Securities Lending and Information Acquisition Stefan Greppmair - - PowerPoint PPT Presentation
Securities Lending and Information Acquisition Stefan Greppmair Stephan Jank Pedro Saffi Jason Sturgess Mannheim Deutsche Bundesbank Cambridge JBS Queen Mary Inquire UK - June 2020 Pedro Saffi (Cambridge) Securities Lending and
Stefan Greppmair Stephan Jank Pedro Saffi Jason Sturgess
Mannheim Deutsche Bundesbank Cambridge JBS Queen Mary
Inquire UK - June 2020
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 1 / 18
Introduction Motivation
◮ How prices incorporate new information is a defining feature of markets ◮ Large literature on how informed investors acquire and trade it:
Grossman & Stiglitz (80), Admati (85), Diamond & Verrecchia (87) Wang (93), Chen, Hong & Stein (01), Banerjee (11), Goldstein & Wang (15)
◮ Why do investors trade differently?
Differences of opinion, investment opportunity sets, private information acquisition, how they process information, etc.
◮ Loads of research show that some investor types (or strategies) have superior ability
In particular, short selling activity forecasts future returns.
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 2 / 18
Introduction Source: Boehmer et al. (10)
Monthly Fama-French-Carhart alphas (abnormal returns)
0% 1% 2% 3% 4%
0.3% 0.8% 1st month Number of months after portfolio creation
Monthly returns on lightly shorted stocks Monthly returns on heavily shorted stocks Cumulative returns on lightly shorted stocks Cumulative returns on heavily shorted stocks
Monthly Fama-French-Carhart alphas (in %) Cumulative Fama-French-Carhart alphas (in %) 6th month 5th month 4th month 3rd month 2nd month
◮ High SI stocks underperform and Low SI stocks outperform ◮ Long-short strategy alpha: Almost 18% per year
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 3 / 18
Introduction Securities Lending
◮ Huge growth of short selling activity
1609: First recorded short by Dutch trader Isaac Le Maire. Oct-2018: Short selling accounts for 42% of trading volume at the NYSE.
◮ SEC’s Definition: “The sale of a security that the seller does not own. In order to deliver
the security to the purchaser, the short seller will borrow the security, typically from a broker-dealer or an institutional investor.”
◮ The market supporting short sales is known as the "securities lending market"
Dec-2018: ISLA reports e2.2 trillion in outstanding loans, with e8.8 billion in fees.
◮ What are the options for a lender that observes an increase in shorting demand?
Increase lending to earn higher fees? Recall shares and sell position completely? Partially?
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 4 / 18
Introduction Securities Lending
benefjcial owner of types of fjrms lend the most assets particularly in certain fjxed income Insurance Companies Banks/Broker Dealers Govt/Sovereign Entities/Central Banks Corporations, LLP, LLC and Foundation & Endowment Mutual/Retail Funds Pension Plans Undisclosed/Other 19% 16% 48% 7% 7% 1% 2%
Exhibit 4 - Lendable assets by fund type
Source: IHS Markit, December 2018
◮ Main lenders: Mutual funds (passive and active) and ETFs
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 5 / 18
Introduction How Do Short Sellers Deliver Alpha?
◮ Literature:
Asquith et al. (05), Cohen et al. (07), Boehmer et al. (08), Diether et al. (08), Boehmer et al. (10), Engelberg et al. (12), Richardson et al. (19)
◮ Where does this skill come from?
Private information uncovered by short sellers: Karpoff & Lou (12), Boehmer et al. (20) Better processing of public information: Engelberg et al. (12) Compensation for risk and liquidity provision: Diether et al. (09), Richardson et al. (19)
◮ Short selling is an important channel to reveal negative information
Do equity lenders react faster to an increase in demand? Do equity lenders’ actions complement or supplement those by short sellers?
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 6 / 18
Hypotheses
1
After an increase in shorting demand, active funds that lend stocks decrease their holdings faster than passive funds.
2
After an increase in shorting demand, active funds that lend a stock react faster than active funds that do not.
Hinges on private signals from sec lending being more informative than public short data.
3
Positions closed by active funds that lend have less negative returns before the exit than those held by non-lenders. After the exit, return are more negative.
