In the light of recent trends showing consistent rise in BankNPAs in India, this paper undertakes a critical analysis of assetquality impairment, and examines factors (macroeconomicvariables) that have a causal effect on Bank Asset Quality,thereby identifying key areas of focus for policy action. Data (onNPAs & other key variables) is collected for a period of 20 years;1997-2016. The Econometric model used is multivariable linearOrdinary Least Squares (OLS) regression.The paper outlines the following:? Literature Review? Data Description? Methodology? Results2018Harini Kapali,Year 3, BSc (Research) Economics,Shiv Nadar University, Delhi NCRContact Details: 81309 85647 [email protected] 2N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentCONTENTSTITLE PAGEAbstract 3Introduction 4Literature Review 5Data Description 7Methodology 9Results 10Conclusion 12References 14Appendix 15Harini 3N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentABSTRACTThe Indian banking sector is considered to be the mainstay of “financial intermediation,monetary policy transmission, credit delivery, and payment systems” in the country. Therefore,sound health of the banking system is inevitable for overall economic stability and growth. Onecrucial determinant of sound functioning of the banking system is the asset quality of banks.Non-performing assets (NPAs), defined as “a credit facility where the interest and/ or instalmentof principal has remained ‘past due’ for a specific period of time”, have profound consequenceson a bank’s liquidity, profitability and sustainability. Rising Bank Non-Performing Assets (NPAs)have been a growing concern in India in recent times, particularly since 2012. Against thisbackdrop, this paper attempts to undertake a critical analysis of the asset quality impairment ofbanks in India, and examines factors (macroeconomic variables) that have a causal effect onBank Asset Quality, thereby identifying key areas of focus for policy action. Data on Bank Non-Performing Assets (used as a proxy for Bank Asset Quality), Gross Domestic Product for India,Gross Domestic Product for World, Wholesale Price Index, Stock Prices, and Bank CreditGrowth, is collected for a period of 20 years; 1997-2016. The Econometric model used here forexamining the link between NPAs and the above macroeconomic variables is multivariablelinear Ordinary Least Squares (OLS) Regression. The extent (magnitude) and direction ofimpact (whether positive or negative) that the five explanatory variables have on Bank NPAs isquantified through STATA computations. These findings could prove crucial in designing policyinitiatives that help salvage the present scenario in the banking sector in India – a scenarioplagued by rising impairment in asset quality. This paper thus adds strength to the hypothesisthat asset quality of banks in India can be managed well, if policy actions aim for a robustdomestic and external market, along with steady growth and reasonable rates of inflation.Harini 4N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentINTRODUCTIONRising Bank Non-Performing Assets (NPAs) have been a growing concern in India in recenttimes, particularly since 2012. Against this backdrop, this paper attempts to undertake a criticalanalysis of the asset quality impairment of banks in India, and examines factors(macroeconomic variables) that have a causal effect on Bank Asset Quality, thereby identifyingkey areas of focus for policy action. Data on Bank Non-Performing Assets (used as a proxy forBank Asset Quality), Gross Domestic Product for India, Gross Domestic Product for World,Wholesale Price Index, Stock Prices, and Bank Credit Growth, is collected for a period of 20years; 1997-2016. Due to non-availability of Bank NPA data prior to 1996, the analysis isconfined to a period of 20 years starting from 1997 till 2016. Data is sourced from The WorldBank, World Development Indicators, RBI Data warehouse, IMF International FinancialStatistics and OECD data. The Econometric model used here is multivariable linear OLS(Ordinary Least Squares) regression.The OLS model estimates a regression equation for NPA as follows:NPA = 18.694 – 1.613 I_GDP + 1.416 W_GDP – 0.372 WPI – 0.238 BCR+ 0.053 BSE + ?The above equation shows that Bank Asset Quality (NPAs) can be explained by a host ofexplanatory macroeconomic variables: GDP of India, GDP of World, WPI, BSE prices, and Bankcredit growth. Policy action that is prudently targeted at managing these indicators would go along way in offsetting some of the damage caused by rising impairment in asset quality.Harini 5N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentLITERATURE REVIEWThe objective in this section is to focus on existing research where macroeconomic variablesthat affect Bank Asset Quality are examined. The following table gives the Literature Review,outlining a summary of the methods and outcomes observed in some such papers.Research PaperMethod, Outcome ; Interpretation of ResultsRe-emerging Stress inthe Asset Quality ofIndian