Discussion of Gerali, Neri, Sessa, Signoretti. Credit and Banking in a DSGE Model

Transcription

1 Discussion of Gerali, Neri, Sessa and Signoretti Credit and Banking in a DSGE Model Jesper Lindé Federal Reserve Board ty ECB, Frankfurt December 15, 2008

2 Summary of paper This interesting paper... Extends the Iacoviello (2005) model with monopolistic competitive banks and allow for endogenous capital accumulation in the banking sector

3 Summary of paper This interesting paper... Extends the Iacoviello (2005) model with monopolistic competitive banks and allow for endogenous capital accumulation in the banking sector Model economy where both households and rms are subject to borrowing constraints

4 Summary of paper This interesting paper... Extends the Iacoviello (2005) model with monopolistic competitive banks and allow for endogenous capital accumulation in the banking sector Model economy where both households and rms are subject to borrowing constraints Two types of households: Patient (non-constrained) and Impatient (constrained)

5 Summary of paper This interesting paper... Extends the Iacoviello (2005) model with monopolistic competitive banks and allow for endogenous capital accumulation in the banking sector Model economy where both households and rms are subject to borrowing constraints Two types of households: Patient (non-constrained) and Impatient (constrained) Kiyotaki-Moore (1997) type of borrowing constraint for the impatient households, banks only willing to lend out fraction m H of IH s housing value

6 Summary of paper This interesting paper... Extends the Iacoviello (2005) model with monopolistic competitive banks and allow for endogenous capital accumulation in the banking sector Model economy where both households and rms are subject to borrowing constraints Two types of households: Patient (non-constrained) and Impatient (constrained) Kiyotaki-Moore (1997) type of borrowing constraint for the impatient households, banks only willing to lend out fraction m H of IH s housing value Impatient entrepreneurs produce intermediate goods using capital and household labor, need to borrow to get desired capital stock

7 Summary of paper This interesting paper... Extends the Iacoviello (2005) model with monopolistic competitive banks and allow for endogenous capital accumulation in the banking sector Model economy where both households and rms are subject to borrowing constraints Two types of households: Patient (non-constrained) and Impatient (constrained) Kiyotaki-Moore (1997) type of borrowing constraint for the impatient households, banks only willing to lend out fraction m H of IH s housing value Impatient entrepreneurs produce intermediate goods using capital and household labor, need to borrow to get desired capital stock Banks only willing to lend out fraction m E of E s networth

8 Summary This interesting paper... Replace standard assumption of perfect competition in the banking sector (e.g. Bernanke, Gertler and Gilchrist, 1999, Iacoviello, 2005) and instead assume that banks have monopolistic competition in setting deposit rates and lending rates

9 Summary This interesting paper... Replace standard assumption of perfect competition in the banking sector (e.g. Bernanke, Gertler and Gilchrist, 1999, Iacoviello, 2005) and instead assume that banks have monopolistic competition in setting deposit rates and lending rates Two type of banks, one type collects deposits, one type lend out to E and IH

10 Summary This interesting paper... Replace standard assumption of perfect competition in the banking sector (e.g. Bernanke, Gertler and Gilchrist, 1999, Iacoviello, 2005) and instead assume that banks have monopolistic competition in setting deposit rates and lending rates Two type of banks, one type collects deposits, one type lend out to E and IH New version of paper allow for endogenous capital accumulation for the lending banks, concave production function in own funding and CB/deposit funding bt H + bt E = kj,t B χb h i 1 1 m j,t + dj,t B χb

11 Summary This interesting paper... Replace standard assumption of perfect competition in the banking sector (e.g. Bernanke, Gertler and Gilchrist, 1999, Iacoviello, 2005) and instead assume that banks have monopolistic competition in setting deposit rates and lending rates Two type of banks, one type collects deposits, one type lend out to E and IH New version of paper allow for endogenous capital accumulation for the lending banks, concave production function in own funding and CB/deposit funding bt H + bt E = kj,t B χb h i 1 1 m j,t + dj,t B χb Central bank provide in nite lending to risk-free rate, therefore no strategic interaction between deposit and lending banks when setting interest rates

