• Selected
• Entire List

Abstract:
The paper compares the forecasting ability of the most popular volatility forecasting models and develops an alternative. The comparison of existing models focuses on four issues: 1) the relative weighting of recent versus older observations, 2) the estimation criteria, 3) the trade-off in terms of out-of-sample forecasting error between simple and complex models, and 4) the emphasis placed on large shocks. Like previous studies, we find that financial markets have longer memories than reflected in GARCH(1,1) model estimates but find this has little impact on out-of-sample forecasting ability. While more complex models which allow a more flexible weighting pattern than the exponential model forecast better on an in-sample basis, due to the additional estimation error introduced by additional parameters, they forecast poorly out-of-sample. With the exception of GARCH models, we find that models based on absolute return deviations generally forecast volatility better than otherwise equivalent models based on squared return deviations. Among the most popular time series models, we find that GARCH(1,1) generally yields better forecasts than the historical standard deviation and exponentially weighted moving average models though between GARCH and EGARCH there is no clear favorite. However, in terms of forecast accuracy, all are dominated by a new, simple, non-linear least squares model, based on historical absolute return deviations, that we develop and test here.

Abstract:
Both bond rating agencies and stock analysts evaluate publicly traded companies and communicate their findings and opinions to investors. This study compares the timeliness of each and explores Granger causality between the two. We find that bond rating downgrades are partially a response to information which both the market and earnings analysts already have and which they have already impounded in prices and earnings forecasts respectively. However, downgrades are also viewed by both earnings analysts and market participants as providing some new information - the market reacts negatively and analysts revise their forecasts sharply downward. Indeed, the reaction of earnings analysts to downgrades exceeds their response to other informational events as documented in previous studies. Controlling for pre-downgrade information releases, we find that analysts reduce their forecasts about 7% in the first month following a downgrade and about 13% over three months. We find that the post-downgrade earnings forecast revisions are partially predictable from the market response to the downgrade implying that those downgrades which are a surprise to market participants and which they view as providing new information are viewed the same way by earnings analysts. On the other hand, we find no evidence that the market response is partially predictable from the pre-downgrade earnings forecast revisions. While the market apparently views upgrades as providing no new information, since there is no market response to the announcement, we find that stock analysts tend to revise their earnings forecasts upward following upgrades. However, these upward revisions are much smaller than the downward revisions following downgrades. Throughout our analysis, we control for the tendency for analysts to be overly optimistic initially and to respond with a lag to public information.

Abstract:
The implied volatility "smile" embodied in option prices has generally been attributed to errors in the Black-Scholes model, specifically the assumption of constant volatility or the assumption of log-normal returns. In other words, the presumption is that, if the implied volatilities were calculated correctly, the smile would disappear. This has led to the development of a plethora of what might be termed "smile consistent" option pricing models, models which generate option prices with smiling B-S implied volatilities when the true (or correctly calculated) smile is flat.

In this paper, we test and reject the hypothesis that the smile in stock index option prices is wholly due to inappropriate distributional assumptions by the Black-Scholes option pricing model. If the true smile is flat, then a trading strategy in which one buys options at the bottom of the incorrect Black-Scholes smile and sells options at the top(s) should not be profitable even on a pre-transaction-cost basis. However, we find that such a strategy is quite profitable. Moreover the profits vary in line with the Black-Scholes model's predictions while they should not if the true smile is flat. Our calculations suggest that roughly half of the observed smile in the stock index options market is due to a smile in the true implied volatilities while about half is due to a difference between the Black-Scholes implied volatilities and the true implied volatilities.

We argue that the true smile persists despite these substantial pre-transaction-cost trading profits, because maintaining the trading portfolio's low risk profile requires frequent re-balancing which quickly eats away the profits. While the trading portfolios are constructed to be both delta and gamma neutral, they quickly lose these properties as the underlying price changes necessitating frequent re-balancing. With daily or more frequent re-balancing, the trading portfolios are not profitable on a post-transaction-cost basis. Consequently, the smile is not evidence of market inefficiency.

implied volatility, options, market efficiency

Abstract:
Dividing the 223 largest traders of heating oil futures between June 1993 and March 1997 into 11 different line-of-business groupings, we document their trading activity. We find substantial and significant differences between the 11 trader types in their propensity to take long, short, or spread positions; the terms-to-maturity of their holdings; their turnover rates; and the size of their positions. We also find that both trading volume and open-interest positions are dominated by potential hedgers but that it is not appropriate to treat traders whom the Commodity Futures Trading Commission classifies as commercials as hedgers.

