Securitisation and the bank lending channel

Transcription

1 Securitisation and the bank lending channel Yener Altunbas (University of Wales, Bangor) Leonardo Gambacorta (Bank of Italy) David Marqués (ECB) 2nd Symposium of the ECB-CFS Research Network on Capital markets and financial integration in Europe - Frankfurt, February 2008 This paper represents the views of the authors only and not necessarily those of the institutions to which they are affiliated

2 The main findings of the paper The development of securitisation had a strong impact on lending supply In normal times securitisation activity shelters banks from the transmission of monetary policy impulses Securitization activity is pro-cyclical and therefore this insulation effect vanishes in bad times and could even reverse in an amplification of the bank lending channel

3 Overview of the presentation 1. Motivation 2. Developments in securitization in the euro area 3. How does securitization affect the monetary transmission mechanism? 4. The data and the econometric model 5. Results 6. What are the effects of a reduction in securitization activity on loan supply in the euro area? 7. Conclusions

4 1. Motivation Since the introduction of the euro we have witnessed a significant increase in securitization activity. In 2006 securitized lending amounted to around one-fifth of new loans in the euro area. This process has altered the fundamental liquidity transformation and monitoring role of banks (Diamond,1984; Holström and Tirole, 1997). It has also amplified bank s ability to supply new loans. How this has changed the monetary transmission? We claim that the new role of banks from originate and hold to originate and distribute has an impact on the bank lending channel.

5 2. Securitisation in the euro area 250, , ,000 Volume (left-hand scale) Number of issues (right-hand scale) ABS issuance in euro has finally taken off since the introduction of the euro that has contributed to financial integration. 100, ,

6 Securitisation activity (percentages of GDP, by country of collateral) Euro Area UK USA Data sources: European Securitization Forum, Bond Market Association, Eurostat. Euro area figures do not include Pfandbriefe; US data exclude Federal Agency securities.

7 2. Securitisation in the euro area (3) Securitisation by type of instrument in 2005 (euro-denominated, volumes, cash funded instruments only) Corporate ABS 6% Consumer ABS 8% CDOs 18% RMBS 53% The bulk of the activity is based on granular mortgage-backed securities (2/3). They are characterized by an high degree of commoditisation and an high standardisation of credit assessment techniques. CMBS 15%

8 2. Securitisation in the euro area (4) Securitisation by country of issuance in 2005 (volumes, CDOs are excluded) Netherlands 27% France 5% other 3% Portugal 5% Spain 27% Germany 11% Securitisation has taken off in most euro area countries and has been particularly strong in those countries experiencing strong increases in housing prices Italy 22%

9 3. How does securitization affect the monetary transmission mechanism?(1) Securitization has reduced the fundamental role of liquidity transformation performed by banks. While the origination of loans remains locally based, securitization can make the funding of previously illiquid loans global by making them tradable. We claim that securitization activity has changed the standard set up of the Bernanke and Blinder model (1988).

10 3. How does securitization affect the monetary transmission mechanism?(2) In the case of a monetary tightening, if the reduction in deposits is not compensated by banks by issuing CDs or bonds, a drop in supplied lending occurs. The bank lending channel literature claims that the drop of supplied lending will be less severe for big, liquid and well-capitalized banks (Kashyap and Stein, 1995; Kishan and Opiela, 2000; Gambacorta and Mistrulli, 2004).

11 3. How does securitization affect the monetary transmission mechanism?(3) We argue that securitization alters the traditional bank lending channel in two ways. Extra liquidity: banks may obtain additional liquidity independently of their securities holdings. This mechanism reduces the effectiveness of the bank lending channel in a way similar to the famous Romer and Romer (1990) critique. Capital relief: by removing loans from their balancesheet, banks can obtain a regulatory capital relief which allows for a positive net effect on the loan supply.

12 4. Data: sources (1) Banks individual data, from Bankscope, from 1999 to 2005 to cover the single currency. Unbalanced sample of around 3,000 banks accounting for a total of about 75% of bank lending in the euro area. Data on individual securitization deals (Bondware and S&P) have been matched with the balance sheet statements of each bank originating the deal.