Securities lending make funds trade more aggressively and amplifies shocks Active funds that lend shares leave “losses on the table”. Consistent with strategic complementarities in trading (Goldstein & Yang (15)).
4
Securities lending allows active lenders to front-run public disclosures about large short sales positions.
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 7 / 18
Data Data Description
◮ German mutual funds’ holdings and securities lending data
Monthly portfolio holdings reported to Bundesbank’s IFS from Dec-14 to Dec-18. Monthly security lending by funds at stock level. Focus on equity funds with $10m+ AUM with at least three years.
◮ Markit’s equity lending data
Aggregate data on equity lending markets. On Loan ≈ 2.5% of market cap.; Fee around 84 bps p.a.
◮ Morningstar: German active and passive mutual funds’ returns and characteristics. ◮ Datastream: Stock data from 23 developed markets as in Fama-French (2012). ◮ Large Short Positions: Reg 236/2012 harmonises short selling regulation in the EU
From Nov/2012, short positions above 0.5% of market cap must be reported. 128,968 large short disclosures involving 6,381 stocks across 28 member states.
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 8 / 18
Data Data Description
More Stats - Table 3
Investment Focus Number of Funds AUM (in billion e) Global 87 2,712 Europe 135 1,934 Germany 57 2,328 Other 35 239 ALL 314 7,213
Fund-Level Lending Obs % Lending All Funds 314 33.12% Active Funds 262 25.57% Passive Funds 52 71.15% Stock Lending Obs Mean Median
25th Pct. 75th Pct. % of AUM Lent 2,551 3.40% 3.52% 0.90% 2.31% 4.70% % of Position MV Lent 21,645 55.44% 36.18% 17.71% 57.67% 92.75%
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 9 / 18
Results Estimation
Yi,j,t = β1SecLendi,j,t−1 × Activej × OnLoani,t−1 + β2SecLendi,j,t−1 × OnLoani,t−1 +β3SecLendi,j,t−1 × Activej + β4SecLendi,j,t−1 + X′
i,j,t−1γ1 + X′ i,j,t−1γ2 × Activej
+αi,t + αi,t × Activej + αj,t + αi,j + ǫi,j,t
◮ Main dependent variables (Yi,j,t):
% Position change for a given stock in a given fund relative to last month. % Position change but fund still owns the stock. D(Exit) = 1 if the fund has sold all shares held of a given firm, zero otherwise.
◮ Control variables: Lagged position change, portfolio weight, Amihud’s ILLIQ.
Fixed effects at stock-time, fund-time, stock-fund, and stock-time-active levels.
◮ Key coefficient is the triple interaction: SecLendi,j,t−1 × Activej × OnLoani,t−1
Tests if active lenders that lend react more to an increase in short sales demand. Lack of change by passive funds is also a nice placebo test.
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 10 / 18
Results Table 4: Equity Lending and Changes in Portfolio Positions %-Position Changes %-Position Changes (w/o exits) Exit Probability (1) (2) (3) (4) (5) (6) (7) (8) (9) SecLendi,j,t−1×Activei×OnLoani,t−1
0.004*** 0.004*** (-6.31) (-5.20) (-4.11) (-2.70) (5.52) (5.11) SecLendi,j,t−1×OnLoani,t−1 0.168*** 0.178*** 0.009 0.007
(5.06) (5.26) (0.59) (0.42) (-5.16) (-5.38) SecLendi,j,t−1×Activej
0.011*** 0.008*** 0.005** (-8.74) (-8.33) (-6.52) (-7.44) (-7.28) (-5.65) (4.01) (3.03) (2.21) SecLendi,j,t−1 0.017
0.070* 0.060 0.071 0.000 0.002* 0.002* (0.13) (-1.50) (-1.32) (1.76) (1.30) (1.56) (0.12) (1.71) (1.66) ILLIQRank
i,j,t−1
0.000 (-2.41) (-4.71) (0.71) Activej×ILLIQRank
i,j,t−1
(-1.21) (-2.65) (-0.59) PosChangei,j,t−1
0.000 (-3.69) (-3.65) (0.13) Activej×PosChangei,j,t−1 0.242*** 0.175***
(5.07) (3.63) (-9.20) PosWeighti,j,t−1
0.005** (-6.47) (-5.63) (2.26) Activej×PosWeighti,j,t−1
(-3.39) (-9.09) (-10.13) Adj.R2 0.217 0.217 0.241 0.190 0.190 0.235 0.252 0.252 0.266 N 1,047,346 1,039,263 996,429 1,005,373 997,647 957,923 1,049,476 1,039,300 996,449 Security×Time FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Security×Time×Active FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Fund×Time FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Security×Fund FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
◮ Active lenders react more aggressively to higher short selling demand.
