First, we compare nested and non-nested models to evaluate whether the spatial lag has
significant effects on flat tax adoption. Second, we estimate Model D from Table 1 as
an ordinary least squares regression, specifying the dependent variable as a top marginal
individual tax rate instead of flat tax change, calculate the multiplier and evaluate short-
term effects. Further research will address the evaluation of spatial interdependence and
diffusion, should the analytical solution for binary model becomes available.
Additionally, because it is difficult to interpret the coefficient on the spatial lag di-
rectly, we use the number of flat tax countries in a given year (Diffusion) instead, which is
highly correlated with the spatial lag. Recall that only one country had gone flat in 1994,
and two more (all in the Baltics) from 1995 to 2000; the count was five in 2001-2002,
seven in 2003, and ten in 2004-2005. We estimate the probability of a flat tax policy
change in a non-flat country, depending on the number of flat countries in the region and
changes in governmental ideology. We estimated effects of two hypothetical changes in
government: from the center to the center-left and to the center-right. We shift the value
of ideology from the mean value of 10.4 by one standard deviation above and below this
mean: to the left government (6.48) and to the right government (14.6).
Using the method of first differences and holding all other variables at their mean
values, we estimate that when the center government is replaced by the left one, the
probability of flat tax change in that country decreases by 0.06 (0.05) with three “flat”
neighbors and by 0.15 (0.1) with seven “flat” countries (standard errors in parentheses).
In contrast, when a right party or coalition comes into power, the probability increases by
0.07 (0.03) and by 0.24 (0.21) percent, given three to seven “flat” neighbors. When the
number of flat countries is nine, the ideological change to the left does not decrease the
probability: it goes up by 0.03 (0.03) percent. Likewise, the change to the right, given nine
24