Editor's Summary

With this issue Economía enters its fifth year. We have published papers on most topics that are relevant to the formulation of economic policy in Latin America. The papers that follow—which were first presented at the panel meeting held in April 2004 and hosted by Harvard University—continue that tradition.

Latin America is clearly the world leader in currency and financial crashes, and these crises entail large costs, sometimes as large as a quarter of GDP. What is much less clear is who bears these costs. Are they equally spread across society, or does one group bear a disproportionate burden? What is the impact on the poor and on the distribution of income? Marina Halac and Sergio Schmukler try to answer these difficult questions in the lead article of this issue. Most research so far shows that crises have distributional effects because the poor and unskilled are the first to lose their jobs, the real value of their meager peso savings plummets with high inflation and devaluation, and social spending is typically cut back in bad times. Halac and Schmukler focus on a different channel: the financial channel. They study financial crises that involve bailouts and ask who pays for these bailouts and how this affects income distribution. Their conclusions are disheartening.

Bailouts have two kinds of distributional effects. Resources are transferred from taxpayers to participants in the banking sector, who get the bailouts. Recipients include bank owners and owners of firms with debts to the banking system—not exactly the poor. Large and small depositors also receive transfers. They are not necessarily rich, but they are not poor either. In Latin America, only households in the upper ranges of the income distribution are likely to have bank accounts. Taxpayers, on the other hand, come from all levels of income. Tax systems in Latin America are hardly progressive, given widespread avoidance of income taxes and heavy reliance on consumption levies that are paid by everyone, including the very poor. In Mexico, for instance, the value added tax (VAT) paid as a percentage of income is roughly constant across all income deciles. The poor also suffer disproportionately when bailouts are financed not by raising taxes, but by cutting spending, since they tend to rely more heavily on transfers and social spending. The conclusion is unavoidable: transfers from nonparticipants to participants in the banking system are transfers "from the relatively poor to the relatively rich."

That is not the end of the story, however. Halac and Schmukler show that large transfers also take place within the financial sector. In the Argentine, Ecuadorian, and Uruguayan crises, large and foreign depositors (or investors with access to foreign-based accounts) obtained compensation or even capital gains, while small depositors suffered capital losses. After the crashes in Chile, Ecuador, and Mexico, large borrowers with close ties to banks also benefited most from the crises and their resolution. The conclusion is again inescapable: "financial redistributions during crises benefit the rich and hurt the poor."

If currency and bank crashes are one perennial concern in Latin America, slow growth is another, at least since the early 1980s. It is now widely accepted that the region's growth performance in the 1990s was disappointing, especially in view of the many structural reforms put in place early that decade. In 1991-2003 Latin America grew not only much less than the countries of East Asia, but also less than the members of the Organization for Economic Cooperation and Development (OECD). So much for convergence, at least among these three groups. Moreover, Latin America also grew less in 1991-2003 than it did in the 1960s and 1970s, the peak years of import substitution.

What accounts for this disappointing performance? What factors explain the difference with East Asia? Are neoclassical growth models up to the task? José de Gregorio and Jong-Wha Lee revisit these thorny questions, employing a variety of econometric strategies. They begin by estimating an extended Solow model with pooled cross-country data. To deal with endogeneity issues they use three-stage least squares, with lagged values of the independent variables and also several other variables as...