How Do Credits Dollarize? The Role of Firm’s Natural Hedges, Banks’ Core and Non-Core Liabilities

Transcription

1. How Do Credits Dollarize ? The Role of Firm’s Natural Hedges, Banks’ Core and Non-Core Liabilities Fatih Yılmaz February 2020 Working Paper No: 20/01
2. © Central Bank of the Republic of Turkey 2020 Address: Central Bank of the Republic of Turkey Head Office Structural Economic Research Department Hacı Bayram Mh. İstiklal Caddesi No: 10 Ulus, 06050 Ankara, Turkey Phone: +90 312 507 80 04 Facsimile: +90 312 507 78 96 The views expressed in this working paper are those of the author(s) and do not necessarily represent the official views of the Central Bank of the Republic of Turkey.
3. How Do Credits Dollarize ? The Role of Firm’s Natural Hedges, Banks’ Core and Non-Core Liabilities∗ Fatih Yılmaz† Central Bank of the Republic of Turkey Abstract. We show that firms’ natural hedges (e.g., export revenues) and banks’ foreign currency (FX) liabilities strongly dollarize credits. In particular, banks’ non-core FX liabilities (e.g., syndications) in average feed credit dollarization almost three times more than their core FX liabilities (e.g., deposits). More importantly, these channels are affected differently by local and global macroeconomic conditions. Keywords: Credit Dollarization, Liability Dollarization, Deposit Dollarization JEL Classification: G21, G32 Özet. Bu çalı¸smada, firmaların dogal ˘ korumalarının (örnegin, ˘ ihracat gelirleri) ve bankaların yabancı para (YP) yükümlülüklerinin kredileri güçlü bir s¸ ekilde dolarize ettigini ˘ göstermekteyiz. Özellikle, bankaların çekirdek dı¸sı YP yükümlülüklerinin (örnegin, ˘ YP sendikasyonlar) çekirdek YP yükümlülüklerinden (örnegin, ˘ YP mevduatlar) neredeyse üç kat daha fazla kredileri dolarize ettigi ˘ görülmekte. Daha da önemlisi, bu kanallar yerel ve küresel makroekonomik ko¸sullardan farklı s¸ ekillerde etkilenmektedir. Anahtar Kelimeler: Kredi Dolarizasyonu, Yükümlülük Dolarizasyonu, Mevduat Dolarizasyonu JEL Sınıflandırması: G21, G32 ∗I would like to thank the seminar participants at the CBRT. I particularly thank Hakan Kara, Yusuf Emre Akgunduz and Salih Fendoglu ˘ for their very valuable comments. The views in this paper are solely the responsibilities of the author and should not be interpreted as reflecting the view of the Central Bank of the Republic of Turkey. † Fatih Yılmaz, Economist, Structural Economic Research Department, Central Bank of the Republic of Turkey, Fatih.Yilmaz@tcmb.gov.tr.
4. How Do Credits Dollarize ? Non-Technical Summary Dollarization of credits is a vital financial stability concern in many emerging economies that is mainly driven by firm and bank tendencies. Relying on their natural hedges (e.g., export revenues), firms prefer FX credits due to their lower cost advantages and longer-term maturity availability. Dollarization of bank liabilities, on the other hand, either via non-core FX liability channel (e.g., FX syndications) or via core FX liability channel (e.g., FX deposits) induces them to transfer the FX risk to borrowers. Using a rich micro data from Turkey, we study how strongly these three channels dollarize firm-bank credits and how they are affected by local and global macroeconomic conditions in this paper. We find that both firm natural hedges and bank FX liabilities derive credit dollarization strongly. When we decompose banks’ total FX liabilities into core and non-core liabilities, we find that banks’ non-core FX liabilities in average feed credit dollarization almost three times more than their core FX liabilities. Differences in the maturity structure between the two sources (e.g., the average maturity of FX deposits is significantly lower than FX syndications) could be one explanation for this result. We also show that the impact of these channels varies depending on the local and global macroeconomic conditions. An increase in the effective Fed funds rate (e.g., tightening of the global liquidity) weakens the effect of non-core FX liability channel on credit dollarization. We also observe a weakening in both non-core FX liability and natural hedge channels during times of high exchange rate volatility in the local currency, while the core FX channels seems to gain some strength. In contrast, during times of positive GDP growth, all the three channels become stronger, although the effect of the core FX channel is moderate. Our results provide an important insight for macroprudential policies aiming to combat credit dollarization. In particular, macroprudential policies may target disciplining deposit dollarization during times of tight global liquidity conditions and/or high exchange rate volatility in the local currency. In contrast, during times of softer global liquidity conditions and/or positive GDP growth, prudential policies can be designed to focus more on natural hedge and non-core FX liability channels. 1
5. How Do Credits Dollarize ? 1 Introduction Dollarization of credits is a vital financial stability concern in many emerging economies. Especially during times of large currency depreciations, foreign currency (FX) liabilities of firms and banks disrupt their balance sheets and can even lead to systemic events. Despite its importance, we have very little empirical evidence on how firm and bank behaviors derive credit dollarization. In particular, firms and banks have different motivations for engaging in such a risky credit relation. Relying on their natural hedges (e.g., export revenues), firms prefer FX credits due to low cost and long-term maturity advantages. For instance, in the case of Turkey, the average interest rate for TL denominated corporate credits is usually almost three times higher than FX credits (Figure 1), while FX credits’ average maturity is significantly longer than TL denominated credits (Figure 2). Dollarization of bank liabilities, on the other hand, either via non-core FX liability channel (e.g., FX syndications, bonds) due to the original sin phenomenon (Eichingreen et. al., 2003) or via core FX liability channel (e.g, FX deposits) induces them to transfer the FX risk to borrowers.1 Overall, such tendencies of firms and banks determine the level dollarization of firm-bank credit relations at the equilibrium. Using a rich micro data from Turkey, we provide empirical evidence on how these three channels, firms’ natural hedges and banks’ core and non-core FX liabilities, dollarize firm-bank credits in this paper. Turkey, as a major emerging economy, provides an ideal laboratory to study credit dollarization. Compare to many other emerging markets, Turkish non-financial corporates carry a relatively higher level of FX credit share (Figure 3), which has been an important concern to financial stability (GFSR, 2018).2 For the analysis, we match the Turkish Credit Registry with firm and bank financial statements. The Credit Registry provides us the currency denomination of firm-bank credit relations, while from financial statements, we observe firms’ natural hedges and banks’ core and non-core FX liabilities. In our main specification, we exploit the heterogeneity in firm-bank credit relations with a large set of controls and fixed effects. In a similar fashion to Khwaja and Mian (2008), we then focus on the firms with multiple bank relations to strengthen the identification. Our results show that both firm natural hedges and bank FX liabilities derive credit dollarization strongly. However, banks’ non-core FX liabilities feed credit dollarization almost three times more than their core FX liabilities. We also show that the impact of these channels varies depending on the local and global macroeconomic conditions. An increase in the effective Fed funds rate (e.g., tightening of the global liquidity) weakens the effect of non-core FX liability channel on credit dollarization. We observe a weakening in both non-core FX liability and natural hedge channels during times of high exchange rate volatility in the local currency. In contrast, during times of positive GDP growth, all three channels become stronger, although the effect of the core FX channel is moderate. These results are robust to different identification techniques, 1 Banks’ net FX open position is highly regulated and cannot exceed certain limits in many emerging economies, including Turkey. 2 It is also worth noting that during the time of this study there was not any major policy change for lending and borrowing of FX funds in Turkey. 2
6. How Do Credits Dollarize ? additional controls and consideration of sub-samples. These results provide an important insight for macroprudential policies aiming to combat credit dollarization. In particular, macroprudential policies may target disciplining deposit dollarization during times of tight global liquidity conditions and/or high exchange rate volatility in the local currency. In contrast, macroprudential policies, on the other hand, may focus more on natural hedge and non-core liability channels during times of softer global liquidity conditions and/or positive GDP growth. This paper contributes to the literature on drivers of credit dollarization by bringing robust empirical evidence from firm-bank level micro data. Most of the papers in this literature provides suggestive evidence from macro data. For instance, Luca and Petrova (2007) presents a simple theoretical framework to identify the role of the aforementioned three channels in deriving credit dollarizaiton at firm-bank level. Yet, they can only test their theoretical findings with aggregated data. Limited number of micro studies in the literature, on the other hand, focuses on only one side of the story at a time, firms’ or banks’ perspectives, separately, which weakens their identification. For instance, Gelos (2003) shows that firms with natural hedges tend to prefer FX credits and similarly, Ozsoz et. al (2015) finds that banks match the currency denomination of their liabilities with their assets. By bringing direct evidence on channels deriving credit dollarization, our paper also complements the recent micro studies looking at the effect of global macroeconomic shocks and monetary policy shifts on banks’ lending decisions (e.g., currency choice) - e.g., Brown et. al. (2014) on Bulgaria and Ongena et. al. (2018) on Hungary. In the next section, we present the details of our micro data. Our estimation procedure is explained in Section 3. In section 4, main research findings are presented and Section 5 concludes. For brevity, we present only the main results in the paper, while the full set of results is available in the Appendix. 2 Data and Descriptive Statistics The firm-bank level monthly Credit Registry is from the Banks Association of Turkey. The annual company balance sheets and income statements are obtained from the Central Bank of Republic of Turkey (CBRT)’s firm data base; monthly bank balance sheets and income statements come from the Banking Regulation and Supervisory Association (BRSA). Manufacturing firms3 with at least 10,000 TL (2,850 USD) average annual real asset value and report financial data at least two consecutive years over the sample period are kept in the analysis. Only the deposit taking banks are employed, as we would like to consider core FX (e.g., deposits) and non-core FX (e.g., syndications) liability channels separately.4 Lastly, we take 3 We can only observe the natural hedges (e.g.,export revenues) of the firms in manufacturing sector. other bank types (e.g., investment and development banks) may also have different motivations (e.g., financing exports, financing development, etc.) in their lending decisions that is beyond the scope of this analysis. 4 Additionally, 3
