A Potential Channel of Credit Crunch Transmission
Internationally active banks have typically played positive roles in many emerging market and developing countries around the world, in terms of the development and stability of the local financial system.1 However, in the context of the current global financial crisis, these foreign banks—typically headquartered in advanced economies but with operations in multiple countries—represent a potential channel of propagation of global financial shocks. As their losses continue to mount and asset quality deteriorates sharply, the global scramble for dollar liquidity and the need to deleverage their balance sheets could lead global banks to reduce their exposures to emerging market and developing countries.
This chapter examines the recent and prospective behavior of such foreign banks with regard to the Latin American and Caribbean (LAC) region. The question is relevant for the LAC economic outlook, to the extent that foreign banks have acted as a significant source of funding for companies and households in many Latin American economies. We show that the nature of the involvement of such banks in LAC has differed in fundamental ways from that seen in other regions, notably emerging Europe, and that this difference has relatively favorable implications for the supply of credit in LAC countries going forward.
At the outset, it is essential to highlight that foreign banks extend credit to LAC and other economies through two distinct routes:
directly, from their headquarters abroad (overseas lending by parent banks or “cross-border flows”). In the terminology of a country’s external balance of payments, credit received from a foreign bank through this traditional route represents a capital inflow, and the accumulation of a liability to nonresidents (i.e., form of external debt).
indirectly, via the activities of their local affiliate banks in host countries (foreign-owned subsidiaries or branches). Since the affiliates are residents of their host countries, the credit they extend locally does not, in itself, represent a balance of payments inflow nor an external debt of the host country. For most LAC countries, this indirect route is the dominant form of foreign banks’ involvement, and is often significant in scale. Indeed, these foreign-owned local affiliates now hold a substantial portion of local banking system assets in a number of LAC countries.2
In this chapter we draw extensively on the international banking statistics reported by the Bank for International Settlements (BIS), which combine information on the two types of lending referred to above, to provide a comprehensive view of a global bank’s exposure to a given country. This ownership-based viewpoint is of interest to the extent that global banks choose to manage their holdings centrally, treating local affiliates as part of a global portfolio. For example, a global bank facing capital or liquidity shortages may instruct its affiliates to curtail their local lending to help improve the group’s consolidated capital asset position, or to transfer liquidity to headquarters. On the other hand, even if some decisions are taken centrally, it is possible that local affiliates follow a distinct business model and that their local lending activities may respond differently—or not at all—to shocks affecting their parent bank. As will be seen, we find strong evidence of such differences in behavior in the most recent period of global financial distress.
To shed light on the potential transmission of the global deleveraging process working through foreign banks, this chapter first looks at some key features of the involvement of foreign banks in the region. It then analyzes econometrically the determinants of foreign bank lending to Latin America, with a special focus on the importance of global liquidity conditions and international banks’ financial soundness. In turn, these results are used to illustrate how foreign banks’ lending to the LAC region could evolve in the period ahead, under certain assumptions for global financial conditions and other key determinants.
The chapter shows that lending by global banks to the different regions of the world retrenched significantly in 2008. To most regions, this deceleration or contraction of lending began in 2008Q2, and deepened through the year. On the other hand, lending to the LAC region only began to be strongly affected in the final quarter of 2008. During that quarter, cross-border loans to LAC contracted sharply, while lending by local affiliates—which plays a much larger role in the region—proved much more resilient. Looking forward, the analysis suggests that tight global financial conditions, coupled with a slow recovery in global banks’ financial health, will weigh against foreign banks’ lending to Latin America. The deepest retrenchment is likely to be in cross-border lending; the fact that such lending generally plays a lesser role in the region will limit the potential impact of its contraction.
Share of Banking Assets Held in Subsidiaries or Branches of Large Foreign Banks 1/
(Percent of total banking system assets, end-2008)
Sources: National authorities; Bankscope and IMF staff calculations.1/ Included in the calculations are the six main foreign banks with global presence. In some countries, the actual share of foreign bank ownership could be higher due to the presence of other international and regional banks.Share of Banking Assets Held in Subsidiaries or Branches of Large Foreign Banks 1/
(Percent of total banking system assets, end-2008)
Sources: National authorities; Bankscope and IMF staff calculations.1/ Included in the calculations are the six main foreign banks with global presence. In some countries, the actual share of foreign bank ownership could be higher due to the presence of other international and regional banks.Share of Banking Assets Held in Subsidiaries or Branches of Large Foreign Banks 1/
(Percent of total banking system assets, end-2008)
Sources: National authorities; Bankscope and IMF staff calculations.1/ Included in the calculations are the six main foreign banks with global presence. In some countries, the actual share of foreign bank ownership could be higher due to the presence of other international and regional banks.Indeed, a number of structural characteristics of foreign bank involvement in the region imply that the retrenchment of foreign bank credit will be less severe than in other emerging market regions, and also compared to LAC’s own past. Foreign banks conduct a higher share of their lending in the region through local affiliates and in domestic currency, mitigating the risk of a homeward flow of foreign banks’ assets. In addition, much of the funding of foreign-owned banks in Latin America has come from domestic sources (mostly deposits), rather than from parent banks’ resources or wholesale funding.3 We show that these features of foreign banks’ operations in the region reduce the risk of contagion from the international liquidity squeeze. In addition, the maturity composition of lending to the LAC region has shifted toward the long term, reducing the vulnerability to a sudden withdrawal of short-term external funding. Taken together, these characteristics reduce the likely impact on LAC of financial distress in mature markets, notwithstanding the overall negative outlook.
Key Features of Foreign Banks’ Lending to LAC
The analysis draws on the BIS Consolidated Banking Statistics, which contains country-level information on the gross claims of international banks to the bank and non-bank sectors in Latin American countries through the last quarter of 2008.4 Such claims include not only bank loans, but also other forms of financing through debt securities and equities; for ease of exposition, we will refer to all these forms together as “lending.” As noted, such lending by BIS-reporting banks is extended internationally by parent banks’ headquarters (“cross-border”) and also locally through their branches and subsidiaries in the recipient country (“local affiliates”).5
Foreign banks’ lending to Latin America has increased significantly in the past five years, both in dollar volumes and as a share of GDP. In terms of the latter, foreign bank assets are most important in Chile, followed by Costa Rica and Mexico, although they are significantly below the median ratio for countries in emerging Europe. In terms of absolute size, Brazil, Mexico, and Chile accounted for almost 80 percent of all outstanding lending by foreign banks to LAC by end-2008.
In Latin America, two-thirds of all foreign banks’ lending in 2008 was disbursed through local affiliates. This share is significantly higher than the median in emerging Europe (50 percent), emerging Asia (37 percent) and Africa and the Middle East (19 percent). This reflects a steady shift in international banks’ business strategy toward LAC, from cross-border lending by bank’s headquarters to lending through local affiliates.6
Share of Foreign Banks’ Lending Extended Through Their Local Affiliates, 2008
(Percent of total)
Sources: Bank for International Settlements; and IMF staff calculationsNote: Regional data correspond to the median across countries.Share of Foreign Banks’ Lending Extended Through Their Local Affiliates, 2008
(Percent of total)
Sources: Bank for International Settlements; and IMF staff calculationsNote: Regional data correspond to the median across countries.Share of Foreign Banks’ Lending Extended Through Their Local Affiliates, 2008
(Percent of total)
Sources: Bank for International Settlements; and IMF staff calculationsNote: Regional data correspond to the median across countries.Deposit-to-Loan Ratios in Foreign-Owned Local Affiliates, 2007 1/
(Percent)
Source: Adler and Cerutti (2009).1/ The deposit-to-loan ratio for each local affiliate is calculated as the sum of demand, time, saving, and foreign currency deposits as a share of their loans to the private sector. For each country, the value reported corresponds to the weighted average of foreign affiliates’ deposit-to-loan ratios, using their loan portfolio as weights.Deposit-to-Loan Ratios in Foreign-Owned Local Affiliates, 2007 1/
(Percent)
Source: Adler and Cerutti (2009).1/ The deposit-to-loan ratio for each local affiliate is calculated as the sum of demand, time, saving, and foreign currency deposits as a share of their loans to the private sector. For each country, the value reported corresponds to the weighted average of foreign affiliates’ deposit-to-loan ratios, using their loan portfolio as weights.Deposit-to-Loan Ratios in Foreign-Owned Local Affiliates, 2007 1/
(Percent)
Source: Adler and Cerutti (2009).1/ The deposit-to-loan ratio for each local affiliate is calculated as the sum of demand, time, saving, and foreign currency deposits as a share of their loans to the private sector. For each country, the value reported corresponds to the weighted average of foreign affiliates’ deposit-to-loan ratios, using their loan portfolio as weights.Share of Foreign Banks’ Lending Denominated in Local Currency, 2008
(Percent of total)
Sources: Bank for International Settlements; and IMF staff calculationsNote: Regional data correspond to the median across countries.Share of Foreign Banks’ Lending Denominated in Local Currency, 2008
(Percent of total)
Sources: Bank for International Settlements; and IMF staff calculationsNote: Regional data correspond to the median across countries.Share of Foreign Banks’ Lending Denominated in Local Currency, 2008
(Percent of total)
Sources: Bank for International Settlements; and IMF staff calculationsNote: Regional data correspond to the median across countries.Lending by foreign banks’ affiliates is mostly financed from domestic deposits, and their dependency on nondeposit funding (parent bank resources or wholesale financing) is particularly low in Brazil. This is in contrast to the situation in emerging Europe, where the flow of new credit in recent years has tended to be predominantly financed by cross-border flows from parent banks.
Reflecting the particular mix of local and cross-border lending, and the domestic sources of funding by local affiliates, LAC is the region with the highest share of domestic-currency-denominated lending in total foreign banks’ lending. For most of the region’s larger economies, half or more of all foreign banks financing is denominated in local currency, with this share exceeding 70 percent in the case of Mexico.
The maturity composition of lending to the LAC region has shifted toward the long term, and refinancing risk remains low compared with developing countries as a whole.7 In addition, lending by foreign banks to the banking sector in ALC (as a share of total lending) is the lowest among all regions considered. For Asia and the Pacific, the proportion is almost 30 percent, while it is 15 percent for Latin America and the Caribbean.
Banks from Spain and the United States are the dominant players in Latin America, jointly accounting for approximately 50 percent of all outstanding financing by foreign banks to Latin American countries.8 These foreign banks have low exposure to emerging Europe.9 Conversely, the most active global banks in emerging Europe (from Austria, Belgium, Sweden and Italy), have a small presence in Latin America.10
Sources of Stability
Four key features of international banks’ lending to the region could mute the possible transmission and amplification of global financial shocks. First, Latin America’s lower reliance on cross-border, foreign-currency-denominated debt may make it less exposed to the risk of a homeward flow of foreign banks’ assets than other regions. Recent empirical evidence has shown that countries in which a larger share of foreign banks lending to firms and households is extended through their local affiliates rather than through external loans tend to enjoy more stable foreign bank financing (Garcia-Herrero and Martinez-Peria, 2007). Because cross-border lending (mostly denominated in foreign currency) is typically funded in international markets, it tends to be highly sensitive to movements in global interbank market conditions. But since lending by foreign-owned affiliates are often funded locally, they may be less sensitive to external shocks.11 (We find evidence of this, as will be discussed.)
