17.1 Getting Started

The quantlib multi is an interactive tool for specifying and analyzing multiple time series models. In the next sections we will try to build a small German money demand model. Our approach is to specify a model in full VAR form. This can be justified as a first step since the full VAR is a quite general model. It is also chosen here for exposition purpose.

Such a model should obviously include the money demand itself and some of its determinants. Money demand is set equal to the money circulating in the economy (henceforth referred as $ M$). Two determinants are included: the transaction volume of the economy which is given by the gross national product (GNP, henceforth referred as $ Y$) and the costs of holding money which is given by an interest rate (henceforth referred as $ I$). From economic theory we know that $ Y$ has a positive whereas $ I$ has a negative effect on $ M$. We might want to find out whether these theoretical considerations are supported by some observed data and whether a full VAR model is a satisfactory description of the relation between the variables involved.

But before starting the analysis we need to prepare the data and start the quantlet 29804 domulti .


17.1.1 Data Preparation

The time series need to be provided as a numeric $ T \times K$ data matrix x, where $ T$ is the length of the time series, and $ K$ its dimension. The first $ K$-dimensional vector of observations is in the first row of x, the second vector of observations is in the second row of x, and so forth. The data matrix must not contain verbal information. Additional data information (variable names, observation period) are given directly to 29842 domulti .

This chapter will use the dataset mts.dat which contains three time series:

The sample begins in 1960, 1-st quarter and ends 1989, 4-th quarter. It contains 120 quarterly observations. $ M$ and $ Y$ are seasonally adjusted and in prices of 1985. It is stated in Billion of German Marks. All data is taken from the OECD.


17.1.2 Starting multi

The quantlet 29891 domulti takes up to five inputs: The data matrix $ x$, the main period of the first observation (1960), the sub period of the first observation (1), the periodicity (4), and the variable names ($ M$, $ Y$, and $ I$).

Call 29894 domulti by

library("multi")
x=read("mts.dat")
domulti(x,1960,1,4,"M"|"Y"|"I")
29898 XAGmts01.xpl

The main menu
\includegraphics[scale=0.7]{mts_main}
will appear which means that we have entered the menu driven part of the multi quantlib. We can start the analysis now.