Forecasting M&A deals with MIDAS count model

This report focuses on the forecast of the number of monthly cross-border deals in the European Union. We propose a new model to improve the forecasting properties of a count model of Foreign Direct Investment deals in EU, by taking into account past trends in high-frequency (daily) deal data and the decomposition of the conditional overdispersion into short-term and long-term components. Our model relies on the dynamic behaviour of the first two moments of the distribution of FDI deals to explain the evolution of parameters 𝜂 and 𝜋 in the Negative Binomial distribution. We test this model with several subsets of M&A deals from 1998 to 2021 obtaining sizable forecast improvements as compared to benchmark INGARCH models.