Second, foreign-owned affiliates’ low reliance on parent credit lines (or potentially unstable commercial obligations) to fund credit growth limits the potential destabilizing effects coming from solvency problems or tight liquidity conditions faced by banks in the major industrialized economies.
Total Foreign Banks’ Lending in Foreign Currency in Emerging Markets
(Percent of GDP, 2008)
Source: IMF staff calculations.Note: the boundaries on this map do not necessarily reflect the IMF’s official position.Total Foreign Banks’ Lending in Foreign Currency in Emerging Markets
(Percent of GDP, 2008)
Source: IMF staff calculations.Note: the boundaries on this map do not necessarily reflect the IMF’s official position.Total Foreign Banks’ Lending in Foreign Currency in Emerging Markets
(Percent of GDP, 2008)
Source: IMF staff calculations.Note: the boundaries on this map do not necessarily reflect the IMF’s official position.Foreign Lending by Bank Nationality
(Percent, 2008Q3)
Sources: Bank of International Settlement, and IMF staff calculationsNote: Percentage computed over the specified lender banks.Foreign Lending by Bank Nationality
(Percent, 2008Q3)
Sources: Bank of International Settlement, and IMF staff calculationsNote: Percentage computed over the specified lender banks.Foreign Lending by Bank Nationality
(Percent, 2008Q3)
Sources: Bank of International Settlement, and IMF staff calculationsNote: Percentage computed over the specified lender banks.Third, a low share of short-term foreign currency lending (especially interbank lending) makes countries of the LAC region less exposed to rollover risk and, thus, to shocks affecting creditor banks.12
Fourth, the large global bank players in emerging Europe have reduced presence in Latin America. In the current crisis, this limits the cross-regional contagion effects that can occur through a “common bank lender effect,” in which losses in a country would cause a global bank to liquidate its assets or cut credit lines to its subsidiaries in order to restore its capital adequacy ratios.13 In the most recent period, bank lending ties have been a major channel of transmission, with western European banks the main source of stress (IMF, World Economic Outlook, April 2009).
The general presence of these sources of resilience, however, does not mean that the LAC region is immune to deceleration or contraction of lending by foreign banks. Moreover, there is heterogeneity within the region, and countries that are more dependent on cross border lending—the component of foreign banks lending that is more sensitive to global funding conditions—may be more affected. Also, the potential knock-on effects on economic activity in each country would likely also depend in part on the depth and structure of financial markets.
What Drives Foreign Bank Lending to the LAC Region?
The turmoil in global credit markets has raised questions about the level and stability of foreign banks’ financing to Latin America. To shed light on how a deepening of the credit crunch could cascade in Latin America, this section looks at the historical determinants of international bank lending to the region from 1999 to 2008, using a multivariate panel regression framework (see the Technical Appendix for details). 14 The model is focused on analyzing the impact of the following three factors on foreign banks’ lending activity to the region: (i) deteriorating global liquidity conditions, (ii) large write downs and weaker balance sheets of major international financial institutions; (iii) the downturn in the lending cycle in advanced economies.
The econometric analysis regresses the quarterly growth in banks’ total lending (i.e., cross-border plus local affiliates) from each foreign country to each country in LAC on the following:15
The TED spread (the spread between the three-month U.S. dollar LIBOR and the three-month U.S. treasury rate) to proxy for liquidity strains in global interbank markets.16
Banks’ lending standards in advanced countries, to control for quantitative bank lending conditions. We use the net percentage of tightening of credit standards in a given quarter based on data from the U.S. Federal Reserve’s Senior Loan Officer Opinion Survey.
Creditor banks’ financial soundness indicators. We use market-based indicators on the expected default frequencies (EDFs) of the banking system specific to each country. These have showed a marked deterioration in recent months, reflecting declining capitalization and earnings, and rising stock price volatilities of publicly listed banks.
Macrofinancial conditions of recipient countries that influence both lending demand and lending supply. We include the short-term interest rate differential, an increase in which would signal a rise in the relative rate of return on investment in LAC. We also control for the real GDP growth rates of the borrower country.
The percentage change in the borrower country exchange rate against the US dollar, to capture valuation effects (and possibly also portfolio reallocations that could occur in response to exchange rate movements).17
A composite indicator of economic, political, and financial risks in each borrower country, to account for the fact that lower perceived economic and institutional risks tends to attract foreign bank investment. We use the International Country Risk Rating, published by the PRS Group.