Abstract:
We document the trading activities over the period from June 1993 through March 1997 of the 223 largest traders of heating oil futures - traders who together account for 58% of the open interest in this market. Dividing these traders into eleven different groupings: refiners, marketers/distributors, commercial banks, investment banks, end users, energy traders, other energy firms, commodity pools, commodity trading advisors, floor brokers, and other non-commercial traders, we explore how their trading activities differ and compare. While past studies have documented the trading of a single trader, a single type of trader, or traders in general, this is the first dis-aggregated study of all large traders in any derivatives market. We find substantial and significant differences between the eleven trader types in: their propensity to take long or short positions, whether they hold outright (naked) or spread positions, the term-to-maturity of the contracts they hold, how long they hold a position, and the size of their positions. We also find that both trading volume and open interest positions are dominated by potential hedgers, rather than speculators, where potential hedgers are defined as traders with substantial positions in the cash and/or forward heating oil market. Finally, we find that it is not appropriate to treat traders which the CFTC identifies as "commercials" as hedgers as is common practice in the finance literature.

Abstract:
Documenting spread and combination trading in a major options market for the first time, we find that spreads and combinations collectively account for over 55% of large trades (trades of 100 contracts or more) in the Eurodollar options market and almost 75% of the trading volume due to large trades. In terms of total volume, the four most heavily traded combinations are (in order): straddles, ratio spreads, vertical spreads, and strangles. These four represent about two thirds of all combination trades. On the other hand, condors, horizontal spreads, guts, iron flys , box spreads, guts, covered calls or puts, and synthetics are very rarely traded while trading is light in collars, diagonal spreads, butterflies, straddle spreads, seagulls, doubles, and delta-neutral combinations. Significant differences in size, cost, and time-to-expirations are found among the various combination types.

Our results confirm that traders use spreads and combinations to construct portfolios which are highly sensitive to some risk factors and much less sensitive to other risk factors. The most popular combination designs are those yielding portfolios which are quite sensitive to volatility and less sensitive to directional changes in the underlying asset value - though they are often not completely delta neutral. Among these, combinations which are short volatility significantly out-number those which were long. Among the minority of combinations which are highly sensitive to the underlying asset price, those with positive deltas significantly outnumber those with negative deltas indicating that traders are using this market to bet on or hedge against an increase in the LIBOR rate.

We find evidence that effective bid/ask spreads are larger on orders exceeding 500 contracts or more than on orders of between 100 and 500 contracts and evidence that effective bid/ask spreads are larger on combinations which short volatility.

option spreads, option combinations, effective spreads, straddles, vertical spreads, effective spreads

Abstract:
Based on the law of large numbers, several options researchers have proposed using (different) weighted averages of the implied standard deviations, ISD, calculated from numerous options with the same expiry to obtain a single best ISD measure. However, most commercial providers use an average (and often an equally weighted average) of just a few at-the-money ISDs. We find that the practitioners' restricted averages forecast slightly better than the broader weighted averages from the academic literature but that neither group forecasts actual volatility very well.

We suggest an adjustment to the extant models which improves their forecasting ability considerably. We also suggest a new weighting scheme which yields better estimates on an out-of-sample basis than any of the existing models (adjusted or unadjusted). However, we also find that because there is little independent noise in individual options ISDs (at least in the S&P 500 options market), there is little gain to averaging. Consequently, individual option ISDs and averages of just a few ISDs forecast almost as well as weighted averages of many ISDs and the weighting scheme choice is relatively unimportant.

Abstract:
Using data from the Eurodollar options on futures market, this paper examines six volatility trades: straddles, strangles, guts, butterflies, iron butterflies, and condors. We argue that straddles and strangles should have lower transaction costs than the other four strategies, and that (when constructed to be delta neutral) straddles, strangles, and guts should have higher vegas and gammas with a straddle's gamma and vega being the highest of the three. Consequently, we predict that in most situations volatility traders should prefer straddles and strangles to the other four strategies and that they should tend to favor straddles over strangles. Consistent with this we find that straddles account for 73.1% of all large volatility trades, strangles 20.8%, and butterflies 4.7% while the other three are rarely traded.

In general we find that most straddles and strangles are designed so that their delta is low and their gamma and vega are high (in absolute terms) but that they are not always constructed so that delta is minimized and vega and gamma maximized. Specifically, we find that most straddle traders choose the closest-to-the-money strike and that most strangle strikes are centered around the underlying asset price. While delta is low and gamma and vega high at these strikes, they may not be the delta minimizing and gamma/vega maximizing strikes. On the other hand, we find that when futures are added to a straddle position it is almost always in the ratio that reduces the delta of the position to zero and that the volatility trader's choice of whether to use a straddle or strangle depends on which can be designed with the lower delta.

There is little evidence that the shape of the smile impacts the strike price choices of straddle and strangle traders or that it impacts the straddle/strangle choice. We do find that the straddle/strangle choice depends on the time to expiration and whether the trader longs or shorts volatility.

volatility, straddle, strangle, option trading

Abstract:
Signaling strategies which sellers of high quality securities, goods, or services employ to differentiate their securities or products from those of lower quality include: (1) developing a reputation for high quality, (2) certification by a respected third party (e.g., underwriters, bond rating agencies, and auditors, for securities, and testing laboratories for goods), (3) warranties (for goods), and (4) information disclosure (such as financial statements for securities and specifications and test results for goods). These signaling strategies are compared using data from the online comic book auction market. This market has a number of important advantages: (1) the information asymmetry is substantial, (2) good measures of reputation are available, and (3) there are many sellers employing different combinations of the four strategies. Of the four strategies, we find certification by a respected third party sends the strongest signal. On average certified comics sell for over 50% more than otherwise equivalent uncertified comics. Moreover it appears that part of this price premium is due to risk reduction, i.e., not all is due to differences in the expected quality of certified and uncertified books. We find that both how positive or negative the seller's reputation is and how well established that reputation is significantly impact the price though reputation is less important than certification. We find no evidence that warranties in the form of money-back guarantees impact the price. Apparently buyers reason that sellers of truly high quality books should seek certification so discount the presence of both warranties and positive reputations. Virtually all sellers disclose information in the form of scans and failure to do so lowers the price a modest amount. Since for a single sale, the seller's reputation is exogenous while the other three strategies are endogenous, we explore how a seller's strategy choices depend on her reputation.

reputation, certification, warranties, information asymmetry, signaling, auctions

Abstract:
We examine whether any second order derivatives other than gamma and any third order derivatives are important in explaining changes in the prices of S&P 500 futures options over one week holding periods. We find that while gamma is normally most important, several other higher order derivatives have considerable incremental explanatory power. Particularly important in accounting for option price changes are the derivatives of delta with respect to volatility and time-to-expiration, and the derivatives of gamma with respect to the asset price, and volatility. The first three are more important for away-from-the-money options, while the fourth is most important for at-the-money options. For shorter-term options, consideration of higher order derivatives reduces the mean absolute unexplained price change by 60% for at-the-money options and by at least 75% for away-from-the-money options. We find that in spite of its theoretical problems and inability to explain the cross-sectional option price pattern (the smile), the Black-Scholes model's Greeks accurately describe the time series option price changes once higher order Greeks are incorporated. We further find that making delta-gamma-vega neutral portfolios of S&P 500 options neutral in terms of these four higher order Greeks leads to a substantial reduction in the risk of an unhedged price change.

Option Greeks, hedging, Delta, Gamma, Vega

Abstract:
We examine whether any second order derivatives besides gamma and any third order derivatives are important in explaining changes in the prices of S&P 500 futures options over one week holding periods. We find that while gamma is normally the most important of the higher order derivatives, several others have considerable incremental explanatory power. Particularly important are the derivatives of delta with respect to volatility and the time to expiration, the third derivative of the option price with respect to the underlying asset price, and the derivative of gamma with respect to volatility. The first three are more important for away-from-the-money options, while the fourth is more important for at-the-money options. In our sample of shorter-term options, consideration of second and third order derivatives reduces the mean absolute unexplained price change 60% in the at-the-money sample and at least 75% in the in- and out-of-the-money samples. We further find that in spite of its theoretical problems and its inability to explain the cross-sectional pattern in option prices (the smile), the Black-Scholes model is quite accurate in its description of the time series pattern once these higher order derivatives are considered.

Options, Greeks, Delta, Gamma

Abstract:
It is well known that for options with the same expiration date, levels of implied volatility differ systematically by strike price in a smile or smirk pattern. We show that (in the equity index options market) information content also differs systematically by strike price displaying a "frown" pattern. Implied volatilities calculated from near-the-money, and particularly from moderately high strike price, options contain considerable information regarding future volatility and efficiently subsume the information in the historical record. Implied volatilities calculated from far-from-the-money options (particularly options with low strike prices) are much less informative and do not incorporate the information in the asset's price history. While the prices of near-the-money and moderately high strike price options appear largely determined by the market's expectations of future volatility, the prices of far-from-the-money and low-strike-price options appear to be largely determined by factors other than the market's volatility expectation. These other determinants of implied volatility are apparently long-lived or persistent since the frown cannot be attributed to transitory factors.

Theories of the smile need to not only explain why volatility levels differ by strike price but should also explain why implied volatilities at some strike prices are largely determined by the market's volatility expectations while others appear relatively independent of these expectations.

Abstract:
Based on data on options on Eurodollar futures, the trading and design of vertical spreads (aka bull and bear spreads) are examined. Reducing the cost and/or increasing the likelihood of profit on long positions appears more important than risk reduction on short positions in most traders' decisions to employ vertical spreads. Somewhat surprisingly, we find that most vertical spreads are not designed so as to maximize their sensitivity to changes in the underlying asset price and minimize their sensitivity to other risks. Likewise, they are not designed so that the signs for theta, vega, and/or gamma are opposite those naked calls or puts with the same delta. Instead, most vertical spread traders display a strong preference for out-of-the-money strikes resulting in low net prices, low transaction costs, and low likelihood of early exercise. We find that when futures are combined with a vertical spread position, it is almost always in a ratio which reduces the position's net delta to zero turning the spread into a volatility trade. Seagulls represent a fairly actively traded but rarely discussed variant of the standard vertical spread design with lower prices and higher deltas than bull and bear spreads.

Option spreads, vertical spreads, bull spreads, trade design

Abstract:
Option pricing models and longer-term value-at-risk models typically require volatility forecasts over horizons considerably longer than the data frequency. These are generally generated from short-horizon forecasts by successive forward substitution. We document deficiencies with the resulting long-horizon volatility predictions generated by GARCH type models, such as GARCH(1,1), EGARCH, and GJR. One, since volatility forecasts for forward periods are functions of forecast volatility for the next period, this recursive procedure keeps the relative weights of recent and older observations the same whether forecasting volatility in the near or distant future. In contrast, we find that older observations are relatively more important in forecasting at long horizons, e.g., more important in forecasting volatility next month than in forecasting volatility tomorrow. Two, forecasts of the return standard deviation - the most appropriate volatility measure for option valuation and value-at-risk models - are strongly positively biased. Three, GARCH(1,1) and GJR forecasts are especially biased following high volatility days. We find that the ARLS model of Ederington and Guan corrects these three deficiencies and generally forecasts better out-of-sample than GARCH, EGARCH, AGARCH and the GJR models across a wide variety of markets and forecast horizons.

GARCH, EGARCH, volatility, options

Abstract:
Ratio spreads in which a trader buys calls (or puts) at one strike and sells an unequal number of calls at a different strike are among the most actively traded option combinations yet are only briefly mentioned in most derivatives texts and have received no attention in the research literature. In texts and other discussions of option spreads and combinations, there is no agreement on when ratio spreads should be used, little guidance on how they should be designed, and no data on how they are actually used and designed. Seeking to fill this gap, we discuss the properties of ratio spreads and document their design and use in the Eurodollar options market.

