Synchronization, Concordance and Similarity between Business and Credit Cycles: Evidence from Turkish Banking Sector

Transcription

1. Synchronization , Concordance and Similarity between Business and Credit Cycles: Evidence from Turkish Banking Sector Mehmet Selman ÇOLAK Abdullah KAZDAL Muhammed Hasan YILMAZ October 2020 Working Paper No: 20/11
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. Synchronization , Concordance and Similarity between Business and Credit Cycles: Evidence from Turkish Banking Sector Mehmet Selman Çolak Abdullah Kazdal Muhammed Hasan Yılmaz1 Abstract In this study, we provide a comprehensive quantification of the co-movement between credit and business cycles in the Turkish case for the period 2007-2020. To this end, we construct synchronization, concordance and similarity index, which aim to measure the time-varying degree of coherence between credit and output dynamics. In specific, these indices are designed to capture the location, momentum and size aspects of the cyclical correlation respectively. Our empirical analysis also covers the cyclical association of 13 different loan sub-categories with the course of the output gap by employing disaggregated data. Overall, index results show that credit-output nexus in the Turkish case present heterogeneities across loan types, sample episodes and cyclical characteristics (location, momentum, and size). We also examine the impact of local and global macroeconomic and financial factors on cyclical coherence by utilizing Tobit regressions. The empirical results indicate that movements in local financial conditions, fluctuations in macroeconomic volatilities, and the course of capital flows are influential determinants of cyclical co-movements. Özet Bu çalışmada Türkiye örneğinde 2007-2020 dönemi için kredi ve iş çevrimleri arasındaki ortak hareketlerin sayısallaştırılması hedeflenmektedir. Bu bağlamda, zamana göre değişen kredi piyasası-ekonomik aktivite ilişki derecesini ölçen senkronizasyon, uyuşma ve benzerlik endeksleri hesaplanmaktadır. İlgili endeksler çevrimsel ilişkinin sırasıyla konum, faz ve boyut özelliklerini yakalamaktadır. Sektör geneli toplam kredi gelişmelerine ek olarak, 13 alt kredi kategorisinin de çıktı açığıyla çevrimsel uyumu incelenmektedir. Genel olarak endeks sonuçları Türkiye ekonomisi özelinde krediler ile iktisadi faaliyet arasındaki çevrimsel ilişkinin kredi türü, örneklem dönemi ve çevrim karakteristikleri (konum, faz ve boyut uyumu) açısından önemli heterojenlikler taşıdığını göstermektedir. Çalışma kapsamında ayrıca yerel ve küresel makro-finansal faktörlerin çevrimsel uyuma olan etkisi Tobit regresyon modelleriyle araştırılmıştır. Ampirik bulgular finansal koşullardaki hareketlerin, makroekonomik oynaklıkların ve sermaye akımlarının seyrinin kredi-iş çevrimi uyumunu anlamlı şekilde etkilediğini ortaya koymaktadır. JEL Classification: G21, E32, C35, C38 Keywords: Credit Cycle, Business Cycle, Synchronization, Filtering, Tobit Regression 1 Central Bank of the Republic of Turkey. E-mail addresses: selman.colak@tcmb.gov.tr, abdullah.kazdal@tcmb.gov.tr, mhy1@st-andrews.ac.uk The views expressed in this paper are those of authors and do not necessarily reflect the official views of the Central Bank of the Republic of Turkey. We gratefully acknowledge suggestions by Selva Bahar Baziki and the anonymous referee.
4. Non-Technical Summary This study aims to analyze the co-movement between credit and business cycles in Turkey from 2007 to 2020 . Initially, the time-invariant relationship is examined with the help of a simple correlation measure and distance indicators. Furthermore, to measure the time-varying degree of coherence between credit and output dynamics, we construct synchronization, concordance, and similarity indices. These indices are designed to capture the position, momentum, and size aspects of the cyclical correlation between credit and output gap series, respectively. In the following step, we investigate how macro-financial factors impact the degree of co-movements. On top of aggregate loans, our analyses are conducted for retail loans as a whole and its subcomponents of general-purpose, vehicle, housing loans and credit cards; and commercial loans broken down to investment, foreign trade, business, SME and large firm loans. The empirical results on the static measures suggest that there exists a lead-lag relationship between credit and output dynamics for which fluctuations in output growth cycle precedes credit cycle and more prominent co-movements seem to occur with 3-4 months as elapsed time. In addition to the static analysis, investigation of the dynamic relationship between credit and output series with the synchronization index reveals that, on average, the perfect synchronization is observed for 80% of the sample period, which demonstrates that total credit and economic activity stand in the same front with respect to the trend, for the majority of the sample period. The highest synchronizations with the output gap are observed for total, housing, and consumer loans; while the lowest ones are found in credit cards, investment, and business loans. The analysis with the concordance index proposes that on average, there is a perfect phase coherence between credit and output gap in nearly 60% of the sample period. Looking at the loan breakdowns, the highest level of concordance is achieved by retail loans mainly driven by housing loans while the lowest concordance with economic activity is observed in credit card and foreign trade loans. The similarity index results measuring the coherence of cyclical sizes suggest that the least discrepancy in the amplitudes of the credit and output gap is observed in consumer and housing loans, and the highest discrepancy is observed in credit cards and investment loans. Our empirical investigation on the determinants of the co-movement between credit and output gap series indicates that financial conditions significantly influence the position coherence for commercial loans (especially foreign trade, business, and SME loans). Similarly, macroeconomic volatilities are detected to affect position-wise suitability between credit and output gap mainly for commercial, general-purpose, investment, SME, and large firm loans. Synchronization seems to be elevated during the episodes characterized by stronger capital inflows for the total loans as well as sub-segments of commercial loans. Estimation results using the concordance index as the dependent variable show that tightening in local financial conditions is significantly stimulating phase coherence for consumer, housing, vehicle, business, and SME loans. On the other hand, in contrast to synchronization, macroeconomic volatilities seem to have a reversed impact on concordance, particularly for sub-components of retail loans. Moreover, capital flows tend to improve phase coherence in a statistically significant way, both for retail and commercial loans. Lastly, Tobit estimation results provide less information for the extent of cyclical similarity between credit and output dynamics. 1
