The relative revenue power of the marginal tax rates in personal income tax schedules: the Revenue Equivalent Marginal Rate Increase (REMI)

ABSTRACT

The aim of this paper is to explore how the revenue power of the marginal tax rates in personal income tax rate schedules can be computed. Analytical expressions to measure the relative revenue effectiveness of marginal tax rates are provided. These expressions distinguish between mechanical and behavioural components as well as differentiating genuine allowances from non-genuine allowances. Finally, an empirical application to Spanish tax return microdata is provided.

KEYWORDS:

• Revenue power
• marginal tax rates
• personal income tax

JEL CLASSIFICATION:

• H24
• H31

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ For example, see Creedy (2011, chapter 13) and Saez, Slemrod, and Giertz (2012)..

² An example will help to better understand the differences between these two ways of computing allowances. Imagine a taxpayer with a taxable base of 61,000 euros and allowance of 28,000 euros who is confronted with the tax schedule shown in Table A1 of the appendix. The tax due under these two alternative ways of computing allowances would be as follows. In the genuine system, we deduct the allowance from the tax base to obtain the taxable income and to this amount we apply the tax band – Taxable income: 61,000–28,000 = 33,000 and Tax due: 8,065.5. In the non-genuine case, the tax base coincides with the taxable income and the allowance generates a tax saving resulting from applying the tax schedule to the allowance separately – Taxable income: 61.000 euros and Tax due: 18,351.5–6,565.5 0 = 11,786. Namely, the tax savings of the genuine allowance reach 10,286 euros ($1,000\cdot 0.45+24,800\cdot 0.37+2,200\cdot 0.30$), whereas in the case of non-genuine allowances the saving is 6,565.5 euros ($12,450\cdot 0.19+7,750\cdot 0.24+7,800\cdot 0.30$). As can be seen, depending on the implemented procedure the tax savings derived from the allowances differ, as each method involves different tax rates.

³ It is a stratified random sample with three levels of stratification: province (49), income level (12 categories) and filing status (jointly or separately). As a result, the final number of strata is 1,176 (49ₓ12ₓ2). The sample size was calculated for a standard error less than 1.5% with a confidence level of 3 per thousand. This sample design provides an accurate estimate of the tax liability.

⁴ The parameter used to incorporate behaviour into the simulations is the tax base elasticity, ${\eta }_{x_j,(1-{\tau }_j)}$. The elasticities assumed for each tax bracket are the following: first and second brackets: 0.05; third bracket: 0.055; fourth bracket: 0.1 and fifth bracket: 0.2. These elasticities are moderate, being in the low range of the elasticities estimated for the Spanish PIT. For some empirical estimations applied to Spain, see Badenes (2001) and Sanmartin (2007), Arrazola et al. (2014) and Sanz et al. (2015).