Bank Lending and Maturity: The Anatomy of the Transmission of Monetary Policy

Transcription

1. Bank Lending and Maturity : the Anatomy of the Transmission of Monetary Policy Selva Bahar BAZİKİ Tanju ÇAPACIOĞLU March 2020 Working Paper No: 20/05
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. Bank Lending and Maturity : the Anatomy of the Transmission of Monetary Policy Selva Bahar Baziki∗ Tanju C ¸ apacıo˘glu∗ Abstract We study the effects of monetary policy decisions on banks’ loan issuance and maturity decisions using a unique matched firm-bank-loan level granular database. We find that changes in the policy rate impact both credit and maturity channels - an increase of 100 basis points reduces commercial loan volumes by 1.6% and maturities by 1.2%, with tighter monetary policy having a larger effect on both. Small banks, banks with relatively weaker capital and liquidity structures, and with weaker access to foreign funding are more sensitive to policy changes. Bank ownership types and loan currency denomination also create asymmetries in responses. Banks reflect these changes to firms with which they have longer established relationships or which have a healthier past credit performance to a lesser extent. A quasi-experimental analysis adds that the intense use of a collateral guarantee scheme has increased maturities at the time of tight monetary policy stance, reversing their long-run negative relationship. These results highlight the importance of the financial regulatory process on banks’ risk taking behavior, search-for yield appetites, identifying areas of potential systemic risk buildup, and finally policy design and coordination. ¨ Ozet Bu c¸alıs¸mada, firma-banka-kredi bazlı mikro veriler kullanılarak para politikasının bankaların kredi arzındaki miktar ve vade tercihleri u¨ zerindeki etkileri incelenmis¸tir. Bulgular politika faizindeki de˘gis¸imlerin kredilerin miktar ve vadeleri u¨ zerinde anlamlı bir etkiye sahip oldu˘gunu ima etmekte olup, sıkılas¸ma d¨onemlerinde daha belirgin olmakla birlikte politika faizindeki 100 baz puanlık bir artıs¸ kredilerin miktarını y¨uzde 1,6 d¨us¸u¨ rmekte, vadesini ise y¨uzde 1,2 kısaltmaktadır. K¨uc¸u¨ k o¨ lc¸ekli, g¨orece daha zayıf likidite ve sermaye yapısına sahip bankalar ile dıs¸ finansmana eris¸im kabiliyeti g¨orece daha sınırlı bankaların politika faizindeki de˘gis¸ikliklere daha duyarlı oldu˘gu bulgulanmaktadır. Ayrıca banka t¨uru¨ ile kredinin para birimi de miktar ve vade u¨ zerinde asimetrik ∗ Central Bank of the Republic of Turkey, Banking and Financial Institutions Department, Macro Financial Analysis Division, Istiklal Caddesi 10, 06100, Ulus, Ankara, Turkey. Phone: (+90) 312 507 5777. Fax: (+90) 312 507 5874. email: firstname.lastname@tcmb.gov.tr. The authors thank the anonymous referee and the Editor for their valuable suggestions and the participants of 2018 Econometric Society North American Summer Meeting and 2019 American Economic Association/ASSA Congress for comments. The views expressed in this paper are those of the authors and do not necessarily represent the official views of the Central Bank of the Republic of Turkey. 1
4. etkilerde bulunabilmektedir . Sonuc¸lar, bankaların politika faizindeki de˘gis¸iklikleri g¨uc¸l¨u ilis¸ki g¨uc¨une sahip oldukları ve daha sa˘glıklı bir kredi gec¸mis¸i olan firmalara daha sınırlı o¨ lc¸u¨ de yansıttı˘gına is¸aret etmektedir. Kredi Garanti Fonu kefaletli kredilerin yo˘gun olarak kullanıldı˘gı d¨oneme dair analizler ise sıkı para politikası durus¸una ra˘gmen bu d¨onemde vadelerin uzun d¨onem ilis¸kisinden farklılas¸arak uzadı˘gını g¨ostermektedir. Bulgular finansal politika yapıcılı˘gın bankaların risk ve getiri odaklı yatırım is¸tahı, sistemik risk birikimi potansiyeline sahip alanların tespiti ile politika dizaynı ve koordinasyonu ac¸ısından o¨ nemini ortaya koymaktadır. JEL Codes: E51, E58, G20, G21, G28 Keywords: Monetary Policy, Transmission Channel, Credit Guarantee Fund, Loan Maturity, Bank Type 2
5. Non-Technical Summary The transmission of monetary policy decisions onto market interest rates constitutes a crucial part of the monetary policy framework . Banks may reflect changes in policy to lending and borrowing rates, but their response can also be multidimensional; as they alter other factors that compose credit conditions such as credit volume or maturity. Thus, the identification of channels and factors that compose this transmission mechanism is of great interest to policy makers. In the aftermath of the global financial crisis, in addition to the use of unconventional monetary policies, many governments have implemented additional tools to increase firms’ access to credit through supportive fiscal measures as well as macroprudential instruments aimed at increasing the long-run resilience of their financial systems. These mixed policy frameworks create a new complex environment which intensifies the question on how the monetary policy transmission mechanism works and interacts with the policies targeting financial stability such as macroprudential instruments or supportive fiscal measures. Using a bank-firm-loan level granular micro-data from Turkey, we document the anatomy of the monetary policy transmission mechanism and its interaction with supportive policies. We find that changes in the policy rate do not only impact volume but also maturity, suggesting the existence of the maturity channel of monetary policy in the transmission mechanism in Turkey. In other words, in response to an increase in the policy rate, both loan amounts are reduced and maturities shorten. An increase of 100 basis points reduces commercial loan amounts by 1.6% and maturities by 1.2%. The reduction in amounts is larger for longer-term borrowing, as a 100 basis points increase lowers long-term borrowing by 3.3% and short term by 1.0%. While both TL and FX denominated loans are affected by the change in the policy rate, the credit channel effect is larger for TL-denominated loans, and maturity effect is larger for FX-denominated loans. Moreover, results suggest that bank heterogeneities play a role in the transmission mechanism; both small and large banks exhibit reactions in the same direction, but smaller banks are more sensitive to the changes in the policy rate. Bank types also have an impact, with heterogeneous responses across; participation banks and private domestic banks are more sensitive than state or foreign private banks in maturities. Banks responses are asymmetric in terms of the monetary policy decision direction as responses are larger in periods of tightening monetary policy compared to easing monetary policy. Banks with relatively stronger capital structures, stronger liquidity structures, and stronger access to foreign funding reduce loan volume and maturities by smaller amounts. Banks reflect these changes by lesser magnitudes to firms with which they have longer established relationships and by larger magnitudes to riskier firms. In addition, we find that the during the initial phase of the treasury supported expansionary policy, the long-run relationship between interest rates and maturities present in the monetary policy transmission mechanism has reversed, with maturities lengthening during a period of tight monetary policy stance. This finding is stronger for banks that have a larger share of scheme loans. 3
6. 1 Introduction A vital part of the monetary policy transmission mechanism is the reflection of changes in the monetary policy rate onto market interest rates . From the perspective of banks, however, the response of financial conditions to monetary policy is multidimensional; they may reflect changes in policy to lending and borrowing rates, but can also chose to change other credit conditions such as credit volume or maturity. In the aftermath of the global financial crisis, in addition to (un)conventional monetary policies, many governments implemented tools to increase firms’ access to credit through supportive fiscal measures in an effort to ease macroeconomic pressures, and macroprudential instruments aimed at increasing the long-run resilience of their financial systems. In the presence of such a multitude of policy changes - monetary, fiscal and macroprudential - the transmission mechanism becomes even more complex, as interactions between these policies may influence market and credit conditions in ways unintended and unforseen by monetary policy decisions. It remains to be investigated if this policy mix creates complementarities that affect monetary policy transmission through changes in banks’ preferences in credit conditions; through less volume or shorter maturity reflected onto borrowers. In this paper, we use a bank-firm-loan level granular micro-data from Turkey, a large emerging economy, with universal coverage which allows us to disentangle demand and supply side factors in investigating these channels. We first document the presence of the maturity channel in addition to credit volume in the transmission mechanism. Then we investigate the effects of bank heterogeneity on this outcome. Finally, in a quasi-experimental setting, we focus on the effects of a widely used collateral guarantee scheme (a policy supporting commercial loan usage through collateral guarantees backed by the Treasury) to investigate how banks’ responses to this expansionary credit policy impacted the transmission of monetary policy. The setup of the scheme, which provided collateral guaranteed loans for longer maturities during a tight monetary policy stance at the time, which presents us with a unique opportunity to test the responsiveness of the transmission mechanism to both policies. Up until the most recent episode, during periods of monetary policy tightening, maturities of new loans issued by Turkish banks have had a negative relationship with policy rates. Figure 1 shows that volume-weighted average maturities in newly issued loans decline in response to higher rates. In other words, during periods of tight monetary policy, or when the policy rate is relatively high, banks reduce the amount dedicated to longer-term loans relatively more. This may be motivated by increasing risk concerns by banks in their maturity transformation at a time of high borrowing costs, as well as changes in consumers’ and firms’ preferences. 4
7. Following higher interest rates , the expected decline in the economic activity - in addition to the higher borrowing costs - may deter both consumption and investment decisions (Matsuyama, 2007). More formally, while the monetary policy rate determines banks’ financing costs (Stein, 1998), the rate also has real effects as it has an impact on both credit supply (Bernanke and Gertler, 1995) and on credit demand by affecting savings and investment decisions of agents (Bernanke and Gertler, 1989; Gilchrist and Zakrajˇsek, 2007). An important problem in the literature is the identification of these two forces; for instance, do changes in credit vary due to for instance consumers’ response to higher borrowing rates (demand channel) or because banks choose to lend less (supply channel). We overcome this problem with the rich nature of the micro-data, and employ controls for demand and supply side factors in the analysis by focusing on firms that work with at least two banks at the same time or by employing controls for bank, firm, time, and currency type in line with the literature. Another component of banks’ risk preferences may be reflected in the degree to which they require and grade collateral. As put forth in Bernanke et al (1996), in the presence of asymmetric information between the borrower and the lender, contracts are based on the net worth of the borrower, and as net worth decreases during economic downturns, the need for and the cost of external financing for the firm increases even more. On the flipside, Jim´enez et al (2014) show that during expansionary periods banks with lower levels of capital to asset ratio issue more credit with less collateral requirements, propagating the risk taking channel. As such, in this setting, a governmental policy guaranteeing collateral for mainly targeting SMEs which are collateral constrained, allows for a chance to observe bank decisions on loan amount and maturities if the risks associated with collateral were insured. Figure 1 Panel A shows that the negative relationship between the policy rate and maturities has disappeared with the intensive use of the collateral guarantee scheme, as the scheme alleviates some of the risk concerns of banks despite higher interest rates. Treating this policy as an exogenous shock to the financial intermediation sector, we will supplement the identification strategy which permits us to separate supply and demand channels, with a quasi-experimental approach around policy change dates that will allow for a causal interpretation of the effect of policies on outcomes of interest. We first investigate the monetary policy transmission mechanism by looking at how monetary policy affects lending decisions across heterogeneous banks. Changes in the central bank policy rate directly impact the local cost of funding for banks not only through their operations with the central bank, but also through the impact the policy rate has over their posted deposit rates.1 There is also a secondary effect in terms of access to bank lending 1 Deposits make up more than 50% of the funding channel of the banking sector. 5
8. from foreign sources ; changes in rates affect the real relative return of local assets and may impact banks’ access to cross-border funding. We find that monetary policy does not only influence new lending amounts but also maturities as well, documenting the presence of the maturity channel. As a result, easing or tightening decisions will affect banks’ commercial loan supply as well as maturity compositions, which should be considered in policy making decisions. In particular, we find that in response to a 100 basis points (bps) increase in the policy rate, commercial loan amounts and loan maturities decrease by 1.6% and 1.2% respectively. The reduction in amounts is larger for longer-term borrowing (A 100bps increase lowers long-term borrowing by 3.3% and short term by 1.0%) which corroborates this finding. The volume*maturity composite factor, which we utilize as a robustness measure of total credit supply in the style of Black and Rosen (2016), declines by 9.5% again in response to an increase of 100bps in the policy rate.2 We find that there are substantial differences between loans issued in different currencies. While both Turkish Lira (TL) and FX denominated loans are affected by the change in the policy rate, the credit channel is larger for TL denominated loans (45% of all corporate loans) while the maturity channel is larger for FX denominated loans (which make up the remaining 55% of corporate loans). FX denominated loans have on average higher maturities, and as such this result ties in with the finding that loans with longer maturities are more sensitive to changes in the policy rate. Bank heterogeneities play a significant role in the utilization of both channels. Both small and large banks exhibit reactions in the same direction, but smaller banks are more sensitive to the changes in the policy rate.3 This makes sense as these banks are less likely to have access to other sources of funding such as eurobond issuance, consortium lending, and have a smaller deposit base. While the transmission mechanism works across all bank types, state and domestic private banks show relatively higher sensitivities than participation and foreign banks in the credit channel, whereas participation and domestic private banks are more sensitive in the maturity channel.4 Banks with weaker capital structures, weaker liquidity structures, and with weaker access to foreign funding reduce loan amounts and maturities by larger amounts (controlling for demand). The type of policy change within the scope of the macroeconomic outlook also have an effect. 2 Jim´ enez et al (2014) find that this should be the lower bound effect, without taking the selection stage into account. 3 This is in line with Kashyap and Stein (1995 and 2000) who tie the result to their weaker access to wholesale funding avenues. 4 The term “participation banks” is used in Turkey to refer to banks that engage in Islamic banking. See Beck et al (2013) for a recent survey of the relevant literature. 6
