A Bayesian methodology to explore the effects of memory loss in currency markets

As time goes by, people tend to establish new norms based on more recent events and information. Old norms and standards are eclipsed by new developments. The same type of behavior appears to operate in financial markets. New and unexpected information can jolt a market, but over time market participants adjust to new patterns of information, of data flow, of market volatility, and even of structural change. On another level, one can ask the question of how much weight to give new information relative to old information. Market participants do this naturally, as they weigh new information, such as just‐released economic data, compare it to their previously held expectations and then revise their expectations of future data releases and potential market reactions. When the question is viewed in this perspective, it seems quite natural to assume that new pieces of information, such as economic data releases, should be given more weight in assessing expectations of future market behavior than releases of the same type of data several months or even several years previously. That is, the information value of certain specific types of information, such as specific economic data releases, probably decays as time passes. This study proposes the use of a dynamic Bayesian technology to explore the effects of time decay in the value placed on specific bits of information that are received at regular intervals through time. The methodology that is being proposed is the main focus of this study and, to clarify this methodology, it is applied to the currency markets asking questions about the time decay pattern of certain economic data. The results reported in this example are purely for illustrative purposes, since they are specific to the data studied, the model specification of currency market behavior, and a wide variety of other assumptions embedded in this specific approach to currency analysis. Nevertheless, the results are interesting, if only because they demonstrate both the basic idea that new information is more valuable than old information and provide a useful and practical method of analyzing the issue of the time decay factor in how markets process information.