Browsing by Author "Haavanlammi, Katja"

The theory about fiscal foresight suggests that, when it is present, the econometric analysis will face substantial challenges; and the shocks received can be nonfundamental if the information in the model is not sufficient. This thesis introduces fiscal foresight into a conventional vector autoregression (VAR) analysis and examines the dynamic effects of unanticipated and anticipated government expenditure shocks with Finnish data. However, since VAR models base only on present and past information, including a forward looking variable into a VAR may cause issues that lead to misspecifications of the model and into wrong conclusions. To measure government expenditure anticipations I create a news variable by using Bank of Finland’s government expenditure forecast data. The idea is that this variable should capture the effects when the future government expenditure increase is anticipated by the forecasters and assumably by the agents in the economy. By including this news variable in the VARs information set the recovered shocks should describe more accurately the identified shocks in the economy. I show with a simple example and with information sufficiency tests that this news variable overcomes the issues that fiscal foresight may cause to the VAR analysis and that this variable makes the VAR analysis more accurate. I will investigate if including this news variable in a conventional VAR model will change the outcomes on government expenditure, tax revenues, private consumption, private investment and output; first when the unanticipated government expenditure increase causes the shock and second when the positive shock comes from the news that can be interpreted to be an anticipated government expenditure increase. In most cases the literature either supports the Keynesian or neoclassical theory since there exist disagreements on the government expenditure effects on private consumption and investment. These effects are also in the centre of my thesis and the first result is that whether or not the news variable is included to the VAR model the unanticipated government expenditure shock does not crowd out private economic activity. Instead the effects are positive and persistent and they support the Keynesian view. Second result is that when the shock to the system comes from the news, the effects on private economic activity are quite the opposite and they support the neoclassical view. This would mean that if government expenditure increase is anticipated by the agents, they will postpone their consumption and investment decisions until the government expenditure increase is implemented and the expenditure starts to return back to its unshocked path. Results all together suggest that government expenditure increase as an expansionary fiscal policy action is effective if the increase is not anticipated by the agents in the economy. If announcements on future government expenditure increases is done, the hoped expansionary effect after implementation can actually retard the economy for a few years.