4. Tax coordination, rents and welfare
The analysis in the two previous sections showed that tax competition among
small jurisdictions has the potential to destroy rents completely, but in that case
it will also cause an underprovision of public goods which could be substantial.
This suggests that an internationally coordinated rise in taxation could be welfare-
improving even if the political process is biased in favour of public sector workers.
Thus, an interesting question is whether tax coordination will raise social welfare
and whether it will do so even if it leads to the emergence of rents? In this main
section we take a closer look at these issues, focusing on the case where individual
jurisdictions are small.
4.1. Tax coordination without rent creation
When the capital tax rate is fixed by some international agreement on tax coordin-
ation, politicians in the individual small country cannot influence k = k (r + τ)
and w = w (r + τ ) since they now take τ as well as r as given. However, they
must still find the politically optimal combination of W and α, subject to the
constraints (2.22) and (2.23). If a political candidate offers to raise the public
sector wage rate by the amount dW, it follows from (2.1), (2.20) and (2.21) that
the resulting marginal political benefit (MPB) in terms of the increase in the
probability of election victory will be
MPB = [αipliu,g + (α - αi) PoUg] dW. (4.1)
Since the tax rate is fixed by international agreement, a rise in the public sector
wage rate can only be financed through a cut in the number of public sector jobs
and hence in public goods provision. According to (2.1), (2.20) and (2.21), the
marginal political cost (the expected loss of votes) associated with a reduction
∣dα∣ in public sector employment is
MPC = {[αiPi + (1 - αi)po] g (α) + Po (ug - up)} ∣dα∣ . (4.2)
In the absence of constraints on wage-setting, an optimising politician will
want to equate the above expressions for the marginal political benefits and costs.
However, in a tax competition equilibrium where condition (3.3) holds, it follows
24