Research seminar run by the Applied Economics research group.
Date/Time: Tuesday 29 November 2016, 17:00-18:00
Venue: Newcastle University Business School, 4.20
Speaker: Dr Jack Fosten, UEA
Dr Fosten will be discussing his paper, including the test used in his research to determine whether 'big data' nowcasting methods, which can produce timely updated predictions of low-frequency variables like Gross Domestic Product (GDP), are monotonically improving as new information becomes available.
Although nowcasting models have become an important tool to many public and private institutions, there has been little work on how to formally evaluate such procedures. Dr Fosten places particular emphasis on models involving estimated factors, since factor-based methods have become a leading case in the high-dimensional empirical nowcasting literature, although the test is still applicable to small-dimensional set-ups like bridge equations and MIDAS models.
The properties of factor estimates when used in nowcast evaluation tests has not yet been explored and is therefore of separate econometric interest. Dr Fosten will provide several contributions to the existing literature. Firstly, to extend the methodology of Chernozhukov et al. (2014) for testing many moment inequalities to the case of nowcast monotonicity testing, which allows the number of inequalities to grow with the sample size. Then he will provide results showing the conditions under which both parameter estimation error and factor estimation error can be accommodated in this high dimensional setting when using the pseudo out-of-sample approach of West (1996).
This illustrates the finite sample performance of the test through Monte Carlo simulations, and conclude with an empirical application of nowcasting U.S. real GDP growth and fi ve GDP sub-components.
Research group: Applied Economics