Information spillovers between derivative markets with differences in transaction costs and liquidity

Abstract

In line with the transactions cost theory, this article shows that the futures market with its higher liquidity and lower transactions costs, leads the options market in the price discovery process. Liquidity and transaction costs are also shown to play a key role in market sensitivity to information, since the futures market s response to shocks is quicker, which means that it receives higher volatility spillovers than does the options market.

Acknowledgements

This article has received financial support from the ERDF and the Spanish Ministry of Education and Science (SEJ2006-14809).

Notes

¹The only exception is to be found in the article by Chng and Gannon (2003) who analysed interactions between implied volatility and underlying futures volatility to obtain out-of-sample volatility forecasts. In this article, both the research proposal and the methodology are clearly different.

²Chan (1992) argues that the index futures market may process market-wide information more efficiently than the cash market, since trading in all index stocks may prove more complicated and more costly.

³The number of futures contracts traded during the period of analysis was 54.84% higher than the options and 1225% higher in monetary terms.

⁴The relative mean spread in the futures was approximately 0.05%, vs. 100 times higher (5.53%) in the options even after excluding outliers. There are no major differences in commissions however. In fact, the market fees were very similar (4.5 euros per equivalent futures and options contract) although the brokerage fees per monetary unit are usually a little lower for futures because the contract size is larger.

⁵In this respect, authors such as Black (1975) or Mayhew et al. (1995) argue that informed investors might be persuaded to trade in the options market because of their lower transaction costs and higher leverage in relation to the underlying assets. Nevertheless, there are more factors to take into account (the concentration of informed trading, the type of underlying asset, …) to fully explain the phenomenon, as recent empirical literature has shown (among others, Easley et al., 1998; Chan et al., 2002; Chen et al., 2005; Schlag and Stoll, 2005 or Pan and Poteshman, 2006) yield mixed results.

⁶For this purpose it is usual to compute option trades that are initiated by buyers or by sellers, the information variable that takes into account the put and call contracts purchased by non-market makers to open new positions or the PIN _i variable, which is a measure of the probability that each trade in stock is informed-based. Obviously, the degree of moneyness is also a key factor in this issue.

⁷The main problems that arise with previous day implied volatility have to do with daily stationarity and intraday stationarity. However, the joint price and volatility estimation suggested in Varson and Selby (1997) requires a large set of transactions for every 15-minute interval, which, in the Spanish market, is very often not possible. In this article, therefore, we have chosen a compromise solution to try to reduce bias deriving from stationarity and data availability.

⁸ The Black formula is similar to the classical Black-Scholes formula for valuing stock options except that the spot price of the underlying is replaced by the forward price. Chng and Gannon (2003) use this model to back out the implied volatility from futures options prices.

⁹To determine the appropriate number of lags (8), following Turkington and Walsh (2000), we have used Schwartz' Bayesian Criterion (SBC) instead of Akaike's information criterion (AIC) since the properties of the SBC are better for use with large samples.

¹⁰An estimation of a GARCH bivariate model yielded similar results to those reported in this article.

¹¹For the market considered, these asymmetries are significant at the univariate level (Blasco et al., 2002). It should be noted that, while financial series may exhibit high persistence or high asymmetry in the variance of their univariate representations, the persistence and the asymmetry may be common across the series, so that linear combinations of the variables show lesser persistence or asymmetry.

¹²The results remain unaltered when the futures (options) market residuals introduced into the options (futures) market variance are previously orthogonalized with the own market residuals, in line with the proposal made by Aragó et al. (2003).