http://www.eviews.com/EViews9/ev95midas.html

Traditional approaches to time-series estimation and forecasting in economics require that the variables be of the same frequency. This often causes a problem since most macroeconomic data is reported at different intervals and frequencies. For example in many countries GDP is reported on a quarterly basis, whereas unemployment figures are reported monthly, and interest rates or stock market prices are reported daily. A frequently used solution to this issue has been to aggregate the higher frequency data into values in the lower frequency. For example, the 3 months of unemployment data in each quarter are averaged to give a single quarterly value. A significant disadvantage to this approach is that through the aggregation you discard data which can lead to less accurate estimation.

Mixed-Data Sampling (MIDAS) is a method of estimating and forecasting from models where the dependent variable is recorded at a lower frequency than one or more of the independent variables. Unlike the traditional aggregation approach, MIDAS uses information from every observation in the higher frequency space.

As an example of using MIDAS in EViews, we will follow the Federal Reserve Bank of St Louis paper "Forecasting with Mixed Frequencies" from November 2010. In one of the applications in this paper, the authors forecast quarterly GDP rates using monthly employment growth data. We will perform a similar study using more recent data.