longitudinal data model with spatial correlation

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

longitudinal data model with spatial correlation

Joao Pedro W. de Azevedo
Dear R users,

I would like to estimate a multivariate space-time series with the following
structure.

Dependent variable: rate of self-employment on region i at time t;

Covariates are: a set of economic (i.e. rate of unemployment, rate of
unemployment squared, sectoral composition) and demographic characteristics
(i.e. gender distribution, proportion of ethnical minorities, educational
level) of region i at time t-1.

The main factor of interest is the relationship between the rate of
unemployment of any given region at period t-1 and the level of
self-employment at period t. The hypothesis being tested is if individuals
are pushed of pulled into self-employment.

I'm using regional data from 100 different counties in the UK (i=100) as my
geographical units (I have both the shape file with the boundaries of this
regions and the centroids for each one of them). And I have over 11 years of
quarterly data (t=46). I've used the spdep package to run a few
cross-section models with spatial auto-correlation, but I would really like
to explore the longitudinal aspect of my dataset.

I would like to know if  you might have any idea of how I could set up a
space-time model using R commands. Any references and/or suggestions will be
more than welcomed.

All the very best,

Joao Pedro