Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University LECTURE 6: DYNAMIC INCONSISTENCY OF MONETARY POLICY, AND HOW TO ADDRESS IT Question: Why is inflation, π, often high? Why π > 0 more often than π < 0? One of several answers: Proclamations of low-inflation monetary policy by central banks are “dynamically inconsistent.” Next question: What institutions can address dynamic inconsistency?

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University If monetary expansion can’t reduce unemployment in the long run, why is inflation so common? Four possible explanations:  Governments think expansion can reduce unemployment in the long run.  They give low weight to price stability, or have high discount rates ( e.g., political business cycle).  Plans to set non-inflationary monetary policy are perceived by the public to be time-inconsistent.  Governments want seignorage, to pay for spending that is not financed by taxes or borrowing.

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Dynamic inconsistency: The Intuition Assume governments, if operating under discretion, choose monetary policy and hence AD so as to maximize a social function of Y & π. –(1) Assume also that the social function centers on >, even though this point is unattainable, at least in the long run. –(2) => Economy is at tangency of AS curve & one of society’s indifference curves. (3) Assume W & P setters have rational expectations –=> π e (& AS) shifts up if rationally-expected E π shifts up –=> π e = E π = π on average. economy is at point B on average. Inflationary bias: π e =E π > 0.. (4) Lesson: The authorities can’t raise Y anyway, so they might as well concentrate on price stability at point C.

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University 1. Barro-Gordon innovation: It can be useful to think of society’s 1 st choice as Y= (& π=0), even if it is unattainable. 3. But π e adjusts upward in response to observed π>0. The LR or Rational Expectations equilibrium must feature π e = π. Result: inflationary bias π>0, despite failure to raise Y above. π πeπe 4. The country would be better off “tying the hands” of the central bank. Result: Y = (no worse on average than under discretion), and yet π=0. ● ● ● 2. If π e would stay at 0, then to get the higher Y it would be worth paying the price of π>0.

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University TIME-INCONSISTENCY OF NON-INFLATIONARY MONETARY POLICY + Policy-maker minimizes quadratic loss function: where the target. => (Romer 11.53) (11.54 )

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Given discretion, the CB chooses the rate of money growth and inflation (assuming it can hit it) where. Take the mathematical expectation: + Rational expectations: => (10.15), the inflationary bias. (11.58)

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University ADDRESSING THE TIME- INCONSISTENCY PROBLEM  How can the CB credibly commit to a low-inflation monetary policy?  Announcing a policy target π = 0 is time-inconsistent, because a CB with discretion will inflate ex post, and everyone knows this ex ante.  CB can eliminate inflationary bias only by establishing non-inflationary credibility,  which requires abandoning the option of discretion.  so public will see the CB can’t inflate even if it wants to.  CB “ties its hands,” as Odysseus did in the Greek myth.

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Addressing the Time-Inconsistency Problem (continued)  Reputation  Delegation. Rogoff (1985): Appoint a CB with high weight on low inflation a′ >> a, and grant it independence. It will expand at only << inflationary bias of discretion.  Binding rules. Commit to rule for a nominal anchor: 1. Price of gold2. Money growth 3. Exchange rate4. Nominal GDP 5. CPI 6. GDP deflator 4.

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University ADDRESSING THE TIME-INCONSISTENCY PROBLEM (continued) Reputations. With multiple periods, a CB can act tougher in early periods, to build a reputation for monetary discipline. –Backus-Driffill (1985) model: people are uncertain if the CB is of hard-money or soft-money “type.” –Then even a soft CB is likely to act tougher, to influence subsequent expectations.

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Alesina & Summers: Central banks that are institutionally independent of their governments have lower inflation rates on average. Delegation

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University “Central Bank Independence, Inflation and Growth in Transition Economies,” P.Loungani & N.Sheets, IFDPS (1995) Transition economies

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Limitations to the Argument for Central Bank Independence 1.Some consider it undemocratic. 2.The argument only works if conservative central bankers are chosen. 3.Although independence measures are inversely correlated with inflation, these measures have been debated and, 4.more importantly, the choice to grant independence could be the result of priority on reducing inflation. 5.As with rules to address time-inconsistency, there is little empirical evidence that it succeeds in reducing inflation without loss of output. 6.As with rules, one loses ability to respond to SR shocks.

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Source: IMF Survey. October 23, Andrea Schaechter, Mark Stone, Mark Zelmer in the IMF, Monetary and Exchange Affairs Dept. Online at: The background papers for the high-level seminar “Implementing Inflation Targets,” held in Washington in March 2000, are available on the IMF Website: Inflation Targeting After 1998, many middle-sized middle-income countries adopted inflation targeting.

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Countries adopting IT experienced lower inflation Gonçalves & Salles, 2008, “Inflation Targeting in Emerging Economies…” JDE

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Introducing disturbances into the Barro-Gordon model à la Rogoff (1985), Fischer (1987), et al AS shocks No effect on inflation bias. Average inflation = Shocks will show up as fluctuations in actual  & y. But discretionary monetary policy can’t offset AS shocks anyway (can only choose the split  vs. y). => Strong case for committing to E  =0. AD shocks Again no effect on inflation bias. Need not show up as fluctuations in actual  & y : If lags in monetary policy < lags in adjustment of W & P, under discretion CB can offset AD shocks. => Choice of rules vs. discretion then becomes choice of eliminating LR inflation bias (E  =0) vs. SR shocks.

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Appendix 1: GLOBAL INFLATION HAS DECLINED SINCE WHY?  Better understanding of costs of inflation and the temporariness of the AS tradeoff ?  Spread of commitment devices such as central bank independence, hard exchange rate pegs (currency boards & monetary unions), & IT?  Rogoff (2003): Globalization & increased competition have reduced  and/or and thereby the inflationary bias ?

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University peak: early 80s

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Continued from previous peak: ≈ 1990 peak: early 90s

Inflation was much more of a chronic problem in the 20 th century than in preceding centuries Source: Carmen Reinhart & Ken Rogoff, “This Time is Different: A Panoramic View of Eight Centuries of Financial Crises,” draft, April 2008

API Professor Jeffrey Frankel, Kennedy School of Government, Harvard University Appendix 2: Targets & Instruments of Policymaking OBJECTIVESInflation, Price Stability Unemployment, Growth INSTRUMENTSOpen market operations Foreign exchange intervention Reserve requirements “Policy interest rate” (short-term i) INTERMEDIATE TARGETS M1 Exchange rate Core CPI… INDICATORSIndex of Leading Indicators Financial Conditions Index (incl. spreads, P/E,...) Length of work-week Consumer Confidence, Business Confidence Purchasing Managers’ Index (ISM, IFO, Tankan,…) …

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University The choice of anchor depends on: 1.Credibility of the commitment 2.Tradeoff: advantage of time-consistent commitment vs. ability to stabilize short-term shocks  Must compare E(Loss) function for M vs. GDP vs. ex.rate vs. P targets)  Original treatment due to Rogoff (1985) 3.Other objectives served (e.g., a peg cuts exchange rate risk) Appendix 3: Comparison of alternate rules (M1 vs. E vs. CPI …)

Copyright 2007 Jeffrey Frankel, unless otherwise noted API Macroeconomic Policy Analysis I Professor Jeffrey Frankel, Kennedy School of Government, Harvard University 6 proposed nominal targets and the Achilles heel of each: