From Funding Liquidity to Market Liquidity: Evidence from the Index Options Market Zhiping Zhou, Chunbo Liu, Cheng Zhang 1 Wuhan University 2 Norwegian School of Economics 3 London School of Economics and Political Science 27 August 2016 1 / 15 Motivation Brunnermeier and Pedersen (2009) Provide a model that elaborates on the relationship between funding liquidity and market liquidity (FL-ML) Show that the two notions are mutually reinforcing, leading to liquidity spirals Duffie (2010, 2012) Financial crisis increases the cost of intermediation and thus lead to increase in trading spreads (2010 AFA presidential address) Investors and issuers of securities would find it more costly to borrow, raise capital, invest, and obtain liquidity (2012) However, the implications of these recent important theoretical findings have not been fully investigated from an empirical point of view 2 / 15 Motivation Chordia, Sarkar, and Subrahmanyam (2005) Explore liquidity movements in stock and Treasury bond markets (1800 trading days) Establish a link between macro liquidity, or money flows, and micro or transactions liquidity Hameed, Kan and Viswanathan (2010) Test the relationship between funding liquidity and market liquidity in the stock market Mancini, Ranaldo, and Wrampelmeyer (2013) Use intraday trading and order data to measure liquidity in the foreign exchange (FX) market Show that negative shocks in funding liquidity lead to significantly lower FX market liquidity Target: this paper analyzes the impact of funding liquidity on index options market liquidity 3 / 15 Results Preview Using daily data on option market liquidity, this paper relates index options market liquidity to measures of funding liquidity as well as liquidity of equity markets We find a positive relationship between option market liquidity and funding liquidity The put option liquidity is more closely related to funding liquidity than call options The FL-ML relationship can be found for short-term options Out-of-the-money and at-the-money options exhibit higher sensitivity to funding liquidity than in-the-money ones The lower the implied volatility (IV), the larger the impact of funding liquidity on market liquidity Consistent with the predictions of Brunnermeier and Pedersen (2009), we find that when funding liquidity is low, option market liquidity becomes sensitive to changes in funding liquidity 4 / 15 Data and variables OptionMetrics IVYDB: daily closing bid and ask quotes, daily volume and open interest on S&P 500 options traded on the CBOE market Time span: Jan 17, 2003 - Jan 30, 2012. Sample size: 2,265 days = 223,447 obs. Option market liquidity measures (Cao and Wei 2010) Proportional Bid-ask spread (PBA) PJ askj −bidj j=1 VOLj ∗ (askj +bidj )/2 PJ j=1 VOLj PJ Trading Volume (VOL) j=1 VOLj PJ Dollar trading volume (DVOL) j=1 VOLj (askj + bidj ) /2 Funding liquidity measures TED spread from Federal Reserve Bank in St.Louis Net acquisition of financial assets by security brokers and dealers from flow of funds account provided by Federal Reserve Statistical Release. 5 / 15 Summary statistics Panel A: Option Liquidity Bid-ask spread Volume Statistics All options Call options Put options Statistics Option bid-ask spread Option volume Option dollar volume Mean 11.84 12.67 11.19 Median 11.40 11.99 10.44 Std. 4.34 5.21 4.62 Mean 2.86 2.54 3.14 Median 2.71 2.37 2.94 Dollar Volume Std. 1.24 1.15 1.47 Mean 7.03 6.32 7.58 Median 5.46 4.99 5.60 Std. 5.87 5.02 6.94 Panel B: The Level of Key Variables in Sub-Periods Jan. 2003 - Jul. 2007 Aug. 2007 - Jun. 2009 Jul. 2009 - Jan. 2012 Mean 12.82 2.28 3.46 Mean 11.68 3.32 8.58 Median 12.37 1.90 2.87 Std. 4.75 1.18 2.08 Mean 9.69 3.62 13.51 Median 9.18 3.46 11.45 Std. 3.52 1.05 7.58 Median 11.54 3.21 7.57 Std. 3.46 0.94 3.93 Compared with call options, puts have a smaller bid-ask spread and a larger trading volume High liquidity for puts might be caused by the high transaction activities of the informed traders during the Great Recession 6 / 15 Econometric Model We utilize an ARMAX model to analyze the relationship between funding liquidity and option market liquidity OLt = α + β1 FLt−1 + β2 VIXt−1 + p X ρi OLt−i + i= q X θj t−j + t j OLt is the option market liquidity on day t FL is the funding liquidity measure and VIX is the proxy for market uncertainty An optimal number of lags p and q is selected based on information criteria We expect β1 to be positive and β2 to be negative 7 / 15 Empirical Results Dependent var. Sample L.VIX L.TED Spread bid-ask All bid-ask All bid-ask Call bid-ask Put -0.109*** (-8.47) 0.649*** (2.76) -0.124*** (-4.49) 1.358* (1.71) -0.001 (-0.80) (4,3) 2264 -0.099*** (-2.73) 0.644 (0.61) -0.132*** (-4.15) 2.036** (2.40) (2,2) 2264 (3,3) 2264 Acq2 ARMAX N OLS 1302 one-σ ↑ in VIX → PBA ↓ by 1.1 bps (25% of one-σ) one-σ ↑ in TED → PBA ↑ by 0.65 bps (18% of one-σ) Options market liquidity declines when liquidity providers face higher funding cost (Brunnermeier and Pedersen 2009) We can only find a positive relationship between PBA spread and TED spread for the subsample of puts Puts are favoured by informed traders to realize their information value 8 / 15 Option liquidity: the effect of maturities L.TED Spread L.VIX N Model Short 1.572* (1.72) -0.154*** (-4.63) 2263 ARMAX(4,3) Medium -0.149 (-0.30) -0.062*** (-2.88) 2263 ARMAX(4,3) Long -0.249 (-0.82) -0.016 (-1.24) 2257 ARMAX(4,3) A reduction of funding liquidity is followed by lower short options liquidity Medium-maturity option liquidity is impacted by VIX Long-maturity option liquidity does not respond to TED or VIX 9 / 15 Option liquidity: the effect of moneyness L.TED Spread L.VIX ARMAX(p,q) N OTM 3.136** (2.16) -0.275*** (-4.53) (2,3) 2264 ATM 1.682* (1.85) -0.123*** (-3.76) (2,1) 2264 ITM 0.490 (1.40) -0.002 (-0.15) (1,2) 2262 DOTM 0.101 (0.43) 0.016 (1.53) (1,1) 2262 DITM 0.410*** (3.30) 0.032*** (5.94) (3,2) 1639 TED spread is positively related to the PBA spread, implying the liquidity of options with different moneyness declines when funding costs are high VIX is negatively correlated with market liquidity of OTM, ATM, and ITM, implying increase in uncertainty is followed by an increase in market liquidity The coefficients become positive for DITM and DOTM. This might due to the fact that investors have to afford a high cost in transactions 10 / 15 Option liquidity: the effect of implied volatility Implied volatility L.TED Spread L.VIX ARMAX(p,q) N (1) IV1 10.008* (1.70) 1.109*** (5.36) (2,3) 1166 (2) IV2 9.518*** (3.67) 0.608*** (7.23) (2,1) 1648 (3) IV3 9.790*** (3.63) -0.021 (-0.13) (2,2) 1919 (4) IV4 3.427 (0.95) -0.447* (-1.84) (3,2) 1788 (5) IV5 -4.226 (-1.03) -0.762*** (-4.22) (3,2) 1279 A monotonic decline in VIX coefficients and an almost monotonic decline in TED spread coefficients from IV1 to IV 5 Investors would like to sell options with high implied volatility levels during the periods of high market uncertainty The liquidity of options with high implied volatility increases 11 / 15 Stock market liquidity Liquidity measure Weight L.vix L.TED Spread BAHL ew High-Low equal -0.155*** (-5.38) 1.093 (1.31) 1.982*** (9.14) BAHL vw High-Low volume -0.160*** (-5.28) 1.244 (1.44) Close equal -0.117*** (-4.31) 1.527* (1.94) Close volume -0.128*** (-4.76) 1.450* (1.77) Volume Dollar -0.138*** (-4.54) 1.391* (1.66) -0.139*** (-4.62) 1.587* (1.90) 2.075*** (8.94) BAClose ew 0.504 (0.74) BAClose vw 25.011*** (3.15) Stock Volume 0.280*** (4.95) Stock Dollar ARMAX(p,q) N (4,3) 2263 (4,3) 2263 (4,3) 2263 (4,3) 2263 (4,3) 2263 0.010*** (5.86) (4,3) 2263 After controlling the VIX and equity market liquidity, there is still a positive relationship between options market liquidity and funding liquidity 12 / 15 The crisis Sample Period L.VIX L.TED Spread ARMAX(p,q) N (1) (2) Dollar volume Pre Post 0.386*** 0.567*** (17.29) (22.40) 0.619 -2.306*** (0.97) (-3.08) (2,1) (2,2) 1140 1124 (3) (4) Volume Pre Post 0.066*** 0.055*** (4.40) (7.35) 0.316 -0.362* (0.72) (-1.76) (2,1) (2,2) 1140 1124 The TED spread is strongly negatively related to the dollar trading volume for the post-crisis period, but its coefficient is not significant for the precrisis period, implying a reduction of funding liquidity is followed by lower dollar trading volume Consistent with the theoretical predictions of Brunnermeier and Pedersen (2009), there are higher comovement in liquidity and higher impact of funding liquidity on market liquidity during the crisis 13 / 15 Conclusions We provide evidence of a positive relationship between funding liquidity and option market liquidity during the periods of high market uncertainty Sub-sample analysis helps uncover interesting phenomenon in the option market The liquidity of puts and calls responds asymmetrically to downward market movements The liquidity of short maturity options is positively related with funding liquidity Options with different Moneyness and implied volatility behave differently Sub-sample analysis of pre- and post- August 2007 means that when funding liquidity is low during the the recessions, market liquidity becomes sensitive to changes in funding liquidity 14 / 15 Future Research One natural extension would be the in-depth examination of the relationship of funding liquidity and options market liquidity using a panel data sample Another area of future research would be to investigate the effect of funding liquidity on the pricing of index options 15 / 15