Creditor-borrower pair fixed effects, to capture unobserved and mostly constant sources of heterogeneity.18
Three-Month Money Market Spreads
(Basis points)
Source: Haver Analytics.Three-Month Money Market Spreads
(Basis points)
Source: Haver Analytics.Three-Month Money Market Spreads
(Basis points)
Source: Haver Analytics.Reported Tightening in U.S. Banks’ Lending Standards 1/
(Net percentage of senior loan officers tightening lending standards)
Source: Board of Governors of the Federal Reserve System, Senior Loan Officer Opinion Survey.1/ The net percentage of tightening is the percentage of senior loan officers who reported tightening minus the percentage of officers who reported easing in credit standards. Higher numbers imply tightening credit standards.Reported Tightening in U.S. Banks’ Lending Standards 1/
(Net percentage of senior loan officers tightening lending standards)
Source: Board of Governors of the Federal Reserve System, Senior Loan Officer Opinion Survey.1/ The net percentage of tightening is the percentage of senior loan officers who reported tightening minus the percentage of officers who reported easing in credit standards. Higher numbers imply tightening credit standards.Reported Tightening in U.S. Banks’ Lending Standards 1/
(Net percentage of senior loan officers tightening lending standards)
Source: Board of Governors of the Federal Reserve System, Senior Loan Officer Opinion Survey.1/ The net percentage of tightening is the percentage of senior loan officers who reported tightening minus the percentage of officers who reported easing in credit standards. Higher numbers imply tightening credit standards.Market Indicator of Banks’ Expected Default Frequency
(Median over banks in each country)
Sources: Moody’s KMV; and IMF staff calculations.Market Indicator of Banks’ Expected Default Frequency
(Median over banks in each country)
Sources: Moody’s KMV; and IMF staff calculations.Market Indicator of Banks’ Expected Default Frequency
(Median over banks in each country)
Sources: Moody’s KMV; and IMF staff calculations.Market Indicator of Banks’ Expected Default Frequency
(Median over banks in each country)
Sources: Moody’s KMV; and IMF staff calculations.Market Indicator of Banks’ Expected Default Frequency
(Median over banks in each country)
Sources: Moody’s KMV; and IMF staff calculations.The estimated model provides a plausible explanation of the factors affecting lending by foreign banks to countries in Latin America. The main findings (based on results presented in the first two columns of the table in the Technical Appendix) are as follows:
There is a strong link between global money market conditions and changes in international banks’ lending to Latin America. Consistent with World Bank (2008) and McGuire and Tarashev (2008), a deterioration in interbank liquidity adversely affects foreign banks’ lending growth to the region. The coefficient estimates suggest that a 10-basis-point increase in the TED spread would lead, on average, to a roughly 1 percentage point reduction in the quarterly growth rate of total lending by global banks.19
A deterioration in banks’ financial health has consistently led to slower growth in international banks’ lending to LAC. A rise in one standard deviation (20 basis points) in banks’ EDF is associated with a 1.4 percentage point average decrease in the quarterly growth rate of foreign lending. This result is consistent with recent work by Čihák and Koeva Brooks (2009) that shows that bank loan supply in the euro area moves in line with banks’ financial soundness.
Changes in banks’ lending standards in advanced economies (proxied by those in the United States) do not seem to have a statistically significant effect on the growth of foreign banks’ credit to Latin America, once we control for other factors.
Exchange rate depreciations are associated with a significant slowdown in foreign banks’ lending during the same quarter. These estimates could be capturing mechanical valuation effects, as well as the impact of currency crises that occurred during the sample period.20
Other explanatory variables have the expected sign and are consistent with theory.21 Higher economic growth is robustly associated with an increase in lending growth to the region during the same quarter. Institutional improvements (as proxied by a more favorable economic and political risk rating) lead to stronger investor confidence and thus attract more foreign bank lending.
Finally, we also estimate a different specification that allows shocks to global liquidity conditions and to parent banks’ financial health to have differential effects across LAC countries, varying according to the share of claims on that country that are denominated in local currency. This share serves as a proxy for the share of local affiliates in foreign banks’ total claims on a country.22 The results, reported in the last two columns of the table in the Technical Appendix, suggest that the transmission of global financial shocks through the foreign bank lending channel is more muted in countries where foreign banks conduct a higher share of their lending in domestic currency. This finding is consistent with results reported by Garcia Herrero and Martinez Peria (2007), who show that countries in which a larger share of foreign bank lending is extended through local affiliates (as opposed to cross-border financing) experience lower volatility in total foreign banks’ lending.
Overall, our results suggest foreign bank lending to Latin America does respond to funding constraints caused by liquidity shortages and heightened counterparty exposure in the global interbank market. Also, increases in foreign banks’ own financial vulnerability can prompt reductions in their financing to the region.23 However, the size of the responses to the above shocks depends on the structure of lending to a given LAC country. As expected, the estimates suggest that the larger effects are on cross-border lending (which is largely denominated in foreign currency), while effects on lending from affiliates (especially in those countries with low dollarization of credit) are smaller.
Foreign Banks’ Lending to LAC: Recent Developments
(Percent change in U.S. dollar values)
Adjusts for exchange rate effects on foreign bank lending denominated in domestic currency.
Foreign Banks’ Lending to LAC: Recent Developments
(Percent change in U.S. dollar values)
Quarterly Change | Annual Change | ||||
---|---|---|---|---|---|
Region/Country | Actual | Valuation Adjusted 2/ | Actual | Valuation Adjusted 2/ | |
Latin America and the Caribbean | -14.0 | -4.8 | -5.1 | 7.0 | |
Brazil | -20.6 | -10.7 | -10.9 | 5.4 | |
Mexico | -16.3 | -1.6 | -11.3 | 3.9 | |
Other South America | -8.1 | -4.4 | 1.8 | 9.0 | |
o/w | Argentina | -10.6 | -6.4 | -5.6 | -1.3 |
Chile | -10.4 | -3.5 | 3.2 | 18.2 | |
Colombia | -7.0 | -10.7 | -11.6 | -15.7 | |
Peru | -4.0 | -3.0 | 25.3 | 27.3 | |
Central America | 0.5 | 0.5 | 13.5 | 13.8 | |
o/w | Costa Rica | -0.8 | -2.3 | -12.1 | 7.4 |
Guatemala | 3.8 | 4.6 | 18.0 | 18.5 | |
Caribbean | -3.7 | -3.6 | 24.2 | 24.3 | |
o/w | Dominican Republic | -5.6 | -5.5 | -3.6 | -3.3 |
Grenada | 2.9 | 2.9 | 13.2 | 13.2 | |
Memo item | |||||
Emerging Europe | -7.5 | -2.2 | 1.0 | 6.1 | |
Africa & Middle East | -8.4 | -5.9 | -0.7 | 5.6 | |
Emerging Asia | -10.5 | -10.0 | -7.1 | -4.9 |
Adjusts for exchange rate effects on foreign bank lending denominated in domestic currency.
Foreign Banks’ Lending to LAC: Recent Developments
(Percent change in U.S. dollar values)
Quarterly Change | Annual Change | ||||
---|---|---|---|---|---|
Region/Country | Actual | Valuation Adjusted 2/ | Actual | Valuation Adjusted 2/ | |
Latin America and the Caribbean | -14.0 | -4.8 | -5.1 | 7.0 | |
Brazil | -20.6 | -10.7 | -10.9 | 5.4 | |
Mexico | -16.3 | -1.6 | -11.3 | 3.9 | |
Other South America | -8.1 | -4.4 | 1.8 | 9.0 | |
o/w | Argentina | -10.6 | -6.4 | -5.6 | -1.3 |
Chile | -10.4 | -3.5 | 3.2 | 18.2 | |
Colombia | -7.0 | -10.7 | -11.6 | -15.7 | |
Peru | -4.0 | -3.0 | 25.3 | 27.3 | |
Central America | 0.5 | 0.5 | 13.5 | 13.8 | |
o/w | Costa Rica | -0.8 | -2.3 | -12.1 | 7.4 |
Guatemala | 3.8 | 4.6 | 18.0 | 18.5 | |
Caribbean | -3.7 | -3.6 | 24.2 | 24.3 | |
o/w | Dominican Republic | -5.6 | -5.5 | -3.6 | -3.3 |
Grenada | 2.9 | 2.9 | 13.2 | 13.2 | |
Memo item | |||||
Emerging Europe | -7.5 | -2.2 | 1.0 | 6.1 | |
Africa & Middle East | -8.4 | -5.9 | -0.7 | 5.6 | |
Emerging Asia | -10.5 | -10.0 | -7.1 | -4.9 |
Adjusts for exchange rate effects on foreign bank lending denominated in domestic currency.
Is a Retrenchment Under Way? The Latest Evidence
BIS banking statistics for end-2008 were released in late April of this year, allowing us to examine how international banks responded in the months following the Lehman Brothers event of September 2008. Because the last quarter of 2008 was characterized by unusually large depreciations of several LAC currencies, the discussion below focuses on movements in lending by foreign banks that exclude currency valuation effects.24
For many countries around the world, total lending by global banks (both cross-border and though local subsidiaries) had begun to slow or even contract in 2008Q2, and this trend continued through the latest data available for end-2008. In Latin America—where previous growth rates of total lending had not been as rapid as elsewhere—significant retrenchment did not occur until 2008Q4, when total lending of foreign banks to the region as a whole contracted by about 5 percent within the quarter. Contractions varied in size, but did occur in most countries of the region (see table). Still, for the year as a whole, lending from foreign banks was up 7 percent.
Growth in Foreign Banks’ Lending to LAC, by Country or Region 1/
(Annual percent change, exchange rate-adjusted)
Sources: Bank for International Settlements; IMF, International Financial Statistics; and IMF staff calculations.1/ Includes cross-border lending and lending by foreign-owned local affiliates in each country.Growth in Foreign Banks’ Lending to LAC, by Country or Region 1/
(Annual percent change, exchange rate-adjusted)
Sources: Bank for International Settlements; IMF, International Financial Statistics; and IMF staff calculations.1/ Includes cross-border lending and lending by foreign-owned local affiliates in each country.Growth in Foreign Banks’ Lending to LAC, by Country or Region 1/
(Annual percent change, exchange rate-adjusted)
Sources: Bank for International Settlements; IMF, International Financial Statistics; and IMF staff calculations.1/ Includes cross-border lending and lending by foreign-owned local affiliates in each country.Growth in Foreign Banks’ Lending, by Region 1/
(Annual percent change, exchange rate-adjusted)
Sources: Bank for International Settlements; IMF, International Financial Statistics; and IMF staff calculations.1/ Includes cross-border lending and lending by foreign-owned local affiliates in each country.Growth in Foreign Banks’ Lending, by Region 1/
(Annual percent change, exchange rate-adjusted)
Sources: Bank for International Settlements; IMF, International Financial Statistics; and IMF staff calculations.1/ Includes cross-border lending and lending by foreign-owned local affiliates in each country.Growth in Foreign Banks’ Lending, by Region 1/
(Annual percent change, exchange rate-adjusted)
Sources: Bank for International Settlements; IMF, International Financial Statistics; and IMF staff calculations.1/ Includes cross-border lending and lending by foreign-owned local affiliates in each country.Channels of Foreign Banks’ Lending to LAC: Differences in Their Behavior
(Annual percent change in U.S. dollar value)
Sources: Bank for International Settlements; and IMF staff calculations.Channels of Foreign Banks’ Lending to LAC: Differences in Their Behavior
(Annual percent change in U.S. dollar value)
Sources: Bank for International Settlements; and IMF staff calculations.Channels of Foreign Banks’ Lending to LAC: Differences in Their Behavior
(Annual percent change in U.S. dollar value)
Sources: Bank for International Settlements; and IMF staff calculations.Within total lending to the region, the behavior of local affiliates and their parent banks diverged dramatically in the last quarter of 2008. The stock of cross-border claims on LAC, which had been growing at annual rates of well over 30 percent in previous quarters, abruptly contracted within the fourth quarter, by almost 20 percent (taking growth for the year down almost to zero).25 On the other hand, lending by local affiliates of foreign banks continued a gradual deceleration, but managed to grow modestly in the fourth quarter with respect to the previous quarter.
What Will Happen in 2009?
