A Micro Data Approach to the Identification of Credit Crunches

Transcription

1 A Micro Data Approach to the Identification of Credit Crunches Horst Rottmann University of Amberg-Weiden and Ifo Institute Timo Wollmershäuser Ifo Institute, LMU München and CESifo 5 December 2011

2 in % in bn. Introduction Given the fall in credit growth and the burdens on banks due to the current financial crisis, the fear increasingly expressed in public debate is that the German economy is experiencing a credit crunch. (Deutsche Bundesbank, Monthly Bulletin, September 2009) Loans of German banks to non-financial corporations Outstanding amounts (chg. ag. prev. year) New business (right scale) 2

3 net percentage Introduction Survey among banks: Results of the ECB Bank Lending Survey Bank Lending Survey tightened eased Credit standards applied to the approval of loans The survey is addressed to senior loan officers of a representative sample of German banks and is conducted four times a year. The banks are asked to respond to the following question: Over the past three months, how have your bank's credit standards as applied to the approval of loans or credit lines to enterprises changed? 3

4 Introduction Survey among firms: Results of the Ifo Credit Constraint Indicator The credit constraint indicator is based on ca. 4,000 responses of firms in industry and trade from the sectors manufacturing, construction, wholesaling and retailing. The firms are asked to respond to the following question: How would you assess the current willingness of banks to extend credit to businesses? The answers to choose from are accommodating, normal and restrictive. The indicator is calculated from the percentages of the firms that indicate that credit access is restrictive. 4

5 Introduction 5

6 Introduction These indicators show that the situation on the credit market in Germany has considerably worsened during the crisis. But are they good indicators for a credit crunch? What is a credit crunch? Economists generally define a credit crunch as a significant contraction in the supply of credit reflected in a tightening of credit conditions. (Udell, 2009) We define a bank credit crunch as a significant leftward shift in the supply curve for loans, holding constant both the safe real interest rate and the quality of potential borrowers. (Bernanke and Lown, 1991) 6

7 Introduction Determinants of the credit supply curve i Cr D Cr S (safe real rate, firm quality, credit crunch) Cr 1. A credit crunch shifts the supply curve to the upper left. 2. Not every shift of the credit supply curve is a credit crunch. credit supply shifts to the upper left, if the opportunity costs of providing risky loans (safe real rate) increase credit supply also shifts to the upper left, if the quality of firms deteriorates Problem: Identification using volumes and interest rates is difficult, since demand- and supply-factors have to be disentangled. 7

8 in % Introduction Not only the loan volume significantly decreased during the crisis, but also the average interest rates of German banks for loans to non-financial corporations 6.5 Loan rates Outstanding amounts New business 8

9 Introduction Determinants of the credit supply curve Since the fall in credit growth comes along with a massive i decline of loan rates, a shift of Cr D the credit demand curve to the lower left is also likely to happen. Cr S (safe real rate, firm quality, credit crunch) Cr This paper proposes an alternative approach to the identification of credit crunches: we use information about the credit supply of banks obtained from a survey. 9

10 Introduction Information of credit supply behavior of banks is obtained from the Ifo Business Survey German firms regularly respond to the following question: How would you assess the current willingness of banks to extend credit to businesses? We interpret the responses to the credit question as information from the point of view of the firms about the banks loan supply conditions. From a theoretical perspective, the survey responses are exclusively used as indicators for the location of the loan supply curve, which allows us to avoid controlling for the demand side of the loan market. 10

11 Introduction Purpose of the paper is to isolate bank industry-specific determinants of credit supply (i.e. credit supply shocks) and to derive a credit crunch indicator that represents shifts in the supply of loans, which can neither be explained by changes in the quality of potential borrowers, nor by variations in the banks opportunity costs of providing risky loans (safe real interest rate). 11

