Abstract
Evidence of herding behaviour in financial markets carries significant implications for investor decision-making. This phenomenon can cause asset prices to deviate from their true value, leading to market instability. Understanding the presence or absence of herding behaviour is vital for guiding investors and safeguarding their interests. This dissertation investigates herding behaviour in three major currency pairs (USD, EUR, and GBP) against a backdrop of ninety-three, ninety-four, and seventy-one other currencies, respectively, during the 2008 Global Financial Crisis. Data was sourced from the Bloomberg database, covering the period from January 1, 2007, to December 31, 2008. To analyse the forex returns distribution comprehensively, two Bayesian regression models were employed. These models accounted for innovations with normal and Student t-distributed characteristics. The findings indicate that herding behaviour was present only in the GBP forex market during market days when the market returns were either extremely lower (bear) or higher (bull), using Bayesian regression models with normally distributed innovations. No evidence of herding was observed in the other currency markets during bear, normal, or bull market days. It is worth noting that the results obtained using the Bayesian model with normally distributed innovations were robust, except for the GBP forex market. These findings contribute valuable insights into herding behaviour within forex markets during a period of financial turmoil, aiding investors and researchers in making informed decisions and assessments.
Keywords: Herding behaviour, forex markets, Bayesian model, currency pairs, financial crisis