CMStatistics 2016: Start Registration
View Submission - CFE
A1401
Title: Extracting fiscal policy expectations from daily stock returns Authors:  Etsuro Shioji - Hitotsubashi University (Japan) [presenting]
Abstract: The ``fiscal foresight'' problem refers to the fact that most fiscal policy changes are announced well before their implementation. As a result, innovations in the actual amount of government spending may not capture the true timing or magnitude of the surprises. To overcome this difficulty, we construct new daily indicators of surprises about public investment spending for Japan. This novel approach combines a detailed analysis of newspaper articles on future policies with information on a cross section of excess returns on stocks of construction companies. In a previous work, we constructed an indicator which was essentially a weighted average of those returns on the days that the news arrived. In contrast, the two new indicators take advantage of heterogeneity within the industry. Degrees of dependence on governmental contracts differ markedly between construction companies. The first indicator simply takes the difference in the average excess returns between the firms that are more dependent on public procurement and the less dependent ones. The second indicator is more elaborate, and is based on the ``Target Rotation'' approach in the factor analysis. It is shown that both serve as good leading indicators of future public investment spending.