A, or Solow residual, measures the quantity of output that does not de-
pend on the production factors. We have computed A for each year of the
sample period and for each region, obtaining a panel-varying TFP. Reorgan-
ising equation (7), the decomposition of labour productivity becomes evident:
(Г) TFP (K) ■
(8)
where γ = i αα. Table 1 shows the average value for each region and
for macro areas for each component of equation (8) in the period 1700-2001.
From this table, we can see a decrease in labour productivity from the North-
West (11% more than the national average) to the South (14% less than the
national average) of the country. A similar gap is estimated for TFP. On the
other hand, the distribution of capital per unit of output seems to be more
homogenous among the macro-areas. Moreover, labour productivity is highly
correlated with TFP (0.80) and little with the ratio K/Y (0.26). Hence, we
believe that in order to explain the Y/L difference, it is more important to
look at TFP rather than the ratio K/Y.
Figure 2 plots the regional time series for TFP, obtained using the growth
accounting methodology. Based on the simple visual inspection of these series
there seems to be a tendency for the series to convergence from 1970 to 1980
and a persistence of the regional gaps over the latter period.
4.2 Results
Tables 2 and 3 show the PUR tests of convergence for two possible measures
of distance in TFP between regions. The first is the simple distance of each
regional series from the cross sectional average, taken as the benchmark, i.e:
1N
yit = In(yit ) - N ∑ ln(Vit ) ■
(9)
The second is a measure of distance that does not make a distinction between
regions below and above the benchmark, and uses the absolute (value of the)
12