Abstract
Selected Issues
An Analysis of Credit Dynamics in Paraguay1
A. Introduction
1. In the last decade, credit in Paraguay has grown rapidly. The credit to GDP ratio increased from 9 percent of GDP in 2006 to 32 percent in 2017, one of the sharpest increases in the credit to GDP ratio in the region.
Credit to GDP Ratio
(In units, 1990–2018)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Sources: BCP and IMF staff calculations.Credit to GDP Ratio
(In units, 1990–2018)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Sources: BCP and IMF staff calculations.Credit to GDP Ratio
(In units, 1990–2018)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Sources: BCP and IMF staff calculations.
Loans to GDP
(In percent change, 2006–2017)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations (based on staffs 2018 GDP estimates).Loans to GDP
(In percent change, 2006–2017)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations (based on staffs 2018 GDP estimates).Loans to GDP
(In percent change, 2006–2017)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations (based on staffs 2018 GDP estimates).2. Rapid credit growth may reflect financial deepening, but it also carries risks. Cross-country evidence shows that a bad outcome after a credit boom has been observed in 121 out of 175 cases (Dell’Ariccia et al., 2012).
3. This paper analyzes what is behind the credit boom. Why has credit grown so sharply? What was the role played by external factors, and what was the role of domestic conditions and policies? To answer these questions, we first examine the dynamics of credit from an aggregate perspective, covering 2006Q1 to 2018Q3. We then take a micro-level perspective and examine bank behavior over the same period.
B. Credit Dynamic: An Aggregate Level Analysis
4. Paraguay has experienced a series of external shocks – agricultural exports almost tripled in dollar terms between 2003 and 2014; the global economic and financial crisis of 2008, and the subsequent easing of global monetary conditions; and a boom-bust in Paraguay’s main trading partners (Brazil and Argentina).
5. There have also been important changes in domestic policies. Macro-policies were strengthened in the aftermath of the banking crisis of the late 1990s. Major measures include adopting an inflation targeting regime, introducing risk-based bank supervision and the adaption fiscal responsibility law.
6. This section tries to quantify the importance of external and domestic shocks for both guaraní-denominated credit and dollar-denominated credit (at constant guarani-U.S. dollar exchange rates), using a simultaneous equation model, between 2006Q1 and 2018Q3. External factors included are the soy bean price in U.S. dollars (proxy for agricultural export prices), the Federal Funds rate, the weighted real GDP growth of major trading partners, and the EMBI spread for Latin America (a proxy for the regional risk-aversion). The monetary policy rate, real GDP growth and the real effective exchange rate have been included as the domestic factors.
7. A structural vector autoregressive model (VAR) with the restriction that domestic variables do not affect external variables was estimated. The impact of shocks to each of these variables on the growth of guaraní-denominated credit and dollar-denominated credit is identified using a Cholesky decomposition of the estimated variance-covariance matrix of the residual (i.e., structural restrictions on contemporaneous correlations among variables). The analysis is further divided into two parts – the first part looks at the determinants of long-run credit dynamics through the variance decomposition of the forecast error; the second part looks at the short-term behavior through the impulse response functions. The estimation results of the model are presented in the appendix follows.
Long-Run Determinants
8. In the long-run, external factors are more important determinants than domestic factors, as shown in the variance decomposition to the forecasting error of the VAR. The four external factors together explain 67 percent of the variation in the guaraní-denominated credit and 64 percent of the variation in the Dollar-denominated credit.
Forecasting Error Variance Decomposition to Guarani Credit
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations.Forecasting Error Variance Decomposition to Guarani Credit
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations.Forecasting Error Variance Decomposition to Guarani Credit
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations.
Forecasting Error Variance Decomposition to US dollar Credit
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Forecasting Error Variance Decomposition to US dollar Credit
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Forecasting Error Variance Decomposition to US dollar Credit
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
9. The quantitative importance of the various factors differs between guaraní- denominated and dollar-denominated credit. The growth of major trading partners accounts for 54 percent of the total variation in the guaraní-denominated credit. Other external factors play a more-or-less even role. This may suggest that the guaraní-denominated credit is used for longer- term purposes, such as investment for expansion of production, that is more dependent on the development of trading partners and less dependent on the other three external factors, which are more cyclical in nature. On the other hand, in the case of the dollar-denominated credit, the role of each external factor in determining the total variation is more balanced. This suggests that the dollar-denominated credit is used for the shorter-term purpose, such as working capital.
Short-term Dynamics: Response to External Shocks
10. The impulse response analysis shows that, a positive shock to the growth of trading partners has a large, immediate and persistent impact on both guaraní-denominated credit growth and dollar-denominated credit growth. The impact of the shock on the guaraní-denominated credit is immediate and persistent – the growth of guaraní-denominated credit raises on impact by 1 percentage point and increase gradually over the next few quarters. The impact of the shock peaks at 6 percent in the fourth quarter after the shock. The impact of the shock on dollar-denominated credit is also immediate but less persistent – the growth of the dollar-denominated credit raises on impact by 2 percentage point. The impact of the shock peak at 6 percent in the third quarter after the shock.
11. A positive shock to the Latin America EMBI spread reduces growth of dollar-denominated credit growth but does not have a statistically significant impact on guaraní -denominated credit growth. The impact of the shock on the dollar-denominated credit is immediate and persistent – the growth of the dollar-denominated credit drops on impact by 1 percentage point and continue to decline gradually over the next few quarters. The impact of the shock peak at -4 percent in the second quarter after the shock. On the other hand, the statistical model cannot reject the null hypothesis that heightened regional risk aversion has no impact on the guaraní-denominated credit growth in the short-term as the impulse response of guaraní-denominated credit growth is not statistically significant at the 10 percent confidence level.
