Friday, March 16, 2018

36 Years of Inequality or: How I Learn to Stop Worrying and Love the Progressive Income Tax

This post continues from the previous post. In the previous post, I wrote that one of the possible remedies for inequality is progressive taxation. Thomas Piketty advocates an extreme form of progressive income tax, up to 80% tax rate for the ultra-rich.

How could it be justified for Indonesia? Credit Suisse's Global Wealth Report 2017 shows that the top 1% richest Indonesians own 45.4% of national wealth, and top 10% richest own 74.8% of national wealth. If this does not seems obscene to you, take a look at the chart here:



The chart above shows Material Power Index (MPI), one of inequality measure developed by Winters (2013). By measuring the wealth of top 40 richest and divide it by per capita GDP, MPI approximates the relative economic imbalances between the ultra rich and the rest of us.

The MPI chart shows that the Indonesian ultra rich are about 500.000-600.000 times richer than average Indonesian. If we can scale the wealth down, theirs (on average) are worth a unit of apartment. The rest of us are (on average) worth a cup of instant coffee. Not even a cup of Starbucks.

Such obscenity probably warrants a more drastic measure, as what Piketty recommends. Although, income tax rate of 80% for the ultra rich is most likely not feasible politically.


The question, then, is whether a more progressive income tax may lead to a reduction in inequality if any, despite not being as high of a rate as what Piketty recommends.

Given that question, it is interesting to see how Indonesia fares.

On Personal Income Tax and Measurement of Its Progressivity

The first caution to be understood is that Indonesia does not have progressive corporate income tax since 2008. Added to that complication are certain rules such as turnover-based single-rate tax (final income tax according to Government Regulation no. 46 Year 2016) and the 50% discount for corporate taxpayer whose turnover does not exceed 50 billion rupiah (under Article 31E of Income Tax Law).

Secondly, the ultimate beneficiary of corporate profit is not the corporation itself, but the people associated with it: workers, managements, creditors, and shareholders – in the form of bonuses, interests, dividends, and/or capital gains. (Hence the economic double taxation that occurs when we tax dividends.) So, even though corporation could consume parts of its profit to expand, it could not be construed that corporation “enjoys” the profit in the traditional sense.

Additionally, in the event of liquidation, the excess value of the company will be distributed to those aforementioned people. Whereas, when a person dies, she can bequeath her excess value in the form of inheritance or estate. Both forms of distribution may indeed sustain or even increase inequality, but such effect can only manifest in personal level instead of corporate.

In light of those difficulties, it is more appropriate to measure the progressivity of personal income tax instead. There are several metrics to measure progressivity of an income tax regime. For example, the top statutory tax rate and the rate in each income bracket could be used roughly (at a glance) as indicators of how progressive the income tax is. US, whose top personal income tax rate of 39.6% applies for taxable income of more than $415,000, could be seen as less progressive than Australia, whose top personal income tax rate of 45% applies for taxable income of more than AUD 180,000.

A more refined measure is outlined by, inter alia, Benabou (2002) and Sabirianova Peter, et al. (2010), in which the latter will be used here. Sabirianova Peter, et al. (2010)'s methodology is as follows: first, we obtain per capita GDP of a country in that year. Next, we create a 100-level income distribution, ranging from 4-400% of the aforesaid per capita GDP. This results in 100 different gross income levels. Then, we employ the relevant tax schedule, i.e. apply the standard deduction for single taxpayer and appropriate tax rate. This will result in the tax liability for that particular income level. Finally, we regress the tax liability to the gross income to obtain the regression coefficient. Such coefficient, called ARP (average rate progression) is our progressivity measure of personal income tax.

If the ARP is zero/statistically insignificant, then the personal income tax is neutral. If the ARP positive and statistically significant, that means the personal income tax is progressive. Conversely, negative and statistically significant ARP means the personal income tax is regressive.

Here how the ARP of Indonesia looks like for the years 1981-2016:



The Effect on Inequality

If we graph Gini ratio and our income tax progressivity measure ARP, it would look like the chart below. I plot ARP on secondary axis because the difference in value range (ARP range from 0.01-0.03 while Gini range from 0.3-0.4) makes it visually difficult to see their fluctuations.


We can visually suspect that in the last 36 years (1981-2016) there seems to be a relationship between Gini and progressivity of income tax.

To gauge the effect of progressive personal income tax to inequality, I follow the methodology of Duncan and Sabirianova Peter (2012). I use Instrumental Variable (IV) estimation using ARP as the independent variable and Gini ratio as the dependent variable.

