By Maurice Obstfeld, Jay C. Shambaugh and Alan M. Taylor
Economists’ Forum, 14 de octubre del 2008
Since the early 1990s, central banks in many emerging markets and developing countries have accumulated foreign reserves at an unprecedented rate. The macroeconomic impact of these official flows has been profound and they have contributed significantly to global imbalances. Providing an explanation for these trends remains a major puzzle in international macroeconomics, and prevailing theories based on trade or debt deliver poor empirical performance. We argue that part of this great reserve accumulation is a response to the threat of financial instability in the context of rapidly expanding financial systems, increasingly mobile capital, and exchange rate objectives. The recent turbulence in global financial markets supports this view.
In the 1980s, the monetary authorities in both industrial and developing countries maintained a fairly steady and consistent level of foreign reserves, at about 4 percent of GDP. After 1990, the trends in the two groups of countries diverged, and a great accumulation of reserves began in the emerging market countries, which by 2005 had accumulated reserves in excess of $2 trillion representing more than 20 percent of their collective GDP (Figure 1).
Why have the central banks of emerging markets and developing countries accumulated such large hoards of foreign reserves? This question remains one of the great puzzles of contemporary international macroeconomics.
For example, in the realm of theoretical work, Jeanne (2007) found it difficult to reconcile the magnitude of current reserve holdings with a model where countries need a buffer to insure against a “sudden stop” in capital flows that might expose them to consumption risk, unless one assumes that the cost of a crisis is implausibly large. In the realm of empirical work, influential papers by Joshua Aizenman and co-authors (2003, 2007) have shown some success. Most empirical models of reserve-to-GDP ratios have tended to rely on the “traditional” explanatory variables such as trade-to-GDP ratios (the rule since the Bretton Woods era being enough reserves to finance three months of imports). Traditional measures such as this were joined in the 1990s by “new” explanatory variables such as the ratio of short-term debt to GDP (based on the so-called Guidotti-Greenspan rule of thumb).
Still, even the augmented empirical models tend to underpredict reserve levels, especially in the more recent years since the wave of emerging-market crises of the late 1990s. That led the IMF (2003) to conclude that existing models could be judged a failure. The motives for the recent reserve build-up therefore remain open to debate and the subject of great controversy. If reserve levels are “too high” relative to any reasonable benchmark, then the case can be made that countries are squandering valuable resources on low yielding assets that could be better invested elsewhere (Summers 2006; Rodrik 2006).
Financial stability and the trilemma
Central banks have multiple responsibilities that go beyond concerns for price stability (and, to a greater or lesser degree, for output and employment). As recently noted by Charles Goodhart on Vox-and as recent events in developed economies have reminded us-central banks also have the frontline responsibility for financial stability. Unlike treasury departments or financial regulators, it is central banks that have the unique capacity to supply liquidity rapidly and plentifully in the event of a banking crisis. How can central banks fulfill their responsibilities under the conditions that prevail in typical emerging market countries today?
In a new paper , we argue that three crucial factors have forcefully coincided since 1990 to expose emerging markets to a much greater risk of crises that take the form of a classic “double drain” from bank deposits to cash, and then from cash to hard currency. These factors are:
1. A continuing desire to maintain a policy of fixed (or tightly managed) exchange rates, or a “fear of floating” (Calvo and Reinhart 2002), whether to provide a transparent and credible nominal anchor, to boost trade (Klein and Shambaugh 2006), or to avert destabilizing balance sheet shocks when liabilities are dollarized.
2. An ongoing trend, related to economic development, toward an increasingly monetized economy with a larger domestic banking and financial system relative to GDP (Demirgüç-Kunt and Levine 2001).
3. A new inclination to shift policy so as to liberalize external financial flows. This liberalizing trend is seen as complementary with a deepening of domestic financial markets (Obstfeld forthcoming).
The potential risk of a financial storm combining these three elements has long been recognized. For examples, as well as some theoretical intuition and potential policy responses, we can look back in history, both distant and recent.
Britain, 1797 and 2007 In 1797 a bank run during the Napoleonic wars saw the Bank of England issue paper credit as a dernier resort to ailing country banks. However, the public desired not paper, but gold, and as the paper came back to Threadneedle Street, the Bank’s stock of gold proved insufficient to cover the drain. The Bank was forced to “stop payment”-that is, suspend the gold standard-and let the exchange rate float. Henry Thornton, in his seminal work Paper Credit, faulted the Old Lady for holding insufficient gold to allay such a “fright”; he also argued that a policy of credit contraction might only make fundamentals worse in the short run, amplifying the risk of crisis. The only safe approach was to accumulate a large war chest of reserves ex ante.
