Equation (4) shows that the capacity for “domestic innovation” depends on the
available human capital stock. The human capital stock independently enhances
technological progress, and, holding human capital levels constant, states with lower initial
productivity levels will experience a faster growth of total factor productivity (assuming both
m and g — m are positive).
The model presented in equation (4) is strictly topological invariant in the sense that
changes in the size, shape and location of the areal units do not have a bearing upon the
results. We therefore incorporate a spatial spillover effect in the available domestic human
capital stock and a distance decay effect in the catch-up term, as follows:
± 1 H
Ymax - Yi
Yi
(5)
(logA -log A0)i= c + gH + r ∑ -Hm + m-ɪ
1d d
m =1 im i ,max
m∈Ji(d)
where states located within the ‘cut-off distance’ d are included in the Ji(d) classes for the
spatial spillover effect, di,max represents the geographical distance of state i to the technology
leader and r the coefficient of human capital accumulation in neighboring regions.
Rearranging and substitution gives:
(log At -logA0)i =c+gHi
1 J1
m--H + r ∑ —H
d i dm
i ,max m =1 im
m∈Ji(d)
1
+m
d
i ,max
Hi
( Y ^
yY2l I,
< Yi J
(6)
Using OLS, we first estimate a model based on equation (2), where the technological
progress is taken into account as in equation (6). For each sector, the previously defined
15