The tools of fiscal redistribution are taxing and spending.

The fourth and fifth regressions in Table 7.1 examine the use the size of public expenditures as a proxy for the main redistribution tools, namely for direct taxation and transfer payments. In these regressions we simply use these variables or similar ones instead of total public expenditures. We find that both variables have a negative effect on inequality, namely that they redistribute income. But the effect of direct taxation is significant, while the effect of social benefits is statistically insignificant. Furthermore, adding social benefits to direct taxation reduces R2, namely the explanatory power of the regression is reduced. This is unsurprising since the variable we use is a small part of transfers, namely unemployment benefits. We conclude, therefore, that the redistributive effect can be assessed much better by use of progressive direct taxation than by social benefits and social subsidies. Below we find that this result is reversed when we focus on the rate of poverty. Note that the absolute effect of direct taxation has a stronger effect on inequality than total expenditures, but this result is unsurprising. Any marginal unit of public spending is only partially financed by direct taxes, while the rest is financed by indirect taxes. Hence direct taxes fluctuate by less than public spending, so their measured effect on inequality should be higher. Overall, regressions (4) and (5) have lower R2, so their explanatory power is lower than that of total public spending. From here on, public spending will be our main measure of the redistribution of income from market to disposable incomes. As a result regression model (4) remains our central and basic regression model.