James G. MacKinnon and Morten Ø. Nielsen, "Numerical Distribution Functions of Fractional Unit Root and Cointegration Tests," Journal of Applied Econometrics, Vol. 29, No. 1, 2014, pp. 161-171. There are two zip files: mn-files.zip and mn-binaries. The file mn-files.zip contains source code and tables to compute critical values and P values. These files are in DOS format, so Unix/Linux users should use "unzip -a". There are 24 data files: frmapp01.txt frmapp04.txt frmapp07.txt frmapp10.txt frmapp02.txt frmapp05.txt frmapp08.txt frmapp11.txt frmapp03.txt frmapp06.txt frmapp09.txt frmapp12.txt frcapp01.txt frcapp04.txt frcapp07.txt frcapp10.txt frcapp02.txt frcapp05.txt frcapp08.txt frcapp11.txt frcapp03.txt frcapp06.txt frcapp09.txt frcapp12.txt These files must either be in the current directory or in a specific directory for the fracdist program to work. On Unix/Linux systems, the specified directory is /usr/local/urcdist. On Windows systems, it is C:/URCDIST. There is one program file: fracdist.f It is written in Fortran (in a dialect that is much closer to f77 than f90) and should be compilable by any Fortran compiler. In particular, the free compilers g95 and gfortran should both work. A very slightly modified version of this program, intended for use on Windows machines, is also included: fracdist.f-win The only differences between the two versions are in the three lines immediately following the three lines that begin with "cwin". As written, the program is interactive. Anyone wishing to compute large numbers of P values or critical values should remove the main program and write a program that calls the routine frval, which in turn calls other routines, as many times as necessary. If you are going to call it many times, be sure to read the comments about the parameter isave. The file mn-binaries includes three compiled versions of fracdist. The one that will be useful to most people is fracdist.exe, a version for 32-bit Windows systems that was compiled with the g95 compiler. It also works on 64-bit Windows systems. This file also contains two compiled versions for Linux systems. As the names imply, fracdist-32 and fracdist-64 are for 32-bit and 64-bit systems, respectively. They were compiled with gfortran and linked statically. We do not guarantee that these will work with all versions of Linux on Intel or AMD systems, but they should work with most modern ones. On a Linux system, it is very easy to compile fracdist. Once gfortran has been installed, simply use the following command line to create an executable called fracdist: gfortran -O3 fracdist.f -o fracdist Note that this will be much smaller than the executables we provide, because it will not be statically linked. The programs and tables provided here may be used freely, so long as the paper is cited in all working papers and published work. An alternative C++ implementation of this program is also available; see https://github.com/jagerman/fracdist/blob/master/README.md for details. Added April 28, 2014: A programming error in fracdist.f has been corrected. The effects of this error are hard to predict. The two source files and the three executables were updated today.