed at building econometric modelsthat take into account that people's behavior patterns will varysystematically with changes in government policies—the rules ofthe game. Most of this research has been conducted by adherentsof the so-called hypothesis of rational expectations. They modelpeople as making decisions in dynamic settings in the face of well-definedconstraints. Included among these constraints are lawsof motion over time that describe such things as the taxes peoplemust pay and the prices of the goods they buy and sell. The hypothesisof rational expectations is that people understand theselaws of motion. The aim of the research is to build models thatcan predict how people's behavior will change when they are confrontedwith well-understood changes in ways of administeringtaxes, government purchases, aspects of monetary policy, and thelike.</p><br><p>The Investment Decision</p><p>A simple example will illustrate both the principle that decisionrules depend on the laws of motion that agents face and the extentthat standard macroeconomics models have violated this principle.Let <i>k<sub>t</sub></i> be the capital stock of an industry and τ<sub><i>t</i></sub> be a tax rate on capital.Let τ<sub><i>t</i></sub> be the first element of <i>z<sub>t</sub></i>, a vector of current and laggedvariables, including those that the government considers whenit sets the tax rate on capital. We have τ<sub><i>t</i></sub> [equivalent to] <i>e<sup>T</sup>z<sub>t</sub></i>, where <i>e</i> is theunit vector with unity in the first place and zeros elsewhere. Let afirm's optimal accumulation plan require that capital acquisitionsobey</p><p>[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1.1)</p><p>where <i>E<sub>t</sub>τ<sub>t+j</sub></i> is the tax rate at time <i>t</i> that is expected to prevail attime <i>t + j</i>.</p><p>Equation 1.1 captures the notion that the demand for capitalresponds negatively to current and future tax rates. Howeverr,equation 1.1 does not become an operational investment scheduleor decision rule until we specify how agents' views about thefuture, <i>E<sub>t</sub>τ<sub>t+j</sub></i>, are formed. Let us supppose that the actual law ofmotion for <i>z<sub>t</sub></i> is</p><p><i>z<sub>t+1</sub> = Az<sub>t</sub></i> (1.2)</p><p>where <i>A</i> is a matrix conformable with <i>z<sub>t</sub></i>. If agents understandthis law of motion for <i>z<<<<sub>t</sub></i>, the first element of which is τ<sub><i>t</i></sub>, then theirbest forecast of τ<sub><i>t+j</i></sub> is <i>e<sup>T</sup>A<sup>j</sup>z<sub>t</sub></i>. We impose rational expectations byequating agents' expectations <i>E<sub>t</sub>τ<sub>t+j</sub></i> to this best forecast. Upon imposingrational expectations, some algebraic manipulation impliesthe operational investment schedule</p><p><i>k<sub>t</sub> = λk<sub>t-1</sub> - αe<sup>T</sup>(I - δA)<sup>-1</sup>z<sub>t</sub></i> (1.3)</p><p>In terms of the list of variables on the right-hand side, equation(1.3) resembles versions of investment schedules that were fit inthe heyday of Keynesian macroeconomics in the 1960s. This isnot unusual, for the innovation of rational expectations reasoningis much more in the ways equations are interpreted and manipulatedto make statements about economic policy than in the lookof the equations that are fit. Indeed, the similarity of standardand rational expectations equations suggests what can be shownto be true generally: The rational expectations reconstruction ofmacroeconomics is not mainly directed at improving the statisticalfits of Keynesian or monetarist macroeconomics models over givenhistorical periods and that its success or failure cannot be judgedby comparing the <i>R</i><sup>2</sup>'s of reconstructed macroeconomics modelswith those of models constructed and interpreted along earlierlines.</p><p>Under the rational expectations assumption, the investmentschedule (equation (1.3)) and the laws of motion for the tax rate andthe variables that help predict it (equation (1.2)) have a commonset of parameters, namely, those of the matrix A. These parametersappear in the investment schedule because they influence agents'expectations of how future tax rates will affect capital. Further, noticethat all of the variables in <i>z<sub>t</sub></i> appear in the investment schedule,since via equation (1.2) all of these variables help agents forecastfuture tax rates. (Compare this with the common econometricpractice of using only current and lagged values of the tax rate asproxies for expected future tax rates.)</p><p>The fact that equations (1.2) and (1.3) share a common set ofparameters (the A matrix) reflects the principle that firms' optimaldecision rule for accumulating capital, described as a function ofcurrent and lagged state and information variables, will dependon the constraints (or laws of motion) that firms face. That is, thefirm's pattern of investment behavior will respond systematicallyto the rules of the game for setting the tax rate τ<sub><i>t</i></sub>. A widely understoodchange in the policy for administering the tax rate can berepresented as a change in the first row of the A matrix. Any suchchange in the tax rate regime or policy will thus result in a changein the investment schedule (equation (1.3)). The dependence ofthe coefficients of the investment schedule on the environmentalparameters in matrix A is reasonable and readily explicable as areflection of the principle that agents' rules of behavior changewhen they encounter changes in the environment in the form ofnew laws of motion for variables that constrain them.</p><p>To illustrate this point, consider two specific tax rate policies.First, consider the policy of a <i>constant</i> tax rate <i>τ<sub>t+j</sub> = τ<sub>t</sub></i> for allj ≥ 0. Then <i>z<sub>t</sub></i> = τ<sub>1</sub>, <i>A</i> = 1, and the investment schedule is</p><p><i>k<sub>t</sub> = λk<sub>t-1</sub> + h<sub>0</sub>τ<sub>t</sub></i> (1.4)</p><p>where <i>h</i><sub>0</sub> = -α/(1 - δ). Now consider an investment tax credit<i>on-again, off-again</i> tax rate policy of the form <i>τ<sub>t</sub> = -τ<sub>t-1</sub></i>. In thiscase <i>z<sub>t</sub> = τ<sub>1</sub>, A = -1</i>, and the investment schedule becomes</p><p><i>k<sub>t</sub> = λk<sub>t-1</sub> + h<sub>0</sub>τ<sub>t</sub></i> (1.5)</p><p>where <i>h</i><sub>0</sub> = -α/(1 + δ). Here the investment schedule itselfchanges as the policy for setting the tax rate changes.</p><p>Standard econometric practice has not acknowledged that thissort of thing happens. Returning to the more general investmentexample, the usual econometric practice has been roughly as follows.First, a model is typically specified and estimated of theform</p><p><i>k<sub>t</sub> = λk<sub>t-1</sub> + hz<sub>t</sub></i> (1.6)</p><p>where <i>h</i> is a vector of free parameters of dimension conformablewith the vector <i>z<sub>t</sub></i>. Second, holding the parameters <i>h</i> fixed, equation(1.6) is used to predict the implications of alternative paths forthe tax rate τ<i><sub>t</sub></i>. This procedure is equivalent to estimating equation(1.4) from historical data when <i>τ<sub>t</sub> = τ<sub>t-1</sub></i> and then using thissame equation to predict the consequences for capital accumulationof instituting an on-again, off-again tax rate policy of theform <i>τ<sub>t</sub> = -τ<sub>t-1</sub></i>. Doing this assumes that a single investmentschedule of the form of equation (1.6) can be found with a singleparameter vector h that will remain fixed regardless of the rulesfor administering the tax rate.</p><p>The fact that equations (1.2) and (1.3) share a common set ofparameters implies that the search for such a regime-independentdecision schedule is misdirected and bound to fail. This theoreticalpresumption is backed up by the distressing variety of instances inwhich estimated econometric models have failed tests for stabilityof coefficients when new data are added. This problem cannot beovercome by adopting more sophisticated and more general lagdistributions for the vector <i>h</i>, as perhaps was hoped in the 1960s.</p><br><p>Are Government Deficits Inflationary?</p><p>A second example that well illustrates our general principles aboutthe interdependence of the strategic behavior of private agents andthe government concerns the inflationary effects of governmentdeficits. We can discuss this matter with the aid of a demandfunction for base money of the specific form</p><p><i>M<sub>t</sub>/pt = α<sub>1</sub> - α<sub>2</sub>E<sub>t</sub>p<sub>t+1</sub>/p<sub>t</sub> α<sub>1</sub> > α<sub>2</sub> > 0</i> (1.7)</p><p>where <i>M<sub>t</sub></i> is the stock of base money at time <i>t, p<sub>t</sub></i> is the price levelat time <i>t</i> and <i>E<sub>t</sub></i>(·) is the value of (·) expected to prevail at time<i>t</i>. Equation (1.7) is a version of the demand schedule for moneythat Phillip Cagan (1956) used to study hyperinflations. It depictsthe demand for base money as decreasing with the expected grossrate of inflation, <i>E<sub>t</sub>p<sub>t+1</sub>/p<sub>t</sub></i>. A variety of theories imply a demandfunction for base money with this property.</p><p>Equation (1.7) is a difference equation, a solution of which is</p><p>[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1.8)</p><p>which expresses the price level at <i>t</i> as a function of the supply ofbase money expected to prevail from now into the indefinite future.We shall use equation (1.8) as our theory of the price level.</p><p>According to equation (1.8), if government deficits are to influencethe price level path, it can only be through their effect on theexpected path of base money. However, the government deficitand path of base money are not rigidly linked in any immutableway. The reason is that the government can, at least to a point,borrow by issuing interest-bearing government debt, and so neednot necessarily issue base money to cover its deficit. More precisely,we can think of representing the government's budget constraintin the form</p><p><i>G<sub>t</sub> - T<sub>t</sub> = M<sub>t</sub> - M<sub>t-1</sub>/p<sub>t</sub>+ B<sub>t</sub> - (1 + r<sub>t-1</sub>)B<sub>t-1</sub>r<sub>t-1</sub> ≥ 0</i> (1.9)</p><p>where <i>G<sub>t</sub></i> is real government expenditures at <i>t, T<sub>t</sub></i> is real taxesnet of transfers (except for interest payments on the governmentdebt), <i>B<sub>t</sub></i> is the real value at <i>t</i> of one-period government bondsissued at <i>t</i>, to be paid off at <i>t</i> + 1 and to bear interest at the netrate <i>r<sub>t</sub></i>. For simplicity, equation (1.9) assumes that all governmentinterest-bearing debt is one period in maturity. Equation (1.9) musthold for all periods <i>t</i>. Again for simplicity, we shall also thinkof equations (1.8) and (1.9) as applying to an economy with nogrowth in population or technical change.</p><p>Under the system formed by equations (1.8) and (1.9), the inflationaryconsequences of government deficits depend sensitivelyon the government's strategy for servicing the debt that it issues.We consider first a strict Ricardian regime in which governmentdeficits have no effects on the rate of inflation. In this regime, thegovernment always finances its entire deficit or surplus by issuingor retiring interest-bearing government debt. This regime can becharacterized by either of the following two equations, which areequivalent in view of equation (1.9):</p><p><i>M<sub>t</sub> - M<sub>t-1</sub> = 0</i> (1.10)</p><p>[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1.11)</p><p>where <i>R<sub>tj</sub> = Π<sup>j</sup><sub>i=0</sub>(1 + r<sub>t+i</sub>)</i>. Equation (1.10) states that the supplyof base money is always constant, while equation (1.11) statesthat the real value of government debt equals the present value ofprospective government surpluses. In this regime a positive valueof interest-bearing government debt signals a stream of futuregovernment budgets that is in surplus in the present value senseof equation (1.11). Increases in government debt are temporary ina sense made precise in equation (1.11).</p><p>In the Ricardian regime, government deficits have no effects onthe price path because they are permitted to have no effects on thepath of base money. For the path of base money to be unaffectedby government deficits, it is necessary that government deficitsbe temporary and be accompanied by exactly offsetting futuregovernment surpluses.</p><p>Since the Ricardian regime may seem remote as a description ofrecent behavior of the US federal government, it is worthwhile torecall that cities and states in the United States are constitutionallyforced to operate under a Ricardian rule, since they have no rightto issue base money.</p><p>There are alternative debt-servicing strategies under which governmentdeficits are inflationary. To take an example at the oppositepole from the Ricardian regime, consider the rule recommendedby Milton Friedman (1948) under which</p><p><i>B<sub>t</sub></i> = 0 for all <i>t</i> (1.12)</p><p><i>G<sub>t</sub> - T<sub>t</sub> = M<sub>t</sub> - M<sub>t-1</sub>/P<sub>t</sub></i> (1.13)</p><p>According to this rule, deficits are always to be financed entirelyby issuing additional base money, with interest-bearing governmentdebt never being issued. In this regime, the time path ofgovernment deficits affects the time path of the price level in arigid and immediate way that is described by equations (1.8) and(1.13). Under this regime it is possible for the government budgetto be persistently in deficit, within limits imposed by equations(1.13) and (1.7). Deficits need not be temporary.</p><p>Bryant and Wallace (1980) and Sargent and Wallace (1981) havedescribed debt-servicing regimes that are intermediate betweenRicardo's and Friedman's. In all versions of these regimes, interest-bearinggovernment debt is issued, but is eventually repaid partlyby issuing additional base money. In the regime studied by Sargentand Wallace, the deficit path <i>G<sub>t</sub> - T<sub>t</sub></i> is set in such a way, and thedemand schedule for interest-bearing government debt is such thateventually the inflation tax must be resorted to, with increases inbase money eventually having to be used to finance the budget.</p><p>In all regimes of the Bryant-Wallace variety, increases in interest-bearinggovernment debt are typically inflationary, at least eventuallybecause they signal prospective increases in base money.Sooner or later these eventual increases in base money will affectthe price level, how soon depending on the coefficients α<sub>1</sub> and α<sub>2</sub>in equations (1.7) and (1.8).</p><p>This discussion indicates that the observed correlation betweenthe government deficits and the price level depends on the debt-repaymentregime in place when the observations were made. Itwould be a mistake to estimate the relationship between the deficitand the price path from time-series observations drawn from aperiod under which a Ricardian regime was in place, and to assertthat this same relationship will hold between the deficit and inflationunder a regime like the one described by Bryant and Wallace.It would be a mistake because private agents' interpretations ofobserved deficits, and consequently the impact of observed deficitson the price level, depend on the debt-servicing regime they imagineto be in place. This can thus be viewed as another exampleof our principle that private agents' rules of behavior depend ontheir perceptions of the rules of the game they are playing.

(Continues...)
Excerpted from Rational Expectations and Inflation by Thomas J. Sargent. Copyright © 2013 by Princeton University Press. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.

"About this title" may belong to another edition of this title.

Publisher
Princeton University Press
Publication date
2013
Language
English
ISBN 10 
0691158703
ISBN 13 
9780691158709
Binding
Hardcover
Edition number
3
Number of pages
392