Exploring what the chosen designs reveal about the motives of the traders, we find that most ratio spreads are designed to have low cost and to be roughly, but not completely, delta neutral. Front-spread designs in which profits are bounded and losses unbounded considerably exceed back-spread designs in which losses are bounded and profits unbounded. Results are mixed on whether ratio spreads are primarily used to exploit expected changes in volatility since while designed so that vega has the hypothesized sign, vega values are generally smaller than could have been achieved with a slightly different design.

Ratio Spreads, option spreads, option trading

Abstract:
Ratio spreads in which a trader buys calls (or puts) at one strike and sells an unequal number of calls at a different strike are among the most actively traded option combinations yet are only briefly mentioned in most derivatives texts and have received no attention in the research literature. While books aimed at traders discuss ratio spreads, there is no agreement on when these spreads should be used, little guidance on how they should be designed, and no data on how they are actually used and designed. We document the design and use of in the Eurodollar options market and explore what the chosen designs reveal about the motives of the traders.

We find that most ratio spreads are designed to have low cost and to be approximately delta neutral. Frontspread designs in which profits are bounded and losses unbounded considerably exceed backspread designs in which losses are bounded and profits unbounded. Results are mixed on whether ratio spreads are used to exploit expected changes in volatility or differences between implied and expected volatility. Though gamma and vega have the hypothesized signs, they are generally smaller than could have been achieved with a slightly different design.

Ratio Spreads, option spreads, option trading

Abstract:
Based on data from eBay auctions of classic comic books, we explore the determinants of the number of bidders in online auctions and the impact of additional bidders on the auction price. We test for winner's curse and examine how secret reserve prices impact both bidding activity and the auction price. In our data set, the number of bidders varies widely from zero in 20.4% of the observed auctions to ten or more in 16.5% but 70% of this variation in bidder numbers is predictable based on the comic being auctioned and attributes of the seller and auction such as third party certification of the comic's condition, the seller's reputation, and the length of the auction.

Each additional bidder tends to increase the realized auction price about 2.4%. A minor part of this increase is apparently either because the auctions have a private value component or because some bidders fail to fully adjust for winners curse. However, most appears due to the inability of bidders to accurately determine the number of bidders they are bidding against. The presence of a secret reserve price sharply reduces the number of bidders in an auction but has little impact on what individual bidders are willing to bid.

Winner's curse, auctions, reserve prices, eBay

Abstract:
Signaling strategies, which sellers of high quality securities, goods, or services employ to differentiate their products from those of lower quality, include: (1) developing a reputation for high quality, (2) certification by a respected third party, (3) warranties, and (4) information disclosure. These signaling strategies are compared using data from the online auction market for classic comic books. This market has several advantages: (1) the information asymmetry is substantial, (2) good measures of reputation are available, and (3) all four signals are common. Third party certification of a comic's quality sends the strongest signal. On average, certified copies sell for roughly fifty percent more than uncertified comics of the same claimed quality, and the percentage differential is higher at the higher grades. It appears that part of the price differential between certified and uncertified comics is due to risk reduction. Books from sellers with positive reputations sell for more than books from sellers with negative reputations, and books from sellers with established reputations sell for more than those from seller's with fledgling reputations. However, reputation has less impact on the price than certification. There is no evidence that books with warranties sell for more than those without. Apparently, buyers either view these warranties as too costly to enforce or they reason that sellers of truly high quality comics should seek certification. We also explore which signals are substitutes or complements, and how choice among the other three strategies depends on the reputation of the seller and other information available to the buyer.

Reputation, certification, warranties, information asymmetry

Abstract:
We document several problems with GARCH type model predictions over the multi-day horizons common to option valuations and value-at-risk models. One, GARCH model forecasts of the return standard deviation - the most common volatility measure and the most appropriate for option valuation and value-at-risk models - are positively biased. Two, the bias is especially severe following high volatility days. Three, in forecasting volatility over longer horizons, the GARCH model puts too little weight on older observations relative to the more recent observations. That is older observations are more important in forecasting volatility next month than in forecasting volatility tomorrow while the GARCH procedure treats them equally at both horizons.

We present a simple unbiased regression estimator of the standard deviation of returns which avoids these problems. We find it forecasts better out-of-sample than GARCH, EGARCH, and historical volatility across a wide variety of markets and forecast horizons.

