The Collateral Channel of Monetary Policy: Evidence from the European Central Bank

Transcription

1 The Collateral Channel of Monetary Policy: Evidence from the European Central Bank European Central Bank Workshop Nuno Cassola and François Koulischer European Central Bank and Université Libre de Bruxelles (ECARES) 17 June 2013 Cassola and Koulischer (2013) The Collateral Channel 1/33

2 Outline Introduction 1 Introduction 2 Descriptive Analysis 3 An Econometric Model of Collateral Policy 4 Results 5 Conclusion Cassola and Koulischer (2013) The Collateral Channel 2/33

3 Introduction Divergence in borrowing cost of small and medium enterprises (SME) in the euro area Source: ECB Cassola and Koulischer (2013) The Collateral Channel 3/33

4 Introduction Survey data suggests the divergence is caused by a fragmentation of financial markets Source: ECB Survey of SMEs April 2013 Cassola and Koulischer (2013) The Collateral Channel 4/33

5 Introduction Banks access to funding vary significantly across countries Source: Bloomberg Cassola and Koulischer (2013) The Collateral Channel 5/33

6 Introduction The fragmentation of euro area financial markets hampers the transmission of monetary policy Weakened transmission of monetary policy during the sovereign debt crisis is a main concern for the ECB Mario Draghi (ECB President, Nov. 2012): The fragmentation of the single financial market has led to a fragmentation of the single monetary policy. (...) This has made di cult the transmission of impulses coming from an accommodative monetary policy through adjustments in interest rates on loans to households and firms by banks. Cassola and Koulischer (2013) The Collateral Channel 6/33

7 Introduction The ECB lowered collateral requirements over Date Measure October 2008 The credit threshold for collateral other than ABS is lowered to BBB- May 2010 Minimum rating threshold of Greek government debt is suspended March 2011 Minimum rating threshold of Irish government debt is suspended July 2011 Minimum rating threshold of Portuguese government debt is suspended December month LTROS with (i) new ABS eligibility requirements; (ii) NCBs may accept additional credit claims as collateral September 2012 Marketable debt instruments denominated in currencies other than the euro and issued and held in the euro area, are eligible Cassola and Koulischer (2013) The Collateral Channel 7/33

8 Introduction How does collateral policy a ect assets pledged to the ECB and banks funding costs? Methodology: Data: assets pledged by banks to the ECB (aggregated at country-level) from January 2009 to September 2011 Econometric model: discrete choice logit model Estimate changes in collateral pledged by banks under various collateral policies Estimate impact of collateral policy on banks funding cost Main finding: Impact of collateral policy varies with bank- and asset-characteristics This suggests that ECB measures to relax collateral requirements contribute to a convergence in banks funding costs Cassola and Koulischer (2013) The Collateral Channel 8/33

9 Outline Descriptive Analysis 1 Introduction 2 Descriptive Analysis 3 An Econometric Model of Collateral Policy 4 Results 5 Conclusion Cassola and Koulischer (2013) The Collateral Channel 9/33

10 ECB collateral policy Descriptive Analysis Source: ECB General Documentation 2010 Cassola and Koulischer (2013) The Collateral Channel 10 / 33

11 Descriptive Analysis Nominal value of assets eligible as collateral Cassola and Koulischer (2013) The Collateral Channel 11 / 33

12 Descriptive Analysis ECB collateral pool (nominal value) Cassola and Koulischer (2013) The Collateral Channel 12 / 33

13 Descriptive Analysis Average price of collateral: High-yield versus Low-Yield countries Cassola and Koulischer (2013) The Collateral Channel 13 / 33

14 Descriptive Analysis Illiquid assets (with low rating) are pledged to the central bank Proportion of government bonds pledged to the ECB for the 10 countries in the sample (weekly) Cassola and Koulischer (2013) The Collateral Channel 14 / 33

15 Descriptive Analysis Riskier banks borrow more from the ECB Cassola and Koulischer (2013) The Collateral Channel 15 / 33

16 Outline An Econometric Model of Collateral Policy 1 Introduction 2 Descriptive Analysis 3 An Econometric Model of Collateral Policy 4 Results 5 Conclusion Cassola and Koulischer (2013) The Collateral Channel 16 / 33

17 An Econometric Model of Collateral Policy The empirical challenge Collateral di ers by asset characteristics (maturity, rating, asset type) Example: 10 ratings, 2 maturities, 9 asset types, guaranteed or not -> 360 asset types The quantity of collateral j pledged by banks could be: q j = j j h j + NX k=1;6=j jkh k Challenge: large number of cross-haircut elasticities to estimate (360 2!) Solution: Use insights from theory to restrict the number of parameters to estimate (similar to discrete choice logit model Mc. Fadden 1973) Cassola and Koulischer (2013) The Collateral Channel 17 / 33

