Research and Shorter Analysis

CLAE staff, visiting scholars, and their coauthors conducted research on topics related to financial crises and depressions, banking, trade, undocumented immigration, and economic measurement. The work included CLAE and Research Department working papers, articles submitted for publication in books and outside journals, and articles in the Dallas Fed’s Southwest Economy publication.

Financial Crises and Depressions

The causes and consequences of financial crises have been important long-time themes of Center for Latin American Economics research. Crises’ consequences for productivity are one of the center’s lines of recent investigation. In “Financial Crises and Total Factor Productivity,” [PDF] Erwan Quintin and Felipe Meza note that measured total factor productivity falls markedly in emerging nations in financial crises. So far, the mechanics that lower productivity at such times have been a puzzle. A possible explanation is that the utilization of capital falls during financial crises. In this paradigm, capital utilization falls because interest rates (the opportunity cost of capital) move markedly above trend at the same time as TFP drops below trend. In a test of this hypothesis for Mexico’s 1994–95 Tequila Crisis, capital utilization in fact turns out to have fallen noticeably in Mexico in 1995.

In models with homogenous capital, this reduction accounts for almost a third of Mexico’s drop in measured TFP. However, the authors show that models that characterize all capital as the same predict counterfactually high energy consumption for Mexico. Applying a measure with heterogeneous capital predicts energy consumption more accurately, but cuts the quantitative importance of capital utilization in half. The reason, it turns out, is that unproductive capital is left idle, with more resources directed to more productive physical capital. Taking account of these, capital utilization seems to account for 15 percent of Mexico’s Tequila Crisis measured TFP drop. Moreover, capital utilization accounts only for a negligible fraction of measured TFP movements in non-crisis quarters.

In “Argentina’s Lost Decade and Subsequent Recovery: Hits and Misses of the Neoclassical Growth Model,” [PDF] Finn E. Kydland and Carlos E. J. M. Zarazaga examine Argentina’s 1980s protracted crisis, depression, and the recovery that followed, from a growth theory perspective. They use the term depression advisedly. By the end of the “lost decade” of the 1980s Argentina’s detrended GDP per capita had fallen 30 percent below ten years earlier. The authors undertake a growth accounting exercise to identify the sources of decline and later growth. A dramatic drop in total factor productivity during the 1980s was followed by a strong turnaround in the 1990–97 period. Output per capita not only grew in contrast to declines in the 1980s, but grew 2.5 times as fast as the 1951–79 average. The authors test to see if standard neoclassical growth theory can explain the 1980s “lost decade” and the recovery of the 1990s. The results suggest that neoclassical growth theory can track Argentina’s economic depression of the 1980s fairly closely. The findings accordingly reject the hypothesis that economic depressions involve a breakdown in rational economic behavior or in the way economic agents form expectations about the future.

The two big misses of the neoclassical growth model are instructive. First, the experiment predicted that labor input should have declined by about 10 percent during the lost decade, while the data measured labor input as having increased by 3 percent. It has often been claimed that employment in provincial governments and state-owned enterprises in Argentina was a covert form of unemployment insurance. Information from some unofficial sources suggests that employment in the provincial and national government may have represented between 10 and 13 percent of the total number of workers during the period of analysis—but this figure ignores employment in the large number of state-owned enterprises of the period. The authors define as an upper bound the increase in Argentine unemployment between the end of 1990 and 1995—when public sector firms were privatized. If this number corresponded to the sum of workers who lost this “hidden unemployment,” then the total number of public sector workers in the lost decade may have approached 20 to 25 percent of total employment. The addition of these redundant workers during the 1980s—only to be fired during the 1990s privatizations—may explain why employment rose rather than fell during the lost decade.

