Upload
lee-mills
View
215
Download
0
Embed Size (px)
Citation preview
Options Trading Activity and Firm Valuation
Richard Roll, Eduardo Schwartz, and Avanidhar
SubrahmanyamUCLA
The Issue Ross (1976) -- options can improve market
efficiency by expanding contingencies covered by traded securities (they help to complete the market). Allocational efficiency.
Also, since informed traders may prefer to trade options rather than stock (more leverage), options may allow agents to trade more effectively on their information, thus improving informational efficiency.
The Issue, contd. Cao and Wei (2007) find that
informational asymmetries play a more dominant role in influencing options liquidity (relative to stocks).
Easley, O’Hara and Srinivas (1998), Chakravarty, Gulen, and Mayhew (2004) find that options order flows contain information about future direction of the underlying stock price.
The Issue, contd. If prices reveal more information, then
resources are allocated more efficiently, which translates to higher firm valuations.
In addition, greater informational efficiency could reduce investment risk because market prices reflect information more precisely.
These arguments suggest that firms with higher options trading volume should be more informationally efficient and thus valued more highly.
A point of clarification The mere listing of an option does not
necessarily imply a valuation benefit. If the options market has insufficient volume,
the valuation benefit from listing would be minor because informed traders see no advantage to trading in options (Admati and Pfleiderer, 1988).
Any valuation benefit of options listing should depend on the amount of trading activity.
To the best of our knowledge, the relation between options trading activity and firm valuation has not been examined previously.
The Analysis We analyze the effect of options trading
volume on firm value after controlling for other variables that may also affect firm value such as firm size, share turnover, return on assets, capital expenditures, leverage and dividend payments.
Following other studies we use a measure of Tobin’s q as the valuation metric.
Findings We find strong evidence that firms with
more options trading volume have higher value.
Firms with more options trading activity in a given period tend to have improved financial performance in the next period. This is consistent with the premise that
options trading, by enhancing information flows, may lead to better corporate resource allocation.
Findings, contd. The results also show that the
effect of options trading on firm valuation is greater in stocks with low analyst following. This indicates that the impact of
options trading on information production is larger in stocks where investment analysis produces comparatively less public information.
Data Options trading data from Option Metrics – 1996
to 2005: 10 years of daily data (we aggregate to total annual options volume for each stock).
Matched with data from Compustat on Tobin’s q and a set of control variables.
Tobin’s q is computed as the sum of the market capitalization of the firm’s common equity, the liquidation value of its preferred stock, and the book value of its debt divided by the book value of the firm’s assets (total firm q). All results go through if we use M/B of equity instead of q.
Control variables A proxy for the firm’s leverage, long-term debt to
total assets, is intended to measure the likelihood of distress, LTD. We expect higher LTD, lower q.
Profitability, ROA, intended to capture the notion that more profitable firms may have more favorable investment opportunities. On the other hand, high ROA may also mean that the firm is in a mature phase, and has limited growth opportunities. The relation between ROA and q is an empirical issue.
Share turnover in the underlying stock: liquidity effects arising from stock trading activity as opposed to options activity.
Controls, contd. A direct measure of investment opportunities
is capital expenditures divided by sales (CapX) —high values should mean greater q.
A dummy variable for whether the firm pays a dividend proxies for capital constraints (firms that pay dividends may have more free cash flow, which may potentially be used to overinvest in marginal projects).
Firm size (market value of firm’s shares).
Number of firms with nonmissing data
Year All firmsPositiveoptionsvolume
1996 6366 13421997 6430 15751998 6157 17171999 5874 16862000 5633 16382001 5150 15032002 4883 15972003 4653 15652004 4603 17052005 4064 1655
Naturalbifurcationof sample
Summary Stats
All Firms Positive Options Volume
Variable Mean Median StandardDeviation
Tobin’s q 1.915 1.151 3.364Options volume
1877 0 23434
Size 2.197 1.885 12.61Share turnover
1.533 0.949 2.636
ROA -0.063 0.026 0.553CapX 0.566 0.040 25.02LTD 0.183 0.114 0.269DivDum 0.319 0 0.466
Variable Mean Median StandardDeviation
Tobin’s q 2.250 1.450 2.929Options volume
6487 394 43223
Size 5.239 1.025 19.92Share Turnover
2.241 1.601 2.468
ROA -0.0072 0.0406 0.2947CapX 0.2631 0.0489 6.1213LTD 0.1852 0.1348 0.2075DivDum 0.3911 0 0.4880
Correlation matrices (all firms)
Tobin’s q
Options Volume Size Share
turnover ROA CapX LTD
Options volume 0.0915
Size 0.0628 0.4136
Share Turnover 0.1055 0.0902 -0.0099
ROA -0.1175 0.0156 0.0401 -0.0786
CapX 0.0100 -0.0013 -0.0030 -0.0015 -0.0102
LTD -0.0464 -0.0154 -0.0100 -0.0555 -0.0806 0.0152
DivDum -0.0960 0.0146 0.1497 -0.1331 0.1475 -0.0125 0.0824
MatureFirms?