12. A positive shock to the soy bean price increases the dollar-denominated credit growth but has no statistically significant impact on the guaraní-denominated credit growth in the short-term (at the 10 percent confidence level). The shock increases the dollar-denominated credit growth by 2 percentage point in the first quarter after the shock and continue to increase to 4 percentage point in the fifth quarter after the shock.
13. A positive shock to the Fed Funds Rate has no statistically significant impact on the guaraní-denominated credit growth and the dollar-denominated credit growth.
Paraguay: Impulse Response of Guaraní-Denominated Credit and Dollar-Denominated Credit to Shocks to External Factors
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff estimates.Paraguay: Impulse Response of Guaraní-Denominated Credit and Dollar-Denominated Credit to Shocks to External Factors
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff estimates.Paraguay: Impulse Response of Guaraní-Denominated Credit and Dollar-Denominated Credit to Shocks to External Factors
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff estimates.Short-term Dynamics: Response to Domestic Shocks
14. A positive shock to the domestic monetary policy rate (i.e, tightening the monetary policy) reduces the guaraní-denominated credit growth but increases the dollar-denominated credit growth. The guaraní-denominated credit growth fell 1 percentage point in the first quarter after the shock. The drop in guaraní-denominated credit growth widens to 3 percentage point in the third quarter after the shock and started to dissipate thereafter. Meanwhile, the dollar-denominated credit growth increases by 2 percentage point on impact of the shock. However, the impact dissipates very quickly.
15. A shock to the real effective exchange rate raises growth of dollar-denominated credit by 1.5 percentage point on impact and then stay around this level for a few quarters after the shock. An explanation might be that an appreciation raises expectations for further appreciation, which makes borrowing in dollars more attractive. The impact of the shock on dollar-denominated credit growth starts to dissipate in the seventh quarter after the shock. The response of the guaraní-denominated credit growth to the shock is not statistically significant.
16. A positive shock to GDP growth has no statistically significant impact on the guaraní-denominated credit growth and the dollar-denominated credit growth.
What drives the rapid credit growth in the last decade?
17. The empirical result suggests that the rapid growth of credit in Paraguay in the last decade was largely due to a favorable external environment–– strong growth of major trading partners high prices of agricultural products improved markedly (which helps support growth of dollar-denominated credit). The impact of these two favorable factors was partly offset by the negative impact of the increase in Latin America EMBI spread.
C. What Drives Credit Growth: A Bank Level Analysis?
18. So far, we have discussed the rapid credit over the past decade. This suggested that there has was one, uninterrupted credit boom. There have, however, been two credit cycles in the last decade.
19. These credit cycles become partly evident if we try to identify the timing of credit booms. Following the methodology discussed in Box 1, we can identify two credit booms. The first credit boom peaked in 2010Q3 – 2011Q3 with credit growth of 60 percent, y/y, and the second credit boom is identified in 2014Q4 – 2015Q4 with credit growing at 67 percent, y/y.2
Paraguay: Impulse Response of Guaraní-Denominated Credit and Dollar-Denominated Credit to Shocks to Domestic Factors
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff estimates.Paraguay: Impulse Response of Guaraní-Denominated Credit and Dollar-Denominated Credit to Shocks to Domestic Factors
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff estimates.Paraguay: Impulse Response of Guaraní-Denominated Credit and Dollar-Denominated Credit to Shocks to Domestic Factors
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff estimates.20. They are even more visible when we look at NPL ratios. Distressed loans increased between late 2010 and 2012, and between 2014 and 2016. Both credit booms saw rising NPLs, increasing provisioning costs, falling profits, and increasing distress loans at the end of the cycle—all major features of a credit boom. While credit grew at a relatively healthy rate after the first credit boom, the second credit boom was followed by a sharper credit decline.
How to Define a Credit Boom?
There is no commonly agreed statistical method on identifying (or filtering) the trend. Given data limitations, we employ two different approaches to the aforementioned two credit datasets to identify credit booms.
The first approach follows Dell’Ariccia et al (2012) to date the credit cycle in Paraguay over the period between 2008Q1 and 2018Q3. We identified boom episodes by comparing the credit-to-GDP ratio (seasonally adjusted) in each quarter to a backward-looking, rolling, cubic trend estimated over the period between quarters t-10 and t. We classified an episode as a boom if either of the following three conditions is satisfied: (i) the deviation from trend is greater than 1.5 times its standard deviation or (ii) the year-over-year growth rate of the credit-to-GDP ratio exceeds 30 percent or (iii) the annualized quarter-over-quarter growth rate of the credit-to-GDP ratio exceeds 30 percent. As noted in Dell’Ariccia et al (2012), the second condition to capture episodes in which aggregate credit accelerates very gradually but credit growth reaches levels that are well above those previously observed in the country. Using these criteria, we identified two major episode of credit boom, 2010Q2–2011Q1, and 2014Q2–2015Q3.
21. Banks behave differently over the different phases of the credit boom-bust cycle. At the beginning of credit booms banks have plenty of capital. Banks are more eager to take on more risk to improve the return on capital. The growth of credit accelerates and exacerbates the credit boom. A rapid expansion of credit may cause deterioration of the quality of the loan portfolio. Distress loan starts to surface in the middle or near the end of the credit boom phrase. Near the end of the boom, as NPLs start to rise, banks’ provisions increase, which reduces profits. Banks become more risk averse. As banks shift from loans to risk-free assets—government bonds, credit volumes decline, which exacerbates the credit bust, and risk weighted capital to asset ratios increase. This phenomenon was particularly strong in mid-2016 as agriculture prices fell. Banks tightened credit volumes, and credit growth collapsed from 30 percent (y/y), attained in mid-2015, to close to zero percent in late 2016.
22. During a credit boom, banks take on riskier assets and rely more on non-traditional funding sources. This trend was more visible during the second credit boom, which featured a larger share of loans in their assets that were financed through non-core sources, including from non-residents.
