The need for high frequency data: estimating monthly GDP

Real Gross Domestic Product is usually computed at quarterly intervals, which makes it uncomfortable to introduce into different types of macroeconomic models, that usually make use of higher frequency data (monthly, weekly) such as inflation, interest and unemployment. Analysts and decision makers may also want to evaluate a close evolution of the aggregated national output, especially during the economic slowdown periods, just as the one which has spread almost worldwide. Moreover, it is well known that the capability of correctly identifying the short-term pattern of an economic phenomenon is directly linked to the frequency of available observations.For economic studies using quarterly data, a low number of observations can cause serious flaws in the quality of quantitative analysis, without even considering the situations when many degrees of freedom are used up in the estimation, this way drastically reducing its power. Another limitation stands in the short sample size for developing countries as result of deep structural changes that occurred in the past couple of decades. One way around this problem is to estimate higher frequency series, using information from the low frequency GDP series and some related series. The paper illustrates and evaluates a Kalman filtering method for forecasting the seasonally adjusted Romanian real GDP at monthly intervals. However the present mixed-frequency method produces monthly GDP forecasts for the first two months of a quarter ahead which are more accurate than one-quarter-ahead GDP forecast based on purely-quarterly data. The purpose of this paper is to achieve this temporarily desegregation for Romanian dates and show few possible applications of these results, such as testing an eventual existence of Taylor Rule based monetary policy among the Central Banks of five economies.