The test statistic obtained, W=178.2, had a
p-value of [0.000] which implies that the region specific effects are
related to the regressors.
There are structural differences between regions in the relationship between
net gaming
expenditure, machine numbers, and the number of clubs and hotels with EGMs.
This makes
it necessary to maintain intercept terms which vary from one area to the
next, and therefore
require the use of fixed effects estimation.
The parameter estimates
The estimated parameters in Equation 4.3 (fixed effects) were used to
calculate the impact on
per capital gaming expenditure of an increase in the number of machines per
thousand
adults. The result obtained suggests that a 10 per cent increase in EGM
numbers contributes
to an 8.09 per cent increase in net gaming expenditure per adult.
A There are no longer negative parameter estimates for numbers of clubs
and hotels with gaming machines. Since the panel data specification is more
comprehensive,
it follows that an increase in the number of gaming clubs or hotels results
in an increase in
player losses, as does the rise in the number of gaming machines.
The random effects model
log(PLOSS) = 4.26 + 0.972 log(PEGMs) + 0.318 log(PCLUBS) + 0.356 log(PHOTELS)
(Eqn 4.4)
(17.9) (17.6) (9.23) (9.55)
where the figures in parentheses are t-ratios, N=60, Time (T)=28, NOBS=1413,
R2=0.547,
and SE=0.594.
As previously discussed, the disadvantage of the random effects
specification is that it
assumes that the (region-specific) effects are uncorrelated with the
explanatory variables. A
Hausman test was used to ascertain whether fixed or random effects is the
preferred
specification (as described in Section 3.2).
Back to gambling
news archive
|