Credit Booms: Composition of Assets and Liabilities
(Share of total assets or liabilities)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Sources: BCP and IMF staff calculations.Credit Booms: Composition of Assets and Liabilities
(Share of total assets or liabilities)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Sources: BCP and IMF staff calculations.Credit Booms: Composition of Assets and Liabilities
(Share of total assets or liabilities)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Sources: BCP and IMF staff calculations.
The Credit Boom-Bust Dynamics
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and staff estimates.The Credit Boom-Bust Dynamics
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and staff estimates.The Credit Boom-Bust Dynamics
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and staff estimates.
Composition of Banks’ Total Assets
(In percent of total assets)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP, Bloomberg, and staff calculations.Composition of Banks’ Total Assets
(In percent of total assets)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP, Bloomberg, and staff calculations.Composition of Banks’ Total Assets
(In percent of total assets)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP, Bloomberg, and staff calculations.
Composition of Banks’ Funding Sources
(In percent of liabilities)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Composition of Banks’ Funding Sources
(In percent of liabilities)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Composition of Banks’ Funding Sources
(In percent of liabilities)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
23. These on-offs in risk aversion are more pronounced in domestic banks.3 Domestic banks increased credit to the agriculture sector by close to 165 percent over 2011- 2015, during which time the soy price increased by 150 percent.4 Following the soy price collapse, domestic banks repaired their balance sheet and reduced their loan portfolio away from the soy sector more than non-domestic banks. Domestic banks also relied less on core funding sources to finance credit and become increasingly reliant on non-core funding sources such as securities.
24. Domestic banks also respond strongly to rises in regional EMBI spreads. As discussed above, higher regional EBMI spreads are negatively associated with lower dollar-denominated credit growth. Bank-by-bank historical data show further that domestic banks are the major drivers as they tend to respond more strongly to a shock in regional EMBI spread, than foreign banks and others, possibly reflecting spillover fears.
25. Among foreign banks, Brazilian and Argentinean banks behave differently. Argentinian banks tended to shrink their balance sheets in Paraguay when financial stress rose. Brazilian owned banks increased their assets and loans in Paraguay when EMBI spreads increased in Brazil. This suggests that Brazilian owned banks were relocating funds away from their domestic market. Brazilian (Argentinian) banks makes up around 19 (11) percent of total assets, and domestic banks close to 50 percent as of end-2017.
Credit Growth by Banks: Credit Boom vs. Non-boom
(In percent growth, y/y average)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP, Bloomberg, and staff calculations.Credit Growth by Banks: Credit Boom vs. Non-boom
(In percent growth, y/y average)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP, Bloomberg, and staff calculations.Credit Growth by Banks: Credit Boom vs. Non-boom
(In percent growth, y/y average)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP, Bloomberg, and staff calculations.
Contribution to Distressed Loans by Banks
(In percent of total loan)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Contribution to Distressed Loans by Banks
(In percent of total loan)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Contribution to Distressed Loans by Banks
(In percent of total loan)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Brazil: Risks and Assets of Brazilian Banks
(In percent change, y/y)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP, Bloomberg and staff calculations.Brazil: Risks and Assets of Brazilian Banks
(In percent change, y/y)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP, Bloomberg and staff calculations.Brazil: Risks and Assets of Brazilian Banks
(In percent change, y/y)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP, Bloomberg and staff calculations.
Argentina: EMBI Spreads and Argentina Banks
(In percent change, y/y)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Argentina: EMBI Spreads and Argentina Banks
(In percent change, y/y)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Argentina: EMBI Spreads and Argentina Banks
(In percent change, y/y)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
D. How Does Monetary and Macroprudential Policy Respond During Credit Booms?
26. The monetary policy rate has not been actively used to target the credit cycle, as credit booms have not been associated with price pressures and inflation pressures have remained low.5 For instance, during the second credit boom, monetary policy remained accommodative considering low inflationary pressures and uncertain international economic context.
27. The macroprudential toolkit in Paraguay is limited. Macroprudential policies can be used to moderate the financial cycle and building financial sector resilience. The authorities did not fully utilize its limited macroprudential toolkit to introduce measures to curb the past two credit booms and/or build financial sector resilience for the aftermath. At the height of the credit boom in 2010, as response to falling CARs, the BCP introduced two prudential policies to curb excessive risk-taking by banks: (i) introduce leverage limits to contain banks’ lending and (ii) regulations tightening FX sales. However, instead of tightening regulations regarding NPLs, the BCP introduced transitory measures that effectively relaxed the definition of NPLs for credits to the agriculture and livestock sectors (See Resolution No. 33, and Resolution No.34, December 2015).
Balance Sheet of BCP
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations.Balance Sheet of BCP
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations.Balance Sheet of BCP
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations.E. Conclusion
28. We discussed driving factors of credit growth in the last decade in Paraguay, at the aggregate level and at the bank level. At the aggregate level analysis, we assessed the impact of external spillovers and domestic factors on the credit development through a structural vector autoregressive model. At the bank level analysis, we identified two credit boom periods in the last decade and found that the behavior of banks contributes to the credit cycles.
29. At the aggregate level analysis, we found that long-run dynamics of credit growth in Paraguay are driven by both external factors and domestic factors, but the external factors play a more important role. Among all external factors considered, we found that the growth of major trading partners of Paraguay plays the most significant role in determining the growth of both guaraní-denominated credit (by more than half of the total variation) and dollar-denominated credit (by close to one-third of the total variation). We also found that investors’ risk-aversion, as measured by the EMBI spreads of Latin America, and soy bean prices, explain a large amount of movement in dollar-denominated credit growth (each factor explains about 18 percent of the total variation). Among all domestic factors considered, the monetary policy rate is the most important driver in explaining the dynamics of credit growth in Paraguay in the long-run.
Paraguay: Available Macroprudential Policies through end-2016
Paraguay: Available Macroprudential Policies through end-2016
| Name of MP | Description | Paraguay |
|---|---|---|
| CCB | A requirement for banks to maintain a countercyclical capital buffer. | ✓ |
| LVR | A limit on leverage of banks share of the bank’s non-risk-weighted capital. | ✓ |
| LLP | Loan loss provision requirements, which include dynamic provisioning and sectoral provisions (e.g. housing loans). | … |
| Capital | Capital requirements for banks, which include risk weights, systemic risk buffers, and minimum capital requirements. | ✓ |
| LTV | Limits to the loan-to-value ratios. | … |
| DSTI | Limits to the debt-service-to-income ratio and the loan-to-income ratio, which restrict the size of debt services or debt relative to income. | … |
| LFC | Limits on foreign currency (FC) lending & rules or recommendations on FC loans. | ✓ |
| RR | Reserve requirements (on domestic or foreign currency) for macroprudential purposes. | ✓ |
| Tax | Taxes and levies applied to specified transactions, assets, or liabilities. | … |
| SIFI | Identification of and additional buffer requirements for global and domestic systemically important financial institutions. | … |
| Liquidity | Regulations for liquidity and funding risks. | … |
| LTD | Limits to the loan-to-deposit (LTD) ratio and penalties for high LTD ratios. | … |
| LFX | Limits on net or gross open foreign exchange (FX) positions, limits on FX exposures and FX funding, and currency mismatch regulations. | ✓ |
| LCG | Limits on growth or the volume of aggregate credit. | … |
| Conservative | Requirements for banks to maintain a capital conservation buffer. | … |
| LoanR | Loan restrictions, that are more tailored than those captured in “LCG”. They include loan limits and prohibitions, which may be conditioned on loan characteristics, bank characteristics, and other factors. | … |
| Other | Macroprudential regulation not included elsewhere, including stress testing, restrictions on profit distribution, and structural measures (e.g., limits on exposures between financial institutions). | … |
Paraguay: Available Macroprudential Policies through end-2016
| Name of MP | Description | Paraguay |
|---|---|---|
| CCB | A requirement for banks to maintain a countercyclical capital buffer. | ✓ |
| LVR | A limit on leverage of banks share of the bank’s non-risk-weighted capital. | ✓ |
| LLP | Loan loss provision requirements, which include dynamic provisioning and sectoral provisions (e.g. housing loans). | … |
| Capital | Capital requirements for banks, which include risk weights, systemic risk buffers, and minimum capital requirements. | ✓ |
| LTV | Limits to the loan-to-value ratios. | … |
| DSTI | Limits to the debt-service-to-income ratio and the loan-to-income ratio, which restrict the size of debt services or debt relative to income. | … |
| LFC | Limits on foreign currency (FC) lending & rules or recommendations on FC loans. | ✓ |
| RR | Reserve requirements (on domestic or foreign currency) for macroprudential purposes. | ✓ |
| Tax | Taxes and levies applied to specified transactions, assets, or liabilities. | … |
| SIFI | Identification of and additional buffer requirements for global and domestic systemically important financial institutions. | … |
| Liquidity | Regulations for liquidity and funding risks. | … |
| LTD | Limits to the loan-to-deposit (LTD) ratio and penalties for high LTD ratios. | … |
| LFX | Limits on net or gross open foreign exchange (FX) positions, limits on FX exposures and FX funding, and currency mismatch regulations. | ✓ |
| LCG | Limits on growth or the volume of aggregate credit. | … |
| Conservative | Requirements for banks to maintain a capital conservation buffer. | … |
| LoanR | Loan restrictions, that are more tailored than those captured in “LCG”. They include loan limits and prohibitions, which may be conditioned on loan characteristics, bank characteristics, and other factors. | … |
| Other | Macroprudential regulation not included elsewhere, including stress testing, restrictions on profit distribution, and structural measures (e.g., limits on exposures between financial institutions). | … |
30. In addition, through the impulse response analysis, we showed that the external factors and the domestic factors have different impact on the short-term dynamic of the credit growth in Paraguay. We found that both guaraní-denominated credit growth and dollar-denominated credit growth move in tandem to the growth of the major trading partners. Secondly, regional risk-aversion could lower the growth of dollar-denominated credit growth but not the guaraní-denominated credit growth. Thirdly, improvement in soy bean price could spur an increase in dollar-denominated credit growth but not guaraní-denominated credit growth. On the domestic side, a tightening of the monetary policy reduces the guaraní-denominated credit growth but increases the dollar-denominated credit growth in the short-term while a real appreciation of guaraní could raises growth of dollar-denominated credit.
31. The empirical result suggests that the rapid growth of credit in Paraguay in the last decade was largely due to a favorable external environment in the last decade –– strong growth of major trading partners and high prices of agricultural commodity prices.
32. At the bank-level analysis, we find evident that banks’ risk on-off behavior contributes to the credit cycles observed in the last decade. We find that banks tend to behave more aggressively (by taking on more riskier assets) during the two credit boom periods (i.e., the period when credit grows faster than the trend level) identified in the last decade. This behavior contributes to the credit boom-bust pattern observed. We also find that the response to external shocks is heterogeneous among banks and is depending on the original of the capital.
References
Caruana, J., 2011, “Why central bank balance sheets matter”, keynote address at the Bank of Thailand-BIS Research Conference entitled “Central bank balance sheets in Asia and the Pacific: the policy challenges ahead”, Chiang Mai, Thailand, 12–13 December, www.bis.org/speeches/sp111216.pdf
Chen, Q., A. Filardo, D. He, and F. Zhu, 2012, “International spillovers of central bank balance sheet policies,” BIS Papers, No. 66p (Basel:.
Cook, D and J. Yetman (2012): “Expanding central bank balance sheets in Emerging Asia: a compendium of risks and some evidence”, BIS Papers, No 66d.
Dell’Ariccia, G, Igan D., Laeven, L., and Tong, H., with Bakker, B. and Vandenbussche, J., (2012), “Policies for Macrofinancial Stability: How to Deal with Credit Booms,” IMF Staff Discussion Note 12/06.
Franks, J., Mercer-Blackman, V., Sab, R. and Benelli, R. (2003), “Paraguay – Corruption, Reform and the Financial System.”
Gadanecz, B, A Mehrotra and M S Mohanty (2014): “Foreign exchange intervention and the banking system balance sheet in emerging market economies”, BIS Working Paper no 445.
Garcia-Herrero, Alicia (1997): “Banking Crises in Latin Amercia in the 1990s: Lessions from Argentina, Paraguay, and Venezuela,”, IMF Working Paper 11/3/97
International Monetary Fund, 1989-2017, Paraguay: Staff Reports for the Article IV Consultation
Annex I. Estimation Result of the Structural VAR
1. This appendix presents the estimation result of the Structural VAR discussed in the main text. The empirical model is specified as follow:
where yt is a vector of dimension (9 X 1) and consists of the Fed Fund Rate (dyfedpolicy), Trading Partner Growth (worldgdpA), Soy Bean Price (dypsoy), Latin America EMBI Spread (dyembi_lac), Domestic Monetary Policy Rate (dypolicy), Real GDP growth (dyngdpr), Real Effective Exchange Rate (dyreer), Guaraní-denominated credit (dycredit_lc) and Dollar-denominated credit (dycreditfcx). The model is estimated with quarterly data from 2006Q1 to 2018Q3. All variables are denominated in term of year-over-year growth rate except for the Fed Fund Rate, Latin America EMBI Spread and Domestic Monetary Policy Rate. These three variables are denominated in year-over-year difference in percentage point εt is vector of dimension (9 X 1). It consists of the shock to each variable in vector yt. The shocks are assumed to be independently normally distributed. A is a square matrix of dimension (9 X 9) that governs the relationship among variables in yt. B is a square matrix of dimension (9 X 9) that governs the impact of shocks on each variable in yt. n is a vector of constants. Restrictions are imposed on both matrix A and matrix B to reflect the prior that the domestic variables do not affect the development of the external variables.
The estimation results of matrix A and matric B are respectively:
Estimation Results of Matrix A
Estimation Results of Matrix A
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | |
|---|---|---|---|---|---|---|---|---|---|
| VARIABLES | dyfedpolicy (t) | worldgdpA (t) | dypsoy (t) | dyembijac (t) | dypoiicy (t) | dyngdpr (t) | dyreer (t) | dycreditjc (t) | dycreditfcx (t) |
| dyfedpolicy (t-l) | 0.93*** (0.05) | -0.03 (0.12) | 3.28 (2.91) | -26.50 (16.47) | 1.91*** (0.51) | -1.68 (1.03) | 3.10** (1.58) | -1.25 (1.58) | 2.22 (1.78) |
| worldgdpA (t-l) | _0.14*** (0.02) | 0.75*** (0.06) | 1.00 (1.50) | 33.59*** (8.36) | -0.86*** (0.29) | 0.94 (0.58) | -0.86 (0.89) | 2.38*** (0.89) | 0.52 (1.00) |
| dypsoy (t-1) | const. (n.a.) | const. (n.a.) | 0.80*** (0.11) | -0.71 (0.55) | 0.03** (0.01) | -0.04* (0.02) | 0.11*** (0.04) | -0.01 (0.03) | 0.02 (0.04) |
| dyembi_lac (t-1) | -0.00** (0.00) | -0.01*** (0.00) | 0.01 (0.02) | 0.70*** (0.12) | -0.00 (0.00) | -0.01*** (0.00) | -0.01 (0.01) | 0.00 (0.01) | -0.03*** (0.01) |
| dypolicy (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | 0.74*** (0.10) | 0.09 (0.21) | 0.70** (0.32) | _0.95*** (0.32) | -0.86** (0.36) |
| dyngdpr (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | 0.21*** (0.07) | 0.21 (0.13) | 0.42** (0.21) | -0.17 (0.21) | 0.17 (0.24) |
| dyreer (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | -0.09** (0.04) | -0.01 (0.08) | 0.46*** (0.12) | -0.19 (0.12) | 0.11 (0.14) |
| dycredit_lc (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | 0.11*** (0.03) | -0.02 (0.06) | 0.28*** (0.09) | 0.73*** (0.09) | 0.21** (0.10) |
| dycreditfcx (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | 0.04 (0.03) | -0.02 (0.05) | -0.04 (0.08) | 0.08 (0.08) | 0.84*** (0.09) |
| Constant | 0.40*** (0.09) | 0.76*** (0.22) | -1.59 (5.10) | -95.47*** (28.98) | -1.55*** (0.53) | 1.79 (1.12) | -3.28* (1.72) | -2.00 (1.71) | -3.46* (1.91) |
| Observations | 50 | 50 | 50 | 50 | 50 | 50 | 50 | 50 | 50 |
| R-squared | 0.93 | 0.88 | 0.66 | 0.62 | 0.80 | 0.37 | 0.76 | 0.93 | 0.94 |
Estimation Results of Matrix A
| (1) | (2) | (3) | (4) | (5) | (6) | (7) | (8) | (9) | |
|---|---|---|---|---|---|---|---|---|---|
| VARIABLES | dyfedpolicy (t) | worldgdpA (t) | dypsoy (t) | dyembijac (t) | dypoiicy (t) | dyngdpr (t) | dyreer (t) | dycreditjc (t) | dycreditfcx (t) |
| dyfedpolicy (t-l) | 0.93*** (0.05) | -0.03 (0.12) | 3.28 (2.91) | -26.50 (16.47) | 1.91*** (0.51) | -1.68 (1.03) | 3.10** (1.58) | -1.25 (1.58) | 2.22 (1.78) |
| worldgdpA (t-l) | _0.14*** (0.02) | 0.75*** (0.06) | 1.00 (1.50) | 33.59*** (8.36) | -0.86*** (0.29) | 0.94 (0.58) | -0.86 (0.89) | 2.38*** (0.89) | 0.52 (1.00) |
| dypsoy (t-1) | const. (n.a.) | const. (n.a.) | 0.80*** (0.11) | -0.71 (0.55) | 0.03** (0.01) | -0.04* (0.02) | 0.11*** (0.04) | -0.01 (0.03) | 0.02 (0.04) |
| dyembi_lac (t-1) | -0.00** (0.00) | -0.01*** (0.00) | 0.01 (0.02) | 0.70*** (0.12) | -0.00 (0.00) | -0.01*** (0.00) | -0.01 (0.01) | 0.00 (0.01) | -0.03*** (0.01) |
| dypolicy (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | 0.74*** (0.10) | 0.09 (0.21) | 0.70** (0.32) | _0.95*** (0.32) | -0.86** (0.36) |
| dyngdpr (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | 0.21*** (0.07) | 0.21 (0.13) | 0.42** (0.21) | -0.17 (0.21) | 0.17 (0.24) |
| dyreer (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | -0.09** (0.04) | -0.01 (0.08) | 0.46*** (0.12) | -0.19 (0.12) | 0.11 (0.14) |
| dycredit_lc (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | 0.11*** (0.03) | -0.02 (0.06) | 0.28*** (0.09) | 0.73*** (0.09) | 0.21** (0.10) |
| dycreditfcx (t-1) | const. (n.a.) | const. (n.a.) | const. (n.a.) | const. (n.a.) | 0.04 (0.03) | -0.02 (0.05) | -0.04 (0.08) | 0.08 (0.08) | 0.84*** (0.09) |
| Constant | 0.40*** (0.09) | 0.76*** (0.22) | -1.59 (5.10) | -95.47*** (28.98) | -1.55*** (0.53) | 1.79 (1.12) | -3.28* (1.72) | -2.00 (1.71) | -3.46* (1.91) |
| Observations | 50 | 50 | 50 | 50 | 50 | 50 | 50 | 50 | 50 |
| R-squared | 0.93 | 0.88 | 0.66 | 0.62 | 0.80 | 0.37 | 0.76 | 0.93 | 0.94 |
Estimation Results of Matrix B
Estimation Results of Matrix B
| dyfedpolicy (t) | worldgdpA (t) | dypsoy (t) | dyembijac (t) | dypolicy (t) | dyngdpr (t) | dyreer (t) | dycreditjc (t) | dycreditfcx (t) | |
|---|---|---|---|---|---|---|---|---|---|
| dyfedpolicy (t) | 0.34 | const | const | const | const | const | const | const | const |
| worldgdpA (t) | 0.14 | 0.80 | const | const | const | const | const | const | const |
| dypsoy (t) | 1.34 | 8.44 | 14.88 | const | const | const | const | const | const |
| dyembi_lac (t) | -28.19 | -56.86 | -27.89 | 71.19 | const | const | const | const | const |
| dypolicy (t) | 0.16 | -0.15 | 0.26 | -0.17 | 1.36 | const | const | const | const |
| dyngdpr (t) | -1.38 | 0.70 | 0.14 | 0.44 | 0.91 | 2.53 | const | const | const |
| dyreer (t) | 0.05 | 0.47 | 2.01 | 0.63 | 0.73 | -0.40 | 4.05 | const | const |
| dycredit_lc (t) | -1.36 | 1.39 | -0.40 | 0.95 | 0.48 | -0.77 | -0.43 | 4.03 | const |
| dycreditfcx (t) | 0.65 | 1.90 | 0.49 | -0.79 | 2.04 | 0.90 | 1.41 | 0.43 | 3.87 |
Estimation Results of Matrix B
| dyfedpolicy (t) | worldgdpA (t) | dypsoy (t) | dyembijac (t) | dypolicy (t) | dyngdpr (t) | dyreer (t) | dycreditjc (t) | dycreditfcx (t) | |
|---|---|---|---|---|---|---|---|---|---|
| dyfedpolicy (t) | 0.34 | const | const | const | const | const | const | const | const |
| worldgdpA (t) | 0.14 | 0.80 | const | const | const | const | const | const | const |
| dypsoy (t) | 1.34 | 8.44 | 14.88 | const | const | const | const | const | const |
| dyembi_lac (t) | -28.19 | -56.86 | -27.89 | 71.19 | const | const | const | const | const |
| dypolicy (t) | 0.16 | -0.15 | 0.26 | -0.17 | 1.36 | const | const | const | const |
| dyngdpr (t) | -1.38 | 0.70 | 0.14 | 0.44 | 0.91 | 2.53 | const | const | const |
| dyreer (t) | 0.05 | 0.47 | 2.01 | 0.63 | 0.73 | -0.40 | 4.05 | const | const |
| dycredit_lc (t) | -1.36 | 1.39 | -0.40 | 0.95 | 0.48 | -0.77 | -0.43 | 4.03 | const |
| dycreditfcx (t) | 0.65 | 1.90 | 0.49 | -0.79 | 2.04 | 0.90 | 1.41 | 0.43 | 3.87 |
Annex II. The Banking Sector Between 1990 and 2006
The annex reviews the development of the banking sector between 1990 and 2006, when Paraguay experienced a protracted banking crisis.
Early Liberalization and the Banking Crisis in 1990s
1. In the early 1990s, the first democratically elected governments after the Stroessner era introduced reforms aimed at liberalizing financial markets. Reforms included the unification of the exchange rates and floating of the guarani, the liberalization of interest rates, the introduction of market-based monetary instruments, the elimination, at least partially, of selective credit controls and the authorization of making foreign currency loans by local banks for selected purposes, as well as the harmonization and gradual reduction of reserve requirements. In addition, the Central Bank of Paraguay (BCP) began to carry out open market operations using its own short-term debt instrument.