9. As expected , volume and maturity responses are larger in periods of tightening compared to easing periods. However, there is an asymmetry across loan currency types in their sensitivity; loans issued in local currency show sensitivities in both channels compared to FX denominated loans which respond more in maturities rather than amounts. Given all these effects, banks do not behave uniformly as they exhibit asymmetries in how they reflect this behavior onto customers. Banks reflect these changes to firms with which they have longer established relationships or which have a healthier past credit performance to a lesser extent. Together with our finding that banks with weaker capital and liquidity structures are more sensitive to changes in the monetary policy stance, the results are harmonious with the findings of Jim´enez et al (2014) where they find banks with lower capital to asset ratios extend more loans to riskier firms in low interest rate environment. We add to their paper with our finding that the maturity channel is also at work. Next, we focus on the example of the collateral guarantee scheme and its impact on the monetary policy transmission mechanism. Gambacorta and Murcia (2017) find that macroprudential tools impact credit growth more when they are used in conjunction with monetary policy that support these policies in the same direction. In the same spirit, we investigate the effects of an expansionary fiscal policy at a time of the tight monetary policy stance. We find that the Credit Guarantee Fund (CGF) scheme which has supported loan growth through collateral guaranteed long-term loans has reversed the long-run relationship; despite the tight monetary policy stance, maturities are lengthening.5 This is especially true for banks that have a larger exposure to CGF supported credit. Previous work on the issue has established the existence of the maturity channel in aggregate data. Bernanke and Blinder (1992) empirically document the link between monetary policy and the volume of loans and deposits. Diamond and Rajan’s (2006) model shows that since the funding avenues are short-term, longer term loans have a higher level of sensitivity to monetary policy decisions. In recent related work, Black and Rosen (2016) build on Diamond and Rajan’s (2006) model to investigate the maturity channel. They use quarterly survey data from the US on commitment and spot lending and find that a 1 percentage point increase in the Fed funds rate reduces lending by 8% and maturities by 3.3%. Our results highlight that in an emerging market setting, credit and maturity channels are equally important, though to a lesser degree in magnitude. Expansionary monetary policy may lead to riskier bank balance sheets through the risk taking channel, as the bank seeks to create higher returns in a low-interest-rate setting (Gambacorta, 5 This finding is in line with Buyukbasaran et al (2019) find that monetary and fiscal policies move in opposite directions in response to policy shocks. 7
10. 2009 ).6 The bank can increase its lending to riskier agents, or it can satiate its desire for riskier lending by increasing the maturity of the loan and elongating the maturity mismatch between its assets and liabilities. In other words, the risk taking channel need not necessarily work through credit risk, it can also work through the exposure to a longer maturity. Our findings add to the literature by showing that during easing periods, an increase in loan volume, i.e a larger exposure, occurs concurrently with longer maturities, a longer exposure.7 We add to the current literature in several dimensions. On the data front, in addition to the clean identification to disentangle supply and demand forces, the granular aspect of the micro-data allows us to examine the effects on loan amount and maturity separately. We are also able to observe the differences in impact across loans denominated in local and foreign currencies, and find that local currency loans have higher responsiveness. We also shed light on the impact of bank characteristics on the transmission mechanism such as capital and asset structures, access to foreign funding, and bank’s ownership type. These bank-level variables have been studied, separately or together, in the literature (Kashyap and Stein, 1995; Gambacorta and Mistrulli, 2004; Jim´enez et al, 2012; Kashyap et al, 1993; Iannotta et al, 2013) but we add to their contributions through the use of micro-level data in analyses that control for the demand side. Finally, we are able to perform a quasi-experimental examination of how a widely used collateral guarantee scheme affects this transmission mechanism. An expansion on the credit channel through the guarantee scheme at the time of tight monetary policy stance provides strong evidence for a discussion of optimal policy design and coordination (Beyer et al, 2017). Normally banks could be expected to prefer firms with stronger credit histories or higher networth (Bernanke et al, 1996; Bernanke and Gertler, 1989) with whom they have a stronger established relationship (Petersen and Rajan, 1994). But if restricted by a policy, the bank could then choose to lend to riskier agents with a search-for-yield motivation. Our results also shed light on this matter and show that relationship banking remains strong even in the transmission mechanism while banks reflect changes in maturities by larger magnitudes to riskier firms. This lends support to the presence of the risk taking channel of monetary policy. 6 Borio and Zhu (2012) find that changes in monetary policy rates may impact agent’s risk appetite or assessment through the risk taking channel which then can impact asset valuations, investor’s search for yield motivations (Rajan, 2006) or habit formation (Campbell and Cochrane, 1999; Longstaff and Schwartz, 1995; Collin-Dufresne et al., 2001). 7 All else equal, a longer maturity could make it possible for the borrower to reduce their monthly debt payments to more affordable levels, hence increasing the probability that the loan would be paid back in full. This improvement in debt service capacity however does not mitigate the term risk associated with longer maturities, the mere existence of a term premium being the prime proof. The fact that maturities shorten in tighter monetary policy periods attests to the fact that the risk factor of longer maturities are more pronounced for banks. For more on the disentanglement of supply and demand side effects in this regard, see Section 5. 8
11. The remainder of the paper is organized as follows . Section 2 summarizes the CGF case study setup, Section 3 summarizes the data we use, Section 4 lays down the empirical strategy, Section 5 presents and discusses the results, and finally Section 6 concludes. 2 The Credit Guarantee Fund Following several external and internal shocks to the geopolitical sphere by mid-2016 and the resulting easing in the economy, several supportive fiscal and financial measures were put in place starting from end-2016 to mid-2017 to increase the credit access of the corporate sector. Most of these measures targeted SME credit access through the use of low-interest loan facilities (TOBB, about TL7.5 billion) as well as some interest-free loans (KOSGEB, about TL11 billion).8 The Credit Guarantee Fund, CGF, backed by the Treasury, launched a largescale TL250 billion collateral support scheme in which SMEs with insufficient collateral would receive a collateral guarantee from the CGF in support of their bank loan application. The program targets lack of adequate collateral, one of the main issues SMEs face in accessing external financing. SMEs do not have access to capital markets and external funding, nor established relationships with financial intermediaries to the same extent as large corporations do, which in turn has real outcomes in terms of firm financials, employment outcomes, and economic growth (Gertler and Gilchrist, 1993; Beck et al, 2005; Beck and Demirg¨uc¸Kunt, 2006). The lack of potential financing avenues makes SMEs more reliant on bank intermediation, making them relatively more sensitive to monetary policy decisions (Gertler and Gilchrist, 1994; Lang and Nakamura, 1995). While the wide network and the accumulated know-how within the banking sector aides SMEs in closing their financing gap, the required posting of collateral to access bank loans could be a further constraint for firms without adequate collateral. Indeed, collateral is listed as a decisive factor in the Bank Loans Tendency Surveys on the approval of a loan, and is a leading constraint in SME access to finance in international surveys (The EBCI Vienna Initiative, 2014; and OECD, 2013). This collateral gap acts as a financial friction for SMEs through two avenues; first, inadequate coverage rates of the SME’s existing collateral may prevent the SME from accessing external financing in the first place. Furthermore, since SMEs have higher levels of information asymmetries and lower established intermediation relationships, they may be facing more 8 TOBB, The Union of Chambers and Commodity Exchanges of Turkey provided this additional facility for TL5 billion which was expanded to TL7.5 billion in April 2018. Dubbed the ”breath credit”, the facility was only made available to its members for loans provided through participating banks with low interest rates (monthly 0,99%). KOSGEB, Small and Medium Enterprises Development Organization of Turkey, an organization related to the Ministry of Industry and Technology, has offered a credit interest support programme to cover all interest, commission and related expenses for loans. For more on these schemes, see the CBRT Financial Stability Report, 2017, TOBB and KOSGEB. 9
12. stringent collateralization requirements relative to more established firms (Berger and Udell 1990, 1995; and Berger et al, 2016)9 Secondly, lack of ample collateral hinders SMEs from rolling over existing firm debt, and periodic revaluation of collateral may require them to pledge additional collateral after the initial issuance of the loan. This constraint is especially binding during episodes of economic downturn when asset values are declining. Compared to more established corporate firms, this puts SMEs at a relative disadvantage; when the economy is slowing down and negatively affecting firm cash flows, firms increasingly need external financing (Gertler and Gilchrist, 1994) and banks have a tendency to switch to less riskier borrowers as the share of loans issued to smaller borrowers decline (in line with flight to quality in Bernanke et al, 1996; and Bernanke and Gertler, 1989). In this setting, the utilization of CGF schemes proves to be effective in overcoming the collateral friction and may have financial and economic additionality as well as economic spillovers conditional on the design and incentive mechanism instilled in the program (Beck et al, 2008; OECD, 2013; Levitsky, 1997). As such, many countries in Central and Eastern Europe have resorted to CGF schemes following economic or financial downturns as credit flow to SMEs declined due to supply-side factors (The EBCI Vienna Initiative, 2014). The recent CGF scheme in Turkey which mostly targets low collateralized SMEs reduces risks associated with SME lending, and provides an opportunity to investigate how the absence of collateral risk impacts bank lending appetite, loan conditions, and through both of these channels, how the transmission mechanism is impacted. Gozzi and Schmuckler (2016) suggest that as the CGF schemes assume risks associated with SME lending, loan conditions can be more favorable. Jim´enez et al (2014) state that in response to a rate fall, uncollateralized loans increase, as a sign of additional risk appetite by banks. Our paper aims to complement this finding by investigating if in a setting of reduced collateral risk, banks satiate their risk appetite by issuing longer term loans. This would be in line with the findings of Rahman et al (2019) who find that collateralization becomes increasingly important with the term of the loan. Internationally, CGF are not only used at the national level in over 70 nations but also by international global banks (Gozzi and Schmukler, 2016). Diverging from common practices elsewhere (Beck et al, 2010), the initial roll-out of the collateral guarantee scheme which we analyse in this paper did not bring any geographical or industry specific restrictions on 9 Steijvers and Voordeckers (2009) provides a review of the literature on collateral and information asymmetries in lending markets. The issue may be of even higher relevance in developing countries as Hanedar et al (2014) show that in developing countries firm-specifics play a large role in determining collateral requirements. The existing literature provides differing views on the effectiveness of collateral in financial intermediation and in mitigating financial risk. Notably Jim´enez and Saurina (2004) show that collateralized loans have higher probability of default. 10
13. recipients ; which gave banks freedom and complete control over which participants they chose into the scheme. The scheme in Turkey was rolled out through established financial intermediaries and introduced a 7% delinquency cap for the collateral guarantee portfolio. Both of these points aimed to improve the asset quality of the CGF portfolio: first by utilizing the sectoral know-how of the intermediaries in selecting eligible firms in with high potential into the system, while at the same minimizing moral hazard problems associated with agency usage with the introduction of the cap.10 In addition, through this aspect of the design, the risk assessment of the loan application was left to the financial intermediary rather than the CGF, which is found to lower default rates, and increases cost effectiveness (Beck et al 2010; Gozzi and Schmuckler, 2016). At the same time, by setting the limit at 7% while NPL ratios for the corporate sector and the SME portfolio in particular had been hovering below 3% and 5% respectively, the scheme allowed for some room for banks to increase their risk appetites and lending. Collateral guarantee coverage was also varied across firms, exporting firms were given 100% collateral guarantee on their loans, whereas other SME loans’ CGF support were capped between 85 to 90% of the value of the loan. This partial coverage limit, together with the cap on default rates, was put in the design to further incentivize banks to better screen applicants. Risk weights on the portion of the loans covered by the scheme were lowered to zero which is estimated to have helped the participating banks’ CAR valuations by 66bps which further stimulated loan growth (Banking Regulation and Supervision Authority, 2018, 2018).11 As put forth by Alper et al, 2016, there is a tight relationship between capital adequacy, profitability and loan growth in Turkish banks: new loan issuances will reduce CAR, the amount of which depends on the risk weights applied, and as such a bank’s loan growth is only sustainable as long as the profitability levels are high enough to support the capital base. Based on this argument which puts CAR as a constraint to high loan growth rates, any regulatory adjustments such as a reduction in risk weights which increases CARs will create additional lending room and appetite for banks, especially those closer to the regulatory lower limits. 10 Firms under liquidation, undergoing bankruptcy procedures, facing legal proceedings due to tax or social security debt, or debt to financial institutions were barred from applying, which further strengthened the asset quality of the porfolio. 11 The fund has been renewed for a total of four times since this analysis. The first renewal of 52,2 Billion TL incorporated some repayments from the initial phase which were issued out as new loans and also introduced region/topic specific limits, such limits set aside for female entrepreneurs. The second renewal in February 2018 used about 50 Billion TL of repayment as well as the 32,5 Billion TL enlargement of the fund, and the usage areas were roughly limited into three: working capital needs, investment, foreign trade which had a higher coverage limit (100%) than the remainder (80-85%). As of June 2018, 55 Billion TL of the 85 allocated for the year had been issued as limits. In the first quarter of 2019, two additional guarantees of 20 Billon TL, which amounted to 25 Billion TL of loan issuance each. To see more on the details of the fund, and other measures that were taken to support corporate credit growth around the same time, please see the CBRT Financial Stability Report (2017) and Baziki and C¸apacıo˘glu, (2020). 11
