The 2007-8 global financial crisis also termed as ‘the great recession’ led to a grave banking panic and threw most of the economies of the world into severe recession. That crisis is attributed as having been caused by several factors. But only few of the factors are associated with the housing and credit markets, the two commonly-cited factors. Insinuated causes include the insolvency of homeowners to meet their mounting mortgage payments, unusually high levels of personal and corporate debt levels, foolhardy financial product innovation, the collapse of vital financial institutions and awful errors of judgment by credit rating agencies in the rating of structured products. Macroeconomic factors have also been added: monetary policy, excessive liberalisation, global trade imbalances and ineffective government regulation.

Without refuting the impairing effects of the above factors, we believe that the origins of the global financial crisis lie in the rational profit-seeking behavior of banks. Indeed, we consider that reason in the course of economic expansion is a vital cause behind the pro- cyclical financial infirmity of any economic system given bank’s heightened sensitivity to a financial shock.

The global financial crisis has spread across the world through a range of financial as well as real economy channels. Cross-border bank lending has also been one of the major financial channels through which stresses in the global financial system are and were spread to other emerging economies. While for some economies in Europe, the transmission conduit for the losses on account of toxic assets was obvious and thereby increased the fragility of bank balance sheets. However, for some Asian economies such as that of India, the story was a bit different as the exposure of their domestic banks was lot less limited to such international toxic assets. Yet even these economies with little exposure did experience a substantial slow down. Hence, toxic assets alone cannot explain the virility of the effect.

The connection of a recession to banking crisis is an intensely problematic research issue. The consequent fall in bank lending experienced in both the exposed or little exposed economies has key implications for growth. However, the sag in credit supply puts upward pressure on interest rate spreads, and thereby leads to an inordinate fall in lending than one might see in a typical recession. What makes the banks squeeze their lending activity during the crisis? It is an oft-repeated question, which deserves re-examination to find credible economic reason(s) applicable across diverse banking systems across the globe. The available literature, though throws up some interesting insights. We attempt to find some more answers from a rather relatively conservative banking system like that of an emerging economy in India. Indian banking similar to much of Asian banking is characterized by sizeable proportion of ownership under government control, and is distinct from the free market and widely-held banks of the Anglo-Saxon model.

This paper is one among the relatively small family of works that underline the relationship between bank behavior and financial instability. It aims to analyze how the financial crisis passed through the banks’ lending behavior. Our endeavor in this study is to understand the impact of financial uncertainty and instability on banks’ behavior on credits more specifically the effect of the crisis on the lending behaviour of the banks.

We begin by presenting in section 2, the theoretical framework illustrating the recent approaches on financial instability placing the banking system at the epicenter of analysis. We analyze more specifically how financial instability affects the bank’s lending behavior during times of economic shocks or crisis. The methodology involving the data and its sources and research design explaining the empirical framework including how to estimate the impact of factors as is done in section 3. The results of the analyses of the findings are presented in section 4. The conclusion and policy implications are offered in section 5.

We use data on bank behavior in India to provide fresh insights into how banks respond to financial crises especially in the area of credit supply. The framework of our empirical analysis is based on a robust database and well-established techniques.

Tables 3, 4 and 5 present the correlations statistics of the determinants employed in the analysis for the pre-recession period, recession period and recovery period. During the Pre- Recession period LnBC has strong positive correlations with; LnAD (0.983), LnINVEST (0.946), LnBORROW (0.811), LnBPLR (0.433) at 0.01% level of significance (Table-3).

LnINVEST (0.916), LnBNIFTY (0.698) at 0.01% level of significance and has negative correlation with; LnMATCAL (-0.659) and LnBPLR (-0.797) at 0.01% level of significance (Table 4). During the Recovery period LnBC has strong positive correlations with; LnAD (0.518), LnINVEST (0.728), LnBPLR (0.422), LnBNIFTY (0.426) at 0.01% level of significance and has negative correlation (Table 5).

(significant) at the 0.01 percent level, indicating that all the variables during these three periods are integrated at the first degree and satisfied the condition for the cointegration test.

Johansen (1988) and Johansen and Juselius (1990) developed this methodology (cointegration) for assessing long-run relationships. The methodology applies maximum likelihood procedure to determine the presence of cointegrating vectors in a set of non- stationary time series.

appropriate lags are carefully selected with a number of multivariate diagnostic tests. The appropriate lag orders can be selected from five information criteria (i.e., the likelihood ratio, the Final Prediction Error (FPE), the Akaike Information Criterion (AIC), the Schwarz Information Criterion (SIC), and the Hannan Quinn Information Criterion (HQIC)) are first determined. For doing cointegration, the study selected lag interval of one (for all three models) as suggested by Schwarz Information criteria (SIC). In this study, both trace and maximum eigenvalue statistics are considered in determining the number of cointegration vectors. In particular, if two test statistics show the same number of cointegration, that number is used. If two statistics do not show the same number of cointegration, the number is selected using trace test. Luintel and Khan (1999) have shown that trace test is more robust than maximum eigen value criteria in testing the cointegration.

Having confirmed that integration of the seven series is of the same order (checked for three periods); we test whether the seven series are cointegrated over the sample period. The numbers in the tables 9, 10 and 11 show the results of the Johansen test. Since the Johansen test is based on vector auto regressive model (VAR), we select one period lag for the model based on SIC. Starting with the null hypothesis of no cointegration among the variables, trace statistics is 220.6801, which is well above the 0.05 critical value. Thus, we reject the null hypothesis of no cointegration among these variables at 0.05 and accept that there is one co- integrating equation.