The model was able to mimic the actual output and capital accumulation swings of the 1980s, but it predicted much more growth for the 1990s than actually occurred. An explanation is government policies that directly or indirectly penalized the accumulation of capital. A related conjecture is based on the possibility of endogenous credit constraints—in which capital constraints are more binding during expansions than during downturns. Investor memories of the sovereign debt default of the 1980s might have discouraged investment during the 1990s. Certainly their reluctance to invest would have been validated in 2001 when Argentina implemented the largest confiscation of bank deposits in its history and then declared a massive default on its sovereign debt obligations.

Over the last decade in Latin America and elsewhere, anticipated or actual government responses to bank asset quality plunges have been increasingly viewed as triggering so-called “sudden stops” in capital flows. The results include large exchange rate depreciations and their consequences. Following a devaluation, the shift in relative prices of nontradables (which fall) and tradables often set off a second round of banking problems if banks lent on the basis of nontradable (such as real-estate) collateral.

These issues have inspired related research at the Center for Latin American Economics. In “Privatization, Competition and Supercompetition in the Mexican Commercial Banking System,” William C. Gruben and Robert P. McComb test the widespread allegations that Mexico’s financial liberalization and bank privatizations of the early 1990s resulted in marked reductions in bank competitiveness. They find that the banks did not become less competitive—in accordance with common belief—but instead commenced behavior consistent with an extreme market share struggle. Starting at about the middle of the privatization period and running at least a year after it ended, the average bank operated at a point where marginal cost exceeded marginal revenue. Such a money-losing arrangement could only take place in the short run.

Gruben and McComb’s findings for the early 1990s are consistent with what followed—as significant declines in bank asset quality came to light. When U.S. interest rates increased by more than 300 basis points in 1994, the commercial banks’ asset problems made Mexico’s central bank reluctant to tighten correspondingly—out of concern that higher interest rates would make even more loans go bad. This inconsistency between U.S. and Mexican monetary policy under a pegged exchange rate regime with the dollar contributed to Mexico’s sudden capital stop in November and December 1994 and a subsequent mega devaluation.

Further pursuing the complications associated with bank liberalizations and privatizations, William C. Gruben, Jahyeong Koo and Robert R. Moore, in their paper “Financial Liberalization, Market Discipline, and Bank Risk,” [PDF] test for connections between such events and bank risk in the presence of so-called “depositor discipline.” If bankers anticipate that their risky lending will be punished by flights of depositors from their banks, it is possible that they will refrain from risky lending. Regardless of country, a bank crisis is not a common event. Special circumstances may be required to trigger the risky lending that can lead to it. One trigger may be the emergence of new markets over which bankers then wage market-share struggles. Bank liberalizations and privatizations can mean new market openings.

The authors present statistical associations between measures of depositor discipline and of bank risk that occurred during periods of liberalization and privatization in six economies. In some of these economies, explicit or implicit government guarantees to make depositors whole and to bail out banks seem to have been anticipated by bankers as removing depositor discipline. In these countries, bankers tended to take larger post-liberalization risks than in countries where depositor discipline existed.

Trade

The importance of trade between the Eleventh Federal Reserve District and Latin America is one of the factors that motivated the original formation of the Center for Latin American Economics. “The Openness-Inflation Puzzle Revisited,” [PDF] by William C. Gruben and Darryl McLeod, addresses an ongoing debate over the connection between international trade and inflation. Research applying data from the 1980s (Romer 1993) shows a strong negative link between trade openness and inflation. Some analysts offer models to suggest that trade openness facilitates currency competition, causing low-inflation good money either to drive out high-inflation bad money or to make the bad money reform. Others claim that trade openness lowers inflation by increasing competition.

Some empirically based work by a Brazilian economist, Cristina Terra (1998), posits that an inverse relationship between openness and inflation is peculiar to the 1980s and, even then, applies only to highly indebted countries relying heavily on the inflation tax. When asset holders set off a balance of payments crisis, the less open the economy, the greater must be the devaluation to produce the trade surplus that is suddenly required. Protected economies don’t trade much or efficiently. What happens to the local currency value of the debt service in a devaluation affects the government’s budget deficit. In the absence of fiscal reform, the increased deficit enforces a rise in the amount of inflation tax the government needs to collect. Because the devaluation is larger in closed economies, those are the ones with the larger increase in the value of the debt in local currency terms and, consequently, greater inflation.