14. Furthermore , the focus on SMEs as well as its loan pool made up of mostly TL loans (85% of total) makes it a prime opportunity to study the effects of the transmission mechanism through the banking sector. 3 Data This paper looks at the transmission mechanism via the credit and maturity channels using a comprehensive credit register data from Turkey.12 With a financial system that is heavily dependent on the banking sector as more than 90% of non-financial lending is done through the banking sector, the setting is ideal to test theories on the credit channel using banking credit register data. In this environment, where very few firms, mostly large corporates, have access to equity financing or direct foreign borrowing, firms’ demand for loans are motivated by monetary policy decisions which directly influence the cost of funds for firms, rather than developments in other channels of funding. Following four years of implicit targeting, the CBRT became a full-fledged inflation targeting central bank from 2006 onwards. While the main lending channel of the CBRT is the weekly repo rate, we use the average of weekly and overnight short-term funding rate as our benchmark interest rate which reflects the direct cost of funding to banks due to motivations established in the literature. Illes et al (2015) identify average short term funding costs to be more relevant for loan rates compared to official policy rates in European countries, and Karagiannis et al (2010) confirm these findings for the Euro Area. Additionally, since 2010, the CBRT has modified its interest rate corridor to better address global volatility conditions through the use of an asymmetric corridor (Binici et al, 2019). This resulted in periods where prevalent market rates differed from the official monetary policy rates. Binici et al (2019) show effective market rates to be more relevant for the transmission of the monetary policy stance than the official rates during this time in Turkey.13 Some of the literature focuses on other forms of financing in the face of difficulties in intermediation. If the corporate sector had access to funding though another route outside of the domestic banking sector, either by issuing bonds or borrowing on the international market would dampen the effect of the transmission mechanism. Bond issuance is not a common practice for the majority of firms in our sample of 2,827,513 firms. As such, the particular setting of the Turkish economy makes it a suitable environment to study the effect of 12 In the credit registry loans that are less than 1000TL (≈ 175USD) are not reported individually but rather in a lump sum. These loans make up a very small fraction of the entire corporate loan porfolio. 13 Binici et al (2019) also point out that this is not an isolated incidence, as interbank rates also differ from official rates announced by the ECB or the Fed. We do not use the BIST O/N directly as the CBRT can influence this rate through market operations. 12
15. monetary policy on credit conditions . Table 1 shows the summary statistics of the monthly data we use in our analysis. In line with the literature, we define relationship between a firm and a bank as the ratio of the firm’s lending from that bank over its total lending for the past 12 months to reflect the intensity of their commercial relationship. In robustness checks, we also entertain the number of years the bank has had a working relationship with the firm, and thus has been able to observe its idiosyncratic changes such as balance sheet item changes in its profitability, employment, size as well as its performance through sector or economy wide shocks the firm has endured. While both approaches produce comparable results, we prefer the first approach as it also presents the bank’s assessment of the profitability of their relationship with more up-to-date information and intensity. We are able to observe all the lending relationships that exist in the banking system, which provides us with the ability to compare and contrast lending by differently exposed banks to the same firm at different points in time. Balance sheets and income statements of banks are available from the CBRT. Our matched dataset comprises of 30 banks. We use loan data from all banks in our sample, which includes private, state-owned, foreign-owned and participation banks and excludes investment banks and development banks, which may have a different business model aligned with risk-taking social welfare goals. At the bank level, we use data on real assets size, loan to assets ratio, deposits to assets ratio, the ratio of capital and liquid assets to total assets, non-performing loans ratio, return on assets ratio, and FX non-core ratio all on a monthly frequency from the CBRT. The definitions of the variables, data sources, and summary statistics are given in Table 1. Macro-economic aggregates in Turkey may affect demand and supply of consumer loans. Hence, we need to control for the business cycles and monetary policy stance in Turkey. This will allow us to better isolate changes in policy rates and implementation of schemes from other changes in economic activity or monetary conditions. At the macro level, we use data on industrial production index (as an indicator of economic activity), consumer price index (CPI), and the real effective exchange rate, all on a monthly frequency from the CBRT.14 The definitions of the variables, data sources, and summary statistics are given in Table 1. 4 Empirical Strategy The initial specification that we use to estimate the impact of monetary policy stance on bank lending behaviour is given in Equation 1. Later, we will build on this specification to 14 To match the monthly frequency of our data, we selected to use the industrial production index as an indicator of GDP, following Modugno et al (2016) among many others. 13
16. investigate the potential channels through which MP affects banks ’ credit supply decisions through both amount and maturity of loans. Our first set of analysis are built on the following equation; Loanb, f ,c,t =β0 + β1 MPt−1 + λMacrot−1 + γBankb,t−1 + ζBank Firmb, f ,t−1 + αb + η f + µc + θt + εb, f ,c,t (1) where the dependent variable Loanb, f ,c,t is the natural logarithmic value of the amount or maturity of bank b’s newly issued commercial loans to firm f at time t, measured monthly, and in currency type c, where c can stand for total loans in all currencies or those denominated in TL or FX separately, where applicable. The main variable we use to measure monetary policy stance, MP, is the CBRT bank funding rate weighted by lending amount to better capture the banking sector’s average funding cost through the central bank. While the CBRT sets an official weekly repo rate as the policy rate, during the period of our investigation, the Bank also used quantity restrictions on its lending due to the interest rate corridor policy or late liquidity window facility. Therefore, the announced policy rate by itself is not a sufficient measure to reflect the true nature of CBRT’s monetary stance. To control for domestic economic conditions and banks’ balance sheet characteristics, we include macro indicators, bank-specific observables and bank-firm match related variables with a one month lag to address endogeneity concerns. Macrot−1 stands for the macro indicators of Turkish economy at time t − 1: industrial production index (as an indicator of economic activity), inflation rate (consumer price index), and reel effective exchange rate, all on a monthly frequency sourced from the CBRT. Bankb,t−1 capture the balance sheet ratios of Turkish bank b at time t − 1 that may have an influence on the credit suply decisions of lender banks observed over the period of interest. At the bank level, we use data on real assets size, loan to assets ratio, deposit to assets ratio, the ratio of capital and liquid assets to total assets, non-performing loans ratio, return on assets ratio and FX non-core ratio. These ratios have monthly frequency and are again sourced from the CBRT. Bank Firmb, f ,t−1 includes the strength of relationship between lender bank b and borrower firm f at time t − 1.15 The definitions of the variables, data sources, and summary statistics are given in Table 1. Moreover, we saturate our model with bank, firm, currency type and time (year) fixed effects to control time-invariant and unobservable factors. In this context, we introduce fixed effects for bank b, αb ; for firm f , η f ; for currency type c, µc ; and finally for the year of time t, θt to capture any year-specific factors we have not been able to control for through our macro 15 When we add the riskiness indicator to the estimation equation, the number of observations decreases significantly due to the unbalanced structure of rating dataset of firms. Hence, the baseline model doesn’t include riskiness indicator of firms as an explanatory bank-firm variable. 14
17. observables . Next, we constrain our sample to firms borrowing from at least two different bank types of state, domestic private, foreign private and participation and create six different dummy variables to reflect these combinations. For example, Dummy1 is identified only for firms borrowing from state and participation banks, and equals zero for state banks and one for participation banks.16 To explore marginal effects of bank ownership structures on credit policies and risk taking practices as a result of monetary policy decisions, we include the interaction of each dummy variable with MP restricting our sample to firms having active loans from these two bank types. The empirical model builds on Equation 1 and is structured as follows: Loanb, f ,c,t =β0 + β1 MPt−1 + β2 DummyX + β3 (MP ∗ DummyX)t−1 + λMacrot−1 + γBankb,t−1 + ζBank Firmb, f ,t−1 + αb + η f ,t + µc + εb, f ,c,t (2) where DummyX represent six different dummy variables to group 4 different bank types in pairs as identified above. Moreover, we add firm-month fixed effects, η f ,t , to restrict our sample to firms borrowing from at least two different bank types in the same month. Henceforth, we focus on how the transmission mechanism differs across bank and bank-firm specifics. In order to get at this result, the separation and identification of supply and demand side factors become crucial. To this end, we control for demand side effects to focus on the supply side factors, and by adding firm-month fixed effects to our baseline specification, Equation 1, we focus on firms that borrow at least from two different banks in the same month. In this context, we select three main indicators for banks; capital ratio, liquidity ratio and FX non-core ratio, and we add the interaction of these variables with the policy rate in order to capture heterogeneities across banks in their policy responsiveness due to these observables (as in Jim´enez et al, 2012, 2014). Similarly, we add firm-month fixed effects, η f ,t , to restrict our sample to firms borrowing from at least two different banks in the same month to utilize the variation among balance sheet structures of lender banks (Khwaja and Mian, 2008). Namely, we estimate: Loanb, f ,c,t =β0 + β1 MPt−1 + β2 Ratiob,t−1 + β3 (MP ∗ Ratio)b,t−1 + λMacrot−1 + γBankb,t−1 + ζBank Firmb, f ,t−1 + αb + η f ,t + µc + εb, f ,c,t 16 The (3) rest of the dummies are for the following pairs, with the value zero for the first and one for the second bank type: Dummy 2 for domestic and participation, Dummy 3 for foreign and participation, Dummy 4 for state and domestic, Dummy 5 for state and foreign, and finally, Dummy 6 for domestic and foreign type banks. The interaction term in Table 7 then identifies the relative responsiveness of the type that has a dummy equals to one over the other. 15
18. where Ratiob ,t−1 are lender banks’ capital, liquidity and FX non-core ratios of lender bank b at time t − 1. The coefficient of interacted term, β3 , gives us the marginal effect of MP on the banks that lend to same firm in the same month. Moreover, MPt−1 and the macro variables, Macrot−1 , having monthly frequency are not identified due to the firm-month fixed effects in this specification. Next, we look at whether the lending preferences and the monetary transmission mechanism examined Equation 1 through 3 differ depending on the depth and duration of the relationship between the lender bank and the borrower firm or the riskiness/rating of firms. To do so, we focus on the variation in credit conditions across firms that work with the same bank in the same month but have varying degrees of relationship intensity or riskiness/rating. We add bank-month fixed effects to our baseline specification, Equation 1, and limit the sample to firms that borrow from the same bank at the same time. The empirical model is structured as follows: Loanb, f ,c,t =β0 + β1 MPt−1 + β2 Indicatorb, f ,t−1 + β3 (MP ∗ Indicator)b, f ,t−1 + λMacrot−1 + γBankb,t−1 + ζBank Firmb, f ,t−1 + αb,t + η f + µc + εb, f ,c,t (4) where Indicatorb, f ,t−1 are the strength of relationship between lender bank b and borrower firm f , or the riskiness/rating of firm f assigned by bank b at time t − 1. Again, bank-specific variables, Bankb,t−1 , and macro variables, Macrot−1 , will not be identified due to the bankmonth fixed effects in this model. Finally, we switch to policy analysis, and use the implementation of the CGF scheme as a case study to investigate its influence on the monetary transmission mechanism in terms of both lending amount and maturity. Therefore, we create a time specific dummy variable, and include the interaction of this dummy variable with the policy rate in our baseline specification, Equation 1. Next, for a window of 6 and 9-month period around the initial implementation date, we identify the marginal effects of CGF scheme on the transmission of monetary policy to treat the CGF scheme as a quasi-experiment. Our estimation equation resembles Equation 3, and is structured as follows: Loanb, f ,c,t =β0 + β1 MPt−1 + β2 A f ter + β3 (MP ∗ A f ter)t−1 + λMacrot−1 + γBankb,t−1 + ζBank Firmb, f ,t−1 + αb + η f ,t + µc + εb, f ,c,t (5) where A f ter is a dummy variable takes the value of 1 from January 2017 onwards. Our window frame of choice is 6 months, given the absence of other policy implementations that could interfere with the results. 16