Considering the null hypothesis of at most one co-integrating relation, based on the trace statistics of 131.3113, which is greater than 0.05 critical value of 83.93712 we reject the null hypothesis of at most one co-integrating equation at 0.05 significance level. Similar is the case with the second and third cointegration relationship, where we find trace statistics is greater than the critical value thus we reject the null hypothesis and infer that there are more than three cointegrating relationship between variables.

The result from Maximum Eigen Statistics table also supports that there is more than three cointegrating relationship. In addition, we find that Maximum Eigen value is greater than the critical value and hence we reject the null hypothesis of at most three cointegration vectors at 0.05 per cent level of significance. Hence based on trace test we may conclude that there are more than three co-integrating equation among these variables during pre-recession.

In the similar vein, the statistics in Table 10 shows the cointegrating relationship between the study variables during recession. In this case, both statistics (trace and maximum eigenvalue) are showing similar results. The trace statistics of 60.93088, which is greater than the critical value (54.07904) at 5 % level of significance (see at most two). Therefore, we are rejecting the null hypothesis that there are at most two cointegrating vectors and may accept the alternative hypothesis of more than two cointegrating relationship among variables.

Cointegrating relationship among variables during recovery period is shown in the Table 11. From these results, it is found that there is more than one cointegrating relationship (based on trace test). In case of trace test, the statistic is greater than the critical value at 0.05 level of significance and rejecting the null hypothesis of at most one cointegrating relationship. However, in case of maximum eigen value criteria it found that there is one co-integrating relationship. However, based on Luintel and Khan (1999) suggestion the study preferred trace test and inferred that there is more than one cointegrating relationship among study variables during recovery period.

The signs of the coefficients like LnBORROW and LnBNIFTY were found to be negative and others are positive during pre-recession (see first column in Table 12). The results show that all the variables are significant. We thus infer that bank credit increases with increase in

LnAD, LnINVEST, LnMATCAL and LnBPLR and decreases with an increase in LnBORROW and LnBNIFTY during pre-recession period.

During recession (see Table 12, column 2), the results show that, except LnMATCAL, all other variables are significant. For LnAD the coefficient is 2.61, signifying that its 0.01 level increase during recession led to more than 2.6 per cent increase in bank credits. In the same way the coefficients for LnBORROW, LnINVEST, LnBNIFTY are 1.17, -2.27(inverse relationship), 0.33 and -2.08(inverse relationship) respectively.

During recovery period (see Table 12 column 3), the variables (LnAD, LnBORROW, and LnBNIFTY) were found to be positively significant indicating that the increase in these variables led to a positive significant increase in bank credits during recovery period. The results show that during this period, LnMATCAL, LnINVEST, and LnBPLR had an inverse or negative relationship with bank credit, explaining the decrease in these factors.

This paper contributes uniquely to the evolving body of literature on bank’s lending behavior during times of financial instability/uncertainty, which mostly prevails during the times of financial crises. Our findings provides answers to the oft-repeated question “What makes the banks squeeze their lending activity during the crisis?”. Broadly, our results are in agreement with the stylized facts on bank behavior in the recent financial crisis, especially on reduced bank lending, increased competition for retail deposits, and reduced monetary policy effectiveness.

We have found that, during the times of recession, particularly in the emerging markets of India, borrowings by banks have a significant positive relationship on the lending behavior of the banks. This is because of the fact that banks are unable to garner the much needed financial resources for lending through deposits in view of the prevailing factors of uncertainty/instability (during recession) which makes the depositors withhold their deposits. Instead depositors look for other real asset investments or to hold cash till the re-appearance of the signs of financial stability in the markets.

Our study has also established that lending rates, even though they were downward during the recession period, were not having positively significant relationship with banks’ lending. This elucidates that the lending rates were required to be reduced drastically in order to infuse bank lending. Further, results indicate that investment activity of the banks had a negatively significant relationship with bank lending. The reason is that during the times of financial instability, the banks tend to tighten their lending activity by stringent screening and rigorous monitoring of their borrowers, thereby increase their investments only if it ensures safe returns (profit-seeking behavior and also due to the fall in the state of confidence of banks and the nature of firm decision structures). This can be discerned from the fact that, while Credit to Deposit Ratio showed a downward trend, Investment to Deposit Ratio showed an upward trend during the recession evidencing the era of shortened lending activity of the banks. Further, we also notice decreasing Cash to Deposit Ratio during recession period which indicates the tight liquidity scenario in view of the financial instability and uncertainty.

We are of the view that failure to curb this trend of sharp decline in bank lending further results in banking crisis (Gentler 2010). Accordingly, we suggest that the bank’s lending could be maintained at the relatively same pace during the times of recession too by increasing the borrowings rapidly either from the Central Banks or from Government supported long term lending institutions. Besides, this measure would also enable the banks to tide over the liquidity crisis that resulted out of the financial crisis.

Author Information: Vigneshwara Swamy is a professor at the Indian Business School, Hyderabad, India. He may be reached via E-mail: vswamypm@gmail.com or Phone: 91-97- 0509-6919. The co-author is S. sreejesh: E-mail: sreejeshibs@gmail.com