Here ARP is assumed to be endogenous, either due to simultaneity problem (see Slemrod and Bakija 2000) or by the existence of confounding variable. If we use OLS (Ordinary Least Squared), the estimates will be biased. So, I estimate ARP using several IVs, such as inflation, interest rate spread, exchange rate depreciation against USD, population, Freedom House index of civil liberties and political rights, and GDP per capita. Some variables are omitted (such as religion and corruption) because they are either irrelevant in non-cross country comparison or having insufficient data.

Based on the IV estimation of the data from 1981-2016, ARP (progressivity of personal income tax) is negatively correlated with Gini ratio (p-value: 0.013; 95% CI: -2.814528 to -0.33558101). To paraphrase: it is likely that the more progressive personal income tax, the lower income inequality will be. To make it easier to understand the result, I presented it in the graph form below:


Conclusions

We can see that as the value of progressivity get bigger (x-axis), Gini ratio is declining (y-axis), i.e. the more progressive personal income tax is, the lower the Gini ratio is. So, it may be the case that progressive personal income tax can reduce income inequality in Indonesia

How should Indonesian government proceed? I personally think that Indonesian government is somewhat ambivalent in this matter. You want to reduce inequality, which necessitates a transfer of wealth by means of taxation. But on the other hand, you do not want to upset the capital owners; the ones which – as the r > g theory suggests – exacerbate inequality.

80% tax rate for the ultra rich is good, if what Indonesian government aims is drastic reduction in inequality. The second best alternative is to gradually increase tax rate, to make it more progressive, and then hoping that inequality does not creep up faster than our tax reform.

References:

Benabou, Roland. 2002. "Tax and Education Policy in a Heterogeneous Agent Economy: What Levels of Redistribution Maximize Growth and Efficiency?" Econometrica 70, 481–517

Husain, Ishrat and Ishac Diwan [eds]. 1990. Dealing with the Debt Crisis. Washington, DC : The World Bank.

Sabirianova Peter, Klara, Steve Buttrick, and Denvil Duncan. 2010. “Global Reform of Personal Income Taxation, 1981-2005: Evidence from 189 Countries.” National Tax Journal, 63(3): 447–478.

Duncan, Denvil and Klara Sabirianova Peter. 2012. Unequal Inequalities: Do Progressive Taxes Reduce Income Inequality? IZA Discussion Papers, No. 6910

Sargan, J. D. 1958. The Estimation of Economic Relationships Using Instrumental Variables. Econometrica 26: 393-415.

Solt, Frederick. 2016. “The Standardized World Income Inequality Database.” Social Science Quarterly 97(5):1267-1281.

Slemrod, Joel B. and Jon Bakija. 2000. Does Growing Inequality Reduce Tax Progressivity? Should it? NBER Working Paper No. w7576. Available at SSRN: https://ssrn.com/abstract=220033

Winters, Jeffrey A. 2013. Oligarchy and Democracy in Indonesia. Indonesia (96), 11-33. doi:10.5728/indonesia.96.0099

Wooldridge, Jeffrey M. 1995. “Score Diagnostics for Linear Models Estimated by Two Stage Least Squares”, in Advances in Econometrics and Quantitative Economics: Essays in Honor of Professor C. R. Rao, ed. G. S. Maddala, P. C. B. Phillips, and T. N. Srinivasan, 66-87. Oxford: Blackwell.

Data Source:
MPI: author's calculation based on Winters (2013)
Gini ratio: Solt (2016), Badan Pusat Statistik
Progressivity (ARP): author's calculation based on Sabirianova Peter, et al. (2010)
Civil liberties and political rights: Freedom House
Interest rate spread (interest on loans minus interest on deposits): Husain and Diwan (1990), World Development Indicator
Inflation (end of period): IMF World Economic Outlook
Per capita GDP (t-1, logged): IMF World Economic Outlook
Population (total, logged): IMF World Economic Outlook
Exchange Rate (Currency annual depreciation rate with respect to USD, end of period): IMF International Finance Statistics

------------------------------------------------------

Appendix

ARP is correctly identified as endogenous using Woolridge (1995) robust regression-based test (p = 0.0040). However there may be misspecification in the equation, based on the result of Sargan (1958) test of overidentifying restriction (p = 0.0224).

Further, I wish to re-calculate the confidence intervals assuming weak instrumental variable. The results, using conditional likelihood ratio (CLR), Anderson-Rubin, Wald, and Lagrange’s K and J tests, are as follow: