Discussion Paper

No. 2018-48 | June 12, 2018
How do taxable income responses to marginal tax rates differ by sex, marital status and age? Evidence from Spanish dual income tax

Abstract

The aim of this paper is to analyze how Spanish taxpayers have responded to the introduction of the dual personal income tax model in 2007. The authors estimate the elasticity of taxable income (ETI) with respect to the marginal net tax rate for different groups of taxpayers by sex, marital status and age, separating the substitution effect from the income effect. For the empirical analysis, they use microdata from the Spanish personal income tax return panel disseminated by the Spanish Institute of Fiscal Studies. The main results show that the 2007 tax reform resulted in a range of elasticity values from 0.41 to 0.43, while the estimated income effect yields a negative value of -0.18. The results for the different taxpayer groups are as follows: the removal of retired people from the sample significantly reduces the ETI; elasticity is higher for women than for men; single people have a considerably higher elasticity than married taxpayers; and the ETI decreases with age. Additionally, the authors find that the marginal cost of public funds increased after the reform, and the top marginal tax rate is above optimal.

JEL Classification:

H21, H24, H31

Assessment

  • Downloads: 217

Links

Cite As

Carlos Díaz-Caro and Jorge Onrubia (2018). How do taxable income responses to marginal tax rates differ by sex, marital status and age? Evidence from Spanish dual income tax. Economics Discussion Papers, No 2018-48, Kiel Institute for the World Economy. http://www.economics-ejournal.org/economics/discussionpapers/2018-48


Comments and Questions


Anonymous - Comments
July 30, 2018 - 23:14

Comments to: “How do taxable income responses to marginal tax rates differ by sex, marital status and age? Evidence from Spanish dual income tax” by Carlos Díaz-Caro and Jorge Onrubia
I really enjoyed reading this paper. The authors present an interesting and complete approach, applying the concept of the elasticity ...[more]

... of taxable income with respect to the marginal net-tax rate to analyse taxpayer´s behavioural responses in the scenario of the dual personal income tax (PIT) reforms. Recent literature has focused in the tax avoidance reason for taxpayer behaviour (Saez et al., 2012) and not only on changes in labour supply, as it was usual not so very long ago. This paper applies the econometric framework proposed by Gruber and Saez (2002) with some methodological variants to make a distinction between the substitution and income effects. The analysis captures overall responses to income taxation, such as income shifting, the choice of different asset types, tax avoidance or labour changes. The paper also takes into account other important concepts, such as the deadweight loss of income taxation, the marginal cost of public funds and the optimal tax rate.
I think that the main contributions of this paper to the existing literature are quite interesting and useful: the elasticity of taxable income (ETI) with respect to the marginal net-tax rate for the semi-dual income tax structure after the Spanish PIT reform; the wide range of EIT for different taxpayer groups allows the control of estimations by gender, age and marital status; the estimation of the deadweight loss and the marginal cost of public funds caused by the reform; and finally, the optimal marginal tax rates according to the estimated elasticities for the population calculation as a whole.
It seems to me a very complete, interesting and useful paper for the study of the joint effects of a given tax reform, which uses an adequate theoretical framework and methodological approach, intelligently applied to the global aspects that the authors want to study. I think that that is a brilliant idea to use, in the dual PIT framework, "weighted marginal tax rates", as a combination of the two statutory marginal rates (from general tax schedule and saving tax schedule).
Finally, the paper includes a singular and global empirical analysis for a semi-dual model of personal income taxation (IRPF) introduced in 2007 (Law 35/2006) in Spain: the elasticities estimations for the whole sample of the Spanish Personal Income Tax Return panel, available by the Spanish Institute of Fiscal Studies (IEF), for the different groups of taxpayers considered, and also the results for deadweight loss, efficiency cost and optimal tax rate using the calculated compensated ETI.
It is a well structured paper, with an exhaustive literature review, well argued, with an adequate methodology. In conclusion, I found this discussion paper very interesting and I would like to recommend its publication as article.


Jorge Onrubia - Reply to anonymous reader
August 21, 2018 - 13:56

Thank you very much for your carefully reading on our paper and for giving your opinion about the same.