19. We add to this analysis a further investigation of whether heterogeneities across banks in terms of CGF scheme exposure are associated with differences in terms of transmission of monetary policy . To this end, we create a bank-specific variable that equals the ratio of the banks’ CGF loans as a share of their total commercial loans to capture the banks’ exposure to CGF scheme. We replace our dummy variable with this ratio, CGFRatiob for each bank b, add the interaction of this variable with the policy rate, and through the inclusion of firmmonth fixed effects, η f ,t , we restrict our sample to banks that lend to the same firm in the same month to compare the differences in lending practices that are due to bank-level differences in CGF exposure. The coefficient of interacted variable, (MP ∗CGFRatio)b , will give us the marginal effects of monetary policy across banks that have lower versus higher exposure to CGF scheme. 5 Results We begin our analysis by examining whether changes in the monetary policy stance impact loan amounts and maturities, and if this response is uniform across different policy stances and types of loans. We then proceed to explore whether bank specifics and borrower side heterogeneities play a role in the degree of responses. Finally, we employ a diff-in-diff strategy around the implementation dates of the CGF policy to investigate if and how the change in the fiscal policy has impacted the rate of responses to monetary policy seen in the previous sections. 5.1 Lending and Maturity Channel To begin, we analyze how changes in the monetary policy rate, MP, is related to changes in commercial loan amounts and maturities. As an initial step, Table 2 presents the estimates of Equation 1 and shows that a higher policy rate is associated with lower loan amounts and shorter maturities, implying the existence of both credit and maturity channels. In columns 2 and 4, where macroeconomic indicators, bank and bank-firm interacted observables as well as bank, firm, currency type and year fixed effects are included show that a 100bps increase in the policy rate reduces commercial loan amounts by 1.6% and maturities by 1.2%. The reduction in commercial loan amounts in response to an increase in the policy rate is the expected outcome of monetary tightening; funds become scarce and more expensive for banks, which reduces the available loan base (Bernanke and Blinder, 1992; Stein, 1998; Black and Rosen, 2016). However, the mechanism does not only work through credit volume, but credit maturity is also shortened in response, further tightening credit conditions (Diamond and Rajan, 2006 and 2011). 17
20. In an effort to capture the impact of monetary policy on aggregate loan supply through changes in loan conditions , and to motivate our focus on the maturity channel, we also test the response of loan amount interacted with maturity to a change in monetary policy. Column 6 shows that in response to a 100bps increase in MP, this interaction term declines by 9.5%. In line with the findings in Black and Rosen (2016) who use the same indicator, the combined response of maturity and volume to monetary tightening indicates a stronger decline in credit supply than either of these factors alone. Together, the results from Table 2 suggest that the common practice of measuring the effect of monetary policy changes on credit supply solely through changes in credit volume is insufficient, and the maturity channel is indeed an important component of credit supply. Loan specifics can also play a role in the way the credit and maturity channels operate in response to changes in monetary policy. To investigate this, the first four columns of Table 3 focus on the effect of the credit channel on short and long term borrowing. Utilizing a full set of controls, Columns 2 and 4 show that a 100bps increase in the policy rate reduces the volume of long-term lending by 3.3%, whereas the reduction is only 1.0% for short-term loans. This asymmetry between the responsiveness of long and short term loans, can be motivated through changes in bank’s cost of funds and risk perceptions. All else equal, loans with longer maturities are regarded as carrying a higher level of risk, since the longer term increases the probability that the loan will not be paid back in full due to higher agency risk. With rising maturity, the bank may also factor in the potential depreciation of the value of the loan due to inflationary pressures, a potential decline in the net worth of the borrower, as well as a rising risk of lower revaluation of the collateral listed (Bernanke and Gertler, 1989). Due to these possible risks, banks may chose to opt out of funding long term investment projects (Matsuyama, 2007). This effect is magnified during a contractionary monetary policy phase, which increases banks’ cost of funds and lowers their access to liquidity, and as a result banks become relatively more keen to fund short-term investments (Black and Rosen, 2016; Diamond and Rajan, 2006 and 2011). The currency composition of lending is another factor that could play a role in the determination of maturity channel. If Turkish banks are funding their foreign-currency lending through foreign loans, then the domestic monetary policy decisions would have a larger impact on TL-denominated loans versus FX loans. To put this to test, we focus on the currency denomination of lending. Corporates’ financial liabilities to banks by currency denomination at the end of the analysis period is divided rather evenly, with 45% of loans denominated in TLs, 18
21. and 55 % in FX.17 This framework will allow us to examine the currency pass-through of monetary policy decisions in terms of responsiveness to monetary policy decisions. Results on Table 3 columns 5 through 12 indicate that monetary policy decisions have larger effects on volume in TL-denominated loans, and on maturities in FX-denominated loans. This ties in with the fact that FX-denominated loans are longer in maturities on average and longer maturity loans are more responsive to changes in the policy rate, as shown above. FX loans are more sensitive to changes in MP since they have longer maturities to begin with, a change in the policy rate increases their maturity risk relatively more. The results show that policy decisions influence long-term lending more than short-term lending, proving that the maturity channel of monetary policy has asymmetric effects. Black and Rosen (2016) find that the credit channel is more than twice as responsive as maturity in the US. In an emerging market setting, we find a relatively stronger response of the maturity channel, as both channels are equally important though to a lesser degree in magnitude. We interpret the reasons behind this difference in outcomes in the following two avenues. First, and more importantly, our data and econometric approach may have played a role in this difference: we use a more detailed loan-level database, and employ an econometric approach which allows us to separate supply and demand channels without having to make assumptions about bank observables, or whether loan issuances are supply or demand driven. Secondly, performing this exercise in an emerging economy may also have added to the relatively higher importance of the maturity channel: emerging economies on average have a higher inflation rate, which may increase the duration mismatch sensitivities of banks. Adding to this sensitivity is the relatively shorter average lifetime of firms in emerging economies, increasing the risk that loans with longer maturities will not be paid back in full (Ayyagari et al, 2013). 18 In fact, in inflationary settings, pricing uncertainties do not only concern the net present value of the loan amount, but also the valuation of the collateral posted.19 However, the collateral channel is not only a concern for emerging economies, in fact, faced with asset price fluctuations, banks in developed economies regularly revalue the collateral posted and may call for additional guarantees during periods of higher uncertainty. Finally, Table 4 tests whether the responses differ across periods of monetary tightening or 17 As of July 2017, corporates’ financial liabilities to banks added up to 799 billion TL in loans denominated in the local currency and 991 billion TL in FX loans, including FX-indexed loans. 18 Meanwhile, lifespan expectancy of even S&P 500 corporations are getting shorter at an increasing pace due to technological disruptions and increased M&A activity (Anthony et al, 2018). This may imply that maturity mismatch may become an even more pressing issue for the global financial sector in the future. 19 In fact, this collateral channel is yet another reason why the CGF backed collateral guarantee schemes has been successful in stimulating loan growth in the countries utilizing it, emerging and developed alike. 19
22. loosening . To this end, we re-run the same regression in Equation 1 for the two different regimes.20 Overall, we see that the monetary policy transmission mechanism works well in Turkey, with an increase(decrease) in MP lowering(increasing) both credit issued and the maturity of the credit.21 However, the effects on loan issuance and maturities are asymmetric across different monetary policy regimes. Comparing the results in columns 1 and 4, we see that tightening periods are at least twice as responsive in both factors; Turkish banks are very responsive in reflecting contractionary policy decisions onto credit conditions, but remain prudently sticky when faced with looser policy rates. This asymmetric response across regimes could be explained by higher agency cost issues in tighter periods which may reduce banks’ risk appetites (Black and Rosen, 2016; Borio and Zhu, 2012). 22 Under a full set of controls, Panel A shows that loan amounts are about twice as responsive to policy changes under tightening periods, and this is mainly driven by a higher level of sensitivity in loans issued in the local currency. Panel B summarizes the results for the maturity channel and shows that FX loans are equally and strongly responsive across regimes, but loans denominated in TL are about four times more responsive to contractionary changes in MP compared to expansionary changes. Across Sector-Wide Heterogeneity Aside from loan-related motivations, bank-specifics could also play a role in the transmission mechanism and path. To this end, we investigate whether bank heterogeneities across size, ownership-type, capitalization, liquidity and access to foreign funding play a role in the results discussed thus far. Table 5 shows the different ways credit and maturity channels work across bank sizes depending on balance sheet assets sizes. We divided the universe of banks into two groups; separating the largest 7 banks from the rest of the sector.23 The results show that the transmission mechanism specified above, the credit and maturity channels work for both small and large banks, however smaller banks are more sensitive in reflecting policy changes in both channels (Columns 2 and 6 versus 4 and 8). Previous literature which has also found smaller banks to be more responsive to changes 20 For ease of comparison across different regimes, we rerun the regression here using a dummy variable strategy in Table A2. This table also reports the estimates on comparisons across currency type, bank size, short versus long term, as well as the comparison of regimes under different loan currencies. 21 In the table, a change of 100bps under the loosening scenario creates movements in loan amounts and maturities with a negative sign. This negative sign for amount and maturity under loosening should be read as a movement in the opposite direction to the lower interest rate, and indicates higher amounts and longer maturities. 22 While we explore credit and maturity channels in this paper, Binici et al (2019) see a similar asymmetric pattern in the loan pricing decisions of banks in Turkey. They interpret this as an indication of monopolistically competitive pricing in credit markets. 23 In terms of asset size, the top 7 banks are clustered at the top, they are the top third of deposit collecting banks in the system, and accounting for about 75% of the sector’s total assets.We have performed a robustness comparing the top 5 to bottom 5 banks as well, results of which are reported in Table A1. 20
23. in MP has tied this result to their costlier fund raising capabilities (Kashyap and Stein, 1995 and 2000). With fewer alternatives to central bank funding, smaller, less capitalized banks with weaker offshore linkages will have no opportunity to smooth changes in the MP with other costs, and as such, are more sensitive to changes in MP. Before we investigate lender-side heterogeneities in more detail, another sector-wide aspect, bank ownership structures is worth exploring as it may also imply different credit policies and risk taking practices. State banks may reflect changes in MP to credit conditions depending on where the economy is on the economic and financial cycle.24 Foreign-owned banks, due to their deeper offshore connections, may have a comparative advantage in FX-lending. As such, we divide the banks in our universe into their ownership groups of state, domestic private, foreign private and participation banks to investigate whether they respond to changes in MP differently.25 We present the results of our analysis with a full set of controls for macroeconomic, bank and firm-bank observables in even numbered columns in Table 6. All bank types reduce lending volumes and maturities in response to an increase in MP. State and domestic private banks’ credit volumes are more responsive to changes in MP compared to participation and foreign private banks, in this respective order. Responsiveness over maturities also show differences across bank types, with participation and domestic private banks showing larger changes compared to state and foreign private banks. Following the argument in Kashyap and Stein (1995, 2000), the fact that foreign private banks are the least responsive in both channels is also suggestive that access to foreign borrowing channels may also play a role in smoothing out responses. We further explore the effects of these bank heterogeneities comparatively in the following section. So far, we have focused on a set of results performed on the pooled loan-level lending data. The results show that there are differences across size and ownership types, however this analysis is silent on whether the results are demand or supply driven. To further test for the robustness of these results, next, we focus on the relative differences across bank-types through an analysis that focuses only on firms that have borrowed from different types of banks at the same time using Equation 2. This approach allows us to control for the demand side through the use of firm-month fixed effects and focuses on the comparative responses of different banks to changes in MP. By using the same firm, we have extracted any effect the firm-specific factors may have on banks’ lending and maturity decisions.26 24 Previous work has shown that state banks have a counter-cyclical lending nature (Bertay et al, 2015). term participation bank is used in Turkey for banks with business models similar to those identified as islamic banking elsewhere. 26 This approach is increasingly employed by empirical studies on lending, Khwaja and Mian, 2008, Jim´ enez et al (2014) among others and rests on a simple assumption that the firm’s demand for funds does not have a 25 The 21