12 Introduction Two-step-procedure short overview Step 1: micro-econometric model In a first step we regress the responses to the credit question on a set of variables that provide information about the creditworthiness of the firm using a panel-data model. In addition to the firm-specific information we include a set of time dummies as regressors into our model. The estimated coefficients on the time dummies are interpreted as additional macroeconomic (opportunity costs) or bank industry-specific factors determining the loan supply decision of the bank. 12

13 Introduction Step 2: credit crunch indicator We separate the variation of lending policies over the business cycle, which is captured by the time dummy coefficients, from changes in the banks opportunity costs of providing risky loans. This is achieved by regressing the estimated time dummy coefficients on the evolution of the safe real interest rate over time using a simple linear regression model. The variation of the time dummy coefficients, which cannot be explained by changes in the opportunity cost, i.e. the residuals of the linear regression, are finally interpreted as bank industryspecific determinants of credit supply that can be used as an indicator for identifying a credit crunch. 13

14 The Micro-econometric Model We are estimating a nonlinear panel-data model the dependent variable y it is a binary choice variable, which measures the firms' perception of the banks' credit conditions is equal to 1, if the firms assess the banks' willingness to lend as restrictive is equal to 0, if the firms indicate normal or accommodating. x it are the regressors α i is an unobserved firm specific effects β are the estimated coefficients, i = 1,2,, N denotes the independent firms, t = 1,2,, T i denotes the observations for the i th unit, and F is the cumulative distribution function of either the logistic distribution (logit model) or the standard normal distribution (probit model) 14

15 The Regressors The regressors x it consist of firm- and sector-specific variables which vary both over time and firms or business sectors which measure the quality of the borrowers time dummies which vary over time and are common to all firms 15

16 The Regressors The firm-specific variables are taken from the Ifo Business Survey. We use the firm s current state of the business and its business expectations for the next six months. A firm can characterize its situation as good, satisfactorily or poor and its expectations as more favorable, unchanged or more unfavorable. Both firm-specific regressors are ordinal variables with three categories, which take a value of 1, if the firm's quality is bad (more unfavorable business expectations, poor state of the business), 2, if the firm's quality is moderate (unchanged business expectations, satisfactory state of the business), 3, if the firm's quality is good (more favorable business expectations, good state of the business). 16

17 The Regressors The firm-specific variables are taken from the Ifo Business Survey. The responses to these questions (in particular the current state of the business) can be viewed as proxies for actual balance sheet figures. This is a major result of the so-called survey of the survey. In this special survey the Ifo Institute examined the factors that form the basis for firms' replies to the monthly business survey. It turned out that for the assessment of the current state of the business and the business expectations for the next six months the firms mainly rely on hard facts, such as the profit situation and the turnover. 17

18 The Regressors In addition to firm-specific variables we also include a sector-specific variable. As a proxy for the sector-specific economic activity, we use the Sector Ifo Business Climate Indicator, which varies considerably across business sectors. The idea here is that the bank also evaluates a firm s creditworthiness on the basis of the performance of the economic activity in the business sector that a firm i is operating in. This variable varies over time, but is identical for all firms producing in a specific business sector. 18

19 The Regressors T 1 time dummies where T = 37 is the number of surveys between June 2003 and November 2010 that are analyzed in the regressions. The idea is that the time dummies capture variations of the banks lending practices over time, which are independent from the bank s assessment of the quality of the firm 19

20 Descriptive Statistics 20

21 The Regression Results If the state of the business is good (=3) or business expectations are more favorable (=3), the probability that a firm perceives the loan supply policy of banks as restrictive decreases. 21

22 The Regression Results If the economic activity in the sector that a firm belongs to increases (business cycle indicator), the probability of a restrictive loan supply policy decreases. 22

23 The Regression Results the first survey in June 2003 is the reference point (=0) Coefficients on the time dummies: the firms access to credit was less restrictive in 2007/08 than in 2003/04 or 2009/10 23