Arguing that currency competition is what explains the negative relation, and not the paradigm that Terra presents, Gruben and McLeod test this relation during the 1990s for a large sample of countries, including most Latin American nations. They find that the inverse relation between trade openness and inflation became stronger in the 1990s—when debt crisis was far less widespread than in the prior decade. They also find that during the 1990s this negative relationship was more pronounced in less indebted countries than in the more indebted countries to which Terra’s paradigm had confined the phenomenon. The results point not only toward a general inverse relation between inflation and openness, but suggest that the continuing process of globalization will enforce lower inflation on most countries—even those that pursued higher inflation in the 1980s. This work is consistent with their previous work on capital account openness and inflation.

In “The Giant in Mexico’s Rearview Mirror,” Erwan Quintin analyzes the widespread concerns over Mexico’s ability to compete internationally with China—especially in the wake of Mexico’s U.S.-linked slowdown since late 2000. Quintin shows that despite complaints about Mexico’s labor competitiveness, Mexican labor costs have not risen any faster than labor productivity and that Mexico has shown a strong commitment to both monetary and fiscal policy discipline. Indeed, he notes, Chinese exports to the United States have not fared much better than Mexico’s in most sectors. Still, he observes, the high cost of electricity and the fiscal uncertainty that plagues the export sector have impeded Mexico’s expansion. Ultimately, Mexico will require structural or competitive reforms in various sectors in order to return to its trajectory of the late 1990s.

William C. Gruben and Sherry L. Kiser, in “Chilean Accord Extends U.S. Free Trade Universe by One,” address another issue—U.S. protectionists’ recalcitrance even over trade agreements with smaller Latin American countries. Even though Chile’s small population of 15 million and its status as the world’s 43rd largest economy would seem to make it a poor candidate for much U.S. protectionist attention, concluding the Chilean–U.S. free trade agreement signed in 2003 not only involved more protracted effort than NAFTA, but it temporarily allows continuing protectionism in several areas. The authors note that, as is common with the United States, the largest elements of remaining protectionism are in agriculture. While many products entered duty-free with the formalization of the accord, Chilean dairy-related exports remain under a quota, and the agreement also permits tariffs on some Chilean fruits to persist for 12 years after the agreement takes effect. Free trade in wine will not take place until 2014. Meanwhile, Chilean protectionist interests in wheat and beets were jubilant about their continued protection against cheaper U.S. products.

Immigration

Determinants and results of immigration from Mexico in particular and from Latin America in general are ongoing research topics at the Center for Latin American Economics. In “The Impact of Illegal Immigration and Enforcement on Border Crime Rates,” [PDF] Roberto Coronado and Pia M. Orrenius investigate the connection between crime rates on the U.S.–Mexico border, illegal immigration, and immigration enforcement since the early 1990s. Over this period, property-related crime declined along the border, but violent crime rates began to rise to the national average. Is there a causal relation, or does some third factor influence the crime statistics? The authors perform econometric tests to find that the volume of illegal immigration is in fact not related to changes in property-related crime—after controlling for other determining factors including economic fluctuations and changes in the level of law enforcement. Similar tests do identify a significant positive correlation between measures of illegal immigration and violent crime, despite increased immigration law enforcement. Over the 1990s, the overall crime rate on the border fell, but the share of violent crimes among overall crimes increased significantly. While immigration law enforcement shows a deterrent effect upon crime in the 1990s, this effect eroded over the decade. Since mid-1999 the marginal impact on crime of an increase in immigration linewatch hours has been zero. The authors conjecture that the correlation between illegal immigration and violent crimes is in fact associated with a third variable, extensive smuggling activity—in particular, violent behavior by drug smugglers whose operations seem to be expanding and whose border-crossing behavior is certainly illegal.

In “Does Immigration Affect Wages? A Look at Occupation-Level Evidence,” [PDF] Pia M. Orrenius and Madeline Zavodny find that an increase in the fraction of workers in a blue-collar occupation group who are foreign born tends to lower the wages of natives in that occupation. However, an increase in the fraction of foreign-born workers in high-skill occupations does not lower the wages of native workers in the same occupations. Additional examinations of the blue collar data indicate that it is immigrants who are adjusting their immigration status within the United States, but not newly arriving immigrants, who negatively affect the wages of low-skilled natives. This finding suggests that immigrants become substitutes for natives only after having spent much time in the United States.

Mark G. Guzman, Joseph H. Haslag and Pia M. Orrenius’ “A Role for Government Policy and Sunspots in Explaining Endogenous Fluctuations in Illegal Immigration” [PDF] mathematically formalizes explanations for why illegal immigration can vary substantially even if there is a constant gap between U.S. and Mexican wages. In this model, volatility in migration flows derives from two sources: a tension between government income transfers that induce migration and law enforcement that discourages it and, separately, the existence of a conditioning uncertainty or sunspot effect. That is, the authors examine the impact of (sunspot) uncertainty in government tax/income transfer policies upon migration as they interact with the effects of smugglers and the border patrol to determine the level of immigration.

Economic Measurement

Measuring economic phenomena is a topic of much concern for analysts at the Center for Latin American Economics because data for Latin American countries are less easily available than for industrial nations. In “Choosing Among Rival Poverty Rates: Some Tests for Latin America,” [PDF] William C. Gruben and Darryl McLeod note that, as of their writing, there is no widely accepted procedure for estimating national poverty rates. They propose an ex post procedure for selecting poverty rates that have desirable properties. They argue that absolute poverty measures, estimated uniformly across countries, ought to be correlated with nonmonetary indicators that reflect the consequences of physical deprivation. These indicators include birth rates, school attendance, measures of malnutrition, and other factors. The authors perform a series of non-nested hypotheses tests to choose among competing poverty and income measures. They screen 66 poverty measures computed for 17 Latin American countries. The tests identify 10 to 15 poverty indicators that meet the standards the authors set forth for usefulness. The authors note that one theme that turns out to be common to the most successful poverty indicators is that scaling to national accounts and to standards of international comparison are important. One commonly used scaling methodology—developed by the U.N. Economic Commission for Latin America—turns out to allow more transparent comparisons than any other.

In “(Mis)reporting Mexico’s Gross Domestic Product,” Franklin Berger addresses problems in seasonally adjusting Mexican (and by extension other) gross domestic product data. Seasonally adjusting any category of data where seasonal factors (such as Easter) are important but do not always occur in the same month or quarter cannot be conveniently performed by the standard approaches applied, say, to Christmas or January. Berger notes that when the U.S. media report economic events in Mexico, they often report GDP and that in some cases reports about Mexican economic fluctuations have been far more negative than would be suggested by the data if properly understood. In many Latin American countries, economic activity declines during the week or so prior to Easter because the entire week, “La Semana Santa,” is celebrated. Based on his own constructions of seasonal adjustments for Mexico and other Latin American countries, Berger identifies increasing statistical sophistication among seasonal adjustment statisticians at Mexico’s central bank (Banco de México) and census bureau (Instituto Nacional de Estadística, Geografía e Informática). Even though their work has resulted in advances in accurate seasonal adjustment, reports in U.S. national business publications have missed these adjustments and refer to year-over-year data that are often misleading as descriptions of current economic events. Berger notes that, in the case of holidays like Easter, even year-over-year data comparisons can be factually wrong without proper seasonal adjustment.

References Romer, David H. (1993), “Openness and Inflation: Theory and Evidence,” Quarterly Journal of Economics 108, 869–903. Terra, Cristina (1998), “Openness and Inflation: A New Assessment,” Quarterly Journal of Economics 113, 641–48.

| |