Using the model estimates presented in the preceding section, and making assumptions about international financial conditions and growth prospects in LAC, we construct forecasts of foreign banks’ lending to Latin America in the coming quarters.26 These forecasts should be viewed only as illustrative projections, particularly in light of current uncertainties.27
According to the projection, the credit crunch in mature markets and other domestic factors will likely cause a steep deceleration in foreign banks’ lending to LAC, with the four-quarter growth rate dropping from about 20 percent in 2008 Q3, to touch bottom at about –1 percent in 2009 Q3. Still, the predicted downturn in lending is smaller and less protracted than seen at the beginning of this decade; this could reflect in part the structural changes in foreign banks’ lending activities to LAC mentioned earlier, as well as the more robust macroeconomic fundamentals of the countries in the region today.
Conclusions
In this chapter, we look at the potential transmission of the global financial shock and deleveraging process to LAC countries, through foreign banks that are actively engaged in the region.
The analysis suggests that foreign banks are likely to lower their total exposure in Latin America, and indeed a deceleration of the growth of such exposure already has been observed, with a contraction occurring in 2008Q4. Going forward, our analysis shows that two key factors shaping the current global financial turmoil—tight interbank liquidity and mounting pressure on major banks’ capital positions—are likely to have negative effects on the availability of foreign bank lending to the region. The deleveraging process could have wider, more adverse implications in countries where cross-border lending has supported more heavily the banking sector.
Still, the impact on foreign bank lending to Latin America will be cushioned by the fact that (1) most lending is funded from a stable domestic deposit base and denominated in domestic currency; (2) cross-border interbank exposure is limited and the maturity structure has moved toward the long term; and (3) the global banks with the largest presences in LAC markets have low exposures to emerging Europe, so that Latin America is not likely to be susceptible to a credit pull-back similar to, or feeding off, strains in emerging Europe. Finally, to the extent that contraction of lending by foreign-owned banks does occur, there is the potential in many LAC countries for strong policy responses—or the behavior of other banks operating in the same country—to act to offset the decrease in international banks’ lending.
Technical Appendix
Empirical model
The baseline empirical analysis is based on a reduced-form model specification given by:
where the dependent variable, FBL, is the quarterly growth rate of Foreign Banks’ Lending (cross-border and by local affiliates) to LAC country i by foreign banking system j in quarter t. The sample spans the period from 1999Q4 through 2008Q4. We consider all 20 BIS-reporting countries, most of which are advanced economies. The 13 LAC borrower countries included in the sample accounted for more than 90 percent of the outstanding foreign banks’ lending in the region at the end of 2008. For the rest of the LAC countries not included in the sample, some of the explanatory variables were not available. Panama was excluded not only due to data limitations, but also because it is a regional banking center and is classified as an offshore center by BIS. It constitutes an outlier in the region, with foreign banks’ claims accounting for 190 percent of GDP.
The dependent variable covers direct cross-border lending by parent banks, and lending by foreign affiliates in each country, in domestic and foreign currency. It includes international banks’ loans to banks in the host country that are not their subsidiaries/branches (such as loans, bank-to-bank credit lines and trade-related credit) and loans to the nonfinancial sector. It also covers portfolio flows (such as holdings of securities) and equity shares in unrelated institutions (in particular, mergers and acquisitions, which are especially important for the region during the sample period). The data used in the analysis reflects an adjustment in 2008Q3 for a one-off operation of ABN AMRO’s sale of its business in Brazil and Uruguay.28
Many growth rate observations of the dependent variable show extreme values, primarily due to countries entering or exiting the reporting population of banks, which can lead to sudden jumps in the outstanding stock of lending vis-à-vis particular countries. To address this, we included a dummy variable for those observations were the dependent variable lied fell in the lower and upper 5 percent of the distribution each year.29 As in McGuire and Tarashev (2008), the inclusion of this dummy significantly increases the regression fit since much of the overall variance in the dependent variable is contained in these observations. The results of the model are robust to the exclusion of this dummy variable, except that the explanatory power of the model is significantly reduced.
As for explanatory variables:
TED is the spread between three-month U.S. dollar LIBOR and the three-month U.S. treasury bill rate (TED spread), and is meant to capture liquidity strains in global interbank markets. As an alternative proxy for global funding conditions, we also use the spread between three-month LIBOR and three-month overnight index swap rates (the LIBOR/OIS spread). This captures uncertainty regarding counterparty credit quality among banks.
EDF represents estimates of banks’ expected default probabilities using data for almost 600 publicly-listed financial institutions in 20 countries. These are calculated by Moody’s KMV using a contingent claims approach that uses equity market information combined with balance sheet data to estimate forward-looking default probabilities. We take the median of bank-level figures to generate time-varying financial soundness measures for each banking system.
LendSt captures the perceived lending standards in the US banking system in a given quarter, based on data from the U.S. Federal Reserve’s Senior Loan Officer Opinion Survey. It is defined as the percentage of senior loan officers who reported tightening minus the percentage of officers who reported easing in credit standards to large and medium-sized firms.
measures the percentage change in quarterly GDP in the LAC borrower country, while CRating is a composite indicator of economic performance and institutional quality for each LAC country, based on the International Country Risk Rating published by PRS Group. A higher value suggests better macroeconomic frameworks and institutions. The variable (rit – rjt) is the short-term nominal interest rate differential between the recipient and lender country. Deprec is the percentage change in the borrower country exchange rate against the U.S. dollar.
The panel data estimation controls for time-specific (λt) and borrower/lender pair-specific (γij) unobserved heterogeneity. The time fixed effects account for regional changes that affect all countries equally. The high degree of commonality observed in the time series behavior of foreign bank lending across countries described in the text suggest that regional factors may be partially driving the behavior of lending. We also control for fixed effects specific to each lender-borrower pair, to account for time invariant and unobserved factors driving cross-country differences in foreign bank lending.
Determinants of Foreign Banks’ Lending to Latin America
1999Q4-2008Q4; Panel OLS with Fixed Effects
Determinants of Foreign Banks’ Lending to Latin America
1999Q4-2008Q4; Panel OLS with Fixed Effects
Dependent Variable: Quarterly Percent Change in Total Foreign Banks’ Lending | |||||
Average effects | Differential effects | ||||
(1) | (1) | (3) | (4) | ||
TED spread | -0.10*** | -0.15*** | |||
(0.03) | (0.03) | ||||
TED spread x share of local-currency lending | 0.14*** | ||||
(0.04) | |||||
-0.13*** | |||||
OIS spread | (0.04) | -0.18*** | |||
(0.04) | |||||
OIS spread x share of local-currency lending | 0.16*** | ||||
(0.05) | |||||
Banks’ EDF | -6.57* | -6.80* | -11.60** | -11.22** | |
(3.52) | (3.58) | (4.71) | (4.71) | ||
Banks’ EDF x share of local-currency lending | 42.13*** | 37.66** | |||
(10.59) | (12.46) | ||||
Tighter lending standards | 0.03 | 0.08 | 0.02 | 0.08 | |
(0.06) | (0.06) | (0.06) | (0.06) | ||
GDP growth of borrower | 0.32** | 0.40** | 0.20* | 0.33* | |
(0.11) | (0.17) | (0.11) | (0.15) | ||
Interest rate differential (Borrower minus Lender) | -0.03 | -0.03 | -0.03 | -0.02 | |
(0.06) | (0.07) | (0.06) | (0.07) | ||
Composite indicator of credit rating | 0.26* | 0.24 | 0.21* | 0.10 | |
(0.14) | (0.18) | (0.10) | (0.11) | ||
Depreciation | -0.12* | -0.13* | -0.07 | -0.07 | |
(0.06) | (0.06) | (0.05) | (0.05) | ||
Share of local currency lending | 16.56 | 18.34 | |||
(12.87) | (10.65) | ||||
Fixed Effects | |||||
Time (quarter) dummies | Yes | Yes | Yes | Yes | |
Borrower/lender dummies | Yes | Yes | Yes | Yes | |
Diagnostics | |||||
Number of observations | 4080 | 4080 | 3572 | 2947 | |
R2 | 0.19 | 0.19 | 0.20 | 0.19 |
Determinants of Foreign Banks’ Lending to Latin America
1999Q4-2008Q4; Panel OLS with Fixed Effects
Dependent Variable: Quarterly Percent Change in Total Foreign Banks’ Lending | |||||
Average effects | Differential effects | ||||
(1) | (1) | (3) | (4) | ||
TED spread | -0.10*** | -0.15*** | |||
(0.03) | (0.03) | ||||
TED spread x share of local-currency lending | 0.14*** | ||||
(0.04) | |||||
-0.13*** | |||||
OIS spread | (0.04) | -0.18*** | |||
(0.04) | |||||
OIS spread x share of local-currency lending | 0.16*** | ||||
(0.05) | |||||
Banks’ EDF | -6.57* | -6.80* | -11.60** | -11.22** | |
(3.52) | (3.58) | (4.71) | (4.71) | ||
Banks’ EDF x share of local-currency lending | 42.13*** | 37.66** | |||
(10.59) | (12.46) | ||||
Tighter lending standards | 0.03 | 0.08 | 0.02 | 0.08 | |
(0.06) | (0.06) | (0.06) | (0.06) | ||
GDP growth of borrower | 0.32** | 0.40** | 0.20* | 0.33* | |
(0.11) | (0.17) | (0.11) | (0.15) | ||
Interest rate differential (Borrower minus Lender) | -0.03 | -0.03 | -0.03 | -0.02 | |
(0.06) | (0.07) | (0.06) | (0.07) | ||
Composite indicator of credit rating | 0.26* | 0.24 | 0.21* | 0.10 | |
(0.14) | (0.18) | (0.10) | (0.11) | ||
Depreciation | -0.12* | -0.13* | -0.07 | -0.07 | |
(0.06) | (0.06) | (0.05) | (0.05) | ||
Share of local currency lending | 16.56 | 18.34 | |||
(12.87) | (10.65) | ||||
Fixed Effects | |||||
Time (quarter) dummies | Yes | Yes | Yes | Yes | |
Borrower/lender dummies | Yes | Yes | Yes | Yes | |
Diagnostics | |||||
Number of observations | 4080 | 4080 | 3572 | 2947 | |
R2 | 0.19 | 0.19 | 0.20 | 0.19 |
Note: This chapter was prepared by Herman Kamil and Kulwant Rai. The authors are grateful to Sebastian Goerlich for his help in interpreting the data from the Bank for International Settlements.
The presence of foreign banks has typically been considered a positive development in emerging market countries, strengthening the financial systems of their host countries (through more efficient allocation of capital, increased competition and more sophisticated financial services, among others). See Claessens, Demirgüç-Kunt, and Huizinga (2001) and World Bank (2008) for an extensive discussion.
However, in a few countries in the region (such as Ecuador, Guatemala, Dominican Republic, and Venezuela), the banking system is largely domestically owned.
This comprises more than 20, primarily OECD, countries (referred to as “BIS-reporting countries”). There are important distinctions between “foreign” and “BIS-reporting” banks, especially in Central America where regionally operating, non-BIS-reporting banks are not uncommon. For ease of exposition, in this chapter we refer to BIS-reporting banks as international, global or foreign banks.