24 The Credit Crunch Indicator A credit crunch is defined as that part of the (leftward) shifts in the loan supply curve that are neither explained by firm- or sector-specific factors (first step) nor by changes in the return of the banks risk free investment alternative, which can be interpreted as the opportunity costs of providing risky loans (second step). The second step focuses on the safe real interest rate. If the safe real interest rate increases, banks invest more of their funds in risk free assets and will consequently reduce their loan supply, everything else being equal (Bernanke and Blinder, 1988). 24

25 The Credit Crunch Indicator Those shifts in loan supply that are not caused by normal determinants of the loan supply curve and that therefore reflect a credit crunch are isolated by regressing the estimated time dummy coefficients on the evolution of the safe real interest rate over time using a simple linear regression model: td t corresponds to the estimated coefficients on the time dummies The opportunity i t costs are measured by the real interest rate on safe government bonds. The real interest rate is calculated by the ECB from AAA rated euro area 10-year central government bonds 25

26 The Credit Crunch Indicator The variation of the time dummy coefficients, which cannot be explained by changes in the safe real interest rate, i.e. the residuals ε t of the linear regression, are finally interpreted as loan supply shocks. i Cr D Cr S (safe real rate, firm quality, ε t ) Cr 26

27 The Credit Crunch Indicator The estimated coefficients on the real interest rate δ are positive and significant, implying that higher opportunity costs explain part of the increase in the time dummies and may therefore contribute to a leftward shift of the loan supply curve. (higher time dummies -> loan supply more restrictive) 27

28 The Credit Crunch Indicator The probability of a credit crunch in the German economy was highest during the years 2003/04, following the economic downturn after the burst of the New Economy Bubble. 28

29 The Credit Crunch Indicator Most surprisingly, in the latest financial crisis, in which banks are much more involved than in previous recessions due to massive write downs of toxic assets, the indications of a credit crunch are much weaker. 29

30 The Credit Crunch Indicator 30

31 The Credit Crunch Indicator During the financial crisis large firms ( 250 employees) are more affected than small firms. 31

32 The Credit Crunch Indicator Small firms have bank relationships with savings banks and credit cooperatives, whereas large firms mainly negotiate credits with private commercial banks and state-owned landes banks. In a special question of the Ifo Business Survey in June 2009 about the firms bank relationships 64% of the firms in our sample provided the requested information about the main lender. For each banking group the Table shows the share of large firms. Private commercial banks and state-owned landes banks have been most hit by the financial crisis. 32

33 The Credit Crunch Indicator Firms that mainly negotiate credits with private commercial banks and stateowned landes banks are most likely affected by a credit crunch. 33

34 Conclusion We estimate a credit crunch indicator based on a regular survey among firms. A credit crunch is defined as a negative shock to the credit supply behavior of banks, which is neither explained by a deterioration of the quality of the borrowers, nor by an increase in the banks opportunity costs of providing risky loans. The indicator reveals that Germany very likely suffered from a severe credit crunch in the years 2003/2004. During the recent financial crisis banks are astonishingly accommodating, which may probably be explained by the massive public sector equity support to banks. 34

35 Conclusion Financial Market Stabilization Fund (as of Nov. 2010) recapitalization (29 bn ) Banks average equity ratio increased during the crisis 29 bn = 0.7% of average total assets of landes banks and private commercial banks in

36 logit model probit model Update Using data until November m1 2005m1 2007m1 2009m1 2011m1 2013m1 creditcrunch_logit_re creditcrunch_probit_re creditcrunch_logit_fe 36

37 logit model probit model Update Using data until March m1 2005m1 2007m1 2009m1 2011m1 2013m1 creditcrunch_logit_re creditcrunch_probit_re creditcrunch_logit_fe 37

38 logit model probit model Update Using data until November m1 2005m1 2007m1 2009m1 2011m1 2013m1 creditcrunch_logit_re creditcrunch_probit_re creditcrunch_logit_fe 38

39 Back-up 39

40 Back-up 40

41 Back-up 41

42 Back-up 42