Download presentation

Presentation is loading. Please wait.

Published byValerie Stoken Modified about 1 year ago

1
Ernesto Talvi** Prepared for Presentation at the Session “Responding to Sudden Stops”, XXVI Meeting of the Latin American Network of Central Banks and Finance Ministries, IADB, Washington DC October 17 th, 2007 Monetary and Fiscal Policies in a Sudden Stop: Is Tighter Brighter? Pablo Ottonello** Federico Sturzenegger*** Boston University*, CERES**, Harvard University and Universidad Torcuato di Tella*** Alberto Ortiz*

2
MOTIVATION During the financial crises of the 1990s, there was a lively debate on the optimal monetary and fiscal policy response: Should a country facing a sudden stop tighten its fiscal and monetary policies to restore credibility and avoid potentially unstable dynamics ? (the IMF’s view)

3
Stanley Fischer (1998), in the context of the Asian 1997 crisis, argued that: “(…) when they approached the IMF, the reserves of Thailand and Korea were perilously low, and the Indonesian rupiah was excessively depreciated. Thus, the first order of business was, and still is, to restore confidence in the currency.” “To achieve this, countries have to make it more attractive to hold domestic currency, which, in turn, requires increasing interest rates temporarily (…)” “At the outset of the crisis, countries needed to firm their fiscal positions, both to make room in their budgets for the future costs of financial restructuring, and --depending on the balance of payments situation -- to reduce the current account deficit.” MOTIVATION

4
Or conversely, should it relax those policies in order to attenuate the output contraction that typically occurs during these events? (IMF’s critics views) During the financial crises of the 1990s, there was a lively debate on the optimal monetary and fiscal policy response: Should a country facing a sudden stop tighten its fiscal and monetary policies to restore credibility and avoid potentially unstable dynamics ? (the IMF’s view)

5
Joseph Stiglitz (2002, 2003), one of the most vocal critics of the IMF view, argued that: “For more than seventy years there has been a standard recipe for a country facing a severe economic downturn. The government must stimulate aggregate demand, either by monetary or fiscal policy – cut taxes, increase expenditures, or loosen monetary policy.(…) The crisis economies of East Asia were clearly threatened with a major downturn and needed stimulation. The IMF pushed exactly the opposite course, with consequences precisely of the kind that one would have predicted.” “(…) these procyclical discretionary fiscal policies exacerbated the downturns still further in country after country.” “When the Fund entered East Asia, it forced countries to raise interest rates to what, in conventional terms, would be considered astronomical levels. (…) The IMF had engineered a simultaneous contraction in aggregate demand and supply.” MOTIVATION

6
This paper attempts to contribute to this debate empirically by studying the fiscal and monetary policy response and their effects on output, for a set of external financial crisis episodes occurred since the 1990s. Or conversely, should it relax those policies in order to attenuate the output contraction that typically occurs during these events? (IMF’s critics views) During the financial crises of the 1990s, there was a lively debate on the optimal monetary and fiscal policy response: Should a country facing a sudden stop tighten its fiscal and monetary policies to restore credibility and avoid potentially unstable dynamics ? (the IMF’s view)

7
aggregate-spread window containing a spike in the EMBI spread exceeding two standard deviations from its mean (which starts when the aggregate EMBI spread exceeds one standard deviation, and ends when it is smaller than one standard deviation). Identification of Systemic Sudden Stop Episodes Sample Countries that are tracked by JP Morgan to construct its global Emerging Market Bond Index, or global EMBI (31 countries) Period Definition of Systemic Sudden Stop (SSS) In similar fashion to Calvo, Izquierdo and Loo-Kung (2005), we define a SSS window as the union of a capital-flow window containing a large fall in capital flows for a given country exceeding two standard deviations from its mean (that starts when the fall in capital flows exceeds one standard deviation, and ends when it is smaller than one standard deviation) that overlaps at any point in time with an

8
ene-70ene-72ene-74ene-76ene-78ene-80ene-82ene-84ene-86ene-88ene-90ene-92ene-94ene-96ene-98ene-00ene-02ene-04 Effective Federal Funds Rate (%) EMBI Sovereign Spread (Bps over US Treasuries) Tequila Crisis Asia-Russian Crises US Monetary Contraction Capital Market Conditions for EMs Fed Fund RateEMBI Spreads Identification of SSS Episodes

