The development of national statistical systems and the improvements made by statistical agencies in compiling and disseminating data to meet the needs of both policymakers and the general public have led to the availability of higher frequency indicators for monitoring changes in economic activity. Although key macroeconomic aggregates, such as GDP, are typically only available at a quarterly frequency, we currently have a relatively wide range of monthly indicators, covering a broad set of economic dimensions. The availability of data at a daily frequency has generally been limited to financial variables, such as stock prices and interest rates.

At the same time, a growing body of literature has focused on the use of higher frequency variables for nowcasting and forecasting the main quarterly macroeconomic variables. The mixed data sampling (MIDAS) regression models introduced by Ghysels, Santa-Clara, and Valkanov (2004) have received considerable attention in the literature. The work of, inter alia, Ghysels et al., 2004, Ghysels et al., 2005 and Ghysels, Santa-Clara et al., 2006, and the growing empirical evidence of its usefulness, have led to a gain in the popularity of MIDAS for forecasting. There is a significant body of literature on the advantages of using MIDAS regressions to improve quarterly macroeconomic forecasts based on monthly and daily data. For instance, Clements and Galvão, 2008 and Clements and Galvão, 2009, Kuzin, Marcellino, and Schumacher (2011), Marcellino and Schumacher (2010), and Schumacher and Breitung (2008) provide evidence of improvements in quarterly forecasts from using monthly data, and Andreou, Ghysels, and Kourtellos (2013), Ghysels and Wright (2009) and Monteforte and Moretti (2013), among others, show forecast improvements from the use of daily data. However, given the limited availability of high frequency economic data, MIDAS forecasting has generally been restricted to daily financial series.

谢谢楼主分享的资料不错啊！

楼主的兴趣很广泛啊