Banks: Macro-Financial Linkages, RBIWorking Paper Series,Shashidhar M. Lokare,March 2014NPAs (Non-Performing Assets) are a crucial indicator of Bank Asset Quality and credit-risk management.This paper undertakes a critical analysis (micro-level sources, bank-group- wise pointers, sectoralanalysis, and macro-financial linkages) of rising asset quality impairment of scheduled commercial banks(SCBs) since 2012. One solution to this issue is restructuring of bank loans, having both positive as wellas negative consequences. The following OLS regression is estimated to examine the relationshipbetween GDP, credit growth, lending rates, stock prices, inflation, and the level of NPAs (dependentvariable).NPAGt = a+ ?1 GDPGt-1 + ?2 CRGDPRt-11 + ?3 MMKTRATEt-1 + ?4 BANKEXt-2 + ?5WGDP t-10+ ?6 WPIINFLt-5 + ?7 D2004Q3 + ?8 D2005Q3 + ?9 D2008Q4 (Quarterly Data: 2001-2013)Asset Quality of IndianBanks: Way Forward,N.S. Vishwanathan,Deputy Governor, RBI,August 2016This RBI Bulletin talks about the huge stock of stressed assets (as a % of gross advances) piled up in thebanking system over the years: incremental accretion to NPAs with credit growth, irrational exuberance,restructured assets, postponing recognition of deterioration in loan portfolios, and the maximum stresspresent in industry ; infrastructure sectors with PSBs facing greatest strain. Elements of efficient creditrisk management avoiding recklessness, like portfolio diversification, spreading risk culture for strongunderwriting, effective pre-disbursement control, and the “3 lines of defence model” (loan officers, creditrisk managers ; internal audit), are discussed. RBI ; Government have issued regulations to dealsquarely with stressed assets; this being crucial for the nation’s economic growth.Asset Resolution &Managing NPAs – What,Why and How, S. S.Mundra, DeputyGovernor, RBI, 1 st CIIBanking Summit,February 2016This paper presents data showing extent of NPAs and states that contrary to popular perception, stress isrelatively much less in priority sector. Restructuring is mostly in larger accounts. Factors contributing toNPAs are: External – Global slowdown, Fall in domestic demand, Policy logjam & Disputed Contracts;Internal – High Leverage, Poor credit appraisal & Financial Instability of Banks & Corporate entities.Unhedged exposures & weak risk management are also causes.Asset Resolution is cited as the solution – Acknowledge (Shift the balance from borrowers to lenders),Redress or recover, and Prevention better than cure. Borrowers & Banks, both need to cooperate.Bank Asset Quality inEmerging Markets:Determinants andSpillovers, IMF WorkingPaper, Reinout De Bockand AlexanderDemyanets, March 2012An OLS regression is used to estimate that the following variables give statistically significant coefficientsand the model explains about 53% of the variation in NPL (Non-performing Loans as a ratio of total loans):negative relationship of NPL with growth in real GDP, capital inflows, trade (goods), nominal exchangerate; and positive relationship of NPL with credit growth & the 1 st lag of NPL (ratio of total loans). Data iscollected for 25 emerging markets for a period of 14 years; 1996 – 2010.Determinants of BankAsset Quality &Profitability – AnEmpirical Assessment,Vighneswara Swamy,IBS Hyderabad, January2012This paper uses data from India for the years 1997-2009 to estimate an OLS of the form:GNPAit = ? + ?1 GDPGRt + ?2 ERt + ?3 MCAPt + ?4 LRt + + ?5 IIPGRt + ?6 INFLAt + ?7 SVGRt + ?8ASSETit + ?9 CARit + ?10 CDRit + ?11 COFit + ?12 ROAit + ?13RUSUBRAit + ?14 CREDGRit + ?15PSCit + ?16 OERit + ?17 ROIit + ?i + ?t + itHere, explanatory variables like GDP, exchange rate, market capitalization rate, lending rate, inflation, IIP,capital adequacy ratio, ROI, credit to deposit ratio, PSC, ROA, etc. explain the variation in Gross NPA(GNPA), calculated as a ratio of Gross Non-Performing Assets to Total Advances. The regression resultsare in tandem with the pattern seen in earlier papers (above) & reinforce the direction of causality betweenHarini 6N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentNPAs and other macroeconomic variables.According to the above analysis of existing research, it is seen that OLS (Ordinary LeastSquares) regression is the most common econometric method used to examine the factorsaffecting Bank Asset Quality. The papers studied build a model to explain NPAs (Non-Performing Assets), used as a proxy for Bank Asset Quality. An asset can turn in to a NPAwhen the borrower defaults on his repayment of interest and/or principal on agreed terms. Theextent of NPAs and their disaggregation over sectors and bank-groups epitomize credit riskmanagement and efficacy in the allocation of resources. NPAs are also an important prudentialindicator to assess the financial health and sound functioning of the banking system, affectingoperational efficiency, profitability, liquidity and solvency position of banks.Macroeconomic variables like real GDP growth, credit-GDP ratio, nominal exchangerate, capital