5. 1 . Introduction Even in the current conjuncture characterized by well-developed and integrated equity and bond markets on a global scale, bank loans remain the most preferred method of financing in emerging market (EM) economies (Dorucci et al., 2009). Following the regulatory and monetary policyrelated measures taken in the recent decade to overcome the repercussions of the Global Financial Crisis, the overall improvement in global liquidity conditions and capital outlook in domestic banking systems shored the credit growth in emerging countries (Eickmeier et al., 2014; Dahir et al., 2019). For households, traditionally, mortgage financing in EMs constitutes a considerable amount of liabilities (in household balance sheets) and they are generally arranged in the form of bank loans (Warnock and Warnock, 2008; Badarinza et al., 2019). While the impact is more pronounced in SMEs, non-financial firms operating in EM economies can face challenges given financing constraints and the availability of collateral as the bank financing is the most important external funding in those countries (Beck et al., 2006; Kira, 2013; Dong and Men, 2014). In this context, bank credit emerges as the most representative indicator for the domestic financial cycle and its compatibleness with economic activity should be inherently highlighted as a proxy of imbalances regarding financial stability. Another factor contributing to the importance of this issue is related to the monetary policy transmission mechanism. Apart from traditional interest rate channel and expectations channel, firm credits and bank lending effectuate important bases of how monetary policy stance is transmitted to macroeconomic aggregates including growth and inflation. Particularly in EMs, credit channel is determined to function considerably so it is expected that the periods, during which credit-business cycle is strongly associated, can also potentially be seen as periods with the more efficient monetary transmission (Caballero and Krishnamurthy, 2004; Çatık and Karakuça, 2012). This issue holds importance for authorities dealing with the estimation of the effect of monetary policies on financial/macroeconomic outcomes. Furthermore, there is rooted empirical evidence in the previous literature asserting the indicative nature of drastic credit movements for the occurrence of banking/financial and real economic crises (Slingenberg and De Haan, 2011; Feldkircher, 2014; Krishnamurthy and Muir, 2017). Hence, analyzing the creditbusiness cycle coherence might improve forecasting practices. Perhaps, more relevant to the Turkish case, countercyclical financial policies aiming to tackle the deceleration in economic activity might directly focus on selective credit extension. In fact, loosening of macroprudential and reserve requirement policies as well as the introduction of new credit facilities including Treasury-backed Credit Guarantee Fund (CGF) guarantees and affordable housing loans) to enhance the credit growth in the Turkish banking sector are observed after 2016. Having a 2
6. quantitative proxy of the coherence between credit and output cycles can provide better information about the effectiveness of such policies . Turkish banking sector serves as a proper case to examine the time-varying nature of business and financial cycle co-movements. Figures 1 and 2 present the credit market outlook across three distinct periods in which economic growth sharply deteriorated from its long-term trend.2 Such periods are especially unique from each other regarding the source, cause, and duration of economic shocks as well as the level of financial development and the policy measures taken afterward. Top-left charts in Figures 1 and 2 take January 2009 as the bottom point of the economic slowdown in the wake of the Global Financial Crisis and represent the fluctuations in main loan categories in terms of both normalized level and month-on-month changes. This period was characterized by decaying external demand, contraction in global liquidity, deteriorated global investor sentiment and abnormal portfolio outflows in emerging markets, which all abridge the financial health of the non-financial companies by tightening the financial conditions and limiting the growth realizations in emerging countries (Frank and Hesse, 2009; Coulibaly et al., 2013; Dimitriou et al., 2013). The case of Turkey is subject to similar results for economic activity, particularly for the real sector firms. Alp and Elekdağ (2011) estimate a structural model to analyze the role of possible factors on the recession in the Turkish economy during this period. Their results confirm that foreign demand and financial uncertainty constitute important parts of the slump in economic growth. Demirhan and Ercan (2018) analyze the export behavior of Turkish manufacturing firms by employing firm-level data and they conclude that this period reduced the export propensity as well as the volume of exports. On the other hand, macroeconomic fundamentals in Turkey were strong before the crisis period as indicated by the relatively lower levels of dollarization, anchored inflation expectations, and subdued external debt (Kılınç et al., 2012). This background combined with expansionary fiscal policy allowed the domestic demanded to rebound considerably aftermath the crisis. The abovementioned divergence between economic agents in terms of the exposure to the economic downturn was reflected in the credit developments. Given the size of the shock, the recovery in total loans took longer compared to other crisis episodes in the recent decades. More importantly, the behavior of loan sub-categories differed in the sense that while the momentum of the commercial loan extensions was stronger than that of retail loans before the peak point, they performed relatively poorly after the point of interest. The stagnation in commercial activities together with 2 We use both annual growth rate and cyclical component of the industrial production index to identify these periods. 3
7. depressed export transactions and investment growth are argued to play important role in this plateau observed in commercial loan growth. [