24. In columns 1 , 4 and 5 identified through firms that borrow from state and other banks, Table 7 shows that loans originating from state-owned banks move relatively less in the direction of the change in MP; in response to a rate hike, loan amounts issued by state banks in the following month are smaller than all other bank types.27 In other words, state banks lend relatively less than all other bank types during tightening periods to the same firm. This could be due to the relative capability of other banks to access international funding sources through their ownership structure and partners. Participation banks are relatively less sensitive in reflecting MP changes to loan amounts (columns 1-3) compared to state and domestic private banks, in line with the findings in previous literature.28 This may stem from their special model of operation; participation banks engage in profit sharing with their credit customers and as such must have a relatively larger amount of liquidity when the economy is heating which they are keen to distribute, a condition that still holds after rising rates in response to the heating economy. Finally, in column 6, domestic private banks are more sensitive to changes in MP compared to foreign private banks, which again can be explained through their relatively higher dependence on local funding. The relative performance of bank types with regards to maturities presents an interesting direction. Participation banks are relatively more sensitive in maturities (columns 1-3) which stands in contrast to their lower relative sensitivity in loan amounts. This suggests that participation banks use credit and maturity channels in their lending policies interchangeably. Faced with a higher MP rate, participation banks, reflect the change in amounts less and maturities more relative to other banks. This can be motivated through the bank balance sheet channel; in a heating economy, through the rising profits of their borrowers, participation banks may be faced with a higher amount of liquidity to lend out. However, faced with rising rates, the net worth of their applicant pool declines, and the bank may want to select suitable borrowers by testing their debt servicing capacities through shorter maturities (Bernanke and Gertler, 1989). Compared to domestic private banks, state banks reflect smaller changes in maturities, smoothing out the transition to a new monetary policy regime (column 4). Finally, column 6 shows that domestic private banks remain more sensitive in maturities compared to their foreign counterparts. To summarize, in Table 7 where we control for the demand side, and compare and contrast different bank types’ lending and maturity channels as a response to monetary policy decisions, we find that banks differ in terms of their sensitivity to these policy changes. Foreign private banks lend relatively larger and longer loans compared to all preference over banks, rather the firm demands funds uniformly across banks. to our identification strategy, the analysis in each column uses firms that borrow from both types of banks in the same month. This means that firms that borrowed from three or more types of banks in the same month are excluded from the analysis, which makes up about 6.5% of our database. 28 Zaheer et al (2013) find that participation banks’ lending is unaffected by contractionary monetary policy, Beck et al (2013) find that they are more likely to continue their lending practices even during crisis episodes. 27 Due 22
25. other bank types that do business with the same exact firm in the same month , which can be motivated by their lower reliance on central bank funding. This suggests that foreign bank presence in an economy may be helpful in smoothing out the transition to a new monetary policy regime. Foreign private, and state banks, on the other hand, smooth the transition to the new policy regime by lending longer maturities to firms compared to other banks which lend to the same firms. Across Lender-Side Heteregoneity So far, Tables 2 to 7 have focused on how existence and depth of monetary policy’s effects on lending and maturity channels have differed across currency denomination, bank size and type, and loosening/tightening periods. Tables 8 and 10 on the other hand, focus on how the responsiveness in the transmission mechanism differs across bank heterogeneities. In order to get at this result, the decomposition of supply and demand factors is of paramount importance, therefore we employ a dual approach to identify each source of variation. First, we focus on firms that borrow at least from two different banks by the inclusion of firmmonth fixed effects as in Equation 3. This approach simply assumes that the firm’s demand for funds does not vary across different banks, allowing us to control for credit demand and ties the results to supply factors. Table 8 investigates how the capital structure of banks influence how they respond to monetary policy changes in terms of lending and maturity. In columns 1 and 2, we show the results controlling for bank and firm time invariant unobservables, whereas columns 3-6 control for firm-month fixed effects as stated. As such, we focus on the results presented in these columns in the discussions below. To interpret the results, for each variable of interest, we have presented the relative responsiveness of the banks in the lowest quantile to those in the highest quantile to a 100bps change in MP. Banks with weaker capital structures are more sensitive to monetary policy decisions, and as such reflect that onto their customers: in the event of a tightening/losening of policy we see banks with weaker capital structures reduce/increase loan amounts and maturities more compared to banks with stronger capital structures. Following an increase of 100bps in the policy rate, weaker banks reduce credit by 0.7% and maturities by 1.8% more compared to stronger banks. Table 9 focuses on the presence of liquid assets rather than capital structure in the same pattern as Table 8 where columns 3-6 again control for firm-month fixed effects. Banks with weaker liquidity structures are more sensitive to monetary policy decisions and respond more to increases/decreases in the rates with lower/higher loan amounts and maturities. Following 23
26. an increase of 100bps in the policy rate , banks with weaker liquidity structures reduce credit by 0.8% and maturities by 1.3% more compared to stronger banks. This finding is in line with the previous literature, which find that small and less liquid banks are more sensitive to changes in MP (Kashyap and Stein (2000)). Heterogeneities remain across banks in both tightening and loosening periods, with the relative responsiveness of less liquid banks being even higher during tighter monetary policy phases. Table 10 focuses on the effect of banks’ non-core FX liabilities on the monetary policy transmission mechanism. Two thirds of the Turkish banking sector’s liabilities (ex. ownership equity) are core liabilities. Non-core liabilities of the sector, on the other hand are 60% foreign sourced. As such, the fact that a bank has access to foreign funding or not has a role to play in its monetary policy transmission depth. To put this to the test, we add the interaction of MP and the share of the bank’s FX-denominated non-core liabilities in total liabilities. Similar to Tables 8 and 9, columns 3-6 control for the demand side effects by including firm-month fixed effects. The results confirm that banks that have limited access to foreign sources of funding, in other words banks that are relatively more dependent on the CBRT for their asset side needs, are more sensitive to the CBRT’s monetary policy decisions, and in the event of a monetary tightening, they reduce loan amounts more but maturities remain rather comparable compared to banks with better foreign linkages. In sum, Tables 8 through 10 show that banks with stronger access to alternative sources of funding reflect policy rate changes less onto their credit conditions. Banks with weaker capital and liquidity structures and less access to foreign funding are more sensitive to policy changes in terms of not only loan amounts but also maturity. This implies that heterogeneity across banks is a source of asymmetry in monetary policy transmission mechanism. The bank balance sheet channel may work as an amplifying mechanism here; less liquid/capitalised banks may prefer safer borrowers and constrain riskier borrowers’ access to loans or longer maturities when faced with tighter monetary conditions and declining corporate net worth. 29 Our contribution is to show that, in addition to the loan amount, the maturity is also an important dimension of the bank balance sheet channel. Relationship and Rating Table 11 looks at whether the lending preferences and the monetary transmission mechanism examined so far differ depending on the depth and duration of the relationship between the lender and the borrower. The related literature signals that bank-firm relationships can be decisive on the amount and cost of loans, reduce the impact of economic and financial cycles, and have real economic outcomes (Petersen and Rajan, 1994; Berger and Udell, 1992; Sette and Gobbi, 2015; Banerjee et al, 2017). In addition to 29 Bernanke and Gertler (1989); Bernanke et al (1996); Matsuyama (2007). 24
27. the method employed in the previous section to control for demand , this section will also focus on demand side factors by controlling for supply. To this end, we use the mechanism outlined in Equation 4 and add bank-month fixed effects to the equations (and focus on banks that lend to at least two different firms, although, this last restriction is trivially satisfied in our sample). Thus, Table 11 will then in turn control for supply and demand side effects to test how the strength of firm-bank relationship impacts the monetary transmission mechanism. First we test the strength of firm-bank relationship and the MP interaction while controlling for supply through bank-month fixed effects to focus on the variation among firms that work with the same bank in the same month but have varying degrees of relationship depth. This approach will allow us to compare and contrast monetary policy responses through loan amount and maturity by banks with different levels of relationship strengths to different firms in the same month. The results in columns 1, and 2 for amount, and columns 4 and 5 for maturity state that out of the banks that lend to firms under a given macroeconomic environment, the bank with the weaker relationship reflects changes in the policy rate at higher magnitudes than banks that have a more established relationship with the said firm, and the effect is more evident in amounts than maturities. The second step controls for the bank-month heterogeneities, and the results in column 3 and 6 show that loans between firms and banks with weak relationships have higher and equally strong responses to an increase in the MP through credit and maturity channels. Banks would be expected to factor in their own internal know-how of the firm and the firm’s established track record in assessing loan applications (relationship channel established in Petersen and Rajan 1994). With rising agency costs in a tight monetary policy regime, this internal information becomes even more valuable, and, compared to the risk assessments they would have to undertake for new applicants, less costly for banks to rely on. As such, banks are able to offer relatively more favorable conditions to their long standing customers by dampening the credit and maturity channel impacts of changes in MP. This policy brings about the potential for a secondary, long-term gain for the bank; anecdotal evidence shows that firms ”return the favor” to banks that have eased credit conditions to them, by transferring their business to these banks. Aside from business/commission gains, and an improvement in its borrower composition with firms that have survived previous cycles, this policy has the potential to provide the bank with a stable source of funding through the firm’s deposits. Another avenue that the bank could respond to changes in the monetary policy decisions (and also the applicant pool) would be to change the composition of the borrower pool to meet its risk appetite. Banks rate firms at the time of loan issuance in a scale of 1 to 5 reflecting increasing risk, which we will use as a risk indicator and include the interaction of 25
28. this variable with the MP in the regressions . In Table 12 we control for supply side factors that could influence these decisions to focus on the firm-specific risk factor in the outcomes. Focusing on columns 3, 6, and 9 which restrict the analysis to banks that lend to different firms, we see that an increase in MP is reflected by larger magnitudes to riskier firms, in both credit and maturity channels. The risk taking channel of monetary policy state that in easing policy regimes, banks may seek riskier lending with profit seeking, search for yield motivations (Borio and Zhu,2012; Adrian and Shin, 2009; Gambacorta, 2009). Here we document the flipside to this argument, that in a tighter policy environment when the pool of applicants are expected to be riskier, firms that are assessed as relatively more risky are issued smaller loans with shorter maturities. 5.2 Credit Guarantee Fund Tables 13 and 14 use the implementation of the CGF guarantee scheme as a case study to investigate its influence on the monetary transmission mechanism through credit and maturity channels. The guarantee policy aiming to liven up the credit channel through subsidized loans with longer term maturity presents a unique opportunity to investigate how the banking sector will adjust credit conditions in response to this non-monetary stimulation. Normally, such an expansionary policy would be expected to increase credit amounts and lengthen maturities. However, the intense implementation of the scheme over the course of 6 months coincided with a monetary contraction phase in response to high inflation, which in contrast is expected to reduce both credit amounts and maturities. Thus in this section, we will investigate how the interaction of an expansionary macroprudential policy and contractionary monetary policy impacts the transmission mechanism. Indeed, as an initial motivation, Figure 1 Panel A shows that during the implementation period of the scheme, the long-run negative relationship between the policy rate and the amount and maturities has disappeared, and tighter monetary policy rates were accompanied by larger and longer loans. For a 6-month long window around the initial implementation date of the CGF scheme, we identify the marginal effects of CGF scheme on the transmission of monetary policy to treat the CGF scheme as a quasi-experiment by employing time dummies and their interaction with the MP decisions.30 30 Our motivation to focus on a 6-month window is two-fold. First, the first 6 months of the year allow us to perform our analysis within the monetary policy contraction phase without any structural breaks and focuses on the most intense utilization of the CGF utilization. Second, given the practice of quarterly bank statement announcements of banks, this half a year window allows banks time to respond to the monetary policy in what is equivalent to two quarterly statements. We have also included a robustness window of 9 months in the lower panels of both Tables 13 and 14, which is still within the contractionary policy phase. In Table 14, the results of this study show smaller responses in amount, and smaller responses in FX-denominated loans in maturity. 26
29. In Table 13 , we observe that the absolute value of the coefficient of monetary policy is very close the coefficient of the interacted term. While the coefficient of MP indicates the effects of monetary policy before the implementation of CGF scheme, the sum of these two coefficients give us the effects of monetary policy after the implementation of CGF scheme on the amount and maturity of commercial loans. And the total effect implies that CGF scheme has changed the channel through which banks transmit the monetary policy decisions onto new loans through loan amounts, or maturity. A 100bps increase in the MP is accompanied by a slight increase in loan amounts, and a very strong lengthening in maturities despite the tight monetary stance of the CBRT during this 6-month period. In particular, the fact that maturities are lengthening could be linked to the treasury guarantees extended for new loans which alleviates banks’ asset quality and maturity mismatch concerns for the new loans issued mostly in TL during this period under the scheme. Since banks differ in the amount of exposure they had to CGF guaranteed loans, we explore the effects of this heterogeneity across the sector in terms of the transmission mechanism next. To this end, we create a measure of banks’ exposure to the scheme as the ratio of banks’ CGF guaranteed loans out of all commercial loans, and compare the relative responsiveness of banks in the top and bottom quantile of this CGF exposure to changes in MP. In Table 14, we add the interaction of this variable with the policy rate, and by adding firm-month fixed effects to our model in even numbered columns, we restrict the sample to banks that lend to the same firm in the same month to compare the differences in lending practices that are due to cross-sectional differences in CGF exposures. The coefficient of interaction gives us the marginal effects of monetary policy across banks that have lower or higher exposure to CGF scheme. The coefficient of this variable suggests that banks with a larger exposure to the CGF scheme, in other words banks that have a larger share of their new lending guaranteed though the CGF scheme reflect monetary policy decisions onto credit conditions by a lesser degree. In other words, banks with a larger exposure to CGF scheme play a more dominant role in the reversal of the relationship compared to banks with a lower exposure to CGF scheme. Again, the overall results are driven by loans issued in the domestic currency, which made up more than %85 of the loans issued in this period. 6 Conclusion The monetary policy rate has a direct effect on banks’ borrowing costs, but also affects the real economy through the effect it has on economic agents’ lending and borrowing decisions. As such, the well functioning of the monetary policy transmission mechanism and the inves27