7. How Do Credits Dollarize ? firm-bank relations that are above 1000 TL (285 USD). 5 Overall, our analysis contains an average number of 20,000 firm-bank credit relations in a given month for an unbalanced panel of 4,396 manufacturing firms with 26 major deposit banks over the period of November 2006 - December 2016. Our sample represents 71 percent of total credits granted to firms in manufacturing sector by deposit banks in Turkey over the sample period (Table 1).6 The average coverage is about 61 percent for total TL credits and 78 percent for total FX credits. Descriptions and summary statistics of key variables are reported in Table 2, while the detailed summary statistics and descriptions of all the variables included in the analysis are available in Table A1 of the Appendix. 3 Estimation Following Luca and Petrova (2007), we estimate a model of credit dollarization at firm-bank level. The model presumes that both banks and firms are risk averse and hence, adopt a minimum variance portfolio (MVP) method in their borrowing and lending decisions. According to this, firms with natural hedges (e.g., export revenues) tend to prefer FX loans to minimize their cost of finance, while banks with FX liabilities are more inclined towards issuing FX loans in order to match the currency denomination of their assets and liabilities: FX Credits Total Credits = α1 ijt + β1 Export Revenues Total Revenues it−1y FX Liabilities Total Liabilities jt−1m + α [Firm Controls]it−1y + β [Bank Controls] jt−1m + θ [Firm-Bank Credit Relation Controls]ijt−1m + f i ∗ b j + dt + eijt (1) The dependent variable, referred as the firm-bank level credit dollarization, is the ratio of FX credits to total credits of firm i with bank j at time t. Firm i’s share of exports in total sales captures natural hedges. Bank j’s FX liabilities are further decomposed into core FX (i.e., share of FX deposits in total liabilities) and non-core FX liabilities (e.g., share of FX securitizations and syndications in total liabilities). In the main specification, besides the main variables of interest, we also control for firm, bank and firm-bank credit relation variables that are lagged for one period.7 Time trend is captured by time fixed effects and unobserved time invariant firm-bank credit relation heterogeneity is saturated with firm-bank fixed effects. Similar to Khwaja and Mian (2008), we re-do the main 5 This is to exclude insignificantly small or zero firm-bank credit relations, which are in total less then 0.005 percent of total firm credits in a given month. 6 The analysis covers only the TL, FX and FX indexed loans that are granted by deposit banks to manufacturing firms. Cross-border lending of foreign financial institutions to non-financial Turkish firms and non-performing loans are excluded. 7 List of key variables is presented in Table 2, while the complete list is available in Table A1 of the Appendix. 4
8. How Do Credits Dollarize ? estimates with firm-time fixed effects that identifies bank supply relying only on firms with multiple bank relations. This does not cost much to our data, as only 3 percent of our sample contains firms with single bank relations. The same idea, bank-time fixed effects, is also applied to identify firm demand, while holding bank supply constant. The robustness of our main estimates is also further tested with sub-samples, additional controls and time structure (e.g., quarterly). We are also interested in how firm and bank tendencies may change depending on local and global macroeconomic conditions. Following the related literature8 , we interact the aforementioned three channels with increase in the effective Fed funds rate, positive real economic growth and the level of exchange rate volatility. The effective Fed funds rate increase is a dummy variable that is equal to one for all increases above 5 basis points, otherwise zero; positive real economic growth is also a dummy variable that is equal to one for years of positive GDP growth in Turkey, otherwise zero and finally, exchange rate volatility is in levels. 4 Results The main results are presented in Table 3, where Column (1) shows the baseline results; bank FX liabilities are further decomposed into core and non-core FX liabilities in Column (2) and this specification is interacted with macroeconomic variables in columns (3)-(5). Only the main variables of interest are presented here, while the full version is available in Table A2 of the Appendix. According to the baseline estimates, firms’ natural hedges and banks’ FX liabilities significantly dollarize firm-bank credits. One standard deviation increase in firm natural hedges (in bank total FX liabilities) is associated with about 2.2 (1.5) percentage point increase in firm-bank level credit dollarization. More importantly, when we decompose banks’ total FX liabilities, we find that one standard deviation increase in core FX liabilities is associated with a 0.6 percentage point increase in credit dollarization that is estimated to be almost three times more for non-core FX liabilities, 1.6 percentage point. One explanation for this result may be the differences in maturity structure between the two sources. For instance, the average maturity of FX deposits held at Turkish banks is less than 3 months, while this number is as high as 68 months for syndicated loans, obtained by Turkish banks from global financial markets.9 Results also show that these tendencies may shift due to local and global macroeconomic conditions. In particular, an increase in the effective Fed funds rate (i.e, tightening of the global liquidity) weakens the non-core FX liability channel significantly, while the core FX liability and firm natural hedge channels do not seem to be affected much. During times of high exchange rate volatility in local currency, we observe similar trends along with a mild reduction in the natural hedge channel. Core liability channel shows some weak tendency towards strengthening 8 For a detailed literature review on macro determinants of credit dollarization, see Hake et. al. (2014). the Chart IV.2.10 in May Financial Stability Report (2019) by the CBRT for average syndication maturity and for the deposit maturity, see the BRSA Monthly bulletin in 2017. 9 See 5
9. How Do Credits Dollarize ? during high volatility times, although it is statistically insignificant. In contrast, during time of positive growth of GDP (Column 4) , we observe a feed to credit dollarization from core and non-core FX channels, while the natural hedge channel seems to also increase, moderately. These results are robust to different considerations. More specifically, the main model is re-estimated with firm-time fixed effects and also, with bank-time fixed effects (Table 4). These estimations are further repeated with sub-samples such as “at least once” (i.e., received FX credit at least once), “exported 50 percent or more” (i.e., exported more than 50 percent of the time during the sample period), “HH FX deposits only” (i.e., employing on household FX deposits only instead of total FX deposits) and finally, “quarterly” (i.e., quarterly data instead of monthly) (Table 5). The baseline estimates remain mostly statistically significant and economically important across all these considerations. 5 Conclusions Our results show that firms’ natural hedges and banks’ liability-asset matching tendencies significantly drive credit dollarization. Among bank FX liabilities, the effect of non-core FX liabilities on credit dollarizaiton appears to be almost three times larger than core FX liabilities. More importantly, these channels are adversely affected by local and global macroeconomic conditions. During times of tight global liquidity conditions, the non-core FX channel is weakened. Similarly, besides the non-core FX channel, high exchange rate volatility in local currency also weakens the natural hedge channel. During times of positive economic growth, all three channels become stronger. These results call upon a more focused macroprudential policy consideration to combat credit dollarization. During times of tight global liquidity conditions and/or high exchange rate volatility in local currency, macroprudential policies may focus more on disciplining deposit dollarization. During times of softer global liquidity conditions and/or positive GDP growth in the domestic economy, the focus may be diverted towards disciplining the effect of natural hedge and non-core liability channels on credit dollarization. References [1] Brown, Martin; Karolin Kirschenmann and Steven Ongena, (2014) “Bank Funding, Securitization, and Loan Terms: Evidence from Foreign Currency Lending” Journal of Money, Credit and Banking, Blackwell Publishing, No 46(7). [2] Eichengreen, B., Hausmann, R., and Panizza, U. (2003) “The pain of original sin. Other people’s money: Debt denomination and financial instability in emerging market economies” The University of Chicago Press, 13-37. Edited by B. Eichengree and R. Hausmann. [3] Gaston Gelos, R., (2003) “Foreign currency debt in emerging markets: firm-level evidence from Mexico” Economics Letters, Elsevier, vol. 78(3), pages 323-327, March. 6
10. How Do Credits Dollarize ? [4] Hake, Mariya and Lopez-Vicente, Fernando Molina, Luis (2014). “Do the Drivers of Loan Dollarization Differ between CESEE and Latin America? A Meta-Analysis” Focus on European Economic Integration, Oesterreichische Nationalbank, (Austrian Central Bank), issue 1, pages 8-35. [5] International Monetary Fund (2018) “A Decade After the Global Financial Crisis: Are We Safer?” Global Financial Stability Report, Chapter 1. [6] Khwaja, Asim Ijaz, and Atif Mian. 2008 “Tracing the Impact of Bank Liquidity Shocks: Evidence from an Emerging Market” American Economic Review, 98 (4): 1413-42 [7] Luca, Alina and Petrova, Iva, (2008) “What drives credit dollarization in transition economies?” Journal of Banking & Finance, Elsevier, 32(5), 858-869. [8] Ozsoz, Emre; Rengifo, Erick W. and Kutan, Ali, (2015) “Foreign Currency Lending and Banking System Stability: New Evidence from Turkey” Central Bank Review, 15, No.2. [9] Ongena, Steven R. G. and Schindele, Ibolya and Vonnák, Dzsamila (2018) “In Lands of Foreign Currency Credit, Bank Lending Channels Run Through?” CFS Working Paper, No. 474. 7
11. How Do Credits Dollarize ? TL Credits 2016m12 2015m12 2014m12 2013m12 2012m12 2011m12 2010m12 2009m12 2008m12 2007m12 2006m12 5 10 percent 15 20 25 Figure 1: Average Insterest Rate for Non-Financial Corporate Credits by Currency Denomination FX Credits Source: Central Bank of Republic of Turkey TL Credits Source: Author’s calculation from Credit Registry 8 FX Credits 2016m12 2015m12 2014m12 2013m12 2012m12 2011m12 2010m12 2009m12 2008m12 2007m12 2006m12 40 50 percent 60 70 80 Figure 2: Share of Long-term Credits (≥1 year) for Non-Financial Corporate Credits by Currency Denomination
12. How Do Credits Dollarize ? 2006 China Korea India Saudi Arabia Thailand Malaysia Chile Poland Brazil Colombia Czech Russian Federation Israel Indonesia South Africa Hungary Argentina Turkey Singapore Mexico 0 20 percent 40 60 80 Figure 3: Non-Financial Corporate FX Debt to Total Debt Ratio for Emerging Economies 2016 Source: Author’s calculation from Institute of International Finance (IIF) Table 1: Sample Representation of the Relevant Population 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Overall Total TL Share Total FX Share Total Share 0.63 0.59 0.69 0.81 0.68 0.64 0.62 0.57 0.54 0.50 0.48 0.61 0.73 0.83 0.85 0.86 0.84 0.80 0.81 0.79 0.73 0.72 0.66 0.78 0.69 0.73 0.79 0.84 0.77 0.73 0.72 0.69 0.64 0.61 0.57 0.71 Share is the total credit amount covered in the sample/relevant population. Relevant population is the credits received by manufacturing firms from deposit banks. 9
13. How Do Credits Dollarize ? Table 2: Variable Descriptions and Summary Statistics of the Main Variables Variable Credit Dollarizationt Export Share (Natural Hedge)t−1y FX Liability Sharet−1m FX Deposit Sharet−1m Non-Core FX Liability Sharet−1m Effective Fed Funds Rate Increaset Positive GDP Growth Ratest Exchange Rate Vollatilityt Definition Obs Mean Std. Dev. Min Max FX Credits/ Credits Export Sales/ Sales FX Liabilities/ Liabilities FX Deposits/ Liabilities Non-Core FX Liability/ Liabilitis Equals to 1 if ∆>5 basis points, otherwise 0 Equals to 1 if growth is positive, otherwise 0 Variance of changes in exchange rates in the last 12 months 2,503,040 48,359 2,965 2,965 2,965 0.35 0.22 0.45 0.28 0.17 0.45 0.29 0.11 0.09 0.09 0.00 0.00 0.15 0.10 0.00 1.00 1.00 0.87 0.86 0.66 122 0.03 0.18 0.00 1.00 122 0.88 0.33 0.00 1.00 122 0.09 0.09 0.01 0.36 Time span of the dependent variable is 2006m11-2016m12. Firm variables’ time span is 2015 - 2015 and they are lagged for one year (y). Bank variables’ (and firm-bank level variables) time span is 2006m11-2016m11 and they are lagged for one month (m). Non-core FX liabilities of banks cover all bank liabilities, excluding deposits. Deposits include firm and household deposits. Exchange rate is defined as the basket of 0.3*Euro+0.7*USD. 10
14. How Do Credits Dollarize ? Table 3: Main Empirical Results 1 2 3 4 5 Bank Liab. X Effective Fed X Positive Real X Exchange Rate Baseline Breakdown Funds Rate Increase Economic Growth Volatility A Export Share 0.0793*** (0.0107) 0.168*** (0.0164) t−1y FX Liability Sharet−1m B FX Deposit Sharet−1m C Non-Core FX Liability Sharet−1m 0.0793*** (0.0107) 0.0796*** (0.0106) 0.0683*** (0.0118) 0.0836*** (0.0110) 0.0930*** (0.0207) 0.208*** (0.0196) 0.0944*** (0.0207) 0.211*** (0.0196) -0.00885 (0.00540) -0.0198 (0.0163) -0.0792*** (0.0147) 0.0520* (0.0286) 0.143*** (0.0257) 0.0125** (0.00578) 0.0458** (0.0201) 0.0748*** (0.0198) 0.0835*** (0.0214) 0.227*** (0.0212) -0.0476* (0.0265) 0.0929 (0.0916) -0.205** (0.0815) YES YES YES YES YES 2,331,834 0.786 YES YES YES YES YES 2,331,834 0.786 YES YES YES YES YES 2,331,834 0.786 A X Macrot B X Macrot C X Macrot Firm Controls Bank Controls Firm - Bank Controls Firm X Bank Fixed Effects Time (year-month) Fixed Effects Observations Adj. R-squared YES YES YES YES YES 2,331,834 0.786 YES YES YES YES YES 2,331,834 0.786 Standard Deviations (STD) and Means (M) to Evaluate the Estimates A Export Share t−1y FX Liability Sharet−1m B FX Deposit Sharet−1m C Non-Core FX Liability Sharet−1m A X Macrot B X Macrot C X Macrot STD M STD M STD M STD M STD M STD M STD M 0.27 0.21 0.09 0.44 0.27 0.21 0.27 0.21 0.27 0.21 0.27 0.21 0.06 0.25 0.08 0.18 0.06 0.25 0.08 0.18 0.060 0.007 0.049 0.009 0.041 0.007 0.06 0.25 0.08 0.18 0.263 0.189 0.098 0.226 0.092 0.166 0.06 0.25 0.08 0.18 0.039 0.018 0.026 0.022 0.015 0.014 Clustered by firm id. *** p<0.01, ** p<0.05, * p<0.1. Dependent variable is the share of FX credits in firm-bank outstanding credit balance. Firm controls include size (log of assets), leverage, profitability, liquidity and fixed asset ratio; bank controls include size (log of assets), leverage, profitability, liquidity and NPL ratio; firm-bank controls cover maturity, sector and the share of biggest sector of the credit relation. All the firm variables are lagged for one year (y). Bank and firm-bank controls are lagged for one month (m). Effective Fed funds rate increase is equal to one for 5 basis point and more increases, otherwise zero; Positive Real GDP Growth is equal to one for positive GDP growth, otherwise, zero and finally, Exchange Rate Volatility is in levels. 11
15. How Do Credits Dollarize ? Table 4: Full Control of Firm and Bank Determinants VARIABLES A Export Share 1 Bank Liability Breakdown 2 X Effective Fed Funds Rate Increase 3 X Positive Growth 4 X Exchange Rate Volatility 0.0217*** (0.00289) 0.0218*** (0.00288) 0.0186*** (0.00319) 0.0228*** (0.00297) B FX Deposit Share C Non-Core FX Share A X Macro -0.000559* (0.000324) 0.00342** (0.00152) 5 Bank Liability Breakdown 6 X Effective Fed Funds Rate Increase X Positive Growth 8 X Exchange Rate Volatility 0.00369*** (0.00125) 0.0131*** (0.00144) 0.00378*** (0.00125) 0.0133*** (0.00145) 0.000261 (0.00181) 0.00763*** (0.00194) 0.00359*** (0.00132) 0.0145*** (0.00157) -0.00108 (0.000779) -0.00298*** (0.000598) 0.00594*** (0.00207) 0.00746*** (0.00184) 0.000231 (0.00251) -0.00300** (0.00125) NO YES YES YES YES YES NO 2,249,801 0.857 NO YES YES YES YES YES NO 2,249,801 0.857 NO YES YES YES YES YES NO 2,249,801 0.857 -0.00171 (0.00104) B X Macro C X Macro Firm Controls Bank Controls Firm - Bank Relation Controls Firm X Bank FE Time FE Firm X Time FE Bank X Time FE Observations Adj. R-squared YES NO YES YES YES NO YES 2,331,834 0.788 YES NO YES YES YES NO YES 2,331,834 0.788 7 YES NO YES YES YES NO YES 2,331,834 0.788 YES NO YES YES YES NO YES 2,331,834 0.788 NO YES YES YES YES YES NO 2,249,801 0.857 Clustered by firm id. *** p<0.01, ** p<0.05, * p<0.1. Dependent variable is the share of FX credits in firm-bank outstanding credit balance. Standardized coefficients are reported, showing the effect of one standard deviation from mean. Table 5: Full Control of Firm and Bank Determinants with Further Robustness VARIABLES A Export Share 1 At Least Once 2 Exported 50 % or More 3 Quarterly 0.0231*** (0.00309) 0.0256*** (0.00340) 0.0215*** (0.00304) B FX Deposit Share C Non-Core FX Share FX Household Deposit Share 4 At Least Once 5 Exported 50% or More 0.00408*** (0.00140) 0.0144*** (0.00160) 0.00440*** (0.00168) 0.0165*** (0.00200) 6 HH FX Deposits Only 0.0128*** (0.00141) 0.00513*** (0.00169) 7 Quarterly 0.00366*** (0.00136) 0.0125*** (0.00155) Firm Controls YES YES YES NO NO NO NO Bank Controls NO NO NO YES YES YES YES Firm - Bank Relation Controls YES YES YES YES YES YES YES Firm X Bank FE YES YES YES YES YES YES YES Time FE YES YES YES YES YES YES YES Firm X Time FE NO NO NO YES YES YES YES Bank X Time FE YES YES YES NO NO NO NO Observations 2,004,952 1,322,492 711,721 1,949,393 1,278,716 2,249,801 682,996 Adj. R-squared 0.765 0.755 0.790 0.842 0.834 0.857 0.861 Clustered by firm id. *** p<0.01, ** p<0.05, * p<0.1. Dependent variable is the share of FX credits in firm-bank outstanding credit balance. Standardized coefficients are reported, showing the effect of one standard deviation from mean. 12
16. How Do Credits Dollarize ? Appendix Variable descriptions and summary statistics of all the variables considered in the study are presented in Table A1. Full set of estimation results from the main specification can be found in Table A2.
17. FX Liabilities / Liabilities FX Deposits/ Liabilities FX HH Deposits/ Liabilities Non-Core FX Liabilities/Liabilitis Log of Assets Short-Term Assets (less than 1 year)/ Assets Liabilities / Assets Operating Profits/Assets NPL/(Performing Loans + NPL) 2,965 2,965 2,965 2,965 2,965 2,965 2,965 2,965 2,965 48,359 48,359 48,345 48,359 48,349 48,340 2,503,040 2,334,267 2,334,267 2,334,267 2,334,267 Obs 0.45 0.28 0.16 0.17 16.84 0.06 0.88 0.01 0.04 0.22 16.94 0.60 0.65 0.07 0.29 0.35 1.22 0.98 0.16 0.16 Mean 0.11 0.09 0.07 0.09 1.46 0.05 0.03 0.01 0.02 0.29 1.43 0.23 0.20 0.09 0.18 0.45 0.44 0.03 0.32 0.33 Std. Dev. 0.15 0.10 0.00 0.00 13.06 0.00 0.69 -0.02 0.00 0.00 8.87 0.00 0.00 -0.93 0.00 0.00 1.00 0.26 0.00 0.00 Min 0.87 0.86 0.69 0.66 19.68 0.34 0.96 0.06 0.17 1.00 23.75 1.97 1.00 0.89 1.00 1.00 10.00 1.00 1.00 1.00 Max Effective Fed Funds Rate Increaset Equals to 1 if ∆>5 basis points, otherwise 0 122 0.03 0.18 0.00 1.00 Positive GDP Growth Ratest Equals to 1 if real GDP growth is positive, otherwise 0 122 0.88 0.33 0.00 1.00 Exchange Rate Vollatilityt Variance of changes in exchange rate in the last 12 months 122 0.09 0.09 0.01 0.36 Time span of the dependent variable is 2006m11-2016m12. Firm variables’ time span is 2015 - 2015 and they are lagged for one year (y). Bank variables’ (and firm-bank level variables) time span is 2006m11-2016m11 and they are lagged for one month (m). Macro variables cover the period of 2006m11-2016m12. Non-core FX liabilities of banks cover all bank liabilities, excluding deposits. Deposits include firm and household deposits and household (HH) deposits include only household deposits. NPL stands for non-performing loans. Exchange rate is defined as the basket of 0.3*Euro+0.7*USD. Macro Variables FX Liability Sharet−1m FX Deposit Sharet−1m FX Household Deposit Sharet−1m Non-Core FX Sharet−1m Assetst−1m Liquidityt−1m Leveraget−1m Profitabilityt−1m NPL Ratiot−1m Bank Variables Export Share (Natural Hedge)t−1y Assetst−1y Leveraget−1y Liquidityt−1y Profitabilityt−1y Capital Intensityt−1y Export Sales / Sales Log of Assets Liabilities / Assets Short-Term Assets (less than 1 year)/ Assets (ROA) Operating Profits/Assets Gross Fixed Assets/Assets FX Credits/ Credits Number sectors that firm i received credit from bank j Share of biggest financing sector that firm i received from bank j Credits with medium-term original maturity/Credits Credits with short-term original maturity/Credits Credit Dollarizationt Number of Financing Sectort−1m Biggest Financing Sector Sharet−1m Original Maturity (12-24 Months)t−1m Original Maturity (24+ Monthst−1m Firm Variables Definitions Firm - Bank Credit Relation Variables Table A1: Variable Descriptions and Summary Statistics How Do Credits Dollarize?
18. How Do Credits Dollarize ? Table A2: Main Results (Full Version) Main Results 1 VARIABLES Baseline A Export Shareit−1y 0.0793*** (0.0107) 0.168*** (0.0164) FX Liability Shareit−1m B FX Deposit Shareit−1m C Non-Core FX Shareit−1m 2 Bank Liability Breakdown 3 X Fed Rate Increase 0.0793*** (0.0107) 0.0930*** (0.0207) 0.208*** (0.0196) Robustness 4 X Positive Growth 5 X Exchange Rate Volatility 6 At Least Once FX 7 Exported 50% or More 8 Only HH FX Deposits Quarterly 0.0796*** (0.0106) 0.0683*** (0.0118) 0.0836*** (0.0110) 0.0822*** (0.0111) 0.0923*** (0.0123) 0.0794*** (0.0107) 0.0788*** (0.0113) 0.0944*** (0.0207) 0.211*** (0.0196) 0.0520* (0.0286) 0.143*** (0.0257) 0.0835*** (0.0214) 0.227*** (0.0212) 0.101*** (0.0231) 0.231*** (0.0219) 0.0941*** (0.0281) 0.252*** (0.0273) -0.00885 (0.00540) -0.0198 (0.0163) -0.0792*** (0.0147) 0.0125** (0.00578) 0.0458** (0.0201) 0.0748*** (0.0198) -0.0476* (0.0265) 0.0929 (0.0916) -0.205** (0.0815) FX Household Deposit Shareit−1m A X Macroit B X Macroit C X Macroit Firm Controls Log of Assetsit−1y Leverageit−1y Liquidityit−1y Profitabilityit−1y Capital Intensityit−1y Bank Controls Log of Assetsit−1m Leverageit−1m Liquidityit−1m Profitabilityit−1m NPL Ratioit−1m Firm-Bank Credit Relation Share of Medium-term Creditsit−1m Share of Long-Term Creditsit−1m Number of Financing Sectorit−1m Biggest Financing Sector Shareit−1m FirmXBank FE Time FE Observations Ad. R-squared 0.198*** (0.0190) 0.126*** (0.0358) 9 0.0914*** (0.0226) 0.208*** (0.0214) 0.0258*** (0.00426) 0.0170 (0.0110) 0.00450 (0.0184) -0.0157 (0.0149) 0.0381* (0.0198) 0.0257*** (0.00426) 0.0170 (0.0110) 0.00444 (0.0184) -0.0160 (0.0149) 0.0379* (0.0198) 0.0257*** (0.00426) 0.0170 (0.0110) 0.00444 (0.0184) -0.0159 (0.0149) 0.0379* (0.0198) 0.0258*** (0.00426) 0.0171 (0.0110) 0.00469 (0.0184) -0.0156 (0.0149) 0.0381* (0.0198) 0.0258*** (0.00426) 0.0171 (0.0110) 0.00474 (0.0184) -0.0151 (0.0149) 0.0382* (0.0198) 0.0301*** (0.00500) 0.0211 (0.0129) 0.00256 (0.0209) -0.0124 (0.0171) 0.0414* (0.0225) 0.0301*** (0.00660) 0.0191 (0.0167) 0.0126 (0.0264) -0.00684 (0.0209) 0.0602** (0.0283) 0.0258*** (0.00426) 0.0169 (0.0110) 0.00432 (0.0184) -0.0161 (0.0149) 0.0378* (0.0198) 0.0259*** (0.00441) 0.0148 (0.0115) 0.00345 (0.0191) -0.0144 (0.0157) 0.0411** (0.0207) -0.0156** (0.00747) 0.417*** (0.0569) -0.106*** (0.0380) -0.427*** (0.119) 0.782*** (0.0820) -0.0215*** (0.00741) 0.383*** (0.0568) -0.0878** (0.0378) -0.363*** (0.119) 0.747*** (0.0813) -0.0214*** (0.00741) 0.384*** (0.0568) -0.0891** (0.0378) -0.374*** (0.119) 0.749*** (0.0813) -0.0214*** (0.00741) 0.370*** (0.0571) -0.0945** (0.0375) -0.364*** (0.119) 0.753*** (0.0812) -0.0213*** (0.00741) 0.379*** (0.0568) -0.107*** (0.0376) -0.373*** (0.119) 0.735*** (0.0811) -0.0233*** (0.00812) 0.423*** (0.0633) -0.0891** (0.0402) -0.389*** (0.134) 0.856*** (0.0902) -0.0167 (0.0102) 0.411*** (0.0786) -0.139*** (0.0536) -0.402** (0.169) 0.939*** (0.110) -0.0230*** (0.00750) 0.396*** (0.0573) -0.0856** (0.0378) -0.373*** (0.120) 0.720*** (0.0814) -0.0200** (0.00803) 0.330*** (0.0626) -0.0911** (0.0388) -0.441*** (0.136) 0.695*** (0.0866) 0.0542*** (0.00348) 0.0880*** (0.00479) 0.156*** (0.00317) 0.536*** (0.0242) 0.0545*** (0.00348) 0.0880*** (0.00479) 0.156*** (0.00317) 0.538*** (0.0243) 0.0545*** (0.00348) 0.0880*** (0.00479) 0.156*** (0.00317) 0.538*** (0.0243) 0.0545*** (0.00348) 0.0881*** (0.00479) 0.156*** (0.00317) 0.538*** (0.0243) 0.0545*** (0.00348) 0.0880*** (0.00479) 0.156*** (0.00317) 0.538*** (0.0242) 0.0625*** (0.00396) 0.101*** (0.00545) 0.158*** (0.00320) 0.552*** (0.0278) 0.0610*** (0.00503) 0.0906*** (0.00711) 0.153*** (0.00398) 0.542*** (0.0388) 0.0547*** (0.00348) 0.0880*** (0.00479) 0.156*** (0.00317) 0.539*** (0.0243) 0.0502*** (0.00330) 0.0835*** (0.00464) 0.122*** (0.00294) 0.425*** (0.0246) YES YES 2,331,834 0.786 YES YES 2,331,834 0.786 YES YES 2,331,834 0.786 YES YES 2,331,834 0.786 YES YES 2,331,834 0.786 YES YES 2,004,952 0.763 YES YES 1,322,492 0.752 YES YES 2,331,834 0.786 YES YES 711,721 0.788 Clustered by firm id. *** p<0.01, ** p<0.05, * p<0.1. Dependent variable is the share of FX credits in firm-bank outstanding credit balance. Effective Fed funds rate increase is equal to one for increases (5 bp and more), otherwise zero; Positive Real Economic Growth is equal to one for positive GDP growth, otherwise, zero and finally, Exchange Rate Volatility is in levels. “At least once FX” refers to firms that received FX credit at least once and “Exported at least 50 % or more” refers to firms that exported at least 50% or more of the time during the sample period. “Quarterly” presents the results with quarterly data.
