enEnglish
CROS

Explaining nowcast errors, by Henriette Druba, Jennifer L. Castle and David F. Hendry

Document date: 
Friday, 12 January, 2018
Language: 
English

Preliminary estimates of macroeconomic aggregates only become available with a time lag and are often unreliable and must be revised, more so during periods of structural change such as the financial crisis and subsequent recession. This article takes the forecast error taxonomy in Hendry and Mizon (2012), and adapts it to nowcasting. The taxonomy provides a framework for thinking about potential problems facing the nowcaster in achieving useful nowcasts. It  considers a wide array of sources of nowcast errors, from estimation uncertainty to model mis-specification. Importantly, the taxonomy incorporates unforeseen changes in parameters, and thus allows for a formal analysis into the consequences of structural breaks for nowcasting. Additionally, the taxonomy is applied to evaluate the impact of the different error sources on the expected nowcast error. This evaluation yields seven insights into the sources of nowcast errors, and into sources of nowcast failure. An empirical example nowcasting
consumption expenditure illustrates the insights from the taxonomy.