Correlation matrices, Firms with Positive Options Volume
Tobin’s q Options Volume Size Share
turnover ROA CapX LTD
Options volume 0.1798
Size 0.1057 0.4679
Share Turnover 0.1514 0.1375 -0.0787
ROA -0.0500 0.0261 0.0734 -0.0613
CapX 0.0155 -0.0028 -0.0068 0.0045 -0.0200
LTD -0.1350 -0.0393 -0.0340 -0.0976 -0.0744 0.0125
DivDum -0.1639 0.0040 0.1823 -0.2945 0.1781 -0.0240 0.0661
Correlations
The correlation between q and options volume is strongly positive (also with share turnover and CapX)
q is negatively related to leverage as well as the dividend dummy
q is negatively related to ROA: mature phase with fewer opp. for growth?
Preamble to main analysis For sample with positive options
volume: sorted into deciles volume each year. For each decile we compute the average value of q across all years.
Indicative that firms with higher options volume have higher q.
Figure 1: Average Tobin’s q vs. options volume
Significant
Sort by size and then by volume
To allow for independent variation in size and volume.
Average q for the resulting 25 portfolios.
q increases with volume for every size group.
Options volume and q by size
Size quintile
1 2 3 4 5
Options volume quintile
1 1.382 1.548 1.770 1.596 1.569
2 1.456 1.603 1.840 1.792 2.278
3 1.714 1.931 2.151 2.316 2.490
4 2.008 2.382 2.782 2.858 3.038
5 2.202 2.764 3.130 3.370 4.283
Strongeffect
Regressions Year-by-year cross-sectional
regressions and then test the significance of the time series coefficients. t-statistics are corrected by procedure Newey/West: residuals of the c-s regressions are likely to be serially correlated due to the autocorrelation in q.
Regression results – all firms
Dependent Variable: Tobin’s q
Time series of annual cross-sections
t-statistic corrected by Newey/West
Variable Coefficient t-statisticOptvol 0.1297 4.83Size 0.9875 2.18Stkturn 0.1259 3.55ROA -0.8672 -3.01CapX 0.0125 2.81LTD -1.0970 -2.73Divdum -0.4326 -4.88
Average adjusted R2=8.55%Average number of firms: 5381
Regression results – firms with positive options volume
Variable Coefficient t-statistic
Optvol 0.1188 3.38
Size 0.7017 2.09
Stkturn 0.1437 2.75
ROA -0.5756 -1.25
CapX 0.0632 1.78
LTD -1.631 -4.31
Divdum -0.7293 -5.31
Average adjusted R2=12.05%
Avg no. of firms: 1557
Summary of results Tobin’s q is positively and significantly
related to options trading; the effect is economically significant, 16% to 23% increase in q for a one sigma increase in options volume, ceteris paribus
q is also negatively related to leverage and the dividend dummy, consistent with proposed hypotheses
Stock trading activity also bears a positive relation with q
Robustness Checks Various checks were performed
and in all cases the central results are unchanged Skewness Panel regression Endogeneity issues Industry effects Additional explanatory variables
Results with log(options volume) for firms with positive options volume
This checks whether the skewness in options volume affects the results; it doesn’t. From now on we use Ln(optvol).
