Funding liquidity, market liquidity and TED spread : A two-regime model Kris Boudt Ellen C.S. Paulus Dale W.R. Rosenthal Discretionary liquidity : Hedge funds, side pockets, and gates Adam L. Aiken Christopher P. Clifford Jesse Ellis Discussant : Jérémy Dudek Monday, March 25 2013 Jérémy Dudek - CREST-Dauphine 1/20
Summary 1 Funding Liquidity versus Market Liquidity 2 Boudt, Paulus and Rosenthal, 2012 3 Aiken, Clifford and Ellis, 2012 4 Comments and Questions Jérémy Dudek - CREST-Dauphine 2/20
Funding Liquidity versus Market Liquidity We distinguish two types of liquidity : Market Liquidity corresponds to the ease with which the asset can be traded. Funding Liquidity corresponds to the ease with which the trader can obtain funding. Brunnermeier, Markus K and Pedersen, Lasse Heje, 2009. Market liquidity and funding liquidity. Review of Financial studies, 22(6), pages 2201-2238. Jérémy Dudek - CREST-Dauphine 3/20
A really hot topic Since the 2007-2008 financial crisis : Some events shed light on funding liquidity problems. Liquidity is at the very center of the crisis as well as : participating to the maintenance of the crisis, a factor of the crisis propagation. A liquidity problem may lead to liquidity spirals. Jérémy Dudek - CREST-Dauphine 4/20
Liquidity Spirals Initial losses Jérémy Dudek - CREST-Dauphine 5/20
Liquidity Spirals Initial losses Increase of leverage ratio Jérémy Dudek - CREST-Dauphine 5/20
Liquidity Spirals Constraint to reduce positions Initial losses Increase of leverage ratio Jérémy Dudek - CREST-Dauphine 5/20
Liquidity Spirals Constraint to reduce positions Initial losses Increase of leverage ratio Decrease of asset s prices Jérémy Dudek - CREST-Dauphine 5/20
Liquidity Spirals Constraint to reduce positions Initial losses Increase of leverage ratio Decrease of asset s prices Losses on existing positions Jérémy Dudek - CREST-Dauphine 5/20
Liquidity Spirals Constraint to reduce positions Initial losses Increase of leverage ratio Decrease of asset s prices Losses on existing positions Jérémy Dudek - CREST-Dauphine 5/20
Viewpoints According to the viewpoint, the provider of funding liquidity may differ : Trader is funded by banks, Boudt, Paulus and Rosenthal, 2012 Fund manager is funded by external investors (clients). Aiken, Clifford and Ellis, 2012 Tools to control for liquidity risk largely differ. Jérémy Dudek - CREST-Dauphine 6/20
Viewpoints Macro / Global Who is interested in? Regulators, market participants (in a more general sense) Funding Liquidity : ease with which a trader can finance his trades Tools Economic policies, market participants action Micro / Fund specific Who is interested in? Fund managers, Risk managers Funding Liquidity : depends on the willingness of fund s clients to withdraw Tools Cash pocket allowing to satisfy first outflows, DLRs Jérémy Dudek - CREST-Dauphine 7/20
Data Macro / Global from March 1998 to December 2011, Bid-Ask spread for the S&P 500, VIX, TED spread, Instruments : tick size change dummy, trend inter trade duration, spread AAA-LIBOR. Funding Liquidity measures : Central Bank measure : excess broker loan rate, Market measure : value weighted average loan rate. Micro / Fund specific New dataset of large institutional investors allowing : to track their portfolios, to determine if a particular hedge fund gated or created a side pocket, to track the performance of the hedge fund following the illiquidity event. Quarterly panel of HFs for each FoFs : 57 unique FoFs, 1.411 unique HFs, 15.133 unique FoF-HF quarterly holdings. Jérémy Dudek - CREST-Dauphine 8/20
Summary 1 Funding Liquidity versus Market Liquidity 2 Boudt, Paulus and Rosenthal, 2012 3 Aiken, Clifford and Ellis, 2012 4 Comments and Questions Jérémy Dudek - CREST-Dauphine 9/20
In this paper They investigate the effect of market liquidity on funding liquidity, more precisely : equity-collateralized funding. They propose two novel measures of funding liquidity, They suggest a separation between stabilizing and destabilizing effect of market liquidity on funding liquidity, They use an econometric method allowing to take into account endogeneity, Hansen (2000), Caner and Hansen (2004). Jérémy Dudek - CREST-Dauphine 10/20
Motivations Arguments : Funding liquidity depends on the willingness of the financiers to provide loans against equity collateral. willingness depends on : quality of the assets, in particular liquidity and volatility. Main contributions : First paper to take into account endogeneity, in the relation between funding liquidity and market liquidity or volatility. Application of a two regimes model in order to capture the non linear relationship, according to the TED spread. Jérémy Dudek - CREST-Dauphine 11/20
Results According to the market wide credit conditions : evidence of a stabilizing and a destabilizing behaviour of financiers. 1 TED spread < threshold calm market, market participants (or central bankers) chose stabilizing rates. 2 TED spread > threshold jittery market, increase of the premium charged fitting the description of destabilizing rates. Jérémy Dudek - CREST-Dauphine 12/20