9
Sample aggregate-spread window containing a spike in the EMBI spread exceeding two standard deviations from its mean (which starts when the aggregate EMBI spread exceeds one standard deviation, and ends when it is smaller than one standard deviation). Countries that are tracked by JP Morgan to construct its global Emerging Market Bond Index, or global EMBI (31 countries) Period Definition of Systemic Sudden Stop (SSS) In similar fashion to Calvo, Izquierdo and Loo-Kung (2005), we define a SSS window as the union of a capital-flow window containing a large fall in capital flows for a given country exceeding two standard deviations from its mean (that starts when the fall in capital flows exceeds one standard deviation, and ends when it is smaller than one standard deviation) that overlaps at any point in time with an Output Performance during SSS Output performance is computed by the peak to trough variation of GDP*. Identification of Systemic Sudden Stop Episodes *If either the peak or trough falls within the SSS window, the contraction is classified as belonging to the period of the SSS. If a country experienced a deceleration but not a contraction, the dating was determined using the HP-filtered cyclical component of output.

10
SSS Episodes and Output PeakTrough Argentina 98 Indonesia Thailand Morocco Turkey 93 Malaysia Russia Mexico Korea Turkey 98 Ecuador Colombia Croatia Argentina 94 Chile Lebanon Brazil 95 Peru Philippines Brazil 97 Poland Dominican Republic Jun-98 Dec-97 Sep-96 Dec-94 Dec-93 Dec-97 Dec-94 Sep-97 Mar-98 Dec-98 Jun-98 Dec-97 Dec-94 Jun-98 Sep-98 Mar-95 Dec-97 Mar-94 Mar-02 Dec-98 Sep-98 Jun-95 Jun-94 Sep-98 Jun-95 Jun-98 Mar-99 Sep-99 Jun-99 Sep-95 Mar-99 Jun-99 Sep-95 Dec-98 Jun-98 Mar-99 Sep-95 GDP Dates Country -20.9% -17.3% -15.1% -13.3% -12.2% -11.0% -10.1% -9.7% -8.5% -8.1% -7.6% -7.1% -5.9% -5.6% -4.6% -3.3% -2.7% -2.4% -2.2% -1.7% 3.2% 6.6% GDP Variation Peak to trough % change Average-7.2%

11
Characterizing Fiscal Policy in SSS: Structural Fiscal Balance Methodology We estimate the Structural Fiscal Balance by adopting the Chilean Fiscal Rule. This rule defines the structural balance as the difference between structural fiscal revenues and observed fiscal expenditures. Structural fiscal revenues are defined as the level of revenues that would have been achieved if output were at its potential level and the copper price were at its long run level.We estimate the Structural Fiscal Balance by adopting the Chilean Fiscal Rule. This rule defines the structural balance as the difference between structural fiscal revenues and observed fiscal expenditures. Structural fiscal revenues are defined as the level of revenues that would have been achieved if output were at its potential level and the copper price were at its long run level. While we cannot directly replicate this rule for other countries we can find a “statistical equivalent” to it. To do so we compute the Lagrange multiplier of the Hodrick-Prescott filter for current revenues in Chile in order to estimate by how much the Chilean authorities smooth their income.While we cannot directly replicate this rule for other countries we can find a “statistical equivalent” to it. To do so we compute the Lagrange multiplier of the Hodrick-Prescott filter for current revenues in Chile in order to estimate by how much the Chilean authorities smooth their income. The Lagrange multiplier that delivers a surplus/deficit that best matches the structural balance reported by the authorities is the one that provides a statistical equivalent to their complex rules. Once the "smoothing" parameter is chosen, the filter is applied to fiscal revenues of the countries included in our sample to compute our measure of structural balance.The Lagrange multiplier that delivers a surplus/deficit that best matches the structural balance reported by the authorities is the one that provides a statistical equivalent to their complex rules. Once the "smoothing" parameter is chosen, the filter is applied to fiscal revenues of the countries included in our sample to compute our measure of structural balance. The Hodrick Prescott (HP) filter chooses the sequence of that minimizes where r* t = adjusted fiscal revenues according to the Chilean Fiscal Rule and g t = total public expenditures, both in percent of GDP. For each country of our sample the Structural Fiscal Balance (sb t ) is defined as:For each country of our sample the Structural Fiscal Balance (sb t ) is defined as: sb t = r* t - g t sb t = r* t - g t ifif, approaches a linear trend. Objective To capture the discretional components of fiscal policy by extracting the effect of cyclical fluctuations on fiscal accounts.To capture the discretional components of fiscal policy by extracting the effect of cyclical fluctuations on fiscal accounts.

