1. Operating procedures and choice of monetary policy instrument. 2. Intermediate targets in policymaking. Literature: Walsh (Chapter 9, pp.

Transcription

1 Monetary Economics: Macro Aspects, 14/ Henrik Jensen Department of Economics University of Copenhagen 1. Operating procedures and choice of monetary policy instrument 2. Intermediate targets in policymaking Literature: Walsh (Chapter 9, pp ) 3. Plan for next lectures Still no info about # of respondents (you know for what) :-( c 2010 Henrik Jensen. This document may be reproduced for educational and research purposes, as long as the copies contain this notice and are retained for personal use or distributed free.

2 Introductory remarks In models so far, the choice variables, or instruments, of the central bank have been Nominal money supply, (nominal interest rate), in ation O.k. for getting various points through In the real world, things are more complicated: What is the actual instrument of a central bank? What is the best, given the uncertainties that inevitable are a part of policymaking?... which operating procedures should be used? In reality, the central bank can only exercise close control over the money base and the (very) shortterm nominal interest rate Even though the money base is the actual instrument, the central bank can very well behave as if the market nominal interest rate is its instrument: It will adjust the base to attain some target value of the nominal interest rate This a case of an interest rate target procedure Alternatively, it could adjust the money base to attain some desired value of a broader money aggregate (M1, M2, M3) This a case of a money supply target procedure 1

3 What is best? General lessons of today: It depends on the ultimate goals of monetary policy (i.e., what are the policy objectives?) It depends on the relative variability of shocks to the economy when these cannot be observed It depends on which variables provide information about the ultimate goal variables Moreover, examples reveal that changes in monetary aggregates (broad money supply, nominal interest rates) tell little about deliberate monetary policy shifts if one does not take into account under which operating procedures the central bank acts (cf. the identi cation problems in the empirical VAR literature) 2

4 Operating procedures and choice of monetary policy instrument The Poole (1970) model of instrument choice Highlights how the relative variances of shocks a ect the optimal choice of policy instrument Here: The money supply or the nominal interest rate Simple IS/LM model (p t = 0 by normalization): y t = i t + u t ; > 0 (9.1) m t = ci t + y t + v t ; c > 0 (9.2) Shocks u t and v t are mean zero, independent shocks with variances 2 u and 2 v, respectively Simple objective of policy: Minimize output variance E [y t ] 2 (9.3) Policy of under either operating target is conducted before shocks u t and v t hit 3

5 When money supply is instrument, y t is solved in terms of m t : Optimal policy: m t = 0. y t = m t + cu t + c v t Output variance: E m [y t ] 2 = c2 2 u v ( + c) 2 (9.4) When the interest rate is the instrument, IS curve gives output immediately: y t = i t + u t Optimal policy: i t = 0. Actual output: y t = u t Output variance: E i [y t ] 2 = 2 u (9.5) 4

6 Comparison: Interest rate operating procedure is preferred i or E i [y t ] 2 < E m [y t ] 2 2 u < c2 2 u v ( + c) c 2 u < 2 v Hence, choose an interest rate targeting procedure: relatively high money demand volatility relatively low aggregate demand volatility Shows how relative variances of macroeconomic shocks matter for optimal choice of instrument 5

7 Extension of the basic Poole model: Monetary base as potential instrument Central banks control the money base, but not, say, M1 Extend model with determination of broad money supply (M1): Here, b t is the money base m t = b t + hi t +! t ; h > 0 (9.7) m t b t is the (log) money multiplier.! t is mean-zero money multiplier shock Money multiplier is increasing in interest rate (banks want to lend more/consumers want to hold less cash => expanding deposits are possible) Note that we now have an LM curve in b t : b t = (c + h) i t + y t + v t! t 6

8 The interest rate as an instrument gives output variance as before; E i [y t ] 2 = 2 u The money base as instrument (again optimal to set b t = 0), yields output as: Associated output variance: y t = (c + h) u t v t +! t + c + h E b [y t ] 2 = (c + h)2 2 u v + 2 2! ( + c + h) 2 The interest operating target is preferred i : 2 (c + h) u < 2 v + 2! Reinforcement of simple Poole result: More volatility on money market/ nancial markets makes a base operating target less attractive Can explain why real-life central banks are using interest-rate operating procedures as money demand is unstable and/or nancial markets are volatile Note: instrument choice is endogenous and will depend on the objectives of the policymaker 7

9 Policy rules and information Normally, the monetary base is the de facto policy instrument, but one may still under an interest rate operating procedure think of the nominal interest rate as an instrument The Poole analysis took an either or perspective; something in between may be optimal Illustrated by money base policy rule, where the money base responds to the observed nominal interest rate: b t = i t (9.8) = 0: A base money operating procedure = h: A (broad) money supply operating procedure! 1: An interest rate operating procedure The LM curve becomes Solution for output (combine (*) with IS-curve): Associated variance: 0 = (c + h + ) i t + y t + v t! t (*) y t = (c + + h) u t (v t! t ) c + h + + E [y t ] 2 = (c + + h)2 2 u v + 2! (c + h + + ) 2 8

10 What is the optimal base rule (in terms of minimizing output variance)? Solve (c + + h) 2 2 u v + 2! min (c + h + + ) 2 Solution is (see the note Equivalence of (9.11)... on the web): = (c + h) + 2 v + 2! Again, dependent upon relative variances and slopes of IS and LM curves. Approaching an interest rate operating procedure requires High money market volatility Low aggregate demand volatility 2 u Note, however, that even 2 v = 2! = 0 does not warrant a pure base rule operating procedure ( = 0) With u t > 0, causing an increase in i, one can do better by contracting b t so as to further increase the nominal interest rate ( < 0) With 2 v; 2! > 0 leaning against the wind may become optimal, as an interest rate increase may re ect either v t > 0 or! t < 0 ( > 0) 9

