Stock markets and their relationship

with the real economy in Latin America

with the real economy in Latin America

DATA AND METHODOLOGY

Econometric Modeling ( ...continuation )

Econometric Modeling ( ...continuation )

**( 5 )**

**( 6 )**

where,

contemporary change in GDP (for each country)

contemporary change in stock exchange index for each country

ε represents the size of the residue derived from the lineal combination of the two cointegrated variables, also known as long term deviation from balance or error correction term.

Granger Causality explains if* x *causes* y*, that is which part of the running value *y* can be explained by the lagged values of *y*. This establishes if adding lag values of *x* improves the explanation for the behavior of *y*. In this case, it is said that *y *is Granger caused by* x*, if *x* influences the prediction of *y*. Equally, this is valid if the coefficients of the lagged variables of *x* are statistically significant. A bilateral causality relationship is frequently found: *x* Granger causes* y*, while *y *Granger causes *x*. The model can be represented in the following way:

**( 7 )**

Here,* l* indicates the number of lags.

Self-regression vector analysis (VAR) is an important addition to contemporary econometric analysis. These models are a generalization of the self-regressive model AR. In general, a variable is explained in terms of its past values and its frequent use is due to the simplification of many assumptions and restrictions of the structural models. In the AR models, the assignation a priori between the endogenous and exogenous variables does not exist and recovers the original dynamic of the time series. It is not necessary to establish beforehand which variables are endogenous or exogenous, as all the variables are endogenous. This is relevant in studies like the present one, where there is always interdependence between the series. The generalized VAR model without Sims restrictions (1980) applied in the present study consists of regressions of each of the variables without lags in relation to all the other variables with various lags based on:

**( 8 )**