Variable Coefficient t-statisticLn(Optvol) 0.1978 3.72Size 0.8288 2.70Stkturn 0.0873 2.35ROA -0.6761 -1.43CapX 0.0569 1.66LTD -1.638 -4.47Divdum -0.7923 -5.57
Average adjusted R2=12.61%Average number of firms: 1557
Panel Regression: pools cross-section and time-series data
Panel Estimates
Variable Coefficient t-statisticLn(Optvol) 0.1097 12.98Size 2.786 26.57Stkturn -0.0857 -7.48ROA -0.0573 -0.22CapX 0.3299 1.03LTD -0.6712 -2.97Divdum -0.9649 -16.07
Balanced Panel, to accommodate serial correlation and cross correlation in the errors, using the Parks (1967) Procedure (see Appendix.) Firms included must be present in all years (502 firms).
Endogeneity One could argue, albeit implausibly, that
high q firms may attract more attention and this may translate to greater options volume (reverse causality).
One simple way to address this issue is to consider the relation between q and one-year lagged options trading volume.
Then we use an Instrumental Variable approach.
Regression results using lagged options volume
Variable Coefficient t-statistic
Lag(Ln(Optvol)) 0.0879 5.56
Size 1.381 2.60
Stkturn 0.1779 2.06
ROA -0.0912 -0.30
CapX 0.0717 1.74
LTD -0.9991 -4.80
Divdum -0.8275 -5.27
Average adjusted R2=10.19%
Average number of firms: 1359
An instrument We need an instrument for options trading
volume that is inherently unrelated to q. Finding such an instrument is a difficult endeavor and inevitably involves an element of subjectivity.
We propose that options volume may be related to the average absolute moneyness, the relative difference between the stock’s market price and the option’s strike price (correlation 0.19).
An alternative instrument for options volume is the total open interest in options within a given year.
IV estimation (2SLS)
Moneyness as instrument Open interest as instrumentVariable Coefficient t-statistic Coefficient t-statisticIV(optvol) 0.3485 2.75 0.1385 2.84Size 0.1839 1.90 1.118 2.91Stkturn 0.1348 3.07 0.1367 2.63ROA -0.7009 -1.53 -0.6277 -1.35CapX 0.0539 1.71 0.0586 1.72LTD -1.551 -4.66 -1.608 -4.47Divdum -0.6509 -7.01 -0.7415 -5.78
Average adjusted R2=13.23% Average adjusted R2=11.53%
Average number of firms: 1557
First equation: q as a function of same variables, using optvol fromSecond equation: optvol as a function of instrument and size.Main result is not due to reverse causality.
Year-by-year coefficients on ln options volume (dependent variable – Tobin’s q)
Are unusual years driving the results; they’re not. Are industry outliers (e.g., the tech bust) are responsible; they aren’t.
Without industry controls With Fama and French (1997) industry controls
Year Coefficient t-statistic Coefficient t-statistic1996 0.1452 4.06 0.1351 3.791997 0.1788 5.52 0.1659 5.181998 0.2236 6.28 0.2029 5.731999 0.5553 8.41 0.5025 7.492000 0.2820 9.01 0.2405 7.542001 0.1524 6.05 0.1410 5.682002 0.0898 4.92 0.0885 5.032003 0.0873 4.80 0.0835 4.802004 0.1123 5.87 0.1085 5.872005 0.1518 7.98 0.1433 7.68
Other robustness checks Results are robust to scaling options
volume by shares outstanding, and to using log transformation of the positive controls.
To test that option trading activity does not proxy for stock riskiness which could potentially affect q: return volatility is not significant in the overall regression for Tobin’s q and the options volume variable remains significant.
Options trading and future firm performance: Further analysis Identify the mechanism by which
options trading enhances firm value. If options trading activity leads to better
corporate resource allocation, then there may be a relation between future firm profitability and options trading.
We regress ROA (our measure of financial performance) on lagged values of options volume and control variables.
Firm performance and options trading (LHS variable=ROA)
Panel EstimatesOne-year lagged variables Coefficient t-statistic
Ln(Optvol) 0.0042 12.35Size 0.0041 0.74Stkturn -0.0019 -4.63ROA 0.4283 9.32CapX 0.0411 1.65LTD -0.1135 -5.98Divdum 0.0456 14.45
Parks procedure to account for autocorrelation
Persistence
Firm performance and options There is a positive relation between
future ROA and current options activity
This supports the information channel: that more options trading is associated with greater informational efficiency, which, in turn, leads to improved resource allocation.
Options Trading and Investment Sensitivity to Stock Price The degree to which managers obtain
information from market prices to make investment decisions can be captured by the sensitivity of corporate investment to market prices.