12 Summary This interesting paper... Replace standard assumption of perfect competition in the banking sector (e.g. Bernanke, Gertler and Gilchrist, 1999, Iacoviello, 2005) and instead assume that banks have monopolistic competition in setting deposit rates and lending rates Two type of banks, one type collects deposits, one type lend out to E and IH New version of paper allow for endogenous capital accumulation for the lending banks, concave production function in own funding and CB/deposit funding bt H + bt E = kj,t B χb h i 1 1 m j,t + dj,t B χb Central bank provide in nite lending to risk-free rate, therefore no strategic interaction between deposit and lending banks when setting interest rates Closely related literature: Andrés and Arce (2007), Aslam and Santoro (2007)

13 Summary Exercises and results Study how propagation of monetary policy and Kydland-Prescott (1982) type of technology shocks are a ected

14 Summary Exercises and results Study how propagation of monetary policy and Kydland-Prescott (1982) type of technology shocks are a ected Find: Financial frictions amplify, monopolistic banks moderate e ects

15 Summary Exercises and results Study how propagation of monetary policy and Kydland-Prescott (1982) type of technology shocks are a ected Find: Financial frictions amplify, monopolistic banks moderate e ects Study e ects of the leverage ratios m E and m H

16 Summary Exercises and results Study how propagation of monetary policy and Kydland-Prescott (1982) type of technology shocks are a ected Find: Financial frictions amplify, monopolistic banks moderate e ects Study e ects of the leverage ratios m E and m H Find: Higher leverage, monetary policy more potent

17 Summary Exercises and results Study how propagation of monetary policy and Kydland-Prescott (1982) type of technology shocks are a ected Find: Financial frictions amplify, monopolistic banks moderate e ects Study e ects of the leverage ratios m E and m H Find: Higher leverage, monetary policy more potent Corollary: Less borrowing constrained entrepreneurs/households and higher competition in banking industry cannot explain Great moderation

18 Summary Exercises and results Grande nale : Simulate a nancial turmoil scenario with the model (qualitative aspects emphasized by authors!) Find: E ects on output in May version of paper and in December version

19 Discussion outline A couple of comments on the model VAR model Empirical implications of the model Financial turmoil scenario Concluding remarks

20 A couple of comments on the model Calibration of the size of the sectors are all of equal size, patient households (0.25), impatient households (0.25), entrepreneurs (0.25) and bankers (0.25), yet only households consume housing, and bankers and entrepreneurs do not provide labor

21 A couple of comments on the model Calibration of the size of the sectors are all of equal size, patient households (0.25), impatient households (0.25), entrepreneurs (0.25) and bankers (0.25), yet only households consume housing, and bankers and entrepreneurs do not provide labor Sensitivity to shrinking the entrepreneurs and banker sectors

22 A couple of comments on the model Calibration of the size of the sectors are all of equal size, patient households (0.25), impatient households (0.25), entrepreneurs (0.25) and bankers (0.25), yet only households consume housing, and bankers and entrepreneurs do not provide labor Sensitivity to shrinking the entrepreneurs and banker sectors How should we think about m E, m H in a world where Bankers make pro ts and consume. In your model, Bankers steady state pro ts and welfare presumably rises when m E, m H increases

23 A couple of comments on the model Calibration of the size of the sectors are all of equal size, patient households (0.25), impatient households (0.25), entrepreneurs (0.25) and bankers (0.25), yet only households consume housing, and bankers and entrepreneurs do not provide labor Sensitivity to shrinking the entrepreneurs and banker sectors How should we think about m E, m H in a world where Bankers make pro ts and consume. In your model, Bankers steady state pro ts and welfare presumably rises when m E, m H increases Would be nice to know how strong these steady state e ects are

24 A couple of comments on the model Calibration of the size of the sectors are all of equal size, patient households (0.25), impatient households (0.25), entrepreneurs (0.25) and bankers (0.25), yet only households consume housing, and bankers and entrepreneurs do not provide labor Sensitivity to shrinking the entrepreneurs and banker sectors How should we think about m E, m H in a world where Bankers make pro ts and consume. In your model, Bankers steady state pro ts and welfare presumably rises when m E, m H increases Would be nice to know how strong these steady state e ects are What are the costs of changing deposit and lending interest rates, i.e. what costs do you have in mind to motive κ > 0?

