Maximo Camacho and Gabriel Perez-Quiros, "This is what the Leading Indicators Lead", Journal of Applied Econometrics, Vol. 17, No. 1, 2002, pp. 61-80. The file codes.zip includes the codes to estimate in-sample linear AR and VAR models (ar.gau and var.gau), nonparametrically GDP (ke.gau) and probabilities of recessions (ake.gau), probit (probit.gau), markov switching (switch.gau) and Logistic Vector Smooth Transition Regression (LVSTR.gau) models. This file also includes the codes to estimate real-time combinations (rtcomb.gau) and to apply the Diebold and Mariano tests (dm.gau). The data are in the zip file cp-data.zip. This incorporate the Excel spreadsheet file cli.xls and the following files, all of which are stored in plain ASCII (DOS text) format. Since the spreadsheet file is a binary file, Unix users should *not* use "unzip -a". cli1.data contains the data for the CLI in real time from the 1968.11 release to the release in 1989.11. The first row gives the date of the release. Additional rows give, for the date indicated in the first column, the value for the CLI in each of the revisions. cli2.data contains the data for the CLI in real time from the 1989.12 release to the 1998.01 release. The first row gives the date of the release. Additional rows give, for the date indicated in the first column, the value for the CLI in each of the revisions. The last release (1998.01) is the one used for the in-sample analysis. The CLI monthly series are transformed into quarterly series as explained in the text. The CLI is released by the Conference Board since October 1996 and by the Bureau of Economic Analysis prior to that date. Each future observation was replaced by zero. The GDP series is in the file gdp.data: chained-weighted GDP quarterly data 1947.1-1998.1 The final file is cli.data: CLI monthly series issued January 1998. It contains data for 1959 through 1997. If you have any questions or problems, please contact Maximo Camacho (mcamacho@um.es) or visit the web page: http://www.um.es/econometria/Maximo