18 An Econometric Model of Collateral Policy Overview of empirical model A commercial bank has a pool of assets ( 1, 2,..., j,..., N ) It chooses an optimal funding and investment scheme by making repeated decisions (eg. 100 decisions for EUR 1 mln each) For each decision, the bank chooses A funding source (private investors / central bank / no borrowing) A collateral j =1,...,N to use if it chooses central bank funding Cassola and Koulischer (2013) The Collateral Channel 18 / 33

19 An Econometric Model of Collateral Policy Bank makes funding-investment choices Let X 1,...,X N be the amounts of collateral pledged and X 0 be the amount borrowed from the private market. For each decision i the bank has the following funding opportunities with marginal profits i No investment: 1i =0 Private funding: 0i = R 0 (q) p k 0 X 0(i),wherek 0 X 0(i) is a cost of funding that depends on the amount borrowed in the private market X 0(i) Central bank funding collateralized by asset j: ji = R 0 (q) h j + v j + p + j k X j(i), X (i) where X j(i) is the quantity of asset j that has already been pledged as collateral as of decision i, X j is the total amount of collateral pledged and k (, ) is an increasing cost function Cassola and Koulischer (2013) The Collateral Channel 19 / 33

20 Choice structure An Econometric Model of Collateral Policy Cassola and Koulischer (2013) The Collateral Channel 20 / 33

21 An Econometric Model of Collateral Policy Marginal profit specification Assume that the cost function is K X j(i), X (i) =ln Xj(i) X (i) X(i) It can be shown (Anderson, de Palma, Thisse 1992 and Verboven, 1996) that the share of asset pledged X j to total funding takes the nested logit form X j P N k=0 X k = PN k=1 exp k 1 T, j exp 1 PN k=0 exp k 1 1. Cassola and Koulischer (2013) The Collateral Channel 21 / 33

22 An Econometric Model of Collateral Policy Problem can be reduced to linear IV estimation Normalize profit of outside good to 0 The expression for shares can be developed to (Berry 1994) i.e. ln X j ln X 0 = j 1 ln NX k=1 k exp 1 ln X j X 0 = j h j + v j + p + ln s jg.!, The intra-group share is endogenous and calls for instruments We use variations in price of other assets -> similar to Berry, Levinsohn and Pakes (1995) Cassola and Koulischer (2013) The Collateral Channel 22 / 33

23 An Econometric Model of Collateral Policy Identification We estimate Potential concerns: ln X j X 0 = j h j + v j + p + ln s jg. 1 Endogenous haircut: the ECB s policy depends on observable asset characteristics 2 Intra-group share s jg : use instrument (price of other collateral) 3 Country, time, country-asset type interaction e ects Cassola and Koulischer (2013) The Collateral Channel 23 / 33

24 Outline Results 1 Introduction 2 Descriptive Analysis 3 An Econometric Model of Collateral Policy 4 Results 5 Conclusion Cassola and Koulischer (2013) The Collateral Channel 24 / 33

25 Results Estimation: overview of variables We estimate ln (X jct ) ln (X 0ct )= jct 1 h jt + 2 rating j + 3 gov guar j + 4 gov guar j rating rating ct + 6 maturity j + 7 holding jct + jct. h jt :assetj s ECB haircut rating j : collateral j s rating gov guar j : is collateral j guaranteed by the government? rating ct : home country s government rating maturity j : maturity of collateral j c + j + cj : Country, time, country-asset type interaction e ects Cassola and Koulischer (2013) The Collateral Channel 25 / 33

26 Results Results Cassola and Koulischer (2013) The Collateral Channel 26 / 33

27 Results Policy counterfactuals: Impact of 1% haircut increase Cassola and Koulischer (2013) The Collateral Channel 27 / 33

28 Results Policy counterfactuals: Impact of 10% haircut increase on non-investment grade government bonds Cassola and Koulischer (2013) The Collateral Channel 28 / 33

29 Results Counterfactual private funding cost in low- and high-yield countries Cassola and Koulischer (2013) The Collateral Channel 29 / 33

30 Results Cost di erential under various haircut policies Cassola and Koulischer (2013) The Collateral Channel 30 / 33

31 Outline Conclusion 1 Introduction 2 Descriptive Analysis 3 An Econometric Model of Collateral Policy 4 Results 5 Conclusion Cassola and Koulischer (2013) The Collateral Channel 31 / 33

32 Conclusion Conclusion: The impact of collateral policy on banks funding cost varies with bank- and asset-characteristics We use an original econometric approach to study the collateral channel of monetary policy Our results can predict the collateral pool under various collateral policy scenarios We find that collateral policy has a di erentiated impact on banks funding cost Suggests that collateral policy could complement the interest rate tool to transmit monetary policy in the Euro area Cassola and Koulischer (2013) The Collateral Channel 32 / 33

33 Conclusion Thank you for your attention Questions? Cassola and Koulischer (2013) The Collateral Channel 33 / 33