25 VAR model Use VAR model to parameterize stickiness in the banking sector Quantify impact of policy on lending and deposit rates by using Cholesky decomposition with the ECB money market rate ordered rst in the VAR

26 VAR model Use VAR model to parameterize stickiness in the banking sector Quantify impact of policy on lending and deposit rates by using Cholesky decomposition with the ECB money market rate ordered rst in the VAR This procedure is not consistent with the timing assumptions in your model

27 VAR model Use VAR model to parameterize stickiness in the banking sector Quantify impact of policy on lending and deposit rates by using Cholesky decomposition with the ECB money market rate ordered rst in the VAR This procedure is not consistent with the timing assumptions in your model Are lending (deposit) rates for newly issued loans (deposits)?

28 VAR model Use VAR model to parameterize stickiness in the banking sector Quantify impact of policy on lending and deposit rates by using Cholesky decomposition with the ECB money market rate ordered rst in the VAR This procedure is not consistent with the timing assumptions in your model Are lending (deposit) rates for newly issued loans (deposits)? Study if you can reject the exible-price speci cation ˆr x t = ˆr ECB t ˆε t ε 1, for linear loan-production technology and x = fd, h, eg.

29 VAR model Use VAR model to parameterize stickiness in the banking sector Quantify impact of policy on lending and deposit rates by using Cholesky decomposition with the ECB money market rate ordered rst in the VAR This procedure is not consistent with the timing assumptions in your model Are lending (deposit) rates for newly issued loans (deposits)? Study if you can reject the exible-price speci cation ˆr x t = ˆr ECB t ˆε t ε 1, for linear loan-production technology and x = fd, h, eg. Can you reject exible-price restriction for ˆε t i.i.d. case by standard LR -test

30 VAR model Use VAR model to parameterize stickiness in the banking sector Quantify impact of policy on lending and deposit rates by using Cholesky decomposition with the ECB money market rate ordered rst in the VAR This procedure is not consistent with the timing assumptions in your model Are lending (deposit) rates for newly issued loans (deposits)? Study if you can reject the exible-price speci cation ˆr x t = ˆr ECB t ˆε t ε 1, for linear loan-production technology and x = fd, h, eg. Can you reject exible-price restriction for ˆε t i.i.d. case by standard LR -test Given ˆε t AR(1), hard to distinguish between exible price and stickiness speci cation without studying the impact of other variables

31 Empirical implications of the model Bank pro ts Standard models with nancial frictions invoke zero pro t condition for banks (e.g. Bernanke, Gertler and Gilchrist, 1999)

32 Empirical implications of the model Bank pro ts Standard models with nancial frictions invoke zero pro t condition for banks (e.g. Bernanke, Gertler and Gilchrist, 1999) This assumption is most likely at odds with the data (de nitely in Sweden)

33 Empirical implications of the model Bank pro ts Standard models with nancial frictions invoke zero pro t condition for banks (e.g. Bernanke, Gertler and Gilchrist, 1999) This assumption is most likely at odds with the data (de nitely in Sweden) In this model, bank pro ts can vary over the business cycle

34 Empirical implications of the model Bank pro ts Standard models with nancial frictions invoke zero pro t condition for banks (e.g. Bernanke, Gertler and Gilchrist, 1999) This assumption is most likely at odds with the data (de nitely in Sweden) In this model, bank pro ts can vary over the business cycle But, no credit losses in the model