Jose Cordero - Comments
July 31, 2018 - 12:34

I believe this paper is relevant both for practitioners and policy-makers. It presents an interesting analysis of the effect of the introduction of the dual income tax model in Spain over the behavior of different types of taxpayers through the estimation of the elasticities of taxable income. To do this, ...[more]

... the authors adopt the methodology proposed by Gruber and Saez (2002) with some improvements and use the elasticities to compute the marginal cost of public funds and revenue maximizing tax rates. The results show a range of elasticity value between 0.41-0.43 and a negative income effect value of 0.18. Both values are considered as normal in the literature of elasticity of taxable income.

From my point of view, it is an interesting contribution to the existing literature about this topic, thus I would recommend its publication in the journal. I only have some minor remarks that the authors should better clarify in a revised version of the manuscript:

- Using the weighted marginal rate based on Onrubia and Sanz (2009) to take into account the two different tax rates in the dual model seems an interesting contribution. However, in my opinion, the authors should also clarify why they do not use it in the previous year (2006), in which there are also two different tax bases.

- In table 6, I can see the estimated optimal tax rate ranges from 45 to 51%. Nevertheless, when the analysis takes into account the different bracket, the optimal tax rate decreased notably with respect to the income. The authors should provide a more detailed explanation for this effect.


Jorge Onrubia - Reply to reader
August 21, 2018 - 13:33

See attached file


Carmen Díaz-Roldán - Comment
July 31, 2018 - 15:38

This paper proposes a useful way for evaluating the optimal tax rate and efficiency cost of tax reforms. In this sense, contributes to the literature on taxation discussing issues that had not been addressed before
But given the peculiar fiscal framework of the Spanish regions, where the Basque Country and ...[more]

... Navarre have a different fiscal regime; it would be useful to explain whether the data used includes or not, data from these regions. As well as the implications, of using or not using such data, on the results obtained in this study.


Jorge Onrubia - Replay to reader (Carmen Diaz)
August 21, 2018 - 19:38

See attached file


Anonymous - Comments
July 31, 2018 - 17:17

This paper seeks to estimate the elasticity of taxable income (ETI) with respect to the marginal net tax rate for different groups of taxpayers by sex, marital status and age. Moreover, they jointly analyse the deadweight loss, the marginal cost of public funds and the optimal tax rates for personal ...[more]

... income taxes.

• Positive contributions:

1. The paper computes specific ETI for very relevant group of citizens, which help to improve the design of tax reforms with the aim of reducing their efficiency costs.

2. The paper applies different approaches to a very interesting case study, which has been highly discussed in the recent years.

3. Authors apply the econometric framework proposed by Gruber and Saez (2002), to separate substitution and income effects.

• Issues to be fixed/improved:

1. In my view, the main issue to be fixed/improved are the policy implications and suggestions, which we could extract from this empirical evidence. So that I consider it would be highly beneficial to expand the Conclusions section.

2. A final minor comment is that tables 5 and 6 should be included before Conclusions.


Jorge Onrubia - Reply to Anonymous Reader 2
August 21, 2018 - 14:17

See attached file


Carmen Marín González - Assessment
August 01, 2018 - 09:14

Assessment of “How do taxable income responses to marginal tax rates differ by sex, marital status and age? Evidence from Spanish dual income tax”. Economics, Discussion Paper No. 2018-48, June 12, 2018.

This paper estimates how the Spanish taxpayers have responded to the introduction of the dual Personal Income ...[more]

... Tax (PIT) model in 2007. The elasticity of taxable income (ETI) with respect to the marginal net tax rate is estimated. The authors have separated the substitution effect from the income effect. The ETI is computed for different groups of taxpayers by sex, marital status and age. Additionally the authors also computed the deadweight loss, the marginal cost of public funds and the optimal tax rates.

From my point of view, the paper is clearly written. The structure of the paper is well organized. The analysis presented is also well motivated in the introduction. The literature review commented is quiet broad. The authors include some sections to ease the reading of this paper. This is the case of the section 2, which introduces to non-Spanish readers the 2007 PIT Reform. The section 3.2 defines the deadweight loss, the marginal cost of public funds and optimal marginal tax rates that can help to non-expert readers. The section 4.1 explains the Spanish PIT return panel to non-users.

