Monetary Stylized Facts Long-run relationships Distinction between correlation and causation Short run relationships Distinction between correlation and causation 1 Monetary Indicators and Output, Inflation in the Long Run Monetary Neutrality Fisher equation 2 1
Long term neutrality: What is it? QTM M V = P Y In the SR If V,P is constant M Y In the LR prices adjust Y long term neutrality of money 3 Long-run relationships: McCandless and Weber (199) 4 2
McCandless and Weber (199) McCandless and Weber (199) 6 3
McCandless and Weber (199) 7 McCandless and Weber (199) 8 4
McCandless and Weber (199) 9 McCandless and Weber (199) 1
Conclusions of McCandless and Weber One-to-one link between inflation and monetary growth in the long run No link between output growth and monetary growth in the long run Some evidence for no long term relationship btw output and money Does not address causation issues! 11 Interest Rates and Real Output Why bother about monetary aggregates neutrality if monetary aggregates are neither policy nor useful indicator variables! Look at SR interest rates! 12 6
In Woodford s (23, p.34) words. once one recognizes that many prices (and wages) are fairly sticky over short time intervals the arbitrariness of the path of nominal prices implies that the path of real activity and the associated path of equilibrium real interest rates are equally arbitrary. It is equally possible, from a logical standpoint, to imagine allowing the central bank to determine, by arbitrary fiat, the path of aggregate real activity, or the path of real interest rates. 13 Interest Rates and Real Output Aksoy&Leon Ledesma (24) If real interest rates proxy the (user cost) price of capital, an exogenous change in the policy variable, nominal rates, should also have long lasting implications in the real output path via standard aggregate demand channels. (e.g. sensitivity of consumption, housing demand or manufacturing investment decisions). Alternatively, by adjusting the nominal interest rate, the policymaker may simply be accommodating changes in the structural conditions in the economy taking into account the term structure of inflation expectations. In that case, even under inflation inertia, equilibrium long term real rates may be disconnected from short term real rates formulated in markets short and long term inflation expectations. In that case we do not expect to see a long term relationship between nominal and real short term interest rates and real output. Otherwise, we do. 14 7
Interest Rates and Real Output Test for long term relationship Short term interest rates are found to be I(1) But do we believe in this unit root tests? 1 To start with: Univariate Properties 16 8
Long term tests Cointegration: requires both output and SR interest rates to be I(1) Pesaran et al (21) Bounds tests: does not require SR interest rates to be I(1) These can be I(1) or I() 17 Full-sample Results 18 9
Stability Issues Three sorts of recursive results Fix endpoint (min. 3 years sample) Fix staring point (min. 3 years sample) Rolling (3 years) 19 U.K. Cointegration Results: Sub-sample Stability (1948-21 sample) 2 1
Figure 2: U.S. Cointegration Results: Subsample Stability (1947-21 sample) 21 Concern standard univariate analysis: difficulties to reject the nonstationarity, power problem Economic intuition: stationary short-term interest rates What to do? 22 11
Bounds Tests Power problems of unit-root tests and theorybased arguments cast doubts about the assumption that both output and the interest rate are I(1) variables Pesaran et al (21) develop a technique to test for the existence of a long-run relationship between two variables irrespective of whether they are I(1) or I(). 23 Pesaran et al. (21) Pre-testing is not an issue. approach is based on the estimation of an unconstrained dynamic error correction representation for the variables involved and testing whether or not the lagged levels of the variables are significant. Test: OLS estimation of a conditional unconstrained ECM: p 1 p 1 y = α + βy + βr + ϕ y + θ r + ω r+ u t 1 t 1 2 t 1 i t i i t i t t i= 1 i= 1 24 12
Statistics an F-statistic test of joint significance of the lagged levels of the variables involved. a t-ratio test for the significance of the lagged level of the dependent variable (y t-1 ). In case that the ECM contains a deterministic trend, the F-test also includes the null of the coefficient on the trend being equal to zero. 2 Bounds two sets of critical values assuming that both regressors are I(1) and that both are I(). Proposed test statistics under the null of (β 1 =β 2 =) against (β 1 and (β 2 or β 2 =)) and its asymptotic distribution of the two polar cases of (r t ) being I() and I(1) These two sets provide a band covering all possible combinations of the regressors into I(), I(1) or mutually cointegrated. Which variable is long-run forcing? Reverse ECM 26 13
Cases to look at Case II: restricted intercepts and no trends. Case III: unrestricted intercepts and no trends (ttest also reported). Case IV: unrestricted intercepts and restricted trends. Case V: unrestricted intercepts and unrestricted trends (t-test also reported). 27 Results 28 14
4. 4. 3. 3. 2. 2. 1. 1... 4. 4. 3. 3. 2. 2. 1. 1... 4. 4. 3. 3. 2. 2. 1. 1... 1 1 2 2 1 1 2 1 1 2 2 6 4 3 2 1 6 4 3 2 1 12 1 8 6 4 2 1 1 2 2 1 1 2 1 1 2 2.6 4.8 4. 3.2 2.4 1.6.8..6 4.8 4. 3.2 2.4 1.6.8. 2. 17. 1. 12. 1. 7.. 2.. 1 1 2 2 1 1 2 1 1 2 2 8 7 6 4 3 2 8 7 6 4 3 2 1 14.4 12.8 11.2 9.6 8. 6.4 4.8 3.2 1.6 1 1 2 2 1 1 2 1 1 2 2 Stability issues (Rolling window estimates) US Treasury Bill Rate FII FIII FIV FV UK Treasury Bill Rate FII FIII FIV FV 29 Stability issues: (fixed starting point) US Treasury Bill Rate FII FIII FIV FV UK Treasury Bill Rate 4. FII 8 FIII 1.8 FIV 8 FV 4. 7 9.6 7 3. 6 8.4 6 3. 7.2 2. 4 6. 2. 4 3 4.8 1. 1. 2 3.6 3. 1 2.4 2. 1 1 2 1 1 2 1.2 1 1 2 1 1 1 2 3 1
4. 4. 3. 3. 2. 2. 1. 1... 4. 4. 3. 3. 2. 2. 1. 1... 6 4 3 2 1 6 4 3 2 1.6 4.8 4. 3.2 2.4 1.6.8..6 4.8 4. 3.2 2.4 1.6.8. 7.. 2.. 8 7 6 4 3 2 1 Stability issues: fixed end point US Treasury Bill Rate FII FIII FIV FV UK Treasury Bill Rate FII FIII FIV FV 31 Fisher Equation Fisher effect (a high correlation between the level of interest rates and inflation) r t = i t -E(π t+1 ) Mishkin (1992) among others Empirical evidence finds no support for a shortrun Fisher effect in which a change in expected inflation is associated with a change in interest rates supports the existence of a long-run Fisher effect in which inflation and interest rates have a common stochastic trend when they exhibit trends. 32 16
Monnet and Weber (21) Annual average interest rates and money growth rates (1961-1998, 31 countries) Correlation btw. money growth and long term interest rates.87 Consistent with Fisher equation 33 Long term neutrality (LTN): Two generations of LTN analysis First generation: St Louis regressions y t = B(L)m t + C(L)x t + ε t Simultaneity issues 34 17
The rational expectation critique y t = φ (m t E t-1 m t ) + e y,t with φ<1 and suppose m t = ρm t-1 + e m,t Then y t = φm t φρm t-1 + e y,t Even if money supply process is stationary ( ρ < 1), the sum of coefficients in the real output equation is inconsistent with neutrality (φ(1-ρ) > ). 3 Second generation: Structural models (SVARs) Geweke (1986), Fisher and Seater (1992), King and Watson (1997) Each based on the idea of long run variation in the money supply process Typically a set of auxiliary assumptions p p y y t = λ ym mt + α j, yy y t j + α j, ym m t j + ε t j= 1 j= 1 m = λ y + t my t p j= 1 α j, my y p t j + α j, mm mt j + j= 1 ε m t 36 18
Second generation: assume a wide range of iterative, λ alternative assumptions on my estimate several observationally equivalent models. A sense of robustness Mixed results λ, γ ym and γ my ym 37 Superneutrality A finding of neutrality does not mean there is no long term superneutrality (see King and Watson (1997) Changes in the growth rate of money supply is found to have a positive impact on long run output. One explanation (Mundell-Tobin) Persistently higher inflation leads to lower real interest rates and portfolio rebalance (from money holdings to capital) Higher investment, higher output 38 19
LR: Conclusions Monetary aggregates and output : (relationship and neutrality) mixed results Monetary aggregates and inflation: there seems to be a long run one-to-one relation Short term interest rates and output: looks like there is no relationship Fisher relation: exists but in the long run 39 Short Run *Friedman, B.M., and K.N. Kuttner (1992), Money, Income, Prices and Interest Rates, American Economic Review, Vol. 82, pp. 472-492. *Sims, C. A., (1972), Money, Income, and Causality, American Economic Review, Vol: 62, pp. 4-2. Sims, C. A., (198), Comparison of Interwar and Postwar Business Cycles, American Economic Review, Vol: 7, pp. 2-277. *Walsh Ch. 1 4 2
Outline Focus: Stylized Facts Information Content of a Monetary Indicator Significance: Granger Causality Tests Importance: Variance Decompositions Identification of Monetary Policy 41 Money and Inflation: The Evidence Inflation is Always and Everywhere a Monetary Phenomenon (M. Friedman) Evidence In every case when π high for sustained period, M growth is high 42 21
Milton Friedman and Anna Schwartz (1963) 1 years of US data "Decisively support treating the rate of change series [of the money supply] as conforming to the reference cycle positively with a long lead" Higher monetary growth is followed by higher growth above trend and higher inflation 43 German Hyperinflation: 1921 23 44 22
Controversial Relationship: Money and Fundamentals (here Inflation, Output) This pattern in the US obvious till the early 198 s (Friedman and Kuttner (1992, AER) After early 8 s collapses. Hardly any statistically significant result found in the literature Similar pattern in other developed countries 4 Graphical Evidence: Money and Fundamentals 2 2 2 2 2 2 1 1 1 1 1 1 - - - -1 6 7 7 8 8 9 9-1 6 7 7 8 8 9 9-1 6 7 7 8 8 9 9 Nominal GNP M1 Real Income M1 Implicit Price Deflator M1 2 2 2 2 2 2 1 1 1 1 1 1 - - - -1 6 7 7 8 8 9 9-1 6 7 7 8 8 9 9-1 6 7 7 8 8 9 9 Nominal GNP Monetary Base (Board of Gov.) Real Income Monetary Base (Board of Gov.) Implicit Price Deflator Monetary Base (Board of Gov.) 2 2 2 2 2 2 1 1 1 1 1 1 - - - -1 6 7 7 8 8 9 9-1 6 7 7 8 8 9 9-1 6 7 7 8 8 9 9 Nominal GNP M1 Currency Component Real Income M1 Currency Component Implicit Price Deflator M1 Currency Component 2 2 2 2 2 2 1 1 1 1 1 1 - - - -1 6 7 7 8 8 9 9 Nominal GNP Federal Funds Rate -1 6 7 7 8 8 9 9 Real Income Federal Funds Rate -1 6 7 7 8 8 9 9 Implicit Price Deflator Federal Funds Rate 46 23
Statistical Evidence Granger Causality Tests VAR s and Variance Decompositions 47 Statistical Issues I: Granger Causality Tests No assumption whatsoever on the causality, endogeneity and controllability issues Reevaluate the statistical relationship between crucial macroeconomic fundamentals and the monetary aggregates Monetary variables first have to pass reliable statistical tests as information variables Only after a statistical connection has been established one can argue for their usefulness As information variables/ indicators Potential monetary instruments 48 24
Information Value Approach As long as movements in money do contain information about future movements in income beyond what is already contained in movements in income itself, monetary policy can exploit that information by responding to observed money growth regardless of whether the information it contains reflects true causation, reverse causation based on anticipations, or mutual causation by some independent but unobserved influence. Friedman and Kuttner (1992). 49 Quarterly Data : 196:1-1998:2 Macrofundamentals studied (in natural log differences): 1. US real income 2. US inflation Financial Variables Considered: Monetary Aggregates: M1, M2, M1 currency component, Two Monetary Base measures Short-term interest rates: Federal funds rate, 3-month commercial paper, 3-month Treasury bill, spread between the two 3-month interest rates 2