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 11 / 18
Results Table 5: Active vs Passive %-Position Changes %-Position Changes (w/o exits) Exit Probability (1) (2) (3) (4) (5) (6) (7) (8) (9) Panel A: Active Funds SecLendi,j,t−1×OnLoani,t−1
0.002*** 0.002*** (-5.06) (-3.75) (-4.18) (-2.74) (3.52) (2.85) SecLendi,j,t−1
0.011*** 0.010*** 0.007*** (-9.44) (-9.62) (-7.82) (-7.45) (-7.34) (-5.60) (4.81) (4.56) (3.65) Adj.R2 0.199 0.199 0.219 0.129 0.130 0.162 0.262 0.261 0.276 N 796,284 790,989 750,386 756,158 751,178 713,657 797,617 791,022 750,404 Panel B: Passive Funds SecLendi,j,t−1×OnLoani,t−1 0.168*** 0.177*** 0.009 0.007
(4.62) (4.80) (0.54) (0.38) (-4.71) (-4.91) SecLendi,j,t−1 0.017
0.070 0.060 0.071 0.000 0.002 0.002 (0.12) (-1.37) (-1.21) (1.61) (1.18) (1.43) (0.11) (1.56) (1.52) Adj.R2 0.786 0.786 0.790 0.930 0.929 0.934 0.532 0.528 0.526 N 251,062 248,274 246,043 249,215 246,469 244,266 251,859 248,278 246,045 Controls No No Yes No No Yes No No Yes Security×Time FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Fund×Time FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Security×Fund FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Spillover Effects - Table 6 Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 12 / 18
Results Table 7: Abnormal Returns around Exits of Highly Shorted Stocks
Stock Lentt−12,t−1 Stock Not Lentt−12,t−1 Difference Event Time 4-Factor Alpha t-stat 4-Factor Alpha t-stat 4-Factor Alpha t-stat
0.088*** (3.44) 0.145*** (9.80)
(-2.24)
0.086*** (3.68) 0.138*** (9.95)
(-2.20)
0.082*** (3.62) 0.128*** (9.77)
(-2.08)
0.069*** (3.40) 0.119*** (9.65)
(-2.42)
0.071*** (3.92) 0.109*** (9.64)
(-2.10)
0.072*** (4.21) 0.098*** (8.82)
(-1.47)
0.055*** (3.49) 0.086*** (9.04)
(-1.93)
0.064*** (5.13) 0.073*** (8.30)
(-0.70)
0.063*** (5.65) 0.065*** (8.59)
(-0.23)
0.060*** (5.73) 0.058*** (8.18) 0.002 (0.20)
0.049*** (5.92) 0.050*** (8.65)
(-0.02)
0.028*** (4.84) 0.026*** (6.18) 0.002 (0.32) 1
(-1.43)
(-0.42)
(-1.09) 2
(-2.43)
(-0.38)
(-1.90) 3
(-2.57) 0.001 (0.13)
(-2.43) 4
(-3.08)
(-1.30)
(-2.09) 5
(-2.75)
(-1.32)
(-1.85) 6
(-2.69)
(-0.71)
(-2.09) 7
(-2.81)
(-0.79)
(-2.23) 8
(-2.58)
(-0.15)
(-2.36) 9
(-2.62) 0.004 (0.24)
(-2.64) 10
(-2.80)
(-0.22)
(-2.57) 11
(-2.31)
(-0.79)
(-1.72) 12
(-2.52)
(-1.23)
(-1.62)
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 13 / 18
Results Table 8: Position Exits of Active Lenders Cumulative 6-Month Alpha Cumulative 12-Month Alpha Variable (1) (2) (3) (4) (5) (6) SecLendi,j,t−1×OnLoani,−1
(-2.23) (-2.27) (-2.68) (-2.30) (-2.41) (-2.06) SecLendi,j,t−1 0.149 0.372 0.468
0.748 (0.17) (0.40) (0.45) (-0.28) (-0.24) (0.39) OnLoani,t−1 0.113 0.119 0.145 0.288 0.307 0.194 (0.76) (0.80) (0.94) (1.09) (1.14) (0.75) Log(FamilyNAV)j,t−1
9.527
4.285 (-0.91) (1.25) (-0.63) (0.25) Log(FundNAV)i,t−1
0.200
(-0.11) (0.10) (-0.84) (-1.62) Log(FundAge)j,t−1 0.660
0.937 20.967 (1.53) (-0.42) (0.94) (1.20) FundRelFlowj,t−1 0.007 0.003
0.030 (0.33) (0.13) (-0.76) (0.63) FundCostsj,t−1
1.125