inflow, inflation, lending rate and stock prices are used to explain NPAs (as a % oftotal advances). The general consensus from all the papers is that NPAs typically have apositive relationship with credit growth, inflation and interest rates, and a negative relationshipwith GDP growth, exchange rate and asset prices. There are observations that show otherwisetoo, with sound economic reasoning behind them.From the findings, it is possible to conclude that the above models provide reliable(statistically significant) estimates of the coefficients explaining the relationship between NPAs /Bank Asset Quality and various macroeconomic variables. Therefore, if the broad outline givenby these models is simulated, it is possible to discern and explain the factors that determineBank Asset Quality in India.Harini 7N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentDATA DESCRIPTIONData is collected for a period of 20 years; 1997-2016, under the following heads:Bank Nonperforming Loans to Gross Loans, %,Not seasonally adjustedNPA**Growth Rate (%) of Constant GDP Per Capita forIndia, 2010 U.S. Dollars, Annual,Not Seasonally AdjustedI_GDPGrowth Rate (%) of Constant GDP Per Capita forWorld, 2010 US Dollars, Annual,Not Seasonally AdjustedW_GDPGrowth (%) of Annual Closing Stock Prices,S&P BSE SENSEXBSEInflation Rate (%) of Wholesale Price Index(2010 = 100)WPIGrowth Rate (%) of Bank Credit BCR** Data on Bank NPAs for India is not available prior to 1996-97.(TABLE 6, Appendix)Harini 8N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentData Summary TableNPA I_GDP W_GDP BSE WPI BCRMean 7.086 5.338 1.634 16.798 5.051 18.125Median 5.542 5.982 1.557 15.255 4.851 16.950Std. Dvtn. 4.621 2.100 1.365 35.078 2.836 6.353Skewness 0.673 -0.411 -1.834 0.094 -0.800 0.768Kurtosis -0.996 -1.055 5.692 -0.447 1.796 -0.003Max 15.700 8.755 3.155 81.033 9.561 30.900Min 2.212 2.017 -2.891 -52.446 -2.737 9.000DATA SOURCES: The World Bank, World Development Indicators,RBI Data Warehouse,International Financial Statistics, IMF Financial Soundness Indicators,OECD dataHarini 9N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentMETHODOLOGYThe method used here for examining the link between macroeconomic variables and NPA ismultivariable linear Ordinary Least Squares (OLS) Regression (FIGURE 1, Appendix). Thismethod is commonly used to build an econometric model that uses various independentvariables to explain the variation in the dependent variable. This method is interpreted here asone examines causality, i.e., factors affecting Bank Asset Quality.In statistics, ordinary least squares (OLS) or linear least squares is a methodused for estimating the unknown parameters in a linear regression model, with the goal ofminimizing the sum of the squares of the differences between the observed responses (valuesof the variable being predicted) in the given dataset and those predicted by a linear function of aset of explanatory variables. OLS entails certain assumptions like approximation to the normaldistribution, homoskedasticity and the absence of multicollinearity and autocorrelation.A two-variable linear OLS regression is of the form:Y = ?0 + ?1 X1 + ?2 X2 + ?,where ?1 and ?2 are the two slope coefficients and ?0 is the intercept coefficient; all 3 areestimated. ? is the error term that explains the variation in dependent variable Y that is leftunexplained by the independent variables X1 and X2. The slope coefficient shows the change inY when there is a 1% change in the corresponding explanatory variable.In this study, the OLS estimate uses NPA (Bank Nonperforming Loans to Gross Loans,%) as a proxy for Bank Asset Quality (dependent variable). Independent variables used includeI_GDP, W_GDP, WPI, BSE, and BCR, as explained in the data description section.Harini 10N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentOLS RESULTSregress NPA I_GDP W_GDP WPI BCR BSENPA Coef. t-statI_GDP -1.613219 -3.11W_GDP 1.415714 2.10WPI -0.3722072 -1.07BCR -0.2381867 -1.54BSE 0.052824 1.66_cons 18.69377 5.62TABLE 2 in the Appendix provides the OLS Regression Results for a linear model estimated asfollows:NPA = ?0 + ?1 I_GDP + ?2 W_GDP + ?3 WPI + ?4 BCR + ?5 BSE + ?Here, ?0 represents the intercept co-efficient, ? represents the error term, and the other ?s areslope coefficients that represent the individual contributions of independent variables to theprediction of the dependent variable.From the STATA output, we see that the model takes the form:NPA = 18.694 – 1.613 I_GDP + 1.416 W_GDP – 0.372 WPI – 0.238 BCR+ 0.053 BSE + ?? D.o.F. = n – k = 20 – 6 = 14? R-Squared is 54.34%, i.e., the above model (the set of independent variables chosen)explains 54.34% of the variation in NPA? Prob > F = 0.0344 (This is the p-value of the model)? p < 0.05; the above model is Statistically SignificantHarini 11N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality Impairment? The 2-tail t-test results show that the coefficients of I_GDP and W_GDP are statisticallysignificant at the 5% level, while the coefficients of