Direct Forecasting for Applied Regional Analysis

The regional-economic literature is vast, and attempts to estimate the time-series dimension of regional relationships therein are disparate. For applied researchers seeking a straightforward approach to time-series estimation within that literature, the scope may seem daunting. In this paper, I provide both an overview of that vast and disparate literature, and a simple path forward for applied work. For the latter, I first estimate spatial impulse response functions from a general time series-autoregressive model, emphasizing direct forecasting techniques. Second, I estimate impulse response functions from a spatial econometric model, the SLX model, showing how one can tease out spatial forecasts from a standard spatial framework. I demonstrate using state-level data on housing prices.