\chapter{Gretl and Ox} \label{chap:gretlOx} \section{Introduction} \label{Ox-intro} \app{Ox}, written by Jurgen A. Doornik \citep[see][]{doornik07}, is described by its author as ``an object-oriented statistical system. At its core is a powerful matrix language, which is complemented by a comprehensive statistical library. Among the special features of Ox are its speed [and] well-designed syntax\dots{}. Ox comes in two versions: Ox Professional and Ox Console. Ox is available for Windows, Linux, Mac (OS X), and several Unix platforms.'' (\url{www.doornik.com}) \app{Ox} is proprietary, closed-source software. The command-line version of the program is, however, available free of change for academic users. Quoting again from Doornik's website: ``The Console (command line) versions may be used freely for academic research and teaching purposes only\dots{}. The Ox syntax is public, and, of course, you may do with your own Ox code whatever you wish.'' If you wish to use \app{Ox} in conjunction with gretl please refer to \url{doornik.com} for further details on licensing. As the reader will no doubt have noticed, most other software that we discuss in this Guide is open-source and freely available for all users. We make an exception for \app{Ox} on the grounds that it is indeed fast and well designed, and that its statistical library---along with various add-on packages that are also available---has exceptional coverage of cutting-edge techniques in econometrics. The gretl authors have used \app{Ox} for benchmarking some of gretl's more advanced features such as dynamic panel models and state space models.\footnote{For a review of \app{Ox}, see \cite{cribari-neto03} and for a (somewhat dated) comparison of \app{Ox} with other matrix-oriented packages such as \app{GAUSS}, see \cite{steinhaus99}.} \section{\app{Ox} support in gretl} \label{sec:Ox-support} The support offered for \app{Ox} in gretl is similar to that offered for \app{R}, as discussed in chapter~\ref{chap:gretlR}. \tip{To enable support for \app{Ox}, go to the Tools/Preferences/General menu item and look under the Programs tab. Find the entry for the path to the \texttt{oxl} executable, that is, the program that runs \app{Ox} files (on MS Windows it is called \texttt{oxl.exe}). Adjust the path if it's not already right for your system and you should be ready to go.} With support enabled, you can open and edit \app{Ox} programs in the gretl GUI. Clicking the ``execute'' icon in the editor window will send your code to \app{Ox} for execution. Figures~\ref{fig:Oxedit} and Figure~\ref{fig:Oxout} show an \app{Ox} program and part of its output. \begin{figure}[htbp] \centering \includegraphics[scale=0.7]{figures/Oxedit} \caption{\app{Ox} editing window} \label{fig:Oxedit} \end{figure} \begin{figure}[htbp] \centering \includegraphics[scale=0.7]{figures/Oxout} \caption{Output from \app{Ox}} \label{fig:Oxout} \end{figure} In addition you can embed \app{Ox} code within a gretl script using a \texttt{foreign} block, as described in connection with \app{R}. A trivial example, which simply prints the gretl data matrix within \app{Ox}, is shown in Listing~\ref{ex:ox-simple}. \begin{script} \caption{Simple example of \app{Ox} usage} \label{ex:ox-simple} \begin{scode} open data4-1 matrix m = {dataset} mwrite(m, "gretl.mat", 1) foreign language=Ox #include main() { decl gmat = gretl_loadmat("gretl.mat"); print(gmat); } end foreign \end{scode} \end{script} The listing illustrates how a matrix can be passed from gretl to \app{Ox}. We use the \texttt{mwrite} function to write a matrix into the user's ``dotdir'' (see section~\ref{sec:named-strings}), then in \app{Ox} we use the function \verb|gretl_loadmat| to retrieve the matrix. How does \verb|gretl_loadmat| come to be defined? When gretl writes out the \app{Ox} program corresponding to your \texttt{foreign} block it does two things in addition. First, it writes a small utility file named \verb|gretl_io.ox| into your dotdir. This contains a definition for \verb|gretl_loadmat| and also for the function \verb|gretl_export| (see below). Second, gretl interpolates into your \app{Ox} code a line which includes this utility file (it is inserted right after the inclusion of \texttt{oxstd.h}, which is needed in all \app{Ox} programs). Note that \verb|gretl_loadmat| expects to find the named file in the user's dotdir. \section{Illustration: replication of DPD model} \label{sec:dpd-replication} Listing~\ref{ex:ox-dpd} shows a more ambitious case. This script replicates one of the dynamic panel data models in \cite{arellano-bond91}, first using gretl and then using \app{Ox}; we then check the relative differences between the parameter estimates produced by the two programs (which turn out to be reassuringly small). This was last tested using version 9.10 of \app{Ox console} for Linux. Unlike the previous example, in this case we pass the dataset from gretl to \app{Ox} as a CSV file in order to preserve the variable names. Note the use of the internal variable \verb|csv_na| to get the right representation of missing values for use with \app{Ox}---and also note that the \verb|--send-data| option for the \texttt{foreign} command is not available in connection with Ox. We get the parameter estimates back from \app{Ox} using \verb|gretl_export| on the \app{Ox} side and \texttt{mread} on the gretl side. The \verb|gretl_export| function takes two arguments, a matrix and a file name. The file is written into the user's dotdir, from where it can be picked up using \texttt{mread}. The final portion of the output from Listing~\ref{ex:ox-dpd} is shown below: % \begin{code} ? matrix oxparm = mread("oxparm.mat", 1) Generated matrix oxparm ? eval abs((parm - oxparm) ./ oxparm) 1.1094e-12 1.6314e-12 1.7885e-13 1.1949e-12 1.8999e-13 1.1359e-13 6.5339e-13 3.1943e-12 1.2108e-12 3.7335e-13 1.7468e-12 3.9536e-12 2.0980e-12 \end{code} \begin{script}[htbp] \scriptinfo{ox-dpd}{Estimation of dynamic panel data model via gretl and \app{Ox}} \begin{scode} open abdata.gdt # 1-step GMM estimation dpanel 2 ; n w w(-1) k ys ys(-1) 0 --time-dummies --dpdstyle matrix parm = $coeff # Write CSV file for Ox set csv_na .NaN store @dotdir/abdata.csv # Replicate using the Ox DPD package foreign language=Ox #include #import main () { decl dpd = new DPD(); dpd.Load("@dotdir/abdata.csv"); dpd.SetYear("YEAR"); dpd.Select(DPD::Y_VAR, {"n", 0, 2}); dpd.Select(DPD::X_VAR, {"w", 0, 1, "k", 0, 0, "ys", 0, 1}); dpd.Select(DPD::I_VAR, {"w", 0, 1, "k", 0, 0, "ys", 0, 1}); dpd.Gmm("n", 2, 99); // GMM-type instrument dpd.SetDummies(DPD::D_CONSTANT + DPD::D_TIME); dpd.SetTest(2, 2); // Sargan, AR 1-2 tests dpd.Estimate(); // 1-step estimation decl parm = dpd.GetPar(); gretl_export(parm, "oxparm.mat"); delete dpd; } end foreign # Compare the results matrix oxparm = mread("oxparm.mat", 1) eval abs((parm - oxparm) ./ oxparm) \end{scode} \end{script} %%% Local Variables: %%% mode: latex %%% TeX-master: "gretl-guide" %%% End: