Cash‐futures arbitrage and forward‐futures spreads in the treasury bill market

Abstract

ersistent discrepancies between implied forward rates on the yield curve P and corresponding futures rates have been widely observed. For instance, in one of our samples, eight week-ahead forward-future spreads averaged nearly 70 discount basis points before 1982 and have since averaged about 30 basis points (see Figure 1). Spreads of these magnitudes are usually rationalized as manifestations of market inefficiencies or transactions cost. However, the observation that the spreads are at least predominantly positive' suggests something more is at work.The observations above appear curious given the presence of leveraged investors (such as securities dealers). securities dealers who seem unlikely to behave inefficiently, are subject to low transactions cost. Assuming they act efficiently, the explanation for substantial forward-futures spreads must be found in the terms at which securities dealers can borrow and lend in order to set up arbitrage positions.As we shall demonstrate, it is these terms which appear to "drive" (i.e., explain the existence and magnitude of) forward-futures spreads. The market requires there be a positive spread between the dealer rep0 rate over the riskfree Treasury bill rate for the same term (the financing cost spread). Further, this spread will systematically determine the size of the forward-futures spread. We find that the predictive power of the financing cost spread is substantial, despite deficiencies in our database which consists of daily averages for repos and Treasury bills and closing prices for futures. This database is, in part, inferior to those using intra-day prices (notably that of Elton, Gruber, and Rentzler (1984)* but has an advantage over this and other studies in that it 'In the case of the spreads calculated for Figure 1, and for those appearing later in this paper, a total of 100 calculated forward minus futures spreads, there are only eight negative ones, whose maximum absolute value is eight discount basis points (see Table 1).

A Comparison of Forward and Futures Prices of an Interest Rate‐Sensitive Financial Asset

“…22See Capozza and Cornell (1979), Cornell and Reinganum (1981), Kawaller and Koch (1984), Gendreau (1985), and Allen and Thurston (1988).…”

Futures and forward price differential and the effect of marking‐to‐market: Australian evidence

“…The initial focus of researchers was on testing the assumption of price equivalence between futures and forward prices in various asset categories, particularly in U.S. markets. Asset categories tested were U.S. Treasury bills (Capozza and Cornell (1979); Elton et al, (1984); Gendreau (1985); Kolb and Gay (1985); Allen and Thurston (1988)), Eurodollars (Meulbroek (1992); Grinblatt and Jegadeesh, (1996)), foreign exchange (Dezhbakhsh, 1994) and silver and copper (French, 1981). The results of these studies clearly show that futures and forward prices of interest rate sensitive instruments such as Treasury bills and Eurodollars are significantly different from each other.…”

“…However, Gendreau (1985) argues that the conflicting results from previous studies are the outcome of different specification of borrowing costs. Allen and Thurston (1988) also attribute futures and forward yield differences to transaction costs. Kamara (1988) attributes the forwards futures price differential in U.S. Treasury bill markets largely to lower default and liquidity premia in the futures market relative to the synthetic forward markets.…”

The theoretical source of autocorrelation in forward and futures price relationships

“…However, there is a consensus that the reported prices of FX forward and futures contracts are not influenced statistically by the presence of asymmetries attributable to differential cross-market states of efficiency, taxation, segmentation, default risk, transactions costs, or liquidity [see Polakoff and Grier (1991); Cornell and Reinganum (1981); and Denis (1976)l. On the other hand, the preceding imperfections are documented extensively by empirical studies of Treasury-bill forwardfutures price relationships [Allen and Thurston (1988); Kamara (1988); Capozza and Cornell (1979);Branch (1978); Lang and Rasche (1978)l.…”

“…In assuming that markets are perfect, CIR use Treasury-bills to represent the financing (reinvestment) costs (returns) arising from futures resettlement. In actual practice, Cappoza and Cornell (1979) note that government securities dealers charge 50 basis points over the Treasury-bill rate for financing costs, although Allen and Thurston (1988) maintain that the term rep0 rate is a more accurate reflection of financing costs in the government bill market.…”