Information and Normal Backwardation as Determinants of Trading Performance: Evidence from the North-Sea Oil Forward Market

In this paper we take advantage of a unique micro-database on forward trading in the international petroleum market, with information on the buyer and seller in each transaction. We utilize transaction-specific data to test directly predictions from the theory of normal backwardation vs. information-based predictions of who profits in these markets. We find that no trader groups make significant profits on interday measures. Within the daYt however t groups likely to have superior information do make significant profits. The results are not supportive of normal backwardation, but are consistent with the time pattern of information dissemination in this market-deals made during the day are widely reported only at day's end. INFORMATION AND NORMAL BACKWARDATION AS DEr'IRMlNANTS OF TRADING PERFORMANCE: Evidence from the North-Sea Oil Forward MarKet Resolution of the longstanding debate over who gains and who loses in financial markets requires data at a disaggregated level. In this paper we take advantage of a unique database with information on the buyer "and seller in each transaction in a forward market to measure trading performance. Our tests are designed to shed light on the potential sources of differential returns to participants in forward and futures markets. We examine trading performance to assess the empirical relevance of two potential sources of differential returns: insurance and information. Behind the insurance view is the theory of "normal backwardation," dating back to 'eynes (1930) and Hicks (1946). The theory treats these markets as arenas wherein risk-averse "hedgers," firms that produce or utilise the physical commodity, purchase insurance from "speculators," who do not. Until the 1980s, the bulk of the research literature on forward and futures markets was based on the normalbackwardation assumption (see, e.g., Kawai 1983).1 An alternative (but not mutually exclusive) explanation of trader performance is based on information. That differential information can affect prices and profits in financial markets has been demonstrated formally in recent research (e.g., Grossman and Stiglitz 1980, Kyle 1985). A literature has developed based on asymmetrically-informed market participants, with the less-informed referred to as "noise traders."Z 1The literature outside this tradition (e.g., Telser and Higginbotham 1977, Telser 1981, Williams 1987), in contrast, is relatively fragmented and sparse. 2Shleifer and Summers (1990) provide an overview of this literature. DETERMINANTS OF TRADING PE~ORMANCE, page 2 The classic theory of normal backwardation yields the testable hypothesis that, on average, speculators should gain and hedgers should lose in these markets. More recently, a literature based on portfolio theory (e.g., the Capital Asset Pricing Model) has pointed out that the traditional normal-backwardation approach implicitly assumes that claims on hedgers' profits (e.g.; equity) cannot be marketed costlessly. Models based on the opposite assumption--that forward and futures markets are perfectly integrated with markets for other assets--yield the conclusion that risk premia are not related to hedging, because speculators can costlessly enter futures and forward markets, and diversify the nonsystematic risk assumed in these markets by combining them in portfolios with other assets. 3 Hirshleifer (1988) integrates the traditional theories of risk premia based on hedging with the portfolio approach, demonstrating that nonmarketability of claims on profits, together with fixed costs of entering asset markets, yield predictions similar to those of the simple normal-backwardation model. Bessembinder (1992) uses monthly aggregate data from a variety of futures markets to test these predictions against those of the portfolio approach. He finds support for Hirshleifer's modern version of the normal-backwardation model, in that net hedging, interacted with residual risk, is associated with returns in futures markets. In contrast, the asymmetric-information view of trader performance predicts that traders with better information will gain at the expense of uninformed or "noise" traders. Thus, if traders who operate in the cash (physicals) market have superior information about future supply and demand conditions, the prediction of the normal-backwardation model can be reversed. 30usak (1973) was the first to attempt to relate returns on futures contracts and the systematic risk emphasized in portfolio theory, finding no significant relationship.