Summary 1 Funding Liquidity versus Market Liquidity 2 Boudt, Paulus and Rosenthal, 2012 3 Aiken, Clifford and Ellis, 2012 4 Comments and Questions Jérémy Dudek - CREST-Dauphine 13/20
This paper Discretionary Liquidity Restrictions Allow to restrict investor liquidity using : 1 gates, 2 side pockets. Study of causes and consequences of DLRs : Focusing on determinants of DLRs, Showing a negative impact on performances, Determining consequences for Hedge Funds investors, Is this a protection for investors as managers claim? Jérémy Dudek - CREST-Dauphine 14/20
Motivations Motivations : > 30% of hedge fund managers used their DLRs, Investors are interested in the impact on the performances. Main contributions : First large, empirical study of hedge fund side pockets and gates, Quantitative analysis of the impact on the performances, Study of the impact on the Hedge Fund family, ability to raise funds after a redemption restriction. Jérémy Dudek - CREST-Dauphine 15/20
Results Adding withdrawal restrictions leads to : no abuses from managers, Potentially efficient rationale for restricting investor liquidity. poor performances following the illiquidity event, impact the ability to raise capital for the Hedge Fund family, Having enacted DLR. Jérémy Dudek - CREST-Dauphine 16/20
Summary 1 Funding Liquidity versus Market Liquidity 2 Boudt, Paulus and Rosenthal, 2012 3 Aiken, Clifford and Ellis, 2012 4 Comments and Questions Jérémy Dudek - CREST-Dauphine 17/20
General Comments Very interesting papers, inside a very hot topic. two distinct approaches of the funding liquidity : focusing on two different scales of the economy. New very complete databases : including both central bank and market data, covering a large sample of Hedge Funds with complete informations on FoF s portfolios. Strong econometric methods to model : the impact of market liquidity on funding liquidity, the impact of DLRs on Hedge Funds performances. Jérémy Dudek - CREST-Dauphine 18/20
Specific Comments and Questions (1/2) Data 1 More information about the consideration of weak instruments, [reporting the value of the F-Test, Staiger and Stock, 1997)] Jérémy Dudek - CREST-Dauphine 19/20
Specific Comments and Questions (1/2) Data 1 More information about the consideration of weak instruments, [reporting the value of the F-Test, Staiger and Stock, 1997)] Model 1 Is fundilliq M t (market based funding liquidity measure) still defined? 1 DS,it = fundilliq M t = log ( N i=1 Trades ) it VWAF it 1 DS,it N i=1 Trades it 1 DS,it 1 if (VWAF i,t 1 < VWAF i,t ) (TBQ i,t 1 < TBQ i,t ) 1 if (VWAF i,t 1 > VWAF i,t ) (TBQ i,t 1 > TBQ i,t ) 0 otherwise Is there any day for which no shift of the demand curve occurs? Jérémy Dudek - CREST-Dauphine 19/20
Specific Comments and Questions (1/2) Data 1 More information about the consideration of weak instruments, [reporting the value of the F-Test, Staiger and Stock, 1997)] Model 1 Is fundilliq M t (market based funding liquidity measure) still defined? 2 κ significant does not mean 2 regimes. Example of a demeaned variable : κ could be equal to 0 (not significantly different from zero), but they could have two distinct regimes around this value. Jérémy Dudek - CREST-Dauphine 19/20
Specific Comments and Questions (1/2) Data 1 More information about the consideration of weak instruments, [reporting the value of the F-Test, Staiger and Stock, 1997)] Model 1 Is fundilliq M t (market based funding liquidity measure) still defined? 2 κ significant does not mean 2 regimes. Results 1 Sum of significant coefficients is still negative, Can we interpret it as a destabilizing effect? Is the sum of coefficients significant? 2 fundilliq M results are not significant. Jérémy Dudek - CREST-Dauphine 19/20
Specific Comments and Questions (2/2) Data 1 Quarterly data leads to small number of observations, only 12 observations for the estimation of liquidity measures. Jérémy Dudek - CREST-Dauphine 20/20
Specific Comments and Questions (2/2) Data 1 Quarterly data leads to small number of observations, only 12 observations for the estimation of liquidity measures. Model 1 How do you define a fund of the control group? based on the probability? based on the value of parameters? 2 Only two quarter lagged returns for matching (Are 2 points enough?). 3 Is it relevant to winsorize before matching? Considering information added by large values in this method. Jérémy Dudek - CREST-Dauphine 20/20
Specific Comments and Questions (2/2) Data 1 Quarterly data leads to small number of observations, only 12 observations for the estimation of liquidity measures. Model 1 How do you define a fund of the control group? 2 Only two quarter lagged returns for matching (Are 2 points enough?). 3 Is it relevant to winsorize before matching? Results 1 What could be the results compared with funds without possible restrictions? Are the managers more conservative knowing they can not restrict redemption? Jérémy Dudek - CREST-Dauphine 20/20