12
Characterizing Fiscal Policy in a SSS: Computation of Structural Fiscal Balance Characterization of Fiscal Policy in SSS Structural Fiscal Impulse in SSS (From Output Peak to Trough) -5.0%-4.0%-3.0%-2.0%-1.0%0.0%1.0% RUS ARG 98 IDN TUR 93 COL THA MYS KOR MEX POL TUR 98 PER BRA 97 ECU CHL HRV PHL ARG 94 Average: -1,1% A positive (negative) value indicates an expansionary (contractionary) fiscal policy. Structural Fiscal Impulse throughout the output peak to trough window:Structural Fiscal Impulse throughout the output peak to trough window: I* t = - sb t I* t = - sb t -3.0%-2.0%-1.0%0.0%1.0%2.0%3.0% 4.0% TUR 93 RUS ECU KOR COL MEX BRA 97 POL IDN TUR 98 PER PHL ARG 94 HRV MYS CHL ARG 98 THA Average: 0,6% Observed Fiscal Impulse in SSS* (From Output Peak to Trough) * Observed Fiscal Balance is defined as fb t = r t – g t,, where r t = fiscal revenues and g t = total public expenditures, both in percent of GDP. Observed Fiscal Impulse is defined as I t = - fb t

13
Characterizing Monetary Policy in SSS: Estimation of Central Bank Reaction Function Methodology We estimate the pre-SSS central bank reaction function by estimating a dynamic stochastic general equilibrium model of a small open economy using Bayesian methods, following Lubik and Shorfheide (2007).We estimate the pre-SSS central bank reaction function by estimating a dynamic stochastic general equilibrium model of a small open economy using Bayesian methods, following Lubik and Shorfheide (2007). where: t R =nominal interest rate, =output, t y t s =nominal exchange rate t =inflation, =inflation, 1 =“Anti-Inflation” coefficient, 2 =“Output Motive” coefficient, 3 =“Fear of Floating” coefficient. R = partial adjustment of the interest rate to target, R t =exogenous policy shock, which can be interpreted as the non-systematic component of monetary policy, Objective To capture the discretional components of monetary policy by eliminating the noise in interest rates movements typically observed during SSS.*To capture the discretional components of monetary policy by eliminating the noise in interest rates movements typically observed during SSS.* *See Calvo (2006)

14
Average Average Source: (a) Lubik and Shorfheider (2007), (b) Ortiz and Sturzenegger (2007). Notes Data for Argentina 98 considers the period 2001:1, 2002:2. Polish data start in 1998:1. Characterizing Monetary Policy in SSS: Central Bank Reaction Function Estimates The “Fear of Floating” coefficient ( ) is significantly larger in our group of EM countries suggesting the exchange rate is a more relevant concern.The “Fear of Floating” coefficient ( ) is significantly larger in our group of EM countries suggesting the exchange rate is a more relevant concern. 3 Argentina Argentina Brazil Chile Colombia Croatia Ecuador Indonesia Korea Malaysia Mexico Peru Philippines Poland Russia Thailand Turkey Turkey Australia (a) New Zeland (a) United Kingdom (a) Canada (a) South Africa (b) The “Anti Inflation” coefficient ( ) is on average similar in our group of EM countries to the control group, but with a greater dispersion in the former.The “Anti Inflation” coefficient ( ) is on average similar in our group of EM countries to the control group, but with a greater dispersion in the former. 1 Similarly with respect to the “Output Motive” coefficient ( ).Similarly with respect to the “Output Motive” coefficient ( ). 2 Central Bank Reaction Function Country Parameter Estimation 1 2 3 The initial reaction of the Central Bank for a given value of the shock and the parameters is larger in our group of EM countries (i.e. a larger (1 - )The initial reaction of the Central Bank for a given value of the shock and the parameters is larger in our group of EM countries (i.e. a larger (1 - ) ) R i

15
CB Reaction Coefficient CB Reaction Coefficient 100% Fear of Floating Coefficient NZL AUS ZAF GBR CAN MYS CHL MEX COL ECU PHL BRA 97 IDN TUR 93 TUR 98 RUS KOR POL PER THA ARG 94 HRV ARG 98 Control Group Control Group Avg.: 0.1 at the outset of the SSS POL MYS CHL KOR PER PHL COL BRA 97 TUR 93 HRV ARG 98 ARG 94 THA ECU MEX TUR 98 RUS IDN ( ) at the outset of the SSS 1 - R 83% Characterizing Monetary Policy in SSS: Central Bank Reaction Function Estimates

16
Monetary Policy Trade-Offs in a SSS: the Typical Shock* *Average of the 18 episodes of the sample 14% 16% 18% 20% 22% 24% 26% t-5t-4t-3t-2t-1tInflation (Annualized rate, in %) Quarter Output Peak Output Trough 15.2% 25.2% t-5t-4t-3t-2t-1tOutput (GDP trough=100) Quarter Output Peak Output Trough -7.5% t-5t-4t-3t-2t-1t Nominal Exchange Rate (GDP trough=100) Quarter Output Trough Output Peak +43.1%