The most recent Bank of England crisis (Northern Rock, 2007) serves to illustrate the unconstrained ability of a developed country central bank to intervene for financial stability purposes when the pound is floating: there is no exchange rate peg to worry about, no reserves to manage. The value of the pound could absorb any shock due to money creation, so the Bank could issue unlimited liquidity support.
Argentina, 1994 and 2001 In December 1994 the Mexican “Tequila Crisis” reverberated across emerging markets and Argentina, then on a pseudo-currency board peg, faced a sudden stop in capital markets and a double drain. The currency stock M0 held steady, but broad money M2 collapsed as depositors fled for Miami and Montevideo. By early 1995 reserves had fallen by about half as the central bank extended liquidity support to domestic banks. The IMF, previously reluctant to lend to Argentina, relented, allowing the country to replenish its reserves. At that point private capital flows resumed, but it was a close shave.
Argentina had no such luck in 2001, when suffering from even worse macroeconomic fundamentals. Over the course of the year to October more than $10 billion in reserves bled out as M2 fled into foreign currency. In November negotiations with the IMF collapsed, no reserves could be borrowed, and the sudden stop was complete. The reserve drain jumped to $1 billion per day, and after 48 hours the capital controls of the corralito put an end to the convertibility plan. By early 2002 the peso was floating at one third to one quarter of its former $1 par value.
These examples illustrate the vexing problems faced by monetary authorities as they navigate between the macroeconomic policy trilemma and the goal of insuring financial stability. Emerging markets with pegs and exposure to sudden stops face a larger risk of crisis the smaller are their reserve holdings, holding constant the size of any flight shock. Under a fixed exchange rate and capital mobility, there is no space for monetary policy autonomy. In previous work we have shown that this applied to interest rate policy (Obstfeld, Shambaugh, and Taylor 2004; 2005). But we now argue that the same constraints affect the reserve cushions needed to weather banking crises.
What the Data Show
We examine a panel of roughly 120 countries from 1980 to 2004 to assess the empirical relevance of these ideas. We find that if we add “financial factor” regressors to a traditional empirical model, they are always economically and statistically significant. Financially open, financially deep countries with pegged exchange rates all tend to hold more reserves. Within the emerging market sample – where much of the puzzle lies – the fixed exchange rate effect is weaker, but financial depth (as measured by the ratio M2/GDP) is highly significant and growing in importance over time. Trade openness is the other robust determinant of reserve demand. We conduct a variety of tests and show that financial depth is important both in the cross-section and in the time series dimension of the data, that our results are not simply a result of a positive effect of reserve growth on money-supply growth, and that our results are robust to considering measures of debt or original sin.
We also consider the predictive power of our arguments. A traditional model of reserves leaves substantial amounts of recent reserve growth unexplained, whereas our model does not. Our model cannot explain all reserve growth (Japan’s behavior in the last decade and China’s since 2004 cannot be explained entirely by financial stability motives and trade), but the puzzle identified in the literature is much smaller when financial factors are considered.
These findings lead us to conclude that recent reserve accumulation is in part explained by concerns for financial stability in the face of open capital markets and pegged exchange rates. In future work we will try to determine whether such accumulation appears sufficient to quell fears or if, instead, it has been too large (and thus accumulation excessive even in the face of financial stability concerns).
Our research suggests that financial development has played a major role in stimulating precautionary reserve accumulation in emerging markets. But we argue that the precaution is not to have a buffer against the inability of domestic residents to issue new external liabilities, the typical “sudden stop” argument. Rather, the buffer is also a safeguard against the sudden wish of domestic residents to acquire new external assets-that is, “sudden flight” (Rothenberg and Warnock 2006; Calvo 2006). The greatest flight risks are posed by the most liquid assets, the liquid liabilities of the banking system, which we measured by M2 and whose growth we found to be highly correlated with reserve growth in recent times.
These forces are by no means small. In our model, the predicted level of reserve purchases by all emerging markets for 2004 was $160 billion. For comparative purposes this sum is about ¼ of that year’s U.S. current account deficit. The emerging markets have large global macro weight and their GDP growth is high. Their M2 growth is higher still. If the risk of sudden flight remains, and if a preference for exchange rate stability remains, policymakers may be expected to continue to scale up their reserves in rough proportion to M2. In that case, we are likely to see even larger reserve accumulation, and larger global imbalances, for many years to come.
Maurice Obstfeld is a Professor of economics at the University of California, Berkeley, and is Director of the Center for International and Development Economics Research. Jay C. Shambaugh is an Associate Professor of economics at Dartmouth College. Alan M. Taylor is a Professor of economics at the University of California, Davis, and is Director of the Center for the Evolution of the Global Economy