30. tigation of factors that promote or hinder this mechanism is of great interest . An important issue in such an analysis would be to disentangle supply and demand side effects. Using a bank-firm-loan level granular micro-data from a large emerging economy with universal coverage which allows us to disentangle demand and supply side factors, we first investigate the presence of the maturity channel in addition to credit volume in the transmission mechanism and investigate the effect of bank heterogeneities on this outcome. Next, in a quasi-experimental setting, we focus on the effects of a widely used policy supporting commercial loan usage through collateral guarantees backed by the treasury - namely a collateral guarantee scheme - to investigate how banks’ responses to this policy impacted the transmission of monetary policy. We find that in response to a rise in the policy rate, loan amounts and loan maturities are reduced. An increase of 100bps reduces commercial loan amounts by 1.6% and maturities by 1.2%. The reduction in amounts is larger for longer-term borrowing, as a 100bps increase lowers long-term borrowing by 3.3% and short term by 1.0%. While both TL and FX denominated loans are affected by the change in the policy rate, the credit channel effect is larger for TL-denominated loans, and maturity effect is larger for FX-denominated loans. Bank heterogeneities play a role in the transmission mechanism as well; both small and large banks exhibit reactions in the same direction, but smaller banks are more sensitive to the changes in the policy rate. Bank types also have an impact, with heterogeneous responses across; participation banks and private domestic banks more sensitive than state or foreign private banks in maturities. Reactions are asymmetric in terms of the monetary policy decision direction, we find that responses are larger in periods of tightening monetary policy compared to easing monetary policy. Banks with relatively weaker capital structures, weaker liquidity structures, and weaker access to foreign funding reduce loan volume and maturities by larger amounts. Banks reflect these changes to firms with which they have longer established relationships or which have a healthier past credit performance to a lesser extent, lending support that the monetary policy risk taking channel exists in our data. Next, we test how an expansionary macroprudential policy supporting commercial loan usage through collateral guarantees backed by the treasury affected loan issuance and the maturity mechanism at a time of monetary tightening. We find that the CGF scheme has changed the long-run relationship observed in the monetary policy transmission mechanism: with maturities lengthening during the tight monetary policy stance. This is especially true for banks that have a larger exposure to CGF scheme loans. 28
31. 7 References Adrian , T., & Shin, H. S. (2009). Money, liquidity, and monetary policy. American Economic Review, 99(2), 600-605. Alper, K., C¸olak, M. S., & Tok, E. (2016). Capital constraints and loan growth: A mechanical analysis (No. 18), CBRT Research Notes in Economics. Anthony, S. D., Viguerie, S. P., Schwartz, E. I., & Van Landeghem, J. (2018). Corporate longevity forecast: Creative destruction is accelerating. Innosight. Ayyagari, M., Demirg¨uc¸-Kunt, A., & Maksimovic, V. (2013). Size and age of establishments: evidence from developing countries (No. 6718). The World Bank. Baziki, S.B., & C¸apacıo˘glu, T. (2020). Macroprudential policies and commercial loans: Lending decisions of banks. Unpublished mimeo. Buyukbasaran, T., Cebi, C., & Yilmaz, E. (2019). Interaction of Monetary and Fiscal Policies in Turkey. CBRT Working Papers (No. 1935). Banerjee, R., Gambacorta, L., & Sette, E. (2017). The real effects of relationship lending (No. 662). BIS Working Papers. Beck, T., & Demirg¨uc¸-Kunt, A. (2006). Small and medium-size enterprises: Access to finance as a growth constraint. Journal of Banking & Finance, 30(11), 2931-2943. Beck, T., Demirg¨uc¸-Kunt, A., Laeven, L., & Levine, R. (2008). Finance, firm size, and growth. Journal of Money, Credit and Banking, 40(7), 1379-1405. Beck, T., Demirg¨uc¸-Kunt, A., & Levine, R. (2005). SMEs, growth, and poverty: crosscountry evidence. Journal of Economic Growth, 10(3), 199-229. Beck, T., Demirg¨uc¸-Kunt, A., & Merrouche, O. (2013). Islamic vs. conventional banking: Business model, efficiency and stability. Journal of Banking & Finance, 37(2), 433-447. Beck, T., Klapper, L. F., & Mendoza, J. C. (2010). The typology of partial credit guarantee funds around the world. Journal of Financial Stability, 6(1), 10-25. Berger, A. N., Frame, W. S., & Ioannidou, V. (2016). Reexamining the empirical relation between loan risk and collateral: The roles of collateral liquidity and types. Journal of Financial Intermediation, 26, 28-46. Berger, A. N., & Udell, G. F. (1990). Collateral, loan quality and bank risk. Journal of Monetary Economics, 25(1), 21-42. Berger, A. N., & Udell, G. F. (1992). Some evidence on the empirical significance of credit rationing. Journal of Political Economy, 100(5), 1047-1077. Berger, A. N., & Udell, G. F. (1995). Relationship lending and lines of credit in small firm finance. Journal of Business, 351-381. Bernanke, B. S., & Blinder, A. S. (1992). The federal funds rate and the channels of monetary transmission. American Economic Review, 82(4), 901-21. 29
32. Bemanke , B., & Gertler, M. (1989). Agency costs, net worth, and business fluctuations. American Economic Review, 79(1), 14-31. Bernanke, B. S., & Gertler, M. (1995). Inside the black box: the credit channel of monetary policy transmission. Journal of Economic Perspectives, 9(4), 27-48. Bernanke, B., Gertler, M., & Gilchrist, S. (1996). The financial accelerator and the flight to quality. Review of Economics and Statistics, 78(1), 1-15. Bertay, A. C., Demirg¨uc¸-Kunt, A., & Huizinga, H. (2015). Bank ownership and credit over the business cycle: Is lending by state banks less procyclical?. Journal of Banking & Finance, 50, 326-339. Beyer, A., Nicoletti, G., Papadopoulou, N., Papsdorf, P., R¨unstler, G., Schwarz, C., Sousa, J., & Vergote, O. (2017). The transmission channels of monetary, macro-and microprudential policies and their interrelations (No. 191). ECB Occasional Paper Series. ¨ u, P. (2019). Monetary transmission with multiple policy rates: Binici, M., Kara, H., & Ozl¨ evidence from Turkey. Applied Economics, 51(17), 1869-1893. Black, L. K., & Rosen, R. J. (2016). Monetary policy, loan maturity, and credit availability. International Journal of Central Banking, 12(1), 199-230. Borio, C., & Zhu, H. (2012). Capital regulation, risk-taking and monetary policy: a missing link in the transmission mechanism?. Journal of Financial stability, 8(4), 236-251. Banking Regulation and Supervision Authority (2018). Development and performance of CGF loans. BRSA Information Notes, June. Campbell, J. Y., & Cochrane, J. H. (1999). By force of habit: A consumption-based explanation of aggregate stock market behavior. Journal of Political Economy, 107(2), 205-251. CBRT Financial Stability Report (2017). Special Topic IV.2: Measures on corporate sector’s access to finance (Vol. 24). 71-76. Collin-Dufresne, P., Goldstein, R. S., & Martin, J. S. (2001). The determinants of credit spread changes. The Journal of Finance, 56(6), 2177-2207. Diamond, D. W., & Rajan, R. G. (2006). Money in a theory of banking. American Economic Review, 96(1), 30-53. Diamond, D. W., & Rajan, R. G. (2011). Fear of fire sales, illiquidity seeking, and credit freezes. The Quarterly Journal of Economics, 126(2), 557-591. Collin-Dufresne, P., Goldstein, R. S., & Martin, J. S. (2001). The determinants of credit spread changes. The Journal of Finance, 56(6), 2177-2207. Gambacorta, L. (2009). Monetary policy and the risk-taking channel. BIS Quarterly Review December. Gambacorta, L., & Mistrulli, P. E. (2004). Does bank capital affect lending behavior?. Journal of Financial Intermediation, 13(4), 436-457. 30
33. Gambacorta , L., & Murcia, A. (2017). The impact of macroprudential policies and their interaction with monetary policy: an empirical analysis using credit registry data (No. 636). BIS Working Paper. Gertler, M., & Gilchrist, S. (1993). The cyclical behavior of short-term business lending: Implications for financial propagation mechanisms. European Economic Review, 37(2-3), 623-631. Gertler, M., & Gilchrist, S. (1994). Monetary policy, business cycles, and the behavior of small manufacturing firms. The Quarterly Journal of Economics, 109(2), 309-340. Gilchrist, S., & Zakrajˇsek, E. (2007). Investment and the cost of capital: New evidence from the corporate bond market (No. w13174). National Bureau of Economic Research. Gozzi, J. C., & Schmukler, S. (2016). Public credit guarantees and access to finance (No. 1122). Warwick Economics Research Paper Series. Hanedar, E. Y., Broccardo, E., & Bazzana, F. (2014). Collateral requirements of SMEs: The evidence from less-developed countries. Journal of Banking & Finance, 38, 106-121. Iannotta, G., Nocera, G., & Sironi, A. (2013). The impact of government ownership on bank risk. Journal of Financial Intermediation, 22(2), 152-176. Illes, A., Lombardi, M. J., & Mizen, P. (2015). Why did bank lending rates diverge from policy rates after the financial crisis? (No. 486). BIS Working Paper. Jim´enez, G., & Saurina, J. (2004). Collateral, type of lender and relationship banking as determinants of credit risk. Journal of Banking & Finance, 28(9), 2191-2212. Jim´enez, G., Ongena, S., Peydr´o, J. L., & Saurina, J. (2012). Credit supply and monetary policy: Identifying the bank balance-sheet channel with loan applications. American Economic Review, 102(5), 2301-26. Jim´enez, G., Ongena, S., Peydr´o, J. L., & Saurina, J. (2014). Hazardous times for monetary policy: What do twenty-three million bank loans say about the effects of monetary policy on credit risk-taking?. Econometrica, 82(2), 463-505. Karagiannis, S., Panagopoulos, Y., & Vlamis, P. (2010). Interest rate pass-through in Europe and the US: Monetary policy after the financial crisis. Journal of Policy Modeling, 32(3), 323-338. Kashyap, A. K., & Stein, J. C. (1995). The impact of monetary policy on bank balance sheets. Carnegie-Rochester Conference Series on Public Policy, 42, 151-195. Kashyap, A. K., & Stein, J. C. (2000). What do a million observations on banks say about the transmission of monetary policy?. American Economic Review, 90(3), 407-428. Kashyap, A. K., Stein, J. C., & Wilcox, D. W. (1993). Monetary policy and credit conditions: Evidence from the composition of external finance. American Economic Review, 83(1), 7898. Khwaja, A. I., & Mian, A. (2008). Tracing the impact of bank liquidity shocks: Evidence from an emerging market. American Economic Review, 98(4), 1413-42. 31
34. Lang , W. W., & Nakamura, L. I. (1995). ‘Flight to quality’in banking and economic activity. Journal of Monetary Economics, 36(1), 145-164. Levitsky, J. (1997). Credit guarantee schemes for SMEs–an international review. Small Enterprise Development, 8(2), 4-17. Longstaff, F. A., & Schwartz, E. S. (1995). A simple approach to valuing risky fixed and floating rate debt. The Journal of Finance, 50(3), 789-819. Matsuyama, K. (2007). Credit traps and credit cycles. American Economic Review, 97(1), 503-516. Modugno, M., Soybilgen, B., & Yazgan, E. (2016). Nowcasting Turkish GDP and news decomposition. International Journal of Forecasting, 32(4), 1369-1384. OECD, (2013). Improving access to finance for SMEs in Central Asia through credit guarantee schemes. Private Sector Development Policy Handbook. Petersen, M. A., & Rajan, R. G. (1994). The benefits of lending relationships: Evidence from small business data. The Journal of Finance, 49(1), 3-37. Rahman, A., Belas, J., Kliestik, T., & Tyll, L. (2017). Collateral requirements for SME loans: empirical evidence from the Visegrad countries. Journal of Business Economics and Management, 18(4), 650-675. Rajan, R. G. (2006). Has finance made the world riskier?. European Financial Management, 12(4), 499-533. Sette, E., & Gobbi, G. (2015). Relationship lending during a financial crisis. Journal of the European Economic Association, 13(3), 453-481. Stein, J. C. (1998). An adverse selection model of bank asset and liability management with implications for the transmission of monetary policy. RAND Journal of Economics, 29(3), 466-486. Steijvers, T., & Voordeckers, W. (2009). Collateral and credit rationing: a review of recent empirical studies as a guide for future research. Journal of Economic Surveys, 23(5), 924946. The EBCI Vienna Initiative, (2014). Credit guarantee schemes for SME lending in Central, Eastern and South-Eastern Europe. Report by the Vienna Initiative Working Group on Credit Guarantee Schemes. Zaheer, S., Ongena, S., & Van Wijnbergen, S. (2013). The transmission of monetary policy through conventional and Islamic banks. International Journal of Central Banking, 9(4), 175-224. 32
35. Figure 1 : Monetary policy rate, TL (Panel A) and FX (Panel B) denominated commercial loan maturities (Percent, Month). The negative relationship between MP and maturities seen in previous tightening periods, especially in TL-denominated issuances shown in Panel A, has disappeared during the intense utilization of the Credit Guarantee Fund scheme. 33
36. 34 2008M01-2017M09 2008M01-2017M09 TurkStat CBRT Other Bank Variable CGF Ratio Riskiness Bank-Firm Variables Relationship Ratio FX Non −Core Ratio Bank Variables Total Assets Credit Ratio Deposit Ratio NPL Ratio ROA Ratio Capital Ratio Liquidity Ratio In f lation Rate ∆ REER CGF loans divided by total TL loans The ratio of loans a firm obtains from a particular bank in the last 12-months to the total loans the firm obtains from all banks during the same period Internal rating of banks on borrower firms The natural logarithm of banks’ total real assets Total loans divided by total assets Total deposits divided by total assets Bank non-performing loans divided by bank total loans Bank net profit divided by total assets Capital divided by total assets Selected FX liquid assets (cash + foreign banks(free) + free Eurobonds) divided by total assets Total FX non-core liabilities divided by total liabilities 2008M01-2017M09 TurkStat Yearly change in industrial production index (used instead of GDP due to discrepancy of frequencies) Yearly change in consumer price index Monthly change in real effective exchange rate based on consumer price index ∆ Industrial Production Index CBRT CBRT weighted average funding rate (effective policy rate) 2017M01-2017M09 2008M01-2017M09 CBRT CBRT 2008M01-2017M09 2008M01-2017M09 2008M01-2017M09 2008M01-2017M09 2008M01-2017M09 2008M01-2017M09 2008M01-2017M09 2008M01-2017M09 2008M01-2017M09 CBRT CBRT CBRT CBRT CBRT CBRT CBRT CBRT CBRT 2008M01-2017M09 2008M01-2017M09 CBRT Independent Variables Turkey (TR) Macro Variables MP 2008M01-2017M09 CBRT The natural logarithm of the amount of new loans (Million TL) granted by Turkish bank b to firm f in currency type c at time t The natural logarithm of the maturity of new loans (day) granted by Turkish bank b to firm f in currency type c at time t Period Dependent Variables The Amount of New Domestic Lending The Maturity of New Domestic Lending Source Definition Variable Names Table 1: Summary Statistics 23 11,993,081 24,564,972 3,834 3,834 3,834 3,834 3,830 3,834 3,834 3,574 117 117 117 117 23,543,406 24,548,967 N 2.25 2.42 0.40 0.20 10.82 0.53 0.50 0.05 0.01 0.16 0.12 8.24 -0.31 3.26 8.96 5.61 4.04 Mean 1.00 0.98 0.40 0.18 1.94 0.18 0.22 0.09 0.02 0.14 0.13 1.79 2.84 8.74 3.12 1.29 1.63 SD 0.39 1.00 0.00 0.00 5.73 0.00 0.00 0.00 -0.72 0.03 0.00 3.99 -11.80 -23.98 4.83 0.00 0.00 Min. 1.58 2.00 0.00 0.10 9.52 0.43 0.47 0.02 0.01 0.10 0.05 7.17 -1.67 0.17 7.04 5.12 3.00 0.25 2.00 2.00 0.24 0.16 10.81 0.60 0.57 0.03 0.01 0.12 0.08 8.17 -0.03 4.17 8.28 5.89 3.91 0.50 3.12 3.00 0.86 0.22 12.42 0.66 0.64 0.05 0.02 0.15 0.14 9.38 1.24 7.52 9.72 6.56 5.00 0.75 4.27 5.00 1.00 0.95 14.08 0.85 0.85 1.00 0.15 0.99 1.18 12.06 7.64 25.71 17.15 8.52 15.10 Max.