13 4. Data: two measures for risk (2) We have inserted two control variables accounting for banks risk. The latter may affect the ability and willingness of a bank to grant credit. 1. Loan-loss provisions as a percentage of loans: ex-post accounting measure of credit risk. 2. Expected default frequency (EDF): forwardlooking estimator of credit risk computed by Moody s KMV.

14 4. The econometric model (3) Our empirical specification builds on the standard set up used in Kashyap and Stein (1995), Ehrmann et al. (2003) and Ashcraft (2007). In addition we incorporate the effect of securitisation and risk. Growth rate in lending Nominal GDP Interest rate changes Securitization activity indicator ln( Loans) + 1 = 0 σ i + κsize t 1 it M t = ζ ln( Loans) * SIZE + ϑliq t 1 t = 0 + ξcap it 1 λ i t = 0 M t + τllp δ ln( GDPN) * LIQ it 1 t ψedp = 0 it 1 t χ i + ε it + 1 = 0 M t β i * CAP M t t = 0 φ i + ηsec t 1 M t + * SEC t 1 + Individual bank variables Bank risk variables: Loan loss provisions and Expected Default Probability i=1,, N and t=1,, T where N is the number of banks and T is the year.

15 5. Results: SEC and loan supply (1) Securitization activity has a direct and positive impact on the average growth rate of supplied lending. On the contrary, banks risk have a negative effect. These effects are conditional upon business (and overall liquidity) conditions: securitization activity tends to be higher in a boom and reverse during an economic slowdown.

16 5. Results: SEC and MTM (2) In normal times securitization reduces the effectiveness of the Bank Lending Channel: banks that are more active in the securitization market are more insulated from monetary policy changes. By how much? Insulation depends on the degree of activism on the market. It is very high if a bank tends to securitize around one quarter of her assets.

17 5. Results: Effect of a one percent increase of the monetary policy rate on bank lending (3) Average bank (SEC=0.02) Median SEC bank (SEC=0.10) Average SEC bank (SEC=0.25) *** *** -0.71*** -0.55*** Effects after one year Long run effect

18 6. What are the effects of a drop in securitization activity on loan supply in the euro area? (1) The sub-prime mortgage crisis has changed market sentiment about the use of structured products. In the past few months the volume of securitization activity has been reduced. 20% 18% 16% 14% 12% 10% 8% 6% 4% 2% Euro area UK US 0%

19 6. What does the model really tell us? Some simulations on the effects of a drop in securitization activity on loan supply (2) There is now a greater uncertainty about the use of these finance products and a general lack of transparency on what is on banks balance sheets. The real important question is: Has the Originate-to-distribute model come to an end? In our opinion the crisis of the OTD model will change the structure of banking, but it is hard to imagine that the industry will abandon securitization. We can imagine a reduction in activity that brings the model back some years. For example, what does it happen to supplied lending if the volume of securitized loans goes back to 2005 (a reduction of 30%) or to the average over the period (-60%)?

20 6. The effects of a drop in securitization activity on loan supply in the euro area: different scenarios (3) 0 Light effect: back to 2005 Strong effect: back to the average values of % -2-3 The volume of s ecuritiza tion a ctivity drops by 30% and the Expected Default P robability for the average bank increases from 0.05% to 0.07% -1.5% The volume of s ecuritiza tion a ctivity drops by 60% and the Expected Default Probability for the average bank increases from 0.05% to 0.14%

21 6. The effects of a drop in securitization activity on loan supply in the euro area: different scenarios (4) The effects would be larger in case the use of bank credit lines by conduits put additional pressure on bank capital. The crucial question is to what extent such a possible reduction in the supply of loans would affect real GDP growth. Preliminary results - obtained using VECM and DSGE models - show that the effects on GDP are very different assuming that the central bank reacts to the credit supply shock or not.

22 7. Conclusions We find that the changing role of banks from originate and hold to originate to distribute reduces in normal times the effectiveness of BLC of monetary policy. However, securitization activity is pro-cyclical and therefore the insulation effect vanishes in bad times and could even reverse in an amplification of the bank lending channel. Simulations show that the impact of the recent turmoil on loan supply also depends on how severe will be the reduction in securitization activity.