19. Central Bank of the Republic of Turkey Recent Working Papers The complete list of Working Paper series can be found at Bank ’s website (http://www.tcmb.gov.tr) Hidden Reserves as an Alternative Channel of Firm Finance in a Major Developing Economy (İbrahim Yarba Working Paper No. 19/36, December 2019) Interaction of Monetary and Fiscal Policies in Turkey (Tayyar Büyükbaşaran, Cem Çebi, Erdal Yılmaz Working Paper No. 19/35, December 2019) Cyclically Adjusted Current Account Balance of Turkey (Okan Eren, Gülnihal Tüzün Working Paper No. 19/34, December 2019) Term Premium in Turkish Lira Interest Rates (Halil İbrahim Aydın, Özgür Özel Working Paper No. 19/33, December 2019) Decomposing Uncertainty in Turkey into Its Determinants (Emine Meltem Baştan, Ümit Özlale Working Paper No. 19/32, December 2019) Demographic Transition and Inflation in Emerging Economies (M. Koray Kalafatcılar, M. Utku Özmen Working Paper No. 19/31, December 2019) Facts on Business Dynamism in Turkey (Ufuk Akçiğit, Yusuf Emre Akgündüz, Seyit Mümin Cılasun, Elif Özcan Tok, Fatih Yılmaz Working Paper No. 19/30, September 2019) Monitoring and Forecasting Cyclical Dynamics in Bank Credits: Evidence from Turkish Banking Sector (Mehmet Selman Çolak, İbrahim Ethem Güney, Ahmet Şenol, Muhammed Hasan Yılmaz Working Paper No. 19/29, September 2019) Intraday Volume-Volatility Nexus in the FX Markets: Evidence from an Emerging Market (Süleyman Serdengeçti, Ahmet Şensoy Working Paper No. 19/28, September 2019) Is There Asymmetry between GDP and Labor Market Variables in Turkey under Okun’s Law? (Evren Erdoğan Coşar, Ayşe Arzu Yavuz Working Paper No. 19/27, September 2019) Composing High-Frequency Financial Conditions Index and Implications for Economic Activity (Abdullah Kazdal, Halil İbrahim Korkmaz, Muhammed Hasan Yılmaz Working Paper No. 19/26, September 2019) A Bayesian VAR Approach to Short-Term Inflation Forecasting (Fethi Öğünç Working Paper No. 19/25, August 2019) Foreign Currency Debt and the Exchange Rate Pass-Through (Salih Fendoğlu, Mehmet Selman Çolak, Yavuz Selim Hacıhasanoğlu Working Paper No. 19/24, August 2019) Two and a Half Million Syrian Refugees, Tasks and Capital Intensity (Yusuf Emre Akgündüz, Huzeyfe Torun Working Paper No. 19/23, August 2019) Estimates of Exchange Rate Pass-through with Product-level Data (Yusuf Emre Akgündüz, Emine Meltem Baştan, Ufuk Demiroğlu, Semih Tümen Working Paper No. 19/22, August 2019) Skill-Biased Occupation Growth (Orhun Sevinç Working Paper No. 19/21, August 2019) Impact of Minimum Wages on Exporters: Evidence From a Sharp Minimum Wage Increase in Turkey (Yusuf Emre Akgündüz, Altan Aldan, Yusuf Kenan Bağır, Huzeyfe Torun Working Paper No. 19/20, August 2019) An Analysis to Detect Exuberance and Implosion in Regional House Prices in Turkey (Evren Ceritoğlu, Seyit Mümin Cılasun, Ufuk Demiroğlu and Aytül Ganioğlu Working Paper No. 19/19, August 2019) A Trade-Based Misallocation Index (Orhun Sevinç Working Paper No. 19/18, August 2019) Invoicing Currency, Exchange Rate Pass-through and Value-Added Trade: An Emerging Country Case (Hülya Saygılı Working Paper No. 19/17, August 2019)