Arbitrage pricing theory (APT)

A theory which aims to estimate returns and, by implication, the correct prices of investments. Intellectually, it is an extension of the capital asset pricing model. It says that the CAP-M is inadequate because it assumes that only one factor - the market - determines the price of an investment, whereas common sense tells us that several factors will have a major impact on its price in the long term. Put those factors into a model and you are making progress.

Thus arbitrage pricing theory (APT) defines expected returns on, say, an ordinary share as the risk-free rate of return plus the sum of the share's sensitivity to various independent factors. (Here sensitivity, as with the CAP-M, is defined by the share's BETA.) The problem is to identify which factors to choose. This difficulty is compounded by academic studies which have come up with varying conclusions about the number and identity of the key factors, although benchmarks for interest rates, inflation, industrial activity and exchange rates loom large in tests.

In practice, the aim of using APT would be simultaneously to buy and sell a range of shares whose sensitivity to the chosen factors was such that a profit could be made while all exposure to the effect of the key variables and all capital outlay were canceled out. To the extent that APT assumes that markets always seek equilibrium, it says that the market would rapidly price away such arbitrage profits.

Alternatively, a portfolio could be chosen which could be expected to outperform the market if there were unexpected changes in one or more key factors used in the model, say industrial activity and interest rates. As such, however, that would be doing little more than betting on changes in industrial production and interest rates and would not have much to do with minimizing risk for a given return. Resolving problems such as these means that APT gives greater cause for thought to academics than to investors.