35 Empirical implications of the model Bank pro ts Standard models with nancial frictions invoke zero pro t condition for banks (e.g. Bernanke, Gertler and Gilchrist, 1999) This assumption is most likely at odds with the data (de nitely in Sweden) In this model, bank pro ts can vary over the business cycle But, no credit losses in the model In the data, bank pro ts most likely strongly a ected by credit losses in periods of nancial turmoil

36 Empirical implications of the model Bank pro ts Bank pro ts excluding credit losses for the 4 big banks in Sweden, accounting for about 80 percent of the market Bank profits excluding credit losses as percentage of GDP (4 quarter moving average)

37 Empirical implications of the model Bank pro ts Credit losses can account for large part of bank pro ts in periods of nancial turmoil Gross bank profits (excluding credit losses) Net bank profits (including credit losses)

38 Empirical implications of the model Bank pro ts: Characterization of the banking crisis in Sweden Sharp decline in bank pro ts was associated with high real interest rates (and sharp fall in production) Net bank profits, left axis Annual real policy rate, right axis Net bank profits (left axis) GDP growth (yony, right axis)

39 Empirical implications of the model Default frequency Standard models with nancial frictions have time-varying default rates (e.g. Bernanke, Gertler and Gilchrist, 1999)

40 Empirical implications of the model Default frequency Standard models with nancial frictions have time-varying default rates (e.g. Bernanke, Gertler and Gilchrist, 1999) In this model, no explicit default rate mechanism

41 Empirical implications of the model Default frequency Standard models with nancial frictions have time-varying default rates (e.g. Bernanke, Gertler and Gilchrist, 1999) In this model, no explicit default rate mechanism Most likely, the lack of this mechanism have undesirable implications for total bank pro ts in the model

42 Empirical implications of the model Default frequency Standard models with nancial frictions have time-varying default rates (e.g. Bernanke, Gertler and Gilchrist, 1999) In this model, no explicit default rate mechanism Most likely, the lack of this mechanism have undesirable implications for total bank pro ts in the model High default frequencies a de ning characteristic of periods with nancial distress (See Jacobson, Lindé and Roszbach, 2008)

43 Empirical implications of the model Default frequency Actual and predicted default rates (using a simple Logit regression) for all Swedish corporates

44 Empirical implications of the model Modelling the interbank market Standard models with nancial frictions have zero spread between the interbank market rate (funding rate) and the policy rate (risk-free rate) (e.g. Bernanke, Gertler and Gilchrist, 1999)

45 Empirical implications of the model Modelling the interbank market Standard models with nancial frictions have zero spread between the interbank market rate (funding rate) and the policy rate (risk-free rate) (e.g. Bernanke, Gertler and Gilchrist, 1999) A model that have the ambition to capture what is currently going on in nancial markets should preferably break this mechanism

46 Empirical implications of the model Modelling the interbank market Standard models with nancial frictions have zero spread between the interbank market rate (funding rate) and the policy rate (risk-free rate) (e.g. Bernanke, Gertler and Gilchrist, 1999) A model that have the ambition to capture what is currently going on in nancial markets should preferably break this mechanism This model actually has a banking intermediation spread, which is a nice feature of the model

47 Empirical implications of the model 3 Month basis spreads in selected countries (in Swedish!)

48 Empirical implications of the model Longer perspective: 3-month interbank - FFR, and 3-month interbank - 3-month Tbill

49 Financial turmoil scenario Use calibrated model to simulate a scenario with tightening of credit conditions Assume set of unexpected shocks in period 1; 2 collateral shocks (both households and entrepreneurs), increase in 2 lending rates and the deposit rate

50 Financial turmoil scenario Use calibrated model to simulate a scenario with tightening of credit conditions Assume set of unexpected shocks in period 1; 2 collateral shocks (both households and entrepreneurs), increase in 2 lending rates and the deposit rate Study the e ects on a broad set of variables, but I miss bank pro ts, house prices and bank intermediation spread

51 Financial turmoil scenario Use calibrated model to simulate a scenario with tightening of credit conditions Assume set of unexpected shocks in period 1; 2 collateral shocks (both households and entrepreneurs), increase in 2 lending rates and the deposit rate Study the e ects on a broad set of variables, but I miss bank pro ts, house prices and bank intermediation spread I suspect that bank pro ts rise, why would otherwise banks "inject" these shocks in the system?