BCR and BSE are significant at the20% level. It is also seen that the coefficient of WPI is just about statistically significant;missing the threshold by just 0.2 points. This discrepancy can be attributed to the factthat the analysis contains 20 data points (due to unavailability of Bank NPA data).? The extent (magnitude) and direction of impact (whether positive or negative) that thefive explanatory variables have on Bank NPAs is quantified above. This input couldprove crucial in designing policy initiatives that help salvage the present scenario in thebanking sector in India – a scenario plagued by rising asset quality impairment.? This paper thus adds strength to the hypothesis that asset quality of banks in India canbe managed well, if there exists a robust domestic and external market, with steadygrowth and reasonable rates of inflation.Harini 12N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentCONCLUSIONThe OLS results obtained are as follows:NPA = 18.694 – 1.613 I_GDP + 1.416 W_GDP – 0.372 WPI – 0.238 BCR+ 0.053 BSE + ?The t-tests and F-test show that the individual coefficients as well as the model, both arestatistically significant. This shows that I_GDP, W_GDP, WPI, BCR & BSE can indeed be usedas explanatory variables to study the causality of Bank NPAs.The coefficient for I_GDP shows that a 1% rise in India’s GDP results in fall in NPAs by1.613 percentage points. Economic theory also reinforces the inverse relationship betweenGDP ; NPAs growth, as general economic slowdown impinges on the performance of banksand financial institutions. On the other hand, the results show a positive relationship betweenWorld GDP growth and NPAs. A robust external environment may encourage borrowers toinvest in profit-making ventures, rather than on repayment of debt; or the positive coefficientmay have arisen due to the analysis being based only on the 20 data points available. Furtherinvestigation could be carried out in this regard.A unit increase in the Wholesale Price Index causes bank NPAs to fall by 0.372 units.High inflation may help borrowers, whose debt is denominated in nominal terms, as it erodesthe real value of debt. Also, there is evidence that banks’ write-off ratio increases after increasein retail price inflation and nominal interest rates (RBI Working Paper, Shashidhar M. Lokare).Here, it is seen that Bank credit growth and NPAs have a negative relationship.Consensus is that the more aggressive banks are in their lending, they may end up pushingriskier loans and thereby end up in higher level of NPAs. However, there is also a contentionthat as banks concentrate more on credit management, they may have developed expertise inmanaging the credit risk and hence may sometimes exhibit lower level of NPAs. Therefore, therole of lending aggressiveness in causing increase in NPAs is still hazy (Determinants of BankAsset Quality and Profitability, Vighneswara Swamy, IBS Hyderabad). Asset prices (measuredby growth rate, %, of closing prices of BSE SENSEX) show a positive relationship with NPAs;1% increase in stock prices leads to 0.053% increase in NPAs.Harini 13N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentTABLE 5 in the Appendix shows a separate linear regression that was estimated by includingLending Rate (%) as one of the explanatory variables. Though 3 of the individual coefficients inthis regression are insignificant, it is seen that the model as well as 3 other coefficients arestatistically significant. This regression shows that Lending Rate and NPAs have a positive co-movement; a unit increase in the lending rate leads to an increase in NPAs by 1.64 units. This isbecause hardening of interest rates makes repayment of loans difficult for borrowers.Additionally, higher interest rates may result in adverse selection of borrowers, with only theriskier ones left in the market (RBI Working Paper, Shashidhar M. Lokare).It is seen that the results obtained are in tandem with the patterns observed in theresearch papers studied, and with economic intuition as well. We could therefore attribute thecausality of Bank Asset Quality to macroeconomic variables like GDP, bank credit, WholesalePrice Index, stock prices and lending rate.Also, the above study could be enhanced if it is considered in the light of currentvalues of macroeconomic variables. It may also help if all the actual data points collected(available in terms of 2010 US $) were converted into 2011 values. This may help reconcile theresults obtained here with today’s numbers; e.g. current growth rate of GDP that is a little over7%. The results could also be improved if data on Bank NPAs is available for a longer period oftime. This analysis is original and true to the best of my knowledge. I apologize for inadvertenterrors that may be found.Harini 14N C Ray Paper Presentation, 2018 Macro-Financial Linkages of Bank Asset Quality ImpairmentREFERENCES? “Re-emerging Stress in the Asset Quality of Indian Banks: Macro-Financial Linkages”,RBI Working Paper Series, Shashidhar M. Lokare,