In order to improve this paper, my contribution to the discussion is referring to the results section.

• I have some doubts about the interpretation of this sentence; In page 13, paragraph 3: “On the contrary, the percentage of income accounted for the main income earner in the case of married couples has a negative impact”. However, from Table 1 this variable takes value 1 for unmarried individuals and married couples with only one wage earner.

• In Table 3, the taxpayer group can be single or married. My question is the treatment of individual and joint files.

• In Section 4.2.4. Optimal Tax rates: it is observed that the optimal marginal tax rate decreases whenever the brackets increase. For the bracket 1, the optimal marginal tax rate would be between 71% and 77%. For the bracket 2, would be around the 61% to 66%. Whereas both tax brackets 4 and 5 have an optimal tax rate of 24% to 26% and 27% to 29%, respectively. I think that the authors should explain why it is recommended to reduce the marginal tax rate of the brackets 4 and 5; meanwhile, the marginal tax rates of the brackets 1 or 2 should be unchanged. This is justified because the fiscal planner has redistributive and progressive goals, so the lower brackets should support a lower tax rate than the higher ones.


Jorge Onrubia - Reply to Carmen Marín (reader)
August 21, 2018 - 21:10

See attached file


Anonymous - Comments on How do taxable income responses to marginal tax rates differ by sex, marital status and age?
August 03, 2018 - 09:03

The paper explains the response of taxpayers to the duel PIT in Spain using the ETI approach. The story and rationale of the paper is easy to follow and it has a good structure.
With regards to the topic, it is a very relevant one for policy makers since ...[more]

... the is still a lot to research on the field of tapayers behaviour to tax reforms. For future research it would also be interesting to use DiD approach with a temporal threshold in the 2007 year.a
However, the methodology applied in the paper is innovative and the data, from the AEAT and IEF is of a very good quality.


Jorge Onrubia - Reply to Anonymous Reader 3
August 21, 2018 - 14:39

Thank you for your kind appreciation of our paper.

The proposal to use a "differences-in-differences" estimator is an option to be considered in this research area (see, among others, Feldstein, 1995, and Saez, 2003, cited in our paper). However, the characteristics of the tax reform analyzed do not ...[more]

... recommend its use. As the literature reviewed notes, the partition of the samples in the treatment and control groups, based on the alterations of the marginal tax rates, would affect practically all taxpayers, since the two tax structures compared are completely different for all of them.


Fidel Picos - Assesment of the paper
August 17, 2018 - 12:23

This paper by Díaz-Caro and Onrubia assesses behavioural responses to the introduction of a semidual Personal Income Tax (PIT) in Spain in 2007, by estimating elasticities of taxable income (ETI) with respect to marginal net tax rates. In my opinion it is a very good piece of research because: * ...[more]

... It takes a comprehensive approach, since the authors deal with the most relevant issues related to the research question.

* It uses an adequate methodology, adapted to the dual nature of the Spanish PIT.

I recommend the publication of this discussion paper as a journal article. I would just make one general comment and several minor ones:

* General comment: although the paper is clear, well organized and well written, I have the impression that it would benefit from a more policy-oriented drafting. Currently it shows a very complete set of results, but I think the paper would improve if the authors manage to build a “story” based on those results. In other words, I would ask for more interpretation of the results.

* Minor comments:

1. In footnote 3 the authors may specify who used quantile splines and why it did not improve the continuity of the distribution of taxable income.

2. It seems that in page 9, last paragraph, the authors do not mention the relevance of employment income, which may be the main source of total income.

3. In page 13, last paragraph of section 4.2.1, I do not understand very well the impact of self-employment income, especially in relation to objective estimation. Maybe the authors may elaborate more on this.

4. In page 14, first paragraph, the authors main explain in more detail how they identify retired and unemployed people in the sample.

5. I have found some minor mistakes and typos about which I can inform privately to the authors.


Jorge Onrubia - Replay to reader (Fidel Picos)
August 22, 2018 - 14:17

See attached file