Several papers have theoretically and empirically analyzed this sensitivity.
But, managers might learn more from market prices when options volume is greater (produces private information).
Corporate Investment Sum of capital expenditures and R&D
expenses scaled by beginning of year book assets.
We look at the interaction variable of q with options volume.
We also include q to capture the baseline effect of market valuation on investment (both lagged) and other controls.
Corporate Investment (LHS)
Panel Estimates
Explanatory variables Coefficient t-statistic
Tobin’s q 0.4168 8.29Ln(Optvol*q) 0.1793 8.33InvAssets 0.4436 6.32Return (t+1) -0.0604 -2.20Cash flow 2.805 8.35
Lagged 1 yr
Corporate Investment Positive sensitivity of investment to
stock price (q). Greater sensitivity to q when options
trading is high. Supports notion that options trading
contributes to information production, which managers use in making corporate investment decisions.
Information Asymmetry Effect of options trading on valuation may
be more pronounced in stocks with greater levels of informed trading.
Difficult to find a measure for level of informed trading in options markets.
We use the PIN (probability of informed trading, computed with stock data) measure of Easley, Hvidkjaer and O’Hara (2002) as a proxy for information asymmetry.
Information Asymmetry Using the structure of a sequential
trade market microstructure model, they derive an explicit measure of the probability of information based trading (PIN) for an individual stock
For stocks with high PINs the effect of option volume on valuation should be greater.
Information Asymmetry
LOVOL LOVOL*PINL
Year Coefficient t-statistic Coefficient t-statistic
1996 0.2297 2.82 0.0645 1.77
1997 0.4388 5.14 0.1334 3.81
1998 0.2948 4.95 0.0650 2.66
1999 0.5390 7.29 0.1563 5.31
2000 0.3795 4.03 0.1150 2.95
2001 0.3227 3.02 0.0907 2.31
The effect of options volume on q is stronger in stocks where more informationis produced by the trading process.
InteractionVariable
Information Asymmetry Options volume variable remains
significant and the interaction of options volume with PIN is positive and mostly significant.
Suggestive evidence that the effect of options volume on q is stronger in stocks where more information is produced by the trading process.
Security analysts and options trading Options volume could proxy for another
measure of information production, the extent of analysts following. If no. of analyst following a company is included in the regressions: not significant whereas option volume remains significant.
Results are also robust to the inclusion of the dispersion of long-term growth forecasts by analysts which is forward looking measure of uncertainty.
A role for analyst following The effect of options in information
production may be greater in stocks with low analyst following, where little public information is produced and trading on private information may be more important. In these cases private information may
play a stronger part in information production.
Testing the impact of analyst following We sort the sample each year into
three groups by analyst following, and label them 0,1,2.
We interact options volume with this indicator variable and include the interaction variable in the regression.
Regression results with interaction variable for analyst following
Variable Coefficient t-statisticLn(Optvol) 0.2573 4.23
Ln(Optvol)*ANALYS -0.0293 -3.78Size 0.9940 2.96
Stkturn 0.0183 2.38ROA -0.6138 -1.37CapX 0.0565 1.67LTD -1.5869 -4.55
Divdum -0.7511 -5.61
Interpretation of results with inclusion of analyst following The impact of options volume on
Tobin’s q is stronger in firms with less analyst following, but it remains significant even for firms with large analyst following.
Suggests private information production is more important in stocks where investment analysts produce less public information.
Bottom Line The amount of options trading is
associated with higher firm valuations. This result is consistent with the dual
notions that more options trading is associated with greater informational efficiency of prices and superior resource allocation.
The results survive when subjected to a variety of robustness checks, including different specifications of volume.
Bottom Line, contd. There is a positive link between future firm
performance and current options volume, suggesting that options trading enhances information production and, in turn, resource allocation.
The role of options volume on valuation is stronger in firms with less analyst following where it is likely that less public information is produced In these stocks, private information
production through trading may be more important for resource allocation
Issues of Interest The key point of our paper is that the degree
to which an option is traded, not its mere listing, is associated with higher valuations.
It would be interesting to consider whether this notion extends to other scenarios.
For example, some countries have futures contracts on individual stocks, and the effect of such contracts on valuation could be ascertained.
Analyzing the impact of index options and futures on market valuation would also be of interest.