11 Analogy between in policy rule and a signal extraction problem (like in Lucas Island model) Central bank observes the nominal interest rate, but not the shocks. Hence, it forecasts shocks based on the signal, the nominal interest rate Formally, assume that the policy rule could respond to shocks: b t = u u t + v v t +!! t Output satis es y t 1 + c a + h = ( v 1) v t + u + h + c u t + (1 +! )! t Output-stabilizing rule: b t = c + h u t + v t! t However, shocks cannot be observed, so estimates of the shocks are made b t = c + h bu t + bv t b! t (9.11 First part) bu t = E [u t ji t ] = u i t ; bv t = E [v t ji t ] = v i t ; b! t = E [! t ji t ] =! i t ; We get b t = c + h u + v! i t (9.11 Second part) The note Equivalence of (9.11)..., see web site, shows that when the s are chosen to minimize squared forecast errors, the coe cient on i t becomes 10

12 Intermediate targets in policymaking Lesson from previous analysis: Forecasts of shocks may determine optimal monetary policy (and forecasts depend on relative variances of shocks) In real life, central banks must rely on best possible information Provides scope for adjusting policy in light of movements in variables that provides good information about movements in goal variables This is intermediate targeting Illustrated in simple AS/IS/LM model with imperfect information about current shocks Policy goal is to minimize in ation variance around a target (here normalized to = 0): V = E [ t ] 2 (9.15 ) 11

13 The AS/IS/LM model: y t = a ( t E t 1 t ) + z t ; a > 0 (AS 9.12) y t = (i t E t t+1 ) + u t ; > 0 (IS 9.13) m t p t = m t t p t 1 = y t ci t + v t ; c > 0 (LM 9.14) Shocks follow AR(1) processes: z t = z z t 1 + e t ; 0 < z < 1; u t = u u t 1 + ' t ; 0 < u < 1; v t = v v t 1 + t ; 0 < v < 1; All innovations, e t, ' t and t, are mean-zero, independent shocks (with variances 2 e, 2 ', 2 ) Policy instrument is the nominal interest rate With the simple policy goal, solution of model is simpli ed as E t 1 t = E t t+1 = 0 AS and IS curves provides solution for in ation as function of interest rate Clearly, optimal shock-contingent interest rate is t = i t + u t z t a (9.16 ) i t = (1=) (u t z t ) (9.17 ) 12

14 If interest rate must be set before period t innovations are realized, the optimal interest rate is Given this policy, actual in ation becomes b t = (1=) ( u u t 1 z z t 1 ) (9.18 ) t (b t ) = ' t Fluctuates with the supply and demand shock innovations (not money demand shocks) Associated in ation variance a e t V (b t ) = 2 ' + 2 e a 2 One can nd the money supply that gives the same in ation and interest rate (all relevant derivations are provided in the note Deriving (9.21)... ; see web-page): c bm t = p t 1 uu t c z z t 1 + v v t 1 Now assume that the central bank when setting the interest rate observes the actual value of m t The individual shocks are still unobservable, but m t provides information about the shocks a signal Adjusting the interest rate such that actual m t becomes equal to bm t may improve policymaking (it takes into account information about the unobservable shocks) m t becomes an intermediate target for policymaking 13

15 Example: 2 = 2 e = 0; only aggregate demand shocks matter ' t > 0 will for given interest rate be re ected in higher output and in ation and higher m t Higher m t thus provides information about u t > 0, and the interest rate should be increased to dampen output and in ation, i.e., to bring back m t to bm t Intermediate targeting is therefore appropriate policy Example: 2 = 2 ' = 0; only aggregate supply shocks matter e t > 0 will for given interest rate be re ected in lower in ation and lower m t Lower m t thus provides information about e t > 0, and the interest rate should be decreased to bring in ation back towards target, i.e., to bring back m t to bm t Intermediate targeting is therefore appropriate policy Example: 2 e = 2 ' = 0; only money demand shocks matter t > 0 will for given interest rate be re ected in higher m t and no change in in ation Higher m t thus provides information about t > 0, and the interest rate should be left unchanged and leave m t di erent from bm t Intermediate targeting would therefore be inappropriate policy 14

16 Overall, the desirability of a certain policy regime, here monetary intermediate targeting, will depend upon the relative variances of the shocks A simple interest rate procedure to keep m t = bm t improves on keeping i t = b t if 2 e and 2 ' are relatively high, and 2 is relatively low Note, the simple Poole model featured and either or choice which could be improved by an optimal policy rule, that optimally processed the new information about shocks Same logic applies here, where one can formulate interest rate rule like i t = b t + x t (9.22) where x t is the new information obtained by observing m t (a linear combination of unobserved shocks) Optimal will, e.g., be decreasing in 2 In this sense new information used by observing the intermediate target is used optimally It will improve over a policy aimed at attaining m t = bm t 15

17 Plan for next lectures Wednesday, April 21 Interest rate policies (I) 1. Price level (in)determinacy 2. The term structure of interest rates 3. Models for monetary policy analysis and the impact of interest rate rule parameters Literature: Walsh (Chapter 10, pp ; pp ; pp ) 16