Volatility, GARCH, forecasting

Abstract:
In many markets, changes in the spot price are partially predictable. We show that when this is the case: 1) although unbiased, traditional regression estimates of the minimum variance hedge ratio are inefficient, 2) estimates of the riskiness of both hedged and unhedged positions are biased upward, and 3) estimates of the percentage risk reduction achievable through hedging are biased downward. For natural gas cross hedges, we find that both the inefficiency and bias are substantial. We further find that incorporating the expected change in the spot price, as measured by the futures-spot spread, into the regression results in a substantial increase in efficiency and reduction in the bias. Accounting for seasonal price patterns results in a smaller efficiency improvement and bias reduction.

hedging, natural gas markets

Abstract:
For short, long, and medium term interest rates, we explore the extent to which future interest rate volatility is predictable based on: 1) recent past volatility, 2) knowledge of when the FOMC will meet, when Treasury auctions will take place, and when important macroeconomic statistics will be released, 3) knowledge of day-of-the-week patterns, and 4) interest rate levels. We find that three of these information sets are consistently important and that conditional volatility varies considerably over time. While macroeconomic announcements, Treasury auctions, and monetary policy actions all impact interest rates, macroeconomic announcements are most important. Announcement patterns are largely responsible for day-of-the-week effects but not for the observed GARCH effects. The impact of monetary policy actions is quite different before and after February 1994 when the FOMC began releasing statements at the end of its meetings. Both volatility persistence and asymmetry are quite different for long and short rates.

interest rates, volatility, GARCH, day-of-the-week effects, macroeconomic announcements

Abstract:
By Jensen’s inequality, a model’s forecasts of the variance and standard deviation of returns cannot both be unbiased. This paper explores the bias in GARCH type model forecasts of the standard deviation of returns, which we argue is the more appropriate volatility measure for most financial applications. For a wide variety of markets, the GARCH, EGARCH, and GJR (or TGARCH) models tend to persistently over-estimate the standard deviation of returns while the ARLS model of Ederington and Guan (2005, JFM) does not. Furthermore, the GARCH and GJR forecasts are especially biased following high volatility days which cause a large jump in forecast volatility which is rarely fully realized.

GARCH, EGARCH, volatility, value-at-risk

Abstract:
According to Jensen's inequality, a forecasting model cannot yield unbiased forecasts of both the variance and standard deviation of returns. We explore the bias in GARCH type model forecasts of the standard deviation of returns, which we argue is a more appropriate volatility measure than the variance for most financial applications since near-the-money option prices and VaR are linear functions of the standard deviation but not the variance. For a wide variety of markets, we find that the GARCH(1,1), EGARCH, and GJR (or TGARCH) models tend to persistently over-estimate the standard deviation of returns while forecasts of the ARLS model of Ederington and Guan (2005a) are unbiased. We further find that the GARCH(1,1) and GJR forecasts, but not the EGARCH and ARLS forecasts, are especially biased following high volatility days because in the former two models, a single high volatility day leads to a large jump in forecast volatility which is rarely fully realized.

GARCH, EGARCH, volatility, value-at-risk

Abstract:
This paper explores differences in the impact of equally large positive and negative surprise return shocks in the aggregate U.S. stock market on: 1) the volatility predictions of asymmetric time series models, 2) implied volatility, and 3) realized volatility. Both asymmetric time series models and implied volatility predict an increase in volatility following large negative surprise returns and ex post realized volatility normally rises as predicted. However, while asymmetric time series models, such as the EGARCH and GJR models, predict an increase in volatility following a large positive return shocks (albeit a much smaller increase than following a negative shock of the same magnitude), both implied and realized volatility generally fall sharply. While asymmetric time-series models predict a decline in volatility following near-zero returns, both implied and realized volatility are normally little changed from levels observed prior to the stable market. Reasons for the differences are explored.