2. These reforms fueled the expansion of the banking sector and bank credit. Prior to 1980, the Paraguayan banking system largely consisted of foreign banks. During the 1980s, several domestic banks and finance companies started to operate, holding a small share of deposits. In the late 1980s and early 1990s, the relaxed entry requirements in the law, resulted in a large increase in the number of domestic banks and finance companies operating in the system. The number of financial intermediaries operated in the country grew to close to 100 in late 1994, at the onset of banking crisis.
Number of Financial Institutions between 1995 and 2003
Number of Financial Institutions between 1995 and 2003
| 1995 | 1996 | 1997 | 1998 | 1999 | 2000 | 2001 | 2002 | 2003 | ||
|---|---|---|---|---|---|---|---|---|---|---|
| Number of banks | 35 | 32 | 33 | 23 | 22 | 22 | 20 | 18 | 14 | |
| Wholly foreign owned | 13 | 14 | 9 | 9 | 9 | 9 | 8 | 7 | 6 | |
| Majority foreign owned | … | … | 9 | 8 | 7 | 8 | 8 | 7 | 5 | |
| Locally owned private banks | 20 | 16 | 13 | 5 | 5 | 4 | 3 | 3 | 2 | |
| Public development bank | 2 | 2 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | |
| Number of finance companies | 68 | 54 | 50 | 36 | 33 | 28 | 25 | 21 | 17 | |
Number of Financial Institutions between 1995 and 2003
| 1995 | 1996 | 1997 | 1998 | 1999 | 2000 | 2001 | 2002 | 2003 | ||
|---|---|---|---|---|---|---|---|---|---|---|
| Number of banks | 35 | 32 | 33 | 23 | 22 | 22 | 20 | 18 | 14 | |
| Wholly foreign owned | 13 | 14 | 9 | 9 | 9 | 9 | 8 | 7 | 6 | |
| Majority foreign owned | … | … | 9 | 8 | 7 | 8 | 8 | 7 | 5 | |
| Locally owned private banks | 20 | 16 | 13 | 5 | 5 | 4 | 3 | 3 | 2 | |
| Public development bank | 2 | 2 | 2 | 1 | 1 | 1 | 1 | 1 | 1 | |
| Number of finance companies | 68 | 54 | 50 | 36 | 33 | 28 | 25 | 21 | 17 | |
3. Macro-economic conditions further contributed to the increase of bank credit. Growth had been strong, and inflation was in double digits prior to the outbreak of the crisis in 1995. High inflation and a cap on lending interest rate resulted in negative real lending rates, which fueled the expansion of bank credit in the early phase of the cycle. Policy measures to alleviate the impact of a negative cotton price shock on the economy in 1992 further contributed to the growth of credit.6
4. Bad banking practices, including excessive concentration and connected lending without appropriate credit-risk analysis, inadequate accounting practices, and inefficient supervisory framework, set the stage for the banking crisis in 1995 (Franks et al, 2002).
5. Banking sector supervision did not keep pace with the expansion of the banking sector (Garcia-Herrero, 1997). Progress in banking supervision during the liberalization process was slow, especially the enforcement of capital requirements and minimum entry capital. Many new small banks and financial companies had very low levels of capital. Adding to the problem, although the Superintendence of Banks was aware of the distressed situation of part of the financial system, it was not allowed to apply the appropriate sanctions by the political authorities. Meanwhile, the BCP continued to give credit to problem banks, while requesting shareholders to provide additional capital. However, this capital never materialized. Thus, when a confidence shock hit the banking sector in mid-1995, it morphed into a full blow banking crisis quickly due to the lack of public confidence in the banking system in general.
6. The banking crisis broke out in mid-1995 triggered by the publication of an accounting discrepancy in the BCP’s books, which shook public confidence. Additionally, the third and fourth largest commercial banks (Banco General and Bancopar), which had been identified by the Superintendence of Banks as having capital deficiencies in 1994, in terms of deposits, were unable to meet their obligations. The BCP intervened the banks to keep them open to the public. The crisis spread quickly in the following few months. Two additional commercial banks and a few other finance companies and saving and loan association fell into distress, requiring BCP intervention. There was a massive withdrawal of deposits from private domestic banks as a group, and from intervened banks in particular.
7. To avoid a run on the entire banking system and a failure of the payments system, the BCP provided large scale liquidity support. Deposits were withdrawn from private domestic and government-owned banks and put into foreign banks that left total deposit in the banking system remained relatively stable. Because of the reluctance of healthier banks, especially foreign banks, to channel their excess liquidity through the interbank market, the BCP’s window was the main source of liquidity for distressed banks. About half of the borrowing from the BCP was used by intervened banks to offset deposit withdrawals, and the other half to cover the call loans outstanding with other financial institutions and short-term external obligations. In addition, pressure started to mount from other organized sectors for special lines of credit to ease the perceived “liquidity shortage” and for longer term credit facilities for investment projects directly or indirectly through banks.
8. The waves of bank failures ended in late 1998 after the closure of three weak banks (including a politically sensitive bank – Workers Bank). By that time, about half of the financial institutions had failed (see table 1). One of the contributing factors to the end of the crisis was the changed way of dealing with insolvent banks by the authorities – instead of extending liquidity support, the government decided to resolve ailing institutions, including politically sensitive ones.
9. Inadequate resources for the supervisory authorities and political interference in dealing forcefully with remaining weak or insolvent banks explained why many of these banks could remain operating and thereby effectively increased the overall costs of their resolution when they were ultimately closed in late 1997 and 1998. The direct cost of the intervention effort led by the central bank was estimated to be around 5 percent of GDP (Garcia-Herrero, 1997). This cost includes i) the liquidity support (amount to 4 percent of GDP) provided by the BCP at end-1995 to intervened and distressed banks; ii) the additional liquidity support of 1 percent of GDP in mid-1996.