(-1.65) (0.64) (-2.31) (-1.72) FundAlphaj,t−1 1.631 0.371
(0.94) (0.13) (-0.59) (-1.45) Fund R2
j,t−1
0.315 (-3.13) (-0.13) (-1.63) (0.95) FundTurnoverj,t−1 0.103** 0.098** 0.114 0.120 (2.38) (2.17) (1.39) (1.55) Adj.R2 0.045 0.049 0.077 0.041 0.052 0.079 N 7,685 7,685 7,232 6,072 6,072 5,696 Time FE Yes Yes Yes Yes Yes Yes Fund FE No Yes Yes No Yes Yes Fund×Time FE No No Yes No No Yes Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 14 / 18
Large Short Disclosures Identification Strategy
◮ How to separate information acquired from securities lending to public price signals?
Decision of mutual funds to reduce positions and by short sellers can be jointly determined. Need exogenous shock to information about short selling to identify effect.
◮ From Nov/2012, EU Reg 235/2012 requires disclosure of short positions above
0.5%.
Public disclosure is informative about future returns. How and when do active lenders react when a large short disclosure happens?
◮ Repeat previous tables, but focus on the six months around disclosures.
D(Disclosure)=1 if there is a new large short position disclosed at month t for stock i
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 15 / 18
Large Short Disclosures Position Changes Around Large Short Disclosures
◮ Active lenders reduce their positions prior to the disclosure... but once the
information is public, there is no longer any difference.
◮ Active funds anticipate large short trades and front-run short sellers.
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 16 / 18
Large Short Disclosures Position Changes Around Large Short Disclosures %-Position Changes %-Position Changes (w/o exits) Exit Probability (1) (2) (3) (4) (5) (6) (7) (8) (9) Disclosurei,t−1×SecLendi,j,t−1
0.009 0.010 0.006 (-1.89) (-2.07) (-1.77) (-2.31) (-2.45) (-2.52) (0.82) (0.90) (0.50) Disclosurei,t×SecLendi,j,t−1
0.001 0.001 0.002 (-1.40) (-1.46) (-1.59) (-1.98) (-1.95) (-1.83) (0.03) (0.09) (0.10) Disclosurei,t+1×SecLendi,j,t−1
0.014 0.015 0.011 (-2.01) (-2.04) (-1.72) (-1.73) (-1.69) (-1.67) (1.09) (1.15) (0.98) Disclosurei,t−1×SecLendi,j,t−1×OnLoani,t−1 0.926*
(1.69) (-0.43) (-2.12) Disclosurei,t×SecLendi,j,t−1×OnLoani,t−1
0.037 0.003 (-0.36) (0.08) (0.51) Disclosurei,t+1×SecLendi,j,t−1−1×OnLoani,t−1
0.007* (-1.93) (-1.30) (1.83) SecLendi,j,t−1×OnLoani,t−1
0.002*** 0.002*** (-5.10) (-3.71) (-4.23) (-2.73) (3.54) (2.79) SecLendi,j,t−1
0.011*** 0.010*** 0.007*** (-9.35) (-9.57) (-7.63) (-7.29) (-7.19) (-5.40) (4.71) (4.46) (3.50) Adj.R2 0.199 0.199 0.219 0.129 0.130 0.162 0.262 0.261 0.276 N 796,284 790,989 750,386 756,158 751,178 713,657 797,617 791,022 750,404 Controls No No Yes No No Yes No No Yes Security×Time FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Fund×Time FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Security×Fund FE Yes Yes Yes Yes Yes Yes Yes Yes Yes
◮ Security lenders are more responsive than non-lenders before disclosures. ◮ Security lending allows fund to obtain info that can anticipate large negative trades.
