background of the theory

The Arbitrage Pricing Theory (APT) is an alternative asset pricing model to the Capital Asset Pricing Model (CAPM). It was developed by economist Stephen Ross in the 1970s as a way to explain the relationship between asset prices and their expected returns.

The APT is based on the idea that an asset’s expected return is determined by multiple factors or variables, rather than just the market risk factor as in the CAPM. These factors could include macroeconomic variables such as interest rates, inflation rates, GDP growth rates, or industry-specific factors such as changes in oil prices or technological advancements.

The APT assumes that investors are risk-averse and seek to maximize their utility by balancing risk and return. It suggests that assets are priced such that there is no opportunity for risk-free arbitrage profits. In other words, if an asset is mispriced relative to its expected return given the relevant factors, investors would immediately exploit the mispricing, leading to market forces driving the asset price back to its equilibrium value.

The APT provides a framework for estimating an asset’s expected return by considering the sensitivity of the asset’s returns to each of the underlying factors. It involves a statistical regression analysis where the returns of the asset are regressed against the different factors to determine their impact on the asset’s expected return.

One of the key advantages of the APT is its ability to capture and explain the cross-sectional variation in asset returns by considering multiple factors. It also allows for a more flexible and diverse set of factors compared to the CAPM, which only considers systematic risk measured by beta.

However, the APT has some limitations. It assumes that the factors are well-defined and observable, which may not always be the case in practice. Estimating the factor sensitivities and their risk premiums accurately can be challenging. Additionally, the APT does not provide specific guidance on how to determine the appropriate factors to include in the model, and their interpretation may vary depending on the specific application.