52 Financial turmoil scenario Use calibrated model to simulate a scenario with tightening of credit conditions Assume set of unexpected shocks in period 1; 2 collateral shocks (both households and entrepreneurs), increase in 2 lending rates and the deposit rate Study the e ects on a broad set of variables, but I miss bank pro ts, house prices and bank intermediation spread I suspect that bank pro ts rise, why would otherwise banks "inject" these shocks in the system? Bank pro ts most likely not rising for the moment, but in many other respects the model appear to have plausible e ects

53 Financial turmoil scenario Use calibrated model to simulate a scenario with tightening of credit conditions Assume set of unexpected shocks in period 1; 2 collateral shocks (both households and entrepreneurs), increase in 2 lending rates and the deposit rate Study the e ects on a broad set of variables, but I miss bank pro ts, house prices and bank intermediation spread I suspect that bank pro ts rise, why would otherwise banks "inject" these shocks in the system? Bank pro ts most likely not rising for the moment, but in many other respects the model appear to have plausible e ects Suggestion, put in combination of nancial and technology shocks that closely match evolution of nancial variables (spreads, LTV-ratios, house prices, bank pro ts, labor productivity) 2007Q3-2008Q3 then do a projection forward based the resulting state 2008Q3

54 Financial turmoil scenario Use calibrated model to simulate a scenario with tightening of credit conditions Assume set of unexpected shocks in period 1; 2 collateral shocks (both households and entrepreneurs), increase in 2 lending rates and the deposit rate Study the e ects on a broad set of variables, but I miss bank pro ts, house prices and bank intermediation spread I suspect that bank pro ts rise, why would otherwise banks "inject" these shocks in the system? Bank pro ts most likely not rising for the moment, but in many other respects the model appear to have plausible e ects Suggestion, put in combination of nancial and technology shocks that closely match evolution of nancial variables (spreads, LTV-ratios, house prices, bank pro ts, labor productivity) 2007Q3-2008Q3 then do a projection forward based the resulting state 2008Q3 What mix of nancial and technology shocks do the model lter out?

55 Concluding remarks Very nice paper with a straightforward approach to introduce monopolistically competitive banks in a DSGE model with nancial frictions

56 Concluding remarks Very nice paper with a straightforward approach to introduce monopolistically competitive banks in a DSGE model with nancial frictions Demand from policy makers to have these channels our models

57 Concluding remarks Very nice paper with a straightforward approach to introduce monopolistically competitive banks in a DSGE model with nancial frictions Demand from policy makers to have these channels our models However, from a macro modellers perspective

58 Concluding remarks Very nice paper with a straightforward approach to introduce monopolistically competitive banks in a DSGE model with nancial frictions Demand from policy makers to have these channels our models However, from a macro modellers perspective Monopolistic competition in the banking sector makes nancial frictions matter less => do I need to have this in my DSGE model to analyze periods with business as usual?

59 Concluding remarks Very nice paper with a straightforward approach to introduce monopolistically competitive banks in a DSGE model with nancial frictions Demand from policy makers to have these channels our models However, from a macro modellers perspective Monopolistic competition in the banking sector makes nancial frictions matter less => do I need to have this in my DSGE model to analyze periods with business as usual? High degree of exogeneity make current model have less appealing implications for some variables (e.g. bank pro ts) => can I use this model to analyze periods with nancial distress? More endogenous propagation seems warranted, otherwise we will need many shocks to account for uctuations in a broader set of macroeconomic variables