The aim of this thesis is to analyse futures contract pricing and the use of futures contracts for hedging. The thesis contributes to the literature in the following ways.
The analysis focuses on the 90 Day Bank Accepted Bill Futures Contract and the Australian All Ordinaries Share Price Index Futures Contract. These contracts have received limited attention in the literature, and so this thesis extends the existing research in this area.
Two alternative methods of analysing the pricing of futures contracts are applied in the thesis, cross-contract regressions and cointegration. Neither of the approaches is evident in the Australian literature for bank accepted bill and share price index futures contracts. They provide a measure of the statistical strength of the cost of carry pricing model rather than attempting to estimate pricing errors.
The cost of carry model exhibits descriptive power for the 90 Day Bank Accepted Bill Futures Contract when using cross-contract regression analysis, though cointegration analysis suggests a change in the market after 1985 with evidence of arbitrage profits across trading days in the period prior to 1986. The introduction of foreign banks in 1985, the dramatic growth in trading volume in both the bill market and the bill futures market after 1985 and the existence of automated trading in the bill market after 1984 are suggested as possible reasons for this change.
The cost of carry model provides little explanatory power in cross-contract regressions for the Australian All Ordinaries Share Price Index Futures Contract. Cointegration analysis, where one cointegrating vector is assumed, suggests the possibility of more than one cointegrating relationship though a specific test for the number of cointegrating vectors provides conflicting evidence. Nevertheless the cost of carry model does not appear to explain share price index futures pricing particularly well.
One problem with the price time series is the use of non synchronous observations. It is difficult to assess the effect of this without access to time matched prices, though it is not expected that this problem will result in biased regressions. It may introduce noise into the analysis and perhaps increase standard errors, thus affecting statistical tests by reducing the chance of rejecting the null.
A number of hedge ratio estimation techniques are applied to the one trading day hedge problem with two important results. Firstly, hedging is not always a variance minimising or expected utility maximising alternative. In a number of cases, especially where the expected utility maximisation criterion is applied, remaining unhedged is preferred to hedging. Secondly, for hedge periods of one trading day, it is often the simple ordinary least squares estimation techniques which exhibit the most consistent performance, and this is most evident where the minimum variance criterion is applied.
The comment on non synchronous observations also applies to the hedging analysis, though in practice the possibility of synchronising positions in the spot and futures market to create a hedged portfolio is somewhat optimistic.