-
Macroeconomic Volatility in Latin America:A Conceptual Framework and Three Case Studies
-
Comments
References
Footnotes
1. However, the country fund statistics suggest that the subsequent contagion to the rest of Latin America does indeed come via New York. See Frankel and Schmukler (1996, 1998).
2. In countries with a flexible exchange rate, such as Mexico more recently, interest rates can actually be even more sensitive to the U.S. Federal Reserve funds rate than they are in countries that use currency boards or dollarization to tie their monetary fates rigidly to the United States, as do Argentina and Panama (Frankel, 1999, table 1.)
3. For example, Lessard and Williamson (1985).
1. See Calvo and Mendoza (1996).
2. Caballero and Krishnamurthy (1999).
3. Eaton and Gersovitz (1981).
4. Calvo (1994); Dornbusch and Werner (1994).
5. See, for example, Paasche (1999); Schneider and Tornell (1999); Edison, Luangaram, and Miller (1998); Calvo (1999); Calvo and Mendoza (2000); Mendoza (2000).
6. Kiyotaki and Moore (1997); Kehoe and Levine (1993); Aiyagari and Gertler (1999); Bernanke and Gertler (1995).
7. Paasche (1999).
8. See Kiyotaki and Moore (1997).
9. Mendoza (2000, appendix).
10. Aiyagari and Gertler (1999).
11. See Mendoza (2000) for details.
12. Mendoza (2000).
13. Calvo (1999).
14. Martin Wolf, editorial, Financial Times, 8 August 1999.
15. Guerra de Luna (1997, 1998) provides data that illustrate this relation dramatically, both in Latin America and elsewhere.
16. Calvo and Mendoza (1996).
17. See Calvo and Mendoza (1996).
18. Calvo and Reinhart (2000).
19. Calvo and Mendoza (1999).
20. Paasche (1999) also supports this conclusion.
Caballero is with the Massachusetts Institute for Technology and the National Bureau of Economic Research.
Footnotes
1. See Caballero and Krishnamurthy (1999) and Aghion, Benerjee, and Piketty (1999) for models in which the correlation between financial development and economic stability is nonmonotonic. However, welfare is monotonically increasing with respect to institutional and financial development.
2. The rewards of successful reforms apparently come in the form of high average growth, a decline in the frequency of crises (although these can be deep), and an increase in the speed at which the economy recovers from such crises. However, the lack of precautionary options is clearly demonstrated by the serious decline in electricity provision in Chile during 1998, which resulted primarily from a severe drought. Shocks that involve the only existing pipeline, be it electricity distribution, financial services, or a road, have a deep impact. While precautionary options and the ability to handle complex scenarios are luxury goods, their rewards come precisely in the form of tamed volatility.
3. The essence of the conceptual framework is an adaptation of that presented in Caballero and Krishnamurthy (1999, 2000). The examples and applications are mostly from Caballero (1999a, 1999b, 1999c), a series of reports prepared for the IDB's Research Department. All of these papers can be downloaded from web.mit.edu/caball/www.
4. On the other hand, as financial development rises so does leverage, and as a result, the vulnerability of the financial system to shocks also increases. Many Latin American economies have suffered at both ends: chronic financial repression and underdevelopment and, when moving away from that, large collapse of the banking system. I return to this issue in the next section.
5. See Caballero and Krishnamurthy (1999) for a fully fleshed model along these lines.
6. In addition to binding microeconomic incentive problems, sovereign risk may be associated with many of these assets, especially in the event of crises. The latter affects foreigners' valuation of these assets even when they acquire the private control rights.
7. These abrupt changes in slopes are only meant to capture as clearly as possible the fact that there are regions where most firms can satisfy their financial needs and the cost of credit is determined by international conditions, and others where it is the domestic availability of international assets that determines such cost.
8. Foreigners or nonspecialists are unable to capture these high returns because at times of crises they only hold and arbitrage claims backed by international collateral. While their arbitrage during normal times keeps the international spread at zero, it is immaterial when the international collateral constraint binds. That is, the interest parity condition shifts until domestic equilibrium, rather than international arbitrage, holds.
9. The international economics literature has long recognized the importance of international collateral and its relation with a country's tradable sector (see Simonson, 1985). Formal models of sovereign debt renegotiation are built around the question of what international lenders can threaten sovereign countries with in the event of default. In this literature, international collateral is typically taken to be some fraction of exports (see Bulow and Rogoff, 1989). Cash revenues from exports can be seized before they make it back into the country. This feature was used by Mexico during the 1994-95 crisis, when its oil revenues were made part of the collateral backing the liquidity package it received.
10. Note that a lower L does not necessarily mean that the explicit domestic rate is lower than in the case with well developed financial markets (for a given supply of international collateral). It essentially means that a lower fraction of investments and loans can be collateralized and is likely to be recouped by the lender.
11. This does not mean that international liquidity is valued less than in the first best. The claim is that it is valued less than the second best indicates.
12. See Caballero and Krishnamurthy (2000).
13. The capital flows reversal during the last crisis can be seen more clearly in the non-financial private sector, where they came down to $2.7 billion in 1998, from $8.2 billion the previous year. Official flows, on the other hand, rose supported by loans from the World Bank and Inter-American Development Bank (IDB).
14. Another interpretation is that the finding is spurious, as the more comprehensive series is polluted by too many no-trades. Although this remains a possibility, aggregate volume data for both indexes do not reveal a pronounced relative decline of transactions in the IGPSA. It is also important to realize what the relative-volatility claim in the text is not about: it does not say that large firms' financing is more distressed than that of smaller firms during crises. Indeed, the reality is quite the opposite, as concerned local banks reallocate their loans toward larger companies. It just says that an important segment of the demand for the shares of prime companies fluctuates with international sentiment about emerging markets.
15. See Caballero (1999a) for details on the bonds included in each sample.
16. The Banco Central de la República Argentina (BCRA) can buy Argentine treasury bonds denominated in dollars (which are counted as reserves) as long as this does not lead to a decline in the ratio of international reserves (net of these bonds) to a base below two-thirds. Government notes in the BCRA rose by about 25 percent from 1994 to 1995 and then declined sharply.
17. The interquartile range (between 25 and 75 percent) of a cross section of nominal interest rates on 30-day peso loans averaged less than 2 percent in the months before the tequila crisis, but it then jumped to more than 16 percent in March 1995. Similarly, the same measure jumped from less than 2 percent to more than 6 percent during the Russian financial crisis.
18. See figure 3 in Powell (1999) for clear evidence on the improved systemic liquidity of the Argentine financial system. As described in that figure, starting in January 1996, liquidity requirements increased steadily from 10 percent of deposits to over 15 percent by March 1999. Excess reserves added a fairly constant 10 percent, and the repo program added yet another 10 percent starting in January 1997.
19. The industries correspond to the stock market subsectors at level of disaggregation 5 of the Datastream classification, which includes 116 potential entries. For Argentina, Chile, Australia, and Mexico, 26, 20, 25, and 24 sectors respectively were represented during the period considered. Similar results were obtained when different measures of dispersion were used.
20. Ideally, these comparisons should be made with ex ante rather than ex post returns.
21. Of course, it would have been better had the government borrowed those resources not to solve its own fiscal imbalances, but to support a financially distressed private sector.
22. The share of loans made by large banks increased from 36 percent in late 1996 to 42 percent at the end of 1997 and 48 percent a year later.
23. Despite the conventional wisdom on the matter, this decline is not purely due to the sharp decline in the prices of primary goods.
24. Outliers not in the sense of measurement error, but in that nominal rigidities are not very relevant for sectors in deep distress and with high turnover rates, such as construction.
25. It is certainly inaccurate to blame foreigners exclusively for the outflows. See, for example, Garber (1998) for a discussion of the role of domestic banks' off-balance sheet activities before the 1994-95 crisis. These activities inflated capital inflows before the crisis and automatically reversed them after the crisis.
26. Much has been said about the positive role the U.S. expansion had in insulating Mexico from a large share of the recent global turmoil. By the same token, it seems unreasonable to solely blame Mexico for its poor growth performance—particularly of exports—in the early 1990s since the United States was not growing much either.
27. See Caballero (1999c) for the details of this analysis and further discussion.
28. Of course, this is not to deny that Mexican banks were vulnerable and had already shown an increasing trend of nonperforming loans before the crisis.
29. See Caballero (1999c) for further detail. Firms with price-earnings ratios greater than 11 at the end of 1992 more than doubled their debt over the next three years, while firms with price-earnings ratios less than 11 increased their debt by less than 50 percent.
30. The proportion of loans over 20 million pesos that were collateralized right after the crisis was around 70 percent of the total for most banks. Gelos and Werner (1999) document that banks' use of collateral increased after privatization.
31. This statement refers to a transitional surge in restructuring, not a steady-state relation between restructuring and banks' performance.
32. In the literature, the latter factor seems to have been underplayed relative to the exchange rate overvaluation as an explanation of depressed growth in Mexico. The overvaluation, the argument goes, was primarily due to the exchange rate-based stabilization program. An alternative—or at least complementary—interpretation, especially for those years well after the initial adjustment to the stabilization program, is that the "overvaluation" was an equilibrium consequence of the massive credit inflows largely made possible by the low U.S. interest rates. Of course, when these flows turned around abruptly, the exchange rate became overvalued.
33. Statistics reported by the Secretaría de Hacienda y Crédito Público (Mexican Ministry of Finance), Instituto Nacional de Estadística, Geografía e Informática (Mexican Census and Statistics Bureau), and Banco de México.
34. Although in the case of Chile, one should add the outstanding debt of the Central Bank, which amounts to about 25 percent of the country's GDP.
35. This is an underestimate, since expenditures (when including those below the line) are also much more volatile in Mexico, as a result of the recurrent bailouts.