Linear Model to estimate y = α+ β y + δ m + v t 4 i= 1 i t i 4 i= 1 i t i t 1 To test A test of causality is whether the lags of one variable enter into the equation of another variable using standard F-tests joint hypothesis of δ 1 =δ 2 =δ 3 =δ 4 = Check F-statistics (equivalently, Wald statistics) ( ) uu ' uu ' / q F = = W/ q uu ' /( n k) 2 26
3 4 27
Controversial Relationship: Money and Fundamentals Affected the making of the monetary policy and monetary research Focus on the federal funds rate as an information variable and a monetary instrument Money supply measures are largely discredited as monetary policy instruments and indicators 6 28
Aksoy Piskorski (24) 7 Empirical Evidence:Variance decomposition 8 29
9 Conclusions Monetary Policy focus has shifted from monetary aggregates to short run interest rates Partially also true for other developed economies No more information content (in terms of significance and size) in Mon Aggregates No more instruments for short run MonPol 6 3
Measuring Monetary Policy: SVAR Approach So far we were looking mainly into simple correlations We need to identify monetary policy! Pioneered by Sims (192, 198) Very large literature now: CEE (1999), Leeper, Sims, Zha (1996) good summaries 61 Measuring Monetary Policy: SVAR Approach Suppose a bivariate system y y u = A( L) + t t 1 yt x t x t 1 u xt Shocks can be thought of linear combinations of iid shocks to output (e y ) and policy (e x ) uyt eyt + θext 1 θ eyt eyt B u = φe xt yt e = = xt φ 1 e xt e + xt 62 31
Measuring Monetary Policy: SVAR Approach Obtaining an estimate of u xt does not tell you much about the policy shocks if φ Identification: Add restrictions on the B matrix that links observalbe VAR residulas to underlying structural disturbances (Sims (972, 88), Bernanke, Blinder (1992), Bernanke Mihov (1998) etc) e.g. φ= (policy shocks are exogenous w.r.t non policy shocks) (Sims) or θ= (policy shocks do not have contemporaneous impact on output) BB (1992) and BM (1998). Or add LR restrictions, e.g. monetary neutrality (Blanchard and Quah (1989)). 63 Restrictions are often difficult justify φ= (policy shocks are exogenous w.r.t non policy shocks) θ= (policy shocks do not have contemporaneous impact on output) depends on the unit of obsrvation bivariate specification is problematic What is the right policy variable? 64 32
Money and Output Sims (1992), FR, GER, J, UK, US Restriction: φ= (policy variable is ordered first, therefore policy shocks are exogenous w.r.t non policy shocks) Short term interest rates as policy variables Other variables: IP, CPI, M, fx rate, commodity prices Finding: hump shaped pattern of output w.r.t. policy shock 6 Christiano et al. (1996, 1998) 66 33
67 Money, Output, Inflation Eichenbaum (1992) Alternative assumptions on the policy variable (M or i) If M is policy variable: M i y If i is policy variable: i y p-level Inclusion of commodity prices and nominal exchange rate solves partially price puzzle Barth and Ramey (21) argues for cost channel 68 34
Critics of the SVAR Impulse responses do not resemble our beliefs SVAR policy residuals (shocks) look quite different than historical accounts of contractionary/ expansionary monetary policies. Policy reaction functions (systematic part) also differ from alternatives (Taylor rule e.g.) Lack of forward looking variables 69 Conclusions Monetary Policy focus has shifted from monetary aggregates to short run interest rates No more information content (in terms of significance and size) in Mon Aggregates No more instruments for short run MonPol VAR s are useful however much criticized approach to account for monetary policy 7 3