It seems almost a tautology that global financial integration leads to international synchronization of business cycles. But economic research—both empirical and theoretical—has found the relationship to be far more nuanced. While many empirical studies have indeed found a positive relation between international financial linkage and cycle synchronization among countries, some recent research on developed-nation ties has discovered that cross-border connections are actually associated with less synchronization when the years under study include few financial crises, such as the pre-2007 period.
Theoretical research to date is inconclusive in the sense that integration could lead to either divergence or convergence of cycles. Much depends on the source of the overall, or aggregate, fluctuations, suggests theory. If the negative shock is to a national banking sector and its efficiency, then problems in one country will likely spread to others, as global banks will also likely pull funds from unaffected countries. In other words, if Citibank Europe goes down, it’s likely that operations of Citibank U.S. will be negatively affected.
But if the crisis is a negative shock to a specific nation’s “real” economy (that is, a nonfinancial sector), then that crisis could actually lead to a divergence in international growth, since banks will tend to pull credit from affected nations and send more of it to untroubled economies, where it’s likely to provide higher returns. So, if Volkswagen is a customer of Citibank Europe and Volkswagen gets into trouble, then Citibank will devote more funds to U.S. firms, improving conditions in the United States.
The issue is quite relevant from the policy perspective. A better understanding of the mechanisms at work could clarify the potential impact on the United States of a euro-area meltdown or aid developing countries in understanding whether more financial integration with the rest of the world is desirable.
To bring greater clarity to the “ambiguous, and sometimes conflicting, answers” from the empirical and theoretical literature, Fabrizio Perri of the Minneapolis Fed, with economists Sebnem Kalemli-Ozcan and Elias Papaioannou, has written “Global Banks and Crisis Transmission,” National Bureau of Economic Research Working Paper 18209, July 2012, and forthcoming in the Journal of International Economics1/ . Their paper takes on two tasks: It analyzes relevant data, and it then creates a model to help explain what the data reveal. In so doing, it tells a consistent and compelling story of the relationships between global financial integration, co-movement in business cycles and banking crises. The study is not the final word on these matters, of course, but it will undoubtedly lead future research in fruitful directions.
Key empirical findings
The economists begin by analyzing a unique database: quarterly data on country-pair bank links from 20 developed nations between 1978 and 2009. The three-decade period is one of international financial calm, by and large, punctuated by several financial crises, particularly that of 2007-09. A critical feature of the data set is that it provides information about indirect banking links as well as direct ties, thereby permitting measurement of the importance of financial exposure between countries through banks in offshore accounts in, for example, the Cayman Islands. Their statistical analysis—running regressions of relevant variables—reveals three central findings:
- When financial markets are calm, the association between banking links and business cycles is significantly negative—consistent with the study mentioned earlier.
- In periods of financial crisis, this negative correlation approaches zero. This suggests that “a financial crisis is an event that induces co-movement” among countries that share financial links, thereby muting the usual negative association.
- During the 2007-09 financial crisis (though not in other crises in the period studied), there was a positive association between business cycle synchronization and exposure to the U.S. financial system. But curiously, indirect links through the Cayman Islands were a powerful explanatory factor in this financial contagion. “The positive correlation between output synchronization and financial linkages to the U.S. emerges only when, on top of direct links to the U.S., we also consider indirect links via the Cayman Islands, the main off-shore financial center of the U.S. economy.”
These findings provide a logical bridge between two separate bodies of research on financial integration: one that looks at business cycles and another that focuses on financial contagion. Financial crises spread contagiously from one country to another through bank connections, it appears, and this creates greater business cycle co-movement among countries that are tightly connected financially. During the recent crisis, many observers believed that the U.S. credit shock spread internationally via bank networks, but empirical evidence for the idea was largely absent. That evidence now exists.
In part, the quality of the study’s data set is what allows the economists to provide this elusive confirmation. Its depth and structure enable them to distinguish the effect of financial connections between individual country pairs from the impact of large shocks common to all nations. With its greater historical range, a better measure of financial integration and solid panel data, the researchers can isolate the specific importance of bilateral financial links.
A model with credit shocks
The second part of the paper is devoted to building a model of international business cycles with banking and then running it quantitatively, to see if, with reasonable parameters, it can generate patterns seen in actual data. The idea is to create a mathematical representation of the economic mechanisms that may be at work in an integrated financial world. If this model can faithfully replicate real-world results, then those mechanisms—and the theory behind them—may in fact be a reasonable explanation, during crisis and calm, of the impact of global banks on national economies.
The economists create an international business cycle model in which global banks allocate funds between, on one hand, households and others who save and, on the other hand, firms and other borrowers who invest those funds—the process referred to as “financial intermediation.” In this model, both banking shocks and productivity shocks can cause economic fluctuations. As the economists write, the model serves two purposes: “to precisely spell a causal link between financial integration and business cycle synchronization” and “to show that our empirical findings can be used to identify sources of output fluctuations, and thus to shed light on the causes of the triggering and spreading of the 2007-2009 crisis.”
They calibrate the model with standard real-world parameters for factors like depreciation rates and capital’s share of output, but also for less standard variables like the degree of financial integration between pairs of countries, the costs incurred by banks in intermediating funds and banks’ share of portfolios devoted to risky assets.
Testing: One, two, three …
With the model built, the economists see how it performs. First they show that when run with both banking shocks and productivity shocks, the model generates plausible business cycles and, indeed, helps explain some features that standard models (without credit shocks) have trouble with. Standard models without credit shocks can’t generate realistic values, for example, for changes in employment relative to gross domestic product or international correlations in consumption.
Then they give it the real test: checking its quantitative results against the empirical results from the first part of their paper. The primary test is to run the same regression equation on the model’s artificial data as they ran earlier with the empirical (real-world) data. If roughly the same relationships appear in both, the model is a good fit and the mechanisms it contains hold explanatory power.
In specific, they compare results for synchronization of GDP growth among countries—business cycle co-movement. During tranquil times, the data show a synchronization coefficient ranging from -0.302 to -0.220. The model generates a coefficient of -0.35—the correct sign (negative) and a close numerical match. During crisis periods, the data’s output coefficient ranged from 0.123 to 0.264. The model: 0.25—an excellent fit.
“The comparison between coefficients,” they write, “suggests that the relation between financial integration and output co-movement implied by our model is statistically close to the one we estimate in the data.” Both model and actual data indicate that when financial times are calm, greater bilateral financial integration leads to diverging business cycles, but when crises hit, this negative relationship is muted, as credit shocks transmit through international banking ties and business cycles synchronize more closely.
Lessons and future research
The model suggests that financial integration is a crucial determinant of synchronization of business cycles. When compared with the statistical relationship seen in real-world data, the model’s estimates are quite close. “Although this does not formally prove that financial integration is indeed a causal driver of international business cycle integration,” the economists observe, “it shows that this hypothesis is entirely consistent with the data patterns.”
A second lesson from their model is that credit shocks are crucial in explaining the tendency for nations with close banking ties to contract simultaneously during crises. “This leads us quite naturally to conclude that indeed large credit shocks to financial intermediaries could have been the underlying source of the global contraction in economic activity that took place during the 2007-2009 global crisis.”
The model also suggests an obvious direction for future research, say the economists: “the analysis of the effectiveness and desirability of policies geared toward reducing capital losses of the financial/banking sector, like the 2008 bailout.” The model indicates that capital losses to banks strongly affect domestic and international economic output; in the future, policymakers might therefore consider measures to prevent or buffer such shocks, or to mitigate their transmission to the broader economy.