36. In the late 1990s, copper exports accounted for about 40 percent of Chilean exports, which is equivalent to about 9 percent of its GDP. Chile has a copper stabilization fund (FEC) aimed primarily at stabilizing fiscal revenues: at the beginning of each year the Budget Office sets a reference price; withdrawals or deposits are made quarterly as a step function of the deviation between actual and reference prices. The largest yearly net deposit to the fund occurred in 1995 and amounted to 5 percent of fiscal revenues, when the price of copper exceeded its average of the 1985-99 period by 22 percent. The largest yearly net withdrawal occurred in 1998 and amounted to around 1.5 percent of revenue when the copper price was 36 percent below the average.
37. The price of copper has trends and cycles at different frequencies, some of which are persistent (see Marshall and Silva, 1998). However, the sharp decline in the price of copper during the current crisis was mostly the result of a transitory demand shock brought about by the Asian crisis. When the latter economies began to recover, so did the price of copper. I would argue that conditional on the information that the current shock was a transitory demand shock, the univariate process used to estimate the present-value impact of the decline in the price of copper in figure 3 overestimates the extent of this decline. The lower decline in futures prices is consistent with this view. The variance of the spot price is six times the variance of fifteen-months-ahead futures prices. Moreover, the expectations computed from the AR process track reasonably well the expectations implicit in futures markets but at the very end, when liquidity premiums may have been a consideration.
38. For Australia, coal represents a bit more than 10 percent of exports; together with wheat and wool, this share rises to around 20 percent. Australian terms of trade were severely hurt by the sequence of crises starting in mid-1997.
39. See Caballero (1999b) for more discussion. Norway also has a similar export concentration and also appeared to have output less sensitive to terms-of-trade shocks.
40. Capital flows were high, matching the high copper price, but the current account was not. The other exception reflects a domestically induced recession, as it resulted from the monetary tightening implemented at the beginning of the new government to offset the inflationary pressures of the preceding political cycle. Capital flows remained high but ultimately led to the accumulation of international reserves rather than financing a current account deficit.
41. These features need not be a problem. As many European economies have demonstrated, banks' credit can do most of the job, but this is less likely when banks are often subject to credit crunches.
42. While excessive churn can be wasteful, it is highly unlikely that Chile's depressed levels are enough to support a solid infrastructure of market makers, able to provide optimal levels of immediacy and liquidity. Moreover, one could argue that the waste associated with normal churn is a cost worth paying to reduce the extent of systemic liquidity crises when these arise.
43. See La Porta and others (1997).
44. While in principle only the real interest rate should matter, in practice several factors justify plotting the relation with respect to nominal interest rates as well. For example, a sudden rise in the flow payment associated with a sharp increase in the nominal rate may induce financial distress on a constrained firm.
45. Banks have also played a more indirect but significant role in the initial rise in interest rates, as they seem to have been one of the main forces behind the attack on the Chilean peso in 1998. Although no public information is available on the subject, the presumption is that they did it to bridge currency mismatches rather than for speculative reasons. If the former is indeed the case, the appropriate response by the Central Bank was not to hike interest rates but to "rent" the reserves to the banks (see the discussion on policy). Decentralizing reserve holdings seems to be a reasonable component of an efficient system for international liquidity management.
46. If the situation persists, local banks will probably borrow abroad to lend to the SMEs. In the short run, however, this mechanism is limited, given uncertainty and the banks' conservatism. In fact, the crowding-out mechanism can be phrased in reverse: the sharp increase in the banks' appetite for quality lowers the equilibrium rate and exacerbates the rationing mechanism.
47. Participant (TIAA-CREF's quarterly news and performance magazine), May 1999, pp. 10-11.
48. In May 2000, Chile opted for the latter—permanently removing capital controls—together with a series of measures aimed at improving corporate governance, integrating with international financial markets, and developing domestic financial markets.
49. See Calvo (1991); Caballero and Krishnamurthy (2000).
50. This also suggests that fiscal adjustments during crises ought to be done on the expenditure side rather than on the tax side. If the latter is unavoidable, the adjustments should be targeted away from the supply side of the economy.
51. The argument against temporary contracts is not that they don't create employment, but rather that they further separate insiders from outsiders by reducing the pressure on insiders and creating a class of temporary, unskilled workers.
52. See Caballero and Krishnamurthy (2000).
53. An additional point against a fully flexible exchange rate system is that it may not be possible for an individual country to allow the development of sufficiently deep currency-risk hedging instruments. Investors may use them to hedge the risk on neighbors' currencies, if these do not have their own deep hedging markets. This was a problem for Mexico around the Brazilian turmoil during 1998-99, and it is a particularly serious concern if the country's financial markets are not deep enough.
54. See Caballero (1999b).
55. Yet another mechanism is to sell domestic assets to foreigners. The issues here are when to sell and, closely related, whether the perceived agency problems (such as the risk of expropriation) are low enough to prevent a steep price discount.
56. The need for intervention here is motivated by another manifestation of the under-provision of international liquidity discussed above: international insurance is undervalued by the private sector. Since long-term external debt is simply short-term debt plus a rollover insurance contract, the private sector will borrow at maturities that are shorter, on average, than is socially optimal. From the point of view of the aggregate economy, the private sector will underinsure with respect to terms-of-trade and external financial shocks.
57. It is important that this rule be predetermined. Part of the recent Chilean recession was arguably created by bickering between the Treasury and Central Bank about who should make the first adjustment.