17
Characterizing Fiscal Policy in a SSS: Computation of Monetary Policy Indices Characterization of Monetary Policy in SSS Countries with a higher (lower) value of the index will have a more contractionary (expansionary) monetary policy than countries with a lower (higher) value of the index, in a SSS episode. I.We construct a Monetary Policy Regime Index at the outset of the SSS, given by*: Inflation/Output Trade-Off Index at the outset of the SSS:Inflation/Output Trade-Off Index at the outset of the SSS: Exchange Rate/Output Trade-Off Index at the outset of the SSS:Exchange Rate/Output Trade-Off Index at the outset of the SSS: II.We construct two separate Monetary Policy indices, given by*: These indices are used as a proxy of how monetary policy (i.e. interest rates) will react during a SSS episode. *Where are computed as deviations with respect to the sample mean. Countries with a higher (lower) value of the index will have a more contractionary (expansionary) monetary policy than countries with a lower (higher) value of the index, in a SSS episode.

18
We first compute as the dependent variable the output performance during SSS, as described by the (output) peak to trough variations ( ). The Impact of Monetary and Fiscal Policy on Output in a SSS: Empirical Strategy We then relate our measures of monetary and fiscal policy to output performance by performing simple OLS regressions : Model IModel I (2.534)(-2.024)

19
Monetary, Fiscal Policy and Output Performance in SSS Fitted GDP Variation Actual GDP Variation ARG 94 ARG 98 BRA 97 CHL COL HRV ECU IDN KOR MYS 97 MEX PER PHL POL RUS THA TUR 93 TUR (Actual vs. Fitted GDP variation) Model I

20
Fiscal Policy and Output Performance in SSS GDP Variation ( Y ) TUR 98 TUR 93 THA RUS POL PHL PER MEX MYS 97 KOR IDN ECU HRV COL CHL BRA 97 ARG 98 ARG (GDP variation and Structural Fiscal Impulse controlled by the rest of policy parameters) Model I: Fiscal Policy Model I: Fiscal Policy

21
GDP Variation ( Y ) TUR 98 TUR 93 THA RUS POL PHL PER MEX MYS 97 KOR IDN ECU HRV COL CHL BRA 97 ARG 98 ARG Model I: Monetary Policy Model I: Monetary Policy (GDP variation and Monetary Policy Index controlled by the rest of policy parameters) Monetary Policy and Output Performance in SSS

22
The Impact of Monetary and Fiscal Policy on Output in a SSS: Empirical Strategy Model IIModel II We first compute as the dependent variable the output performance during SSS, as described by the (output) peak to trough variations ( ). We then relate our measures of monetary and fiscal policy to output performance by performing simple OLS regressions : Model IModel I (2.534)(-2.024) (2.768) (-1.920) (-1.850)

23
Monetary Policy and Output Performance in SSS GDP Variation ( Y ) Inflation/Output Trade-Off Index Inflation/Output Trade-Off Index (GDP variation and Inflation/Output Trade-Off controlled by the rest of policy parameters) ARG 94 ARG 98 BRA 97 CHL COL HRV ECU IDN KOR MYS 97 MEX PER PHL POL RUS THA TUR 93 TUR Model II: Model II: (GDP variation and Exchange Rate/Output Trade-Off controlled by the rest of policy parameters) Exchange Rate/Output Trade-Off Index Exchange Rate/Output Trade-Off Index TUR 98 TUR 93 THA RUS POL PHL PER MEX MYS 97 KOR IDN ECU HRV COL CHL BRA 97 ARG 98 ARG GDP Variation ( Y ) Model II: Model II:

24
Monetary and Fiscal Policies in SSS: An Evaluation Is Tighter Brighter? It is Not. Is Looser Mightier? Maybe Yes, Maybe Not.

25
Ernesto Talvi** Prepared for Presentation at the Session “Responding to Sudden Stops”, XXVI Meeting of the Latin American Network of Central Banks and Finance Ministries, IADB, Washington DC October 17 th, 2007 Monetary and Fiscal Policies in a Sudden Stop: Is Tighter Brighter? Pablo Ottonello** Federico Sturzenegger*** Boston University*, CERES**, Harvard University and Universidad Torcuato di Tella*** Alberto Ortiz*

Similar presentations

© 2017 SlidePlayer.com Inc.

All rights reserved.

Ads by Google