37. Table 2 : The Effects of Monetary Policy on the Amount and Maturity of Turkish Banks’ New Domestic Lending (1) Dependent Variable: Currency Type (2) (3) Amount (4) Maturity (5) (6) Amount*Maturity ALL ALL ALL ALL ALL -0.017*** [0.000] 4.301*** [0.005] -0.016*** [0.000] 0.441*** [0.052] -0.012*** [0.000] 6.023*** [0.001] -0.012*** [0.000] 3.136*** [0.036] -0.137*** [0.001] 27.102*** [0.012] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio No No No Yes Yes Yes No No No Yes Yes Yes No No No Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.580 21,613,511 0.584 20,918,862 0.454 20,770,666 0.466 20,104,059 0.408 20,770,666 0.421 20,104,059 -1.70 -1.60 -1.20 -1.20 -13.70 -9.50 MPt−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP: ALL -0.095*** [0.001] -27.095*** [0.271] Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity or amount*maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”) or not included (”No”). *** Significant at 1%, ** significant at 5%, * significant at 10%. 35
38. 36 -1 .00 -1.00 0.654 13,106,276 Yes Yes Yes Yes Yes Yes Yes -0.010*** [0.000] 1.118*** [0.060] (4) -3.00 0.609 8,059,104 Yes Yes Yes Yes No No No -0.030*** [0.001] 4.258*** [0.011] -3.30 0.615 7,848,330 Yes Yes Yes Yes Yes Yes Yes (5) -1.80 0.559 20,109,578 Yes Yes n/a Yes No No No -0.018*** [0.000] 4.208*** [0.005] Amount -0.033*** [0.001] 4.143*** [0.105] Long Term (3) TL -1.60 0.563 19,543,565 Yes Yes n/a Yes Yes Yes Yes -0.016*** [0.000] -0.414*** [0.054] (6) -0.30 0.601 1,503,933 Yes Yes Yes Yes No No No -0.003*** [0.001] 5.554*** [0.018] (7) FX -0.90 0.601 1,375,297 Yes Yes Yes Yes Yes Yes Yes -0.009*** [0.001] 5.037*** [0.142] (8) -1.20 0.487 19,370,097 Yes Yes n/a Yes No No No -0.012*** [0.000] 6.029*** [0.002] (9) TL -1.10 0.499 18,830,010 Yes Yes n/a Yes Yes Yes Yes -0.011*** [0.000] 3.205*** [0.037] (11) -2.20 0.413 1,400,569 Yes Yes Yes Yes No No No -0.022*** [0.000] 6.100*** [0.007] Maturity (10) FX -1.80 0.417 1,274,049 Yes Yes Yes Yes Yes Yes Yes -0.018*** [0.000] 3.006*** [0.107] (12) Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”) or not included (”No”) or not applicable (”n/a”). *** Significant at 1%, ** significant at 5%, * significant at 10%. %∆ with 100 bps increase in MP: 0.651 13,591,469 Yes Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects -0.010*** [0.000] 4.185*** [0.006] No No No R2 Number of Observations (2) Short Term TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Constant MPt−1 Maturity/Currency Type Dependent Variable: (1) Table 3: The Effects of Monetary Policy on the Amount and Maturity of Turkish Banks’ New Domestic Lending across Different Maturity and Currency Types
39. Table 4 : The Effects of Monetary Policy on the Amount and Maturity of Turkish Banks’ New Domestic Lending across Different Phases (1) (2) (3) (4) Dependent Variable: (6) Amount Policy Phase: Currency Type (5) Loosening Tigthening ALL TL FX ALL TL FX -0.010*** [0.000] 1.132*** [0.073] -0.011*** [0.000] 0.391*** [0.077] -0.010*** [0.001] -2.157*** [0.160] -0.020*** [0.001] 0.018 [0.095] -0.021*** [0.001] -0.918*** [0.101] -0.011*** [0.001] -2.562*** [0.205] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects Yes Yes Yes Yes Yes Yes n/a Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes n/a Yes Yes Yes Yes Yes 0.600 10,997,632 0.583 10,266,164 0.608 731,468 0.619 8,884,619 0.604 8,324,725 0.626 559,894 -1.00 -1.10 -1.00 -2.00 -2.10 -1.10 MPt−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP: Dependent Variable: Maturity Policy Phase: Currency Type Loosening Tigthening ALL TL FX ALL TL FX -0.006*** [0.000] 3.803*** [0.051] -0.004*** [0.000] 3.933*** [0.054] -0.019*** [0.001] 3.502*** [0.147] -0.017*** [0.000] 2.236*** [0.069] -0.016*** [0.000] 2.293*** [0.072] -0.019*** [0.001] 1.939*** [0.192] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects Yes Yes Yes Yes Yes Yes n/a Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes n/a Yes Yes Yes Yes Yes 0.484 10,575,505 0.518 9,897,097 0.433 678,408 0.504 8,532,708 0.537 8,014,426 0.455 518,282 -0.60 -0.40 -1.90 -1.70 -1.60 -1.90 MPt−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP: Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”) or not included (”No”) or not applicable (”n/a”). *** Significant at 1%, ** significant at 5%, * significant at 10%. 37
40. 38 -1 .10 -2.00 0.641 3,398,004 Yes Yes Yes Yes Yes Yes Yes -1.90 0.578 17,520,858 Yes Yes Yes Yes No No No ALL (4) -1.40 0.583 17,520,858 Yes Yes Yes Yes Yes Yes Yes -0.014*** [0.000] -2.386*** [0.062] Large -0.019*** [0.000] 4.284*** [0.006] ALL Amount (3) -1.60 0.463 3,858,651 Yes Yes Yes Yes No No No ALL (6) -1.50 0.481 3,192,044 Yes Yes Yes Yes Yes Yes Yes (7) -1.10 0.469 16,912,015 Yes Yes Yes Yes No No No ALL (8) -1.20 0.478 16,912,015 Yes Yes Yes Yes Yes Yes Yes -0.012*** [0.000] 2.534*** [0.042] Large -0.011*** [0.000] 6.073*** [0.002] ALL Maturity -0.015*** [0.000] 1.974*** [0.075] Small -0.016*** [0.000] 5.492*** [0.004] ALL (5) Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”) or not included (”No”). *** Significant at 1%, ** significant at 5%, * significant at 10%. %∆ with 100 bps increase in MP: 0.625 4,092,653 Yes Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects R2 Number of Observations No No No ALL (2) -0.020*** [0.001] 4.107*** [0.093] Small -0.011*** [0.001] 4.095*** [0.012] ALL TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Constant MPt−1 Bank Size Currency Type Dependent Variable: (1) Table 5: The Effects of Monetary Policy on the Amount and Maturity of Turkish Banks’ New Domestic Lending across Different Bank Sizes
41. Table 6 : The Effects of Monetary Policy on the Amount and Maturity of Turkish Banks’ New Domestic Lending across Different Bank Types Currency Type (1) (2) (3) (4) ALL ALL ALL ALL Dependent Variable: (5) (6) (7) (8) ALL ALL ALL ALL Amount Bank Type State Domestic Foreign Participation -0.025*** [0.001] 4.127*** [0.010] -0.022*** [0.001] -1.567*** [0.217] -0.020*** [0.001] 4.191*** [0.008] -0.021*** [0.001] 0.733*** [0.071] 0.000 [0.001] 3.428*** [0.013] -0.005*** [0.001] -0.274** [0.123] -0.018*** [0.001] 4.057*** [0.011] -0.016*** [0.002] 1.122*** [0.265] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio No No No Yes Yes Yes No No No Yes Yes Yes No No No Yes Yes Yes No No No Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.569 6,071,821 0.583 6,071,821 0.567 8,853,399 0.571 8,853,399 0.708 4,873,760 0.711 4,873,699 0.652 1,814,531 0.702 1,119,943 -2.50 -2.20 -2.00 -2.10 0.00 -0.50 -1.80 -1.60 MPt−1 Constant R2 Number of Observations %∆ with increase in the MP by 100 bp: Dependent Variable: Maturity Bank Type State Domestic Foreign Participation -0.007*** [0.000] 6.280*** [0.003] -0.015*** [0.000] 1.206*** [0.096] -0.014*** [0.000] 5.228*** [0.003] -0.021*** [0.000] 3.666*** [0.073] -0.003*** [0.000] 5.906*** [0.005] -0.001** [0.000] 4.429*** [0.070] -0.022*** [0.001] 5.575*** [0.006] -0.034*** [0.001] 3.253*** [0.203] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio No No No Yes Yes Yes No No No Yes Yes Yes No No No Yes Yes Yes No No No Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.547 5,969,955 0.553 5,969,955 0.384 8,510,303 0.395 8,510,303 0.565 4,632,289 0.568 4,632,228 0.597 1,658,119 0.686 991,573 -0.70 -1.50 -1.40 -2.10 -0.30 -0.10 -2.20 -3.40 MPt−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP: Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”) or not included (”No”). *** Significant at 1%, ** significant at 5%, * significant at 10%. 39
42. Table 7 : The Effects of Monetary Policy on the Amount and Maturity of Turkish Banks’ New Domestic Lending across Different Bank Types Currency Type (1) (2) (3) (4) (5) (6) ALL ALL ALL ALL ALL ALL Dependent Variable: Bank Type Amount Participation vs. State Participation vs. Domestic Participation vs. Foreign Domestic vs. State MPt−1 - - - - - - Dummy - - - - 0.010** [0.004] 0.005* [0.003] -0.002 [0.003] 0.661*** [0.032] 0.010*** [0.001] 0.005*** [0.001] 0.047*** [0.007] 0.008*** [0.001] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Firm*Month) Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.706 1,032,137 0.685 3,413,355 0.703 1,275,632 0.693 4,598,838 0.734 2,012,297 0.692 4,918,364 Foreign vs. State Dummy*MPt−1 R2 Number of Observations Dependent Variable: Bank Type Foreign vs. State Foreign vs. Domestic Maturity Participation vs. State Participation vs. Domestic Participation vs. Foreign Domestic vs. State MPt−1 - - - - - - Dummy - - - -0.024*** [0.003] -0.019*** [0.002] -0.007** [0.003] 3.383*** [0.073] -0.015*** [0.001] 0.382 [112.656] 0.001 [0.001] 0.095*** [0.008] 0.007*** [0.001] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Firm*Month) Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.546 940,062 0.507 3,164,406 0.561 1,135,342 0.513 4,339,763 0.573 1,863,404 0.498 4,588,077 Dummy*MPt−1 R2 Number of Observations Foreign vs. Domestic Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”), not included (”No”). Fixed effects or variables spanned by another set of fixed effects are marked with (”-”). *** Significant at 1%, ** significant at 5%, * significant at 10%. 40
43. Table 8 : Bank Lending Channel of Monetary Policy on Amount and Maturity across Banks Having Different Capital Ratio (1) (2) (3) Dependent Variable: Policy Phases (4) (5) (6) ALL Loosening Tigthening Amount ALL ALL -0.049*** [0.001] 0.324*** [0.006] -0.886*** [0.053] 4.363*** [0.008] -0.032*** [0.001] 0.181*** [0.006] 1.345*** [0.062] 0.108** [0.054] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio No No No Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Firm*Month) Fixed Effects ALL - - - - 0.271*** [0.011] -0.673*** [0.102] 0.142*** [0.012] 1.044*** [0.121] 0.191*** [0.016] 0.966*** [0.157] 0.115*** [0.018] 1.323*** [0.212] Yes Yes Yes No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.580 21,613,454 0.584 20,918,862 0.677 8,235,589 0.683 7,756,593 0.684 4,028,625 0.684 3,367,431 1.66 0.93 1.39 0.73 0.98 0.59 ALL ALL ALL ALL Loosening Tigthening -0.056*** [0.001] 0.498*** [0.006] -2.472*** [0.054] 6.193*** [0.008] -0.034*** [0.001] 0.321*** [0.006] 2.004*** [0.064] -2.879*** [0.057] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio No No No Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Firm*Month) Fixed Effects MPt−1 MPt−1 * Capital Ratiot−1 Capital Ratiot−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP by lower (25%) versus higher (75%) capitalized banks: Dependent Variable: Policy Phases MPt−1 MPt−1 * Capital Ratiot−1 Capital Ratiot−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP by lower (25%) versus higher (75%) capitalized banks: Maturity - - - - 0.553*** [0.013] -2.342*** [0.113] 0.357*** [0.013] 1.439*** [0.134] 0.474*** [0.019] 0.596*** [0.173] 0.393*** [0.021] 0.524** [0.227] Yes Yes Yes No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.455 20,770,609 0.468 20,104,059 0.486 7,724,717 0.496 7,265,954 0.498 3,774,735 0.492 3,156,985 2.56 1.65 2.84 1.83 2.43 2.02 Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”), not included (”No”) or spanned by another set of effects (”-”). ”-” also indicates dropped variables due to the fixed effects. *** Significant at 1%, ** significant 41at 5%, * significant at 10%.
44. Table 9 : Bank Lending Channel of Monetary Policy on Amount and Maturity across Banks Having Different Liquidity Ratio (1) (2) (3) Dependent Variable: Policy Phases (4) (5) (6) ALL Loosening Tigthening Amount ALL ALL -0.021*** [0.000] 0.029*** [0.001] -0.251*** [0.012] 4.310*** [0.006] -0.021*** [0.000] 0.040*** [0.001] 0.897*** [0.016] 0.466*** [0.052] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio No No No Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Firm*Month) Fixed Effects ALL - - - - 0.054*** [0.003] -0.123*** [0.034] 0.080*** [0.003] 0.757*** [0.038] 0.066*** [0.005] 0.892*** [0.054] 0.093*** [0.005] 0.689*** [0.065] Yes Yes Yes No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.581 20,918,892 0.584 20,918,862 0.681 7,756,623 0.683 7,756,593 0.684 4,028,625 0.684 3,367,431 0.27 0.37 0.50 0.75 0.62 0.87 ALL ALL ALL ALL Loosening Tigthening -0.009*** [0.000] 0.033*** [0.001] 0.304*** [0.011] 5.805*** [0.005] -0.010*** [0.000] 0.045*** [0.001] 1.539*** [0.015] -2.404*** [0.056] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio No No No Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Firm*Month) Fixed Effects MPt−1 MPt−1 * Liquidity Ratiot−1 Liquidity Ratiot−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP by lower (25%) versus higher (75%) liquid banks: Dependent Variable: Policy Phases MPt−1 MPt−1 * Liquidity Ratiot−1 Liquidity Ratiot−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP by lower (25%) versus higher (75%) liquid banks: Maturity - - - - 0.109*** [0.004] -0.203*** [0.038] 0.144*** [0.004] 0.646*** [0.042] 0.122*** [0.005] 1.027*** [0.060] 0.150*** [0.006] 0.209*** [0.070] Yes Yes Yes No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.460 20,104,089 0.467 20,104,059 0.490 7,265,987 0.495 7,265,954 0.497 3,774,735 0.492 3,156,985 0.31 0.42 1.02 1.34 1.14 1.40 Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”), not included (”No”) or spanned by another set of effects (”-”). ”-” also indicates dropped variables due to the fixed effects. *** Significant at 1%, ** significant at 5%, * significant at 42 10%.
45. Table 10 : Bank Lending Channel of Monetary Policy on Amount and Maturity across Banks Having Different FX Foreign Funding Ratio (1) (2) (3) Dependent Variable: Policy Phasese (4) (5) (6) ALL Loosening Tigthening Amount ALL ALL -0.021*** [0.000] 0.025*** [0.002] 0.178*** [0.020] 4.327*** [0.006] -0.017*** [0.000] 0.011*** [0.002] -0.494*** [0.022] 0.461*** [0.052] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio No No No Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Firm*Month) Fixed Effects ALL - - - - 0.082*** [0.005] -0.423*** [0.044] 0.081*** [0.005] -1.078*** [0.047] 0.090*** [0.007] -1.219*** [0.068] 0.077*** [0.007] -0.805*** [0.074] Yes Yes Yes No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.580 21,613,454 0.584 20,918,862 0.677 8,235,589 0.683 7,756,593 0.684 4,028,625 0.684 3,367,431 0.29 0.13 0.95 0.93 1.04 0.89 ALL ALL ALL ALL Loosening Tigthening -0.015*** [0.000] 0.022*** [0.002] -0.689*** [0.017] 6.096*** [0.003] -0.013*** [0.000] 0.012*** [0.002] -1.036*** [0.020] 3.150*** [0.036] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio No No No Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Firm*Month) Fixed Effects Yes Yes Yes Yes No MPt−1 MPt−1 * FX NonCore Ratiot−1 FX NonCore Ratiot−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP by lower (25%) versus higher (75%) foreign funded banks: Dependent Variable: Policy Phases MPt−1 MPt−1 * FX NonCore Ratiot−1 FX NonCore Ratiot−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP by lower (25%) versus higher (75%) foreign funded banks: Maturity - - - - 0.009* [0.005] 0.042 [0.048] -0.007 [0.005] -0.360*** [0.052] -0.017** [0.007] -0.669*** [0.075] 0.056*** [0.008] -0.471*** [0.081] No No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 0.455 20,770,609 0.466 20,104,059 0.486 7,724,717 0.495 7,265,954 0.497 3,774,735 0.492 3,156,985 0.25 0.14 0.10 -0.08 -0.20 0.65 Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio and NPL Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”), not included (”No”) or spanned by another set of effects (”-”). ”-” also indicates dropped variables due to the fixed effects. *** Significant at 1%, ** significant at 5%, * significant at 43 10%.
