“A Third-Generation Framework:

Reconsidering the Relationship between Central Bank Independence and Fixed Exchange Rates”

David H. Bearce, University of Colorado Boulder and

Ana Carolina Garriga, University of Essex

​

​

​

​

​

​

​

​

​

​

​

​

​

​

Bernhard, Broz, and Clark (2002): study CBI and FIX together “as a response to same economic problem” (i.e., inflation control) to understand if they function more “institutional substitutes (where the presence of one negates the need for another) or complements (where each reinforces the effect of the other)”.  

Hypotheses to be Tested

Research Design

​

All hypotheses tested using same sequence of models

​

1. LDV only

​

2. Country FE

​

3. Country + Year FE

​

4. LDV + Country + Year FE

​

​

Main results using de jure measures for CBI and FIX: CBIlegal (Garriga 2016) and FIXdejure (IMF)

both scaled 0-1 for comparability

​

​

Robustness checks using de facto measures: CBtorinv (Dreher, Sturm, and de Haan 2008, 2010) and

FIXdefacto (Ilzetzki, Reinhart, and Rogoff 2019)

​

​

Consistent set of controls: Democracy, Regional, Inflation Target, GDP, Population, lnGDPpc, Trade Openness, Capital Openness

H1: more democratic governments select greater CBI.

OLS coefficients with robust standard errors clustered on the country in parentheses. Control variables included but not reported. Statistical significance: * p<0.05 (two-tailed).

​

​

Weak results robust to using Freedom House measure of Democracy (Table A1)

Find contrary results using the de facto measure of CBI (CBtorinv) as the DV (Table A2).

Table 2: Estimates of CBIlegal on Democracy (Polity).

H2: FIX associated with lower inflation at least in less democratic regimes.

OLS coefficients with robust standard errors clustered on the country in parentheses. Control variables included but not reported. Statistical significance: * p<0.05 (two-tailed).

​

​

Non-results robust to using a de facto measure of FIX (Table A3).

Table 3: Estimates of Inflation on FIXdejure conditioned on Democracy (Polity).

H3: there is a positive association between FIX and CBI.

OLS coefficients with robust standard errors clustered on the country in parentheses. Control variables included but not reported. Statistical significance: * p<0.05 (two-tailed).

​

​

Non-results robust to reversing the DV and IV and

when using de facto measures of FIX and CBI (Table A4).

Table 4: Estimates of FIXdejure on CBIlegal.

H4: CBI and FIX are associated with reduced inflation when combined.

OLS coefficients with robust standard errors clustered on the country in parentheses. Control variables included but not reported. Statistical significance: * p<0.05 (two-tailed).

​

Also no evidence of combined inflation reduction when using de facto measures of FIX and CB1 (Table A5).

Table 5: Estimates of Inflation on FIXdejure*CBIlegal interaction.

The Macrofoundations Underlying the 3rd Generation Framework

​

​

1. CBI and FIX generally used to achieve different macroeconomic objectives:

​

H5: CBI more associated with inflation control than FIX, while

​

H6: FIX more associated with external currency stability than CBI.

​

​

2. Tension between these macroeconomic objectives and these monetary institutions:

​

H7: Inflation control and external currency stability pull monetary policy in opposite directions,

​

H8: CBI associated with increased inflation in the presence of FIX, and

​

H9: FIX associated with reduced external currency stability in the presence of CBI.

​

H5: CBI more associated with inflation control than FIX.

OLS coefficients with robust standard errors clustered on the country in parentheses. Control variables included but not reported. Statistical significance: * p<0.05 (two-tailed).

​

Even weaker results for FIX when using de facto measures (Table A7).

Table 6: Estimates of Inflation on FIXdejure and CBIlegal.

​

H6: FIX more associated with external currency stability than CBI.

OLS coefficients with robust standard errors clustered on the country in parentheses. Control variables included but not reported. Statistical significance: * p<0.05 (two-tailed).

​

Results robust when using de facto measures of FIX and CB1 (Table A8).

Table 7: Estimates of Exchange Rate Stability on FIXdejure and CBIlegal.

​

H7: Inflation control and exchange rate stability pull monetary policy in opposite directions

OLS coefficients with robust standard errors clustered on the country in parentheses. Control variables included but not reported. Statistical significance: * p<0.05 (two-tailed).

Table 8: Estimates of Interest Rate on Inflation and Exchange Rate Stability.

H8: CBI associated with increased inflation in the presence of FIX.

OLS coefficients with robust standard errors clustered on the country in parentheses. Control variables included but not reported. Statistical significance: * p<0.05 (two-tailed).

Table 5: Estimates of Inflation on FIXdejure*CBIlegal interaction.

H9: FIX associated with reduced external currency stability in the presence of CBI.

OLS coefficients with robust standard errors clustered on the country in parentheses. Control variables included but not reported. Statistical significance: * p<0.05 (two-tailed).

Table 10: Estimates of FIXdefacto on FIXdejure Conditioned on CBIlegal.

What does this framework mean for political science research on monetary institutions?

​

​

​

1. A dilemma between using monetary policy for domestic price stability and external currency stability.

​

​

​

2. CBI and FIX used by different sides to impose their monetary policy preferences.