Implied volatility, asymmetric volatility, GARCH, EGARCH, volatility

Abstract:
Option pricing models and longer-term VaR models typically require volatility forecasts over horizons considerably longer than the data frequency. These are generally generated from short-horizon forecasts by successive forward substitution. We document deficiencies with the resulting long-horizon volatility predictions generated by GARCH type models, such as GARCH(1,1), EGARCH, and GJR. Since volatility forecasts for forward periods are functions of forecast volatility for the next period, this recursive procedure keeps the relative weights of recent and older observations the same whether forecasting volatility in the near or distant future. In contrast, we find that older observations are relatively more important in forecasting at longer horizons, e.g., more important in forecasting volatility next month than in forecasting volatility tomorrow. We find that the ARLS model of Ederington and Guan (2005) and a modified version of the standard EGARCH model in which the parameter values vary with the forecast horizon forecast better out-of-sample than the GARCH(1,1), EGARCH, and GJR models across a wide variety of markets and forecast horizons.

GARCH, EGARCH, options, volatility, volatility forecasting, value-at-risk

Abstract:
Utilizing monthly aggregate flow data for U.S. equity mutual funds over 1986-2008, we document several new findings on investor behavior. First, we find a strong negative relationship between changes in expected market volatility as measured by the VIX index and net equity fund flows. Second, we document several asymmetries in mutual fund investors’ buy-sell decisions. We find that the negative volatility - net flow relationship is entirely due to the effect of volatility on outflows. When volatility increases, equity fund inflows actually increase - just not as much as outflows. In contrast, returns on equity funds only impact inflows, i.e., when returns increase aggregate inflows rise but outflows do not slow. These findings imply that mutual fund investor purchase decisions are primarily driven by returns while redemption decisions are primarily driven by risk perceptions. Suggesting further compartmentalization in investor decision making, we also find an asymmetry in the investor reaction to risk in that both inflows and outflows increase significantly (with a much stronger effect on outflows) as the VIX increases, but there is little effect on either outflows or inflows when the VIX declines. Lastly, we find evidence of month-of-year flow patterns consistent with: 1) tax minimization, and 2) a tendency to reevaluate and rebalance portfolios at the turn-of-the-year.

mutual funds, VIX, implied volatility

Abstract:
This paper explores the risk structure of interest rates. More specifically, we ask whether yields on industrial and commercial bonds indicate that market participants base their evaluations of a bond issue`s default risk on agency ratings or on publically available financial statistics. Using a non-linear least squares procedure, we relate the yield to maturity to Moody`s rating, Standard & Poor`s rating, and accounting measures of credit worthiness such as coverage and leverage. We find that market yields are significantly correlated with both the ratings and with a set of readily available financial accounting statistics. These results indicate (1) that market participants base their evaluations of an issue`s credit worthiness on more than the agencies` ratings and (2) that the ratings bring some information to the market above and beyond that contained in the set of accounting variables. In addition, our results suggest that the market views Moody`s and S&P`s ratings as equally reliable measures of risk. Although the accounting measures also affect yields on new or recently reviewed issues, our analysis suggests that the market may pay more attention to the accounting measures and less to the ratings if the rating has not been reviewed recently.

Abstract:
This paper represents an extension and integration of recent empirical and theoretical research on default risk and taxability. The purpose of the paper is to develop and test a model of interest rate spreads which incorporates both the effect of taxes and differences in default probabilities in a theoretically correct manner. There is an important fundamental difference between our approach to explaining yield spreads and the approach most commonly taken in literature. Unlike nearly all of the previous work, we do not begin with a yield spread model, i.e.,one which begins by examining differences in yields, but rather begin with an expected return or pricing model, which can then be expressed in the yield spread format. This is a fundamental difference in approaches which we feel leads to a superior theoretical formulation which can then be tested empirically without many of the problems inherent in the alter-native approach. The theoretical model is a simple extension of earlierwork on default by Bierman and Hass (1975) and Yawitz (1977), altered appropriately to take explicit account of tax effects. While there is a considerable literature that analyzes the effect of taxability on rate spreads, we are unaware of any previous study that considers tax consequences in the event of default, a rather surprising omission.

Abstract:
The adjusted mean absolute deviation is proposed as a simple-to-calculate alternative to the historical standard deviation as a measure of historical volatility and an input to option pricing models. We show that this measure forecasts future volatility consistently better than the historical standard deviation across a wide variety of markets. Moreover, it forecasts as well as or better than the GARCH(1,1) model.