46. 44 0 .17 0.582 21,613,511 0.43 0.584 20,918,862 Yes Yes Yes Yes No 0.26 0.576 20,668,655 Yes Yes Yes - 0.003*** [0.000] -0.254*** [0.003] - ALL (3) 0.09 0.460 20,770,666 Yes Yes Yes Yes No No No -0.004*** [0.000] -0.001*** [0.000] -0.286*** [0.003] 6.046*** [0.005] ALL (4) 0.00 0.467 20,104,059 Yes Yes Yes Yes No Yes Yes -0.004*** [0.000] 0.000 [0.000] -0.294*** [0.003] -2.437*** [0.056] ALL Maturity (5) 0.17 0.447 19,836,408 Yes Yes Yes - -0.002*** [0.000] -0.270*** [0.003] - ALL (6) 1.55 (8) 3.18 0.421 20,104,059 Yes Yes Yes Yes No Yes Yes -0.128*** [0.003] 0.037*** [0.002] -3.031*** [0.020] -25.812*** [0.417] ALL ALL (9) 1.46 0.412 19,836,408 Yes Yes Yes - 0.017*** [0.002] -2.753*** [0.020] - Amount*Maturity 0.418 20,770,666 Yes Yes Yes Yes No No No -0.128*** [0.002] 0.018*** [0.002] -2.778*** [0.020] 26.479*** [0.041] ALL (7) Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”), not included (”No”) or spanned by another set of effects (”-”). ”-” also indicates dropped variables due to the fixed effects. *** Significant at 1%, ** significant at 5%, * significant at 10%. %∆ with 100 bps increase in MP by weaker (25%) versus stronger (75%) bank-firm relationship: R2 Number of Observations Yes Yes Yes Yes No Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Bank*Month) Fixed Effects Yes Yes -0.018*** [0.000] 0.005*** [0.000] -0.287*** [0.003] 0.471*** [0.052] -0.017*** [0.000] 0.002*** [0.000] -0.247*** [0.003] 4.385*** [0.005] No No ALL Amount (2) ALL TR Macro Variables Bank Variables Constant Relationship Ratiot−1 MPt−1 * Relationship Ratiot−1 MPt−1 Currency Type Dependent Variable: (1) Table 11: The Effects of Monetary Policy on the Amount and Maturity of Turkish Banks’ New Domestic Lending across Firms Having Different Relationship Ratio
47. 45 -1 .10 0.575 10,266,188 -1.10 0.576 10,090,942 Yes Yes Yes Yes No -0.90 0.568 9,760,617 Yes Yes Yes Yes -0.009*** [0.000] 0.135*** [0.002] - ALL (3) -0.30 0.436 9,814,462 Yes Yes Yes Yes No No No No -0.006*** [0.001] -0.003*** [0.000] 0.071*** [0.002] 5.741*** [0.010] ALL (4) -0.20 0.443 9,647,143 Yes Yes Yes Yes No Yes Yes Yes -0.011*** [0.001] -0.002*** [0.000] 0.063*** [0.002] -0.075 [0.103] ALL Maturity (5) -0.10 0.422 9,313,066 Yes Yes Yes Yes -0.001*** [0.000] 0.048*** [0.002] - ALL (6) -2.20 0.384 9,814,462 Yes Yes Yes Yes No No No No (8) -1.10 0.576 10,090,942 Yes Yes Yes Yes No Yes Yes Yes -0.054*** [0.001] -0.011*** [0.000] 0.128*** [0.002] 3.813*** [0.086] ALL ALL (9) -2.10 0.382 9,313,066 Yes Yes Yes Yes -0.021*** [0.002] 0.050*** [0.019] - Amount*Maturity -0.245*** [0.006] -0.022*** [0.002] 0.066*** [0.017] 27.113*** [0.076] ALL (7) Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate .Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2008:M01-2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”), not included (”No”) or spanned by another set of effects (”-”). ”-” also indicates dropped variables due to the fixed effects. *** Significant at 1%, ** significant at 5%, * significant at 10%. %∆ with 100 bps increase in MP by non-risky (25%) versus riskier (75%) firms: R2 Number of Observations Yes Yes Yes Yes No Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects (Bank*Month) Fixed Effects Yes Yes Yes -0.054*** [0.001] -0.011*** [0.000] 0.128*** [0.002] 3.813*** [0.086] -0.050*** [0.001] -0.011*** [0.000] 0.124*** [0.002] 4.931*** [0.010] No No No ALL Amount (2) ALL TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Constant Riskinesst−1 MPt−1 * Riskinesst−1 MPt−1 Currency Type Dependent Variable: (1) Table 12: Firm Borrowing Channel of Monetary Policy on Amount and Maturity across Firms Having Different Rating
48. Table 13 : The Effects of Monetary Policy and Macroprudential Policy (CGF Loans) on the Amount and Maturity of Loans (Diff-in-Diff Estimations) (1) (2) (3) Window: (5) (6) 6-Months Dependent Variable: Currency Type (4) Amount Maturity ALL TL FX ALL TL FX -2.744*** [0.082] -0.346*** [0.013] 0.349*** [0.010] 9.331*** [0.774] -2.812*** [0.084] -0.357*** [0.014] 0.357*** [0.011] 9.394*** [0.810] -2.190*** [0.295] -0.240*** [0.050] 0.284*** [0.038] 10.232*** [2.223] -1.319*** [0.075] -0.141*** [0.012] 0.184*** [0.010] -1.162 [0.732] -1.437*** [0.077] -0.157*** [0.013] 0.201*** [0.010] -0.878 [0.777] -0.818*** [0.227] -0.113*** [0.038] 0.098*** [0.029] 9.407*** [1.557] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects Yes Yes Yes Yes Yes Yes n/a Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes n/a Yes Yes Yes Yes Yes 0.688 3,977,074 0.678 3,821,181 0.695 155,893 0.585 3,871,723 0.605 3,728,171 0.578 143,552 0.30 0.00 4.40 4.30 4.40 -1.50 A f ter MPt−1 A f ter * MPt−1 Constant R2 Number of Observations %∆ with increase in the MP by 100 bp after CGF: Window: 9-Months Dependent Variable: Currency Type Amount Maturity ALL TL FX ALL TL FX -0.443*** [0.028] -0.045*** [0.004] 0.059*** [0.004] 8.896*** [0.431] -0.411*** [0.030] -0.044*** [0.004] 0.056*** [0.004] 9.042*** [0.456] -0.938*** [0.097] -0.082*** [0.012] 0.127*** [0.012] 8.542*** [1.166] -0.018 [0.025] -0.059*** [0.003] 0.023*** [0.003] 4.709*** [0.389] -0.047* [0.026] -0.066*** [0.003] 0.028*** [0.003] 5.095*** [0.421] -0.410*** [0.074] -0.025*** [0.009] 0.045*** [0.009] 7.792*** [0.791] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Currency Type Fixed Effects Time (Year) Fixed Effects Yes Yes Yes Yes Yes Yes n/a Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes n/a Yes Yes Yes Yes Yes 0.668 5,788,612 0.656 5,546,286 0.680 242,326 0.558 5,627,870 0.577 5,404,672 0.549 223,198 1.40 1.20 4.50 -3.60 -3.80 2.00 A f ter MPt−1 A f ter * MPt−1 Constant R2 Number of Observations %∆ with 100 bps increase in MP after CGF: Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate.Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2016:M04-2017:M09. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”), not included (”No”), not applicable (”n/a”) or spanned by another set of effects (”-”). ”-” also indicates dropped variables due to the fixed effects. *** Significant at 1%, ** significant at 5%, * significant at 10%. 46
49. 47 Yes Yes Yes Yes Yes Yes Yes No TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Bank Fixed Effects Currency Type Fixed Effects Firm Fixed Effects Time (Year) Fixed Effects (Firm*Month) Fixed Effects Yes n/a Yes Yes No Yes Yes Yes 0.31 0.704 2,820,540 -0.081*** [0.004] 0.002* [0.001] 3.427*** [0.440] -1.076 [1.444] 1.69 0.728 1,991,574 TL (4) Yes n/a Yes Yes Yes -0.31 0.707 832,020 -0.002 [0.002] 0.459*** [0.106] - 0.92 0.701 592,807 0.006** [0.002] 1.487*** [0.193] - Amount -0.307*** [0.009] 0.011*** [0.002] 5.415*** [0.780] -16.732*** [2.577] TL (3) Yes Yes Yes Yes No Yes Yes Yes 2.92 0.713 109,454 -0.021 [0.013] 0.019*** [0.005] -4.836*** [1.313] 18.489*** [4.403] 5.08 0.732 72,209 -0.155*** [0.033] 0.033*** [0.007] -4.271 [3.003] 12.704 [10.140] FX (5) Yes Yes Yes Yes Yes 2.00 0.717 53,952 Yes Yes Yes Yes No Yes Yes Yes 0.77 0.603 2,858,018 -0.118*** [0.003] 0.005*** [0.001] 6.400*** [0.398] -6.851*** [1.293] 0.46 0.631 2,014,713 9-Months 0.013 [0.008] -0.641** [0.316] - 4.62 0.714 35,288 ALL (7) -0.429*** [0.008] 0.003** [0.002] 10.162*** [0.776] -38.260*** [2.554] 6-Months 0.030*** [0.010] -0.823 [0.751] - FX (6) Yes Yes Yes Yes Yes 0.77 0.528 889,686 0.005** [0.002] 0.406*** [0.097] - 0.92 0.531 630,775 0.006** [0.002] 1.764*** [0.187] - ALL (8) Yes n/a Yes Yes No Yes Yes Yes 0.77 0.621 2,757,583 -0.125*** [0.003] 0.005*** [0.001] 7.089*** [0.419] -8.653*** [1.359] 0.46 0.649 1,948,484 TL (10) Yes n/a Yes Yes Yes 0.92 0.554 801,723 0.006*** [0.002] 0.432*** [0.104] - 0.92 0.557 572,086 0.006** [0.003] 1.700*** [0.193] - Maturity -0.438*** [0.008] 0.003* [0.002] 10.291*** [0.799] -38.846*** [2.623] TL (9) Yes Yes Yes Yes No Yes Yes Yes 1.08 0.610 100,435 -0.001 [0.010] 0.007* [0.004] -3.272*** [0.906] 14.828*** [3.034] 3.08 0.634 66,229 -0.096*** [0.026] 0.020*** [0.005] -3.297 [2.295] 14.347* [7.738] FX (11) Yes Yes Yes Yes Yes -0.92 0.615 48,187 -0.006 [0.006] -0.442* [0.243] - 0.92 0.611 31,407 0.006 [0.008] -0.474 [0.601] - FX (12) Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Credit Ratio, Deposit Ratio, Capital Ratio, Liquidity Ratio, ROA Ratio, NPL Ratio and FX Non-Core Ratio. Analysis covers the period of 2016:M04-2017:M09. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”), not included (”No”), not applicable (”n/a”) or spanned by another set of effects (”-”). ”-” also indicates dropped variables due to the fixed effects. *** Significant at 1%, ** significant at 5%, * significant at 10%. Yes Yes Yes Yes Yes 0.00 0.46 0.000 [0.002] 0.371*** [0.101] - %∆ with increase in the MP by 100 bp by banks having lower (25%) versus higher (75%) CGF ratio: -0.076*** [0.004] 0.003*** [0.001] 2.662*** [0.422] 0.820 [1.391] 0.92 0.701 658,887 0.006*** [0.002] 1.542*** [0.179] - ALL 0.707 931,123 Window: 1.85 0.732 2,063,783 -0.304*** [0.009] 0.012*** [0.002] 5.288*** [0.761] -16.922*** [2.523] ALL (2) 0.711 2,929,994 R2 Number of Observations Constant CGF Ratiot−1 MPt−1 * CGF Ratiot−1 MPt−1 %∆ with increase in the MP by 100 bp by banks having lower (25%) versus higher (75%) CGF ratio: R2 Number of Observations Constant CGF Ratiot−1 MPt−1 * CGF Ratiot−1 MPt−1 Window: Currency Type Dependent Variable: (1) Table 14: The Effects of Monetary Policy and Macroprudential Policy (CGF Loans) on the Amount and Maturity of Loans
50. 48 Yes Yes Yes 0 .763 6,876 -6.80 Bank Fixed Effects Firm Fixed Effects Time (Year) Fixed Effects R2 Number of Observations ∆% with increase in the MP by 100 bp: -5.20 0.768 6,875 Yes Yes Yes Yes Yes Yes -2.80 0.565 10,123,175 Yes Yes Yes No No No (4) -1.60 0.573 10,123,175 Yes Yes Yes Yes Yes Yes -0.016*** [0.001] -0.062 [0.130] Large -0.028*** [0.000] 4.344*** [0.008] Amount (3) -13.40 0.638 6,747 Yes Yes Yes No No No (6) -14.70 0.648 6,746 Yes Yes Yes Yes Yes Yes (7) -1.00 0.497 9,828,951 Yes Yes Yes No No No (8) -1.80 0.509 9,828,951 Yes Yes Yes Yes Yes Yes -0.018*** [0.000] 8.465*** [0.078] Large -0.010*** [0.000] 6.126*** [0.002] Maturity -0.147*** [0.015] 14.594*** [1.329] Small -0.134*** [0.011] 6.619*** [0.100] (5) Note. – This robustness table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Capital Ratio, Liquidity Ratio, Credit Ratio, Deposit Ratio, ROA and NPL Ratio. Analysis covers the period of 2008:M01 – 2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”) or not included (”No”). *** Significant at 1%, ** significant at 5%, * significant at 10%. No No No (2) -0.052*** [0.020] 5.825*** [1.250] Small -0.068*** [0.016] 5.859*** [0.275] TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Constant MPt−1 Bank Size Dependent Variable: (1) Table A1: The Effects of Monetary Policy on the Amount and Maturity of Turkish Banks’ New Domestic Lending across Different Bank Sizes
51. 49 Dummy 1 .00 0.592 20,918,862 Yes Yes Yes -0.30 0.587 19,882,251 Yes Yes Yes Yes Yes Yes -0.013*** [0.002] -0.015*** [0.000] -0.003*** [0.000] 0.267*** [0.054] Loosening vs Tightening Amount (3) -0.30 0.567 18,590,889 Yes Yes Yes Yes Yes Yes -0.013*** [0.002] -0.016*** [0.000] -0.003*** [0.000] -0.578*** [0.056] TL (Loosening vs Tightening) (4) 0.00 0.601 1,291,362 Yes Yes Yes Yes Yes Yes 0.007 [0.007] -0.009*** [0.001] 0.000 [0.001] -2.303*** [0.121] FX (Loosening vs Tightening) (5) 1.70 0.584 20,918,862 Yes Yes Yes Yes Yes Yes -1.190*** [0.051] -0.030*** [0.000] 0.017*** [0.000] 1.446*** [0.062] Small vs Large (6) -0.90 0.473 20,104,059 Yes Yes Yes Yes Yes Yes 0.743*** [0.004] -0.012*** [0.000] -0.009*** [0.000] 2.949*** [0.035] TL vs FX (7) -0.70 0.468 19,108,213 Yes Yes Yes Yes Yes Yes 0.076*** [0.002] -0.007*** [0.000] -0.007*** [0.000] 3.147*** [0.038] Loosening vs Tightening (8) -0.70 0.501 17,911,523 Yes Yes Yes Yes Yes Yes 0.081*** [0.002] -0.006*** [0.000] -0.007*** [0.000] 3.249*** [0.039] TL (Loosening vs Tightening) Maturity (9) -0.20 0.419 1,196,690 Yes Yes Yes Yes Yes Yes -0.031*** [0.006] -0.020*** [0.001] -0.002*** [0.001] 3.056*** [0.111] FX (Loosening vs Tightening) (10) 0.40 0.466 20,104,059 Yes Yes Yes Yes Yes Yes 0.058 [0.039] -0.015*** [0.000] 0.004*** [0.000] 3.079*** [0.050] Small vs Large (11) Note. – The table reports estimates from ordinary least squares regressions. The dependent variable is the natural logarithm of amount or maturity of Turkish banks’ new domestic lending to firms. Table 1 contains the definition of all variables and the summary statistics for each included variable. For each column, the Dummy variable takes the value 0 for the first group stated in the column title. TR Macro Variables are yearly change in industrial production index, inflation rate and monthly change in reel effective exchange rate. Bank Variables include the lagged values of Bank Total Assets, Capital Ratio, Liquidity Ratio, Credit Ratio, Deposit Ratio, ROA and NPL Ratio. Analysis covers the period of 2008:M01 – 2016:M12. Coefficients are listed in the first row, robust standard errors are reported in the row below, and the corresponding significance levels are placed adjacently. Fixed effects or control variables are either included (”Yes”), not included (”No”). *** Significant at 1%, ** significant at 5%, * significant at 10%. -1.70 0.595 20,918,862 R2 Number of Observations ∆% marginal difference with increase in the MP by 100 bp: Yes Yes Yes Bank Fixed Effects Firm Fixed Effects Time (Year) Fixed Effects Yes Yes Yes 0.911*** [0.004] -0.015*** [0.000] 0.010*** [0.000] 0.270*** [0.051] 0.599*** [0.002] -0.010*** [0.000] -0.017*** [0.000] 1.562*** [0.051] Yes Yes Yes TL vs FX (2) Short-Term vs Long-Term TR Macro Variables Bank Variables Bank-Firm Relationship Ratio Constant Dummy*MPt−1 MPt−1 Dependent Variable: (1) Table A2: The Effects of Monetary Policy on the Amount and Maturity of Turkish Banks’ New Domestic Lending across Different Breakdowns
52. 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) Do Local and Global Factors Impact the Emerging Markets’s Sovereign Yield Curves? Evidence from a Data-Rich Environment (Oğuzhan Çepni, İbrahim Ethem Güney, Doruk Küçüksaraç, Muhammed Hasan Yılmaz Working Paper No. 20/04, March 2020) The Role of Imported Inputs in Pass-through Dynamics (Dilara Ertuğ, Pınar Özlü, M. Utku Özmen, Çağlar Yüncüler Working Paper No. 20/03, February 2020) Nowcasting Turkish GDP with MIDAS: Role of Functional Form of the Lag Polynomial (Mahmut Günay Working Paper No. 20/02, February 2020) How Do Credits Dollarize? The Role of Firm’s Natural Hedges, Banks’ Core and Non-Core Liabilities (Fatih Yılmaz Working Paper No. 20/01, February 2020) 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)