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 17 / 18
Conclusion
◮ New channel through which prices incorporate short seller’s information ◮ Active funds use securities lending information to change portfolio holdings
After increases in shorting, active lenders reduce holdings more than active non-lenders Passive funds do not exhibit this behavior
◮ Informed exits front-run negative price revelation from shorting demand
For highly-shorted stocks, active lenders exhibit relatively lower losses Lenders leave “losses-on-the-table”: Avoid almost 7% underperformance after exit Modest effects at fund-level since highly-shorted stocks are small % of portfolio
◮ How to distinguish between sec lending and public information on shorting?
Use EU Reg 235/2012 on disclosure of large short positions Active lenders reduce their positions relatively more prior to the disclosure Active funds anticipate large short trades and potentially front-run short sellers Short sellers and security lenders have complementary roles
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 18 / 18
Appendix
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 1 / 4
Appendix
Back
Variable Mean Median
25th Pct. 75th Pct. Obs. Panel A: Portfolio Variables Position Change (in %)
0.00 25.14 0.00 0.00 1,131,356 Position Weight (in %) 1.08 0.63 1.34 0.23 1.43 1,178,351 D(Position Exit) 0.04
D(Stock Lent) 0.02
D(Stock Lent (past 12 months)) 0.05
Panel B: Stock Variables On Loan (% of Market Cap.) 2.54 1.03 4.03 0.35 2.89 158,350 Short Fee (bps p.a.) 84.31 19.47 285.76 11.58 32.72 152,731 Return (in %) 0.61 0.47 9.43
5.40 162,089 Carhart 4-Factor Alpha (in %) 0.08
9.11
4.28 152,104 D(Disclosed Short Position) 0.17
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 2 / 4
Appendix
Back
Variable Mean Median
25th Pct. 75th Pct. Obs. Panel C: Fund Variables D(Fund Lends (past 12 months)) 0.27
Gross Alpha Carhart 4-Factor) (in %)
0.00 1.54
0.83 12,586 Net Alpha (Carhart 4-Factor) (in %)
1.54
0.72 12,607 Fund Costs (excl. Transaction Fees) (in %) 1.36 1.46 0.69 0.96 1.71 12,625 Net Asset Value (in billion e) 0.57 0.10 1.60 0.03 0.35 12,757 Fund Flows (in % of AUM) 0.60
29.08
0.52 12,743 Fund Age (in years) 18.76 16.33 13.16 9.83 21.83 12,757 Fund Turnover (in %) 4.43 1.92 8.19 0.08 5.40 12,747 Panel D: Fund Family Variables D(Family Lends (past 12 months)) 0.19
Family Net Asset Value (in billion e) 45.73 12.38 71.97 3.47 49.89 1,545
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 3 / 4
Appendix
Back
%-Position Changes %-Position Changes Exit Probability (w/o exits) (1) (2) (3) (4) (5) (6) SecLendi,j,t−1×OnLoani,t−1
0.003*** 0.003*** (-5.05) (-4.05) (-3.65) (-2.43) (4.15) (3.84) FundLendi,j,t−1×OnLoani,t−1
0.002*** 0.002*** (-2.10) (-2.78) (-0.03) (-0.40) (3.27) (3.84) FamilyLendi,t−1×OnLoani,t−1
0.013 0.029 0.001 0.001 (-0.41) (-0.33) (0.30) (0.57) (0.72) (0.86) SecLendi,j,t−1
0.010*** 0.0070*** (-9.40) (-7.50) (-7.36) (-5.57) (4.38) (3.43) Adj.R2 0.199 0.219 0.130 0.162 0.261 0.276 N 790,989 750,386 751,178 713,657 791,022 750,404 Controls No Yes No Yes No Yes Security×Time FE Yes Yes Yes Yes Yes Yes Fund×Time FE Yes Yes Yes Yes Yes Yes Security×Fund FE Yes Yes Yes Yes Yes Yes
Pedro Saffi (Cambridge) Securities Lending and Information Acquisition Inquire UK Seminar 4 / 4