Abstract:
There is information asymmetry between the seller andbuyer about the quality of comic books offered on eBay online auctions. Thisstudy examines the different signaling strategies that sellers use to remedythe information asymmetry. Four types of common signals are considered: thirdparty certification, reputation, warranties, and information disclosure. Datawere compiled about online comic book sales in 3,664 eBay auctions endingbetween January and June 2001 for 30 collectible comic books. The goal was to test how signaling actions taken by a seller to resolveinformation asymmetry affect price. Certified comics are found to sell forabout 58 to 59 percent more than uncertified comics. Reputation mattersconsiderably less than certification. How much is known about the seller, andwhether it is good or bad, affects the price. Warranties seem not to affectprice. Failing to provide scans of comics reduces the price by about 11 to 13percent. More specifically, it was found that price increases as proportion ofnegative feedback declines and uncertainty about the proportion declines. Sincereputation is a long-run strategy, three other signals are examined.Experienced sellers with well-established reputations are cognizant of thebenefits of providing certification. Bidders penalize failure to seekcertification when doing so would be in the interest of an honest seller;certification substantially affects price even when there is no reason to doubtthe quality of uncertified comics. While cost of certification is substantial,many sellers pay for it. Reputation matters much more when a comic book has notbeen certified. (TNM)

Signaling theory, Firm image, Credibility, Reliability, Electronic commerce, Information asymmetry, Valuation, Information utilization, Certification, Prices

Abstract:
We test and reject the hypothesis that managers call in-the-money convertibles when they view a decline in the value of the firm as likely. Inconsistent with this view, we find that insiders generally buy equity before conversion-forcing calls. Also, analysts tend to raise their earnings forecasts following a call. Thus, our evidence supports the alternative hypothesis that the price decline immediately following conversion-forcing calls is a purely transitory decline caused by the anticipated increase in the supply of equity. Indeed, our evidence confirms that the initial price decline is reversed in the weeks following the announcement.

Abstract:
We examine how prices in interest rate and foreign exchange futures markets adjust to the new information contained in scheduled macroeconomic news releases in the very short run. Using 10-second returns and tick-by-tick data, we find that prices adjust in a series of numerous small, but rapid, price changes that begin within 10 seconds of the news release and are basically completed within 40 seconds of the release. There is some evidence that prices overreact in the first 40 seconds but that this is corrected in the second or third minute after the release. While volatility tends to be higher than normal just before the news release, there is no evidence of information leakage. In our analysis, we correct for the biases created by bid-ask spreads and tick-by-tick data.

Abstract:
This paper examines (1) whether implied volatility is an unbiased informationally efficient predictor of actual future volatility and (2) its predictive power. If markets are efficient and the option pricing model is correct, then the implied volatility calculated from option prices should be an unbiased and informationally efficient estimator of future volatility, that is, it should correctly impound all available information including the asset's price history. However, numerous studies have found that implied volatility is not informationally efficient and that historical volatilities have incremental predictive power -- often out-predicting implied volatilities. For the S&P 500 options on futures we find the following. One, at least part of the apparent inefficiency of implied volatility from past studies stems from measurement error which biases estimates of the importance of implied volatility downward and of the importance of historical volatility upward. Once we correct for this error, there is no significant inefficiency. Two, implied volatility has strong predictive power -- considerably stronger than found by previous equity index studies. Three, stock market volatility prediction results are quite sensitive to (1) the forecasting horizon and (2) whether the data period covers the October 1987 stock market crash.

Abstract:
We model and examine the impact of information releases on market uncertainty as measured by the implied standard deviation (ISD) from option markets. Distinguishing between scheduled and unscheduled announcements, we hypothesize that since the timing, although not the content, of scheduled announcements is known a priori, the pre-release ISD will impound the anticipated impact of important releases on price volatility and that the ISD will normally decline post-release as this uncertainty is resolved. Conversely, we hypothesize that the unexpected high volatility caused by major unscheduled releases will cause market participants to adjust upward their estimates of likely volatility over the remaining life of the option resulting in an increase in the ISD. Our evidence supports both hypotheses. The ISD's which we consider are from the T-Bond, Eurodollar, and Deutschemark options markets. The scheduled news releases which we examine are macroeconomic news releases such as the employment report and the PPI. We also find that the observed tendency for the ISD to fall on Fridays and rise on Mondays is due to the weekday pattern of scheduled news releases.