Incorporating short data into large mixed-frequency VARs for regional nowcasting

Interest in regional economic issues coupled with advances administrative data is driving the creation of new data. Many these series could be useful for nowcasting activity, but they suffer from a short (albeit constantly expanding) time which makes incorporating them into models problematic. Regional already challenging because release delay on tends to greater than that at national level, and "short" imply "ragged edge" both beginning end sets, adds further complication. In this paper, via an application UK, we develop methods include wide range mixed-frequency VAR model. These hitherto unexploited VAT turnover We address problem ragged edge our sample by estimating factors using different missing algorithms then incorporate find nowcasts output growth are generally improved when condition factors, only produced before (UK-wide) published.