Show Me Listing

Online Ebooks (PDF) Library

• Home
• Popular
• Latest
• Contact

Home » Economic » Econometrics

Computing power and the power of econometrics

Source: dss.ucsd.edu
Topic: Econometrics

Sort Desciption: nities for the discipline of econometrics a process of methodological ... ways that the practice of econometrics has been transformed by advances in ...

Content Inside: Computing power and the power of econometrics April 12 2006 James D. Hamilton Department of Economics 0508 University of California San Diego La Jolla CA 92093-0508 jhamilton@ucsd.edu Prepared in celebration of the 50th anniversary of the Econometric Institute at Erasmus University 0 1 Introduction. I bought my first computer in 1984 an IBM PC-AT. It was a marvelous machine with a clock speed of 6 MHz a hard disk that could hold 20 megabytes of data and capable of something like 8000 floating point operations in a single second. And every single year since then the computers available for anyone to buy have become even more marvelous. Weve gone from measuring speed in flops (floating point operations per second) to megaflops (one million flops) gigaflops (a billion) and the now-standard teraflops (one trillion or 10 12 ). I carry around in my pocket today a tiny jump drive that by itself can hold more data than 40 of those PC-ATs and transfer it almost instantly to any computer. The phenomenal progress in computing power has generated tremendous new opportu- nities for the discipline of econometrics a process of methodological advance that is ongoing with the technological progress in computing. In this essay I call attention to a few of the ways that the practice of econometrics has been transformed by advances in computing and will continue to evolve in the years to come. 1 2 Very large data bases. 2.1 High-frequency data. Huge improvements in data storage and access technology have opened up new classes of data sets that can be analyzed. One very interesting area of research concerns high-frequency data as we have gone from quarterly or monthly data sets to looking now at daily or even tic-by-tic data that capture every single transaction during the day even on exchanges where such transactions can be huge in number. One of the things one sees immediately in such data sets is that a much richer concept of seasonality is needed than that with wh ...

©2007 Show Me Listing - Designed by Free CSS Templates

RSS Feed Sitemap Privacy Policy