The BIS consolidated banking statistics track banking systems’ consolidated worldwide foreign claims netting out intra-group positions with their affiliates worldwide. This is an important advantage of this data set since it allows a measure of net exposure to a borrower country. See McGuire and Wooldridge (2005) for a detailed discussion on the structure of the BIS consolidated banking statistics.
This shift in foreign banks’ lending strategy was based, in part, in the acquisition by foreign-owned local affiliates of large local banks, with an already significant local deposit base. This has been particularly important, for example, in Peru, with the entry of Scotiabank (acquiring the third-largest bank in the system) and the reentry of Banco Santander (acquiring a medium-sized local bank) in the late 1990s. For a detailed account of foreign banks’ mergers and acquisitions in the region, see Pozzolo (2008).
Information about residual maturity is available only for cross-border and local affiliates’ lending denominated in foreign currency. For the region as a whole, the share of total foreign banks’ claims (denominated in foreign currency) with short maturity dropped from 54 percent in 1997 to 42 percent in 2008.
Canadian banks account for the largest share of foreign bank assets in the Caribbean. Foreign bank claims on Central America, on the other hand, are relatively diversified between U.S., U.K., and other western European banks.
For a detailed analysis of international banks’ exposure to emerging Europe, see Maechler and Ong (2009).
The exceptions are banks from Germany and France.
As discussed in Garcia-Herrero and Martinez-Peria (2007), because local lending activities require paying higher fixed and irreversible costs, it seems reasonable to expect these flows to be more stable and less responsive to negative shocks than cross-border lending.
The short-term maturity of foreign currency lending by Spanish banks to emerging markets is 40 percent, while that of U.S. banks is close 80 percent (McGuire and Tarashev, 2008). This suggests that U.S. banks can in principle adjust large portions of their exposures more easily than Spanish banks can.
The BIS consolidated banking statistics were reported semiannually until 1999, and quarterly thereafter.
The 13 borrower countries from ALC included in the sample are Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, Mexico, Peru, Venezuela, and Uruguay. McGuire and Tarashev (2008) conduct a similar exercise for 19 emerging markets.
In an alternative specification, we use the difference between the three-month U.S. dollar LIBOR and the three-month overnight index swap rates (the LIBOR/OIS spread), which captures banks’ perceptions of the creditworthiness of other financial institutions. When perceived counterparty risk increases, as it has during the current financial market turmoil, banks may become more reluctant to lend to each other, thus possibly reducing available funds to acquire assets abroad.
BIS data are end-period values, expressed in U.S. dollars; changes in these values incorporate valuation changes (exchange rate changes, marking to market of securities, and write-downs of nonperforming loans) and so may differ from net lending flows. Currency valuation effects can at times be significant, especially in countries where local-currency-denominated lending represents a significant portion of the total.
Foreign bank presence tends to be higher in countries with common language, similar legal systems and banking regulations, and geographical proximity (Claessens and van Horen, 2008). For example, in Latin America and the Caribbean, 60 percent of foreign banks are headquartered in the United States and Spain, whereas in Europe and Central Asia more than 90 percent of foreign banks are headquartered in the European Union.
In the alternative specification, an increase in the LIBOR/OIS spread by 10 basis points is expected to lead to a decrease in the growth of lending to LAC by 1.3 percent. This negative effect of an increase of the LIBOR/OIS spread is consistent with results reported in World Bank (2008) for a larger group of emerging market countries, although the absolute value of the estimated impact for LAC countries is significantly smaller than those reported in the World Bank study.
The literature suggests that borrowing conditions are likely to tighten for a country that experiences a currency collapse given the balance sheet effects due to currency mismatches.
The exception is the coefficient on the nominal interest rate differential between the borrower and lender countries. However, its estimated value was insignificant.
These two shares are positively correlated across countries, since cross-border lending is rarely extended in local currency, and lending by local affiliates is often (though not always) denominated mainly in local currency. Data availability limitations preclude directly estimating separate models for cross-border lending and lending through local affiliates. Such a data breakdown is only available since 2005, and on an aggregate country-level basis.
Using data through 2000, Goldberg (2001) shows that movements in U.S. bank lending to Latin American countries are closely tied to economic conditions in the parent country.
Adjusting for exchange rate valuation effects can be important not only in light of the sharp depreciations in the last semester of 2008 (especially in Brazil and Mexico, which together account for about two-thirds of foreign bank claims on the LAC region), but also because these depreciations were preceded by sustained periods of domestic currency appreciation.
In addition to the factors discussed earlier, it is possible that recent bank support or rescue programs in advanced economies may be accelerating the curtailment of cross-border bank flows. In particular, banks receiving public support may feel pressure to expand domestic lending at the expense of their foreign operations.
To condition on the future values of the explanatory variables in the model, we use IMF forecasts on the TED spread. Countries’ GDP growth forecasts are based on the IMF’s April 2009 World Economic Outlook. We assumed that lender country banks’ EDFs and borrower country risk ratings remain constant at their April 2009 levels. We also assume that nominal exchange rates will hold constant, at their April 15 levels, for the rest of the year.
It also should be noted the model was estimated over a period that was mostly tranquil, but it is being used to forecast lending growth in the wake of a shock of unprecedented magnitude. That said, the model did reasonably well in predicting the
Since in both cases the sales were to a local subsidiary of a Spanish bank (also reporting to the BIS), the amount in question was reported in total claims of Spanish banks in Brazil and Uruguay, in December 2008.
Foreign banks’ credit to smaller countries constitutes a very small share of total foreign credit, and thus tends to exhibit large variations over a small base, reducing the statistical power of our tests.