Global Imbalances, Reserve Holdings, and the Trilemma
Fondation Banque de France 2006 Proposal Maurice Obstfeld
University of California, Berkeley & CEPR
Jay C. Shambaugh
Dartmouth College
Alan M. Taylor
University of California, Davis, LBS & CEPR
1
Outline
Accumulation of reserves by central banks poses a major problem in economic theory and policy Hot debates
Contribute to “Global Imbalances” Part of the“global savings glut” Sustain U.S. “exorbitant privilege” Key force in “Bretton Woods 2”
Is there a pattern? Will it continue?
• E.g.: Emerging Asia increases exchange rate flexibility; will reserve holdings go up or down? Not obvious… 2
What is going on?
Global imbalances: U.S. & them
CURRENT ACCOUNTS 2005 ($bn)
$600 $400 $200 $24 $0 -$200 -$400 -$600 -$800 -$805 -$1,000 $153 $68 $159 $73
Source: Martin Wolf, Larry Summers
$380
$328
(T O T A L A S IA )
EURO ZO NE
C H IN A
REST OF W ORLD AND D IS C R E P A N C Y
US
O IL EXPO R TERS
REST OF E M E R G IN G A S IA
JA P A N
3
“It’s Official”: Official Settlements Balance not Non Reserve Financial Account
CHART 17. BALANCE OF PAYMENTS OF EMERGING MARKET ECONOMIES ($bn)
$800.0
Source:IMF
$600.0 $400.0 $200.0 $0.0 -$200.0 -$400.0 -$600.0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Change in reserves
(-)
Current account balance Other private flows
Direct investment, net
4
Recent literature
Empirical models stress “buffer stock” (precautionary motives)
Cover for X months of imports Cover for Y months of short-term liabilities
• Old story: traditional BOP crisis • New story: “sudden stop” & Greenspan-Guidotti rule
+ Control for volatility of external shocks + Other controls
• Frenkel and Jovanovic (1981); Frenkel (1978); Lane and Burke (2001); Flood and Marion (2002); Aizenman and Marion (2004); Aizenman and Lee (2005)
Model performance is disappointing
5
What is missing?
The Trilemma Extant models take no account of exchange rate regime, capital controls, and monetary policy and the interaction of the three
The Trilemma says that a central bank can choose 2 out of 3 from
• Fixed exchange rate • Autonomous monetary policy • Capital mobility
Trilemma does not suggest corner solutions but tradeoffs across policy options
6
Trilemma matters
We have shown that trilemma does bind. (REStat 2005/IMF Staff Papers 2004)
Code all known exchange rate regimes (de facto and de jure) since 1870 (peg/float) Code capital control regimes (open/closed) “Open+peg” regimes cannot move their interest rates independently of the base country Other regimes enjoy more autonomy
Result
Can the trilemma affect reserve holdings too? We think so
7
Trilemma and reserves?
Floaters
In theory no reserves needed In reality dirty floaters need reserves
• Coding of ER regimes needed—Shambaugh, Reinhart & Rogoff
Closed pegs
In theory no reserves needed In reality controls leaky, so need reserves
• controls coding needed—Edwards (2005), others
Open pegs
In theory need reserves In reality depends on “precautionary factors” and monetary policy
• buffer stock variables should matter more in this case
8
Trilemma and reserves?
Suggests we need to consider the tradeoffs being made and the extent a country is trying to break out of the trilemma Suggests we need a model that allows for interaction effects
Trilemma indicators Standard buffer stock variables + new financial variables (Greenspan-Guidotti rule)
E.G. Reserves may be unnecessary with perfect capital controls and useless with no capital controls as intervention becomes powerless, BUT useful with intermediate amounts
9
We may also need to allow for nonlinearities….
Trilemma and reserves?
What about sterilization?
• If countries think they can sterilize (and want to) they may need more reserves • But this may only be when there are (some) controls.
What about transmission mechanisms?
• Countries able to adjust home nominal interest rate R to changes in base country nominal interest rate R* via rapid changes in domestic credit (rather than reserves) need less buffer
What about “room for maneuver”?
• Wider “target zone” — less reserves • Using zone (deviation of R from R*)— more reserves • Factors discussed above are not easy to measure
10
The dependent variable
Use IMF data on reserves
• BOP data on flows known to be seriously flawed • IMF stock data are only data on reserve holdings for a large sample of countries
Can the IMF stock data be trusted?
Known weaknesses may not affect us much
• levels data weak only for China and some gulf oil exporters; unclear if bias is changing over time
In our discussion with IMF and other experts we have learned that
• still the best data out there, and cannot be improved upon (honesty/secrecy of central banks) • levels data are quite accurate for most countries with the few exceptions noted above 11
Empirics
A simple buffer stock model can be the starting point Reserves/X = a + b1 Volatility + b2 IRgap
Where Reserves are scaled by some “X” factor (GDP, M2, etc) Where Volatility uses a measure of the volatility of a monetary base Where IR gap measures the cost of holding reserves (home interest rate minus some base rate) Real and financial openness can be added as well Country fixed effects can be used to pick up numerous other effects
12
Preliminary work
In preliminary work, we find some support for the basic structure of the empirical framework, but much is missing As noted above, simply including dummies for trilemma measures is unlikely to be sufficient… and it does not appear to be in preliminary work
13
Preliminary work cont.
Interacting precautionary variable with trilemma variables appears to be important Including financial variables or accumulated CA surpluses may be helpful as well
Again, with interactions across ER, Capital Control, and monetary policy interactions
14
Conclusion
Important topic, growing literature Missing important link to the trilemma Interactions matter, but ignored so far Not an obvious effect, will take work to identify the appropriate empirical model as well as to classify countries’ intent Importance of interactions also seen in other recent work
Eg Aghion, Bacchetta, Ranciere and Rogoff
15