Discussion of “A Global Village without borders

Download Report

Transcript Discussion of “A Global Village without borders

“Productivity Shocks, Budget Deficits and
the Current Account”
Bussiere, Fratzscher, Muller
Discussion comments from
John Rogers, FRB
April 29, 2006
Findings
• Productivity shocks are an important
determinant of OECD countries
current account balances.
• Government budget balances are not
important.
• Suggests there is little evidence of a
“twin deficits” story to current
account determination.
Related Literature
• Twin Deficits papers
The Salad Days:
1980s papers (Summers, Bernheim, Roubini, Evans);
Mostly skepticism more recently:
VAR papers;
DGE models (Normandin, Erceg et. al.);
Cross-country estimation (Chinn-Prasad)
• Productivity papers
Glick-Rogoff, Kollman, Nason-Rogers, Gregory-Head
• Bussiere, Fratzscher, Muller
-- Extension of Glick-Rogoff
-- Very well-written, a pleasure to read
Theory Section
• Glick-Rogoff (1995) model with a new wrinkle
• Motivates role for budget balance;
mechanism not taken too seriously
• Add rule-of-thumb consumers to G-R
- -> gov’t budget balance empirically unimportant for CA
- -> Issler-Vahid (JME, 2001)
use same mechanism more successfully
(closed economy VECM as in KPSW)
- -> why no permanent vs. temporary (fiscal policy) distinction?
Ahmed (JME, 1986 and 1987), others
Data
•
Annual, 1960 – 2003
•
Two separate panels
- G-7
- 21 OECD countries
•
Global, country-specific productivity
– OECD measure
– Solow residuals:
(i) Form Solow residual for each country;
(ii) Take (GDP) weighted average across countries;
(iii) Country-specific is deviation from global avergage.
•
Regressions
• Similar to Glick-Rogoff (1995)
(1) ΔCA(t) = rCA(t−1) + λΔS(t) + γ1I(t−1) + γ2ΔAc(t) + γ3ΔAg(t)
(2) ΔI(t) = (β1 − 1) I(t−1) + β2ΔAc(t) + β3ΔAg(t)
-- S primary gov’t surplus (the wrinkle)
-- Ac (Ag) country-specific and global productivity
-- Panel estimates and country-by-country (G-7)
• Of interest:
-- γ2 < 0 γ3 = 0 (standard implications; BFM confirm)
-- λ important? (BFM: No)
Robustness
• Productivity computed with principal component analysis
– Nice, but I’m not sure this identifies common world productivity
 Provide details
 Relate to Gregory and Head (JME, 1999) [more below]
• Drop certain countries from the sample
• Cyclically adjusted fiscal balance vs. unadjusted
Why Aggregate the Data?
•
Best answer may be comparability with GR.
•
But using average productivity  e.g., a large movement in any individual country
raises the average even if all other countries productivity unchanged.
Suppose there is a country specific component to TFP.
- If it is idiosyncratic white noise, aggregating will wash it out.
- If ~ random walk and independent across the economies, aggregating fine.
- If ~ ARMA(p,q) and correlated across economies, aggregating implies
mismeasurement of aggregate TFP. (Literature on “common features”.)
Alternative: Common trend-common cycle model
- Vahid and Engle (JAE, 1993), Engle and Issler (JME, 1995).
- Test for number of permanent and transitory components in G-7 TFPs.
- Use estimates to test response of G-7 CA and I to TFP shocks.
- Method measures simultaneous productivity changes in all seven countries.
- Fluctuations may be of any magnitude and still be entirely country-specific.
- Common movements fundamentally different from cross-country averages.
Conjecture: common features exist in G-7 transitory country specific TFPs because G-7
output growth rates appear to share common features.
Suggested Alternative Approach
• Follow Gregory and Head (JME, 1999)
- Dynamic factor model of Productivity, Investment and CA
Zjt = aZjWt + uZjt,
Ijt = aIjWt + bjuZjt+ uIjt
CAjt = aCAjWt + cZjuZjt + cIjuIjt + uCAjt
• Current account decomposed into four components:
– world-wide component, W;
– country-specific productivity (uZ) and investment (uI) fluctuations;
– residual associated with fluctuations in neither productivity nor investment.
• Results:
– Country-specific productivity fluctuations little impact on CA
(consistent with GR)
– Country-specific investment fluctuations of primary importance for CA
Other Alternatives
• Chinn-Prasad (JIE, 2003)
– Empirical determinants of “medium-horizon” CA
fluctuations
– More satisfying econometrically than GR regressions
– Gov’t budget balance consistently positive and
significant
• Nason-Rogers (JMCB, 2002)
– Emphasis on joint dynamics of Investment and CA
– Use a collection of restrictions and several justidentified VARs of I and CA
– Focus on shocks
Conclusion
• BFM find no evidence for Twin Deficits
story for CA determination.
• Suggestions:
– (1) examine time-variation in estimates
– (2) relate more to existing literature