10. The crisis reduced the share of private domestic banks. There were increases in the share of deposits, assets and net worth held by foreign-owned banks, as well as government-owned banks. By end-December 2002, 85 percent of total deposits in the banking system was in branches of foreign banks or majority foreign-owned banks, and 83 percent of total assets were held by foreign banks or majority foreign-owned banks. In sum, the impact of the crisis was uneven across different types of banks and the banking system became more concentrated and segmented.
11. At a micro-level, the increased share of foreign wholly-owned and majority owned banks contributed to the slow recovery of credit. Local firms (which tend to be smaller local firms, in asset size, than foreign firms operated in Paraguay) used to obtain credit from domestic private banks, before the crisis, could no longer get credit from these institutions due to the higher interest rate, reflecting the lack of deposit base at these institutions. However, these borrowers were also not able to obtain credit from foreign owned banks as they were perceived to be too different from the usual borrowers at foreign-owned banks (large international corporations), and foreign-owned banks did not want to take the risk of accepting borrowers from the failed institutions (mostly small and local borrowers).
The Economy After the Banking Crisis
12. The economy was in recession between 1998 and 2002 as it was hit by a mix of external and domestic shocks––the Argentinean financial crisis, a drought and a foot-and-mouth disease. The result was an increase in the non-performing loan ratio. Credit to the private sector fell from the peak reached in 1997.
Non-Performing Loan to Total Loan Ratio
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations.Non-Performing Loan to Total Loan Ratio
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations.Non-Performing Loan to Total Loan Ratio
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: BCP and IMF staff calculations.13. The Argentinean financial crisis in 2002 had a direct impact on the financial system in Paraguay. In mid-2002, there was a deposit run on Banco Aleman, owned by an Argentine-Uruguayan conglomerate, began after the parent institutions run into difficulties. The ailing bank was the third-largest bank by asset back then (about 11 percent of banking system assets). The authorities responded swiftly by intervening in the bank, paying out small deposits with a deposit guarantee provided by the central bank, and transferring most other assets and liabilities to other banks in the system. As the ability of most banks to absorb these shocks improved a lot since 1998, there were no wide spread failure of banks, except for the public development bank (BNF).
14. Rapid growth came back afterwards, in part, due to good macro-policies. Between 2003 and 2008, in the context of several IMF programs, there was an improvement of macro-economic policies and an opening of the economy. Inflation and exchange rate volatility came down as the central bank shifted to inflation targeting, while prudent fiscal policy reduced government debt significantly. The global agricultural commodity price boom led to a rapid growth in real incomes and a surge in domestic demand and a credit boom.
15. The quality of banking supervision improved as major revisions to Banking laws were passed, which bought banking supervision closer to those prescribed in the Basel Core Principles. Supervisors were granted more resources, more independence and more sanctioning capacity. Management capacity of bank supervisors and corporate governance were also improved. The soundness of the banking sector improved as banking practices become better regulated as a result of these reforms.7 For example, individual bank generally held more capital than the amount required legally; 8 provision to NPLs had increased considerably.
Provision to NPLs
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: Data extracted from past AIV reports and Frank et al (2002)Provision to NPLs
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: Data extracted from past AIV reports and Frank et al (2002)Provision to NPLs
(In percent)
Citation: IMF Staff Country Reports 2019, 112; 10.5089/9781498312318.002.A003
Source: Data extracted from past AIV reports and Frank et al (2002)Lessons Learned after the Banking Crisis
16. The credit boom in the early 1990s was largely the result of financial market liberalization that stimulated the expansion of the banking sector, coupled with favorable economic conditions. However, the loose entry requirements allowed weak banks to join. In addition, poor enforcement of banking laws allowed banks to evade capital requirements. This resulted in a banking crisis with half of the banks failed and a collapse of the credit. Credit growth stagnated after the banking crisis. Credit growth then resumed around 2006 and kept growing until 2015.
The fall of soybean price in 2015 triggered a decline in credit growth but, unlike the previous crisis, there was no banking crisis, due to the improved resilience of the banking sector. This improved resilience is a result of a series of efforts to improve the soundness of the banking sector since the previous banking crisis, which allows it to handle the credit increase more comfortably. The modernization of the bank supervision framework and increased capacity of the supervisors contributed to the improvement. Banks that remained after the previous banking crisis were generally healthier financially and adopted safer banking practices.
Prepared by Alex Ho with Marie Kim (currently on leave from the IMF). Comments from participants in a presentation at the Central Bank of Paraguay during the 2019 Article IV consultation mission to Paraguay are gratefully acknowledged.
These credit booms have been typically shorter than traditional booms -about a year compared to the cross¬country average of three years – and did not result in a financial crisis and/or severe downturn. The credit-to-GDP grew on average by 15 percent per year during these boom years, or about five times its median growth in non-boom years, which is in line with credit booms observed in other countries (see Dell’Ariccia et al., 2012).
As of end-2017, Paraguay has total 16 banks, of which 8 banks are domestic owned, 4 banks are foreign owned, 3 banks are foreign branches, and there is one state bank. Domestic banks have become more dominant players in the market, explaining close to 50 percent of total assets at end-2017, from 28 percent of total assets in early 2008. On the contrary, foreign branches have become less important explaining 4 percent of total assets at end-2017 from 13 percent in early 2008.
Not only soy but total agriculture exports matter. Credit to the soy sector explains about 49 percent of total agriculture sector, on average from 2008–2016. Since the definition of agriculture sector loan has changed in 2016 and is difficult to compute the relevant agriculture price export index, this paper mainly focuses on the soy sector.
The main monetary policy tool in Paraguay is the policy interest rate, which has been the principal tool used to target inflation since 2010, coupled with a flexible exchange rate, to balance between growth and price stability.
The authorities provided subsidized credit, through commercial banks, to the agricultural sector.
For example, banks faced strengthened capital requirement; stricter information requirements for the granting of loans, more stringent criteria for portfolio classification; higher provisioning requirements for the nonperforming loan portfolio.
See discussions in Paraguay’s 2007, 2009 and 2010 Article IV reports.
