Mapping Income Inequality in the US Using IRS SOI Data

States with a higher proportion of dual similar-income joint filers with fewer children might hit the penalty more often (like a 2 doctor or lawyer household), and states with higher proportions of single or highly disparate income joint filing households (with a stay-at-home or marginally employed spouse) with more children would hit the bonus more often.

But as you mention that effect is usually dominated by the effect of personal exemptions, and not by the fact that tax bracket thresholds between individual and joint filers don't line up perfectly while normalizing for number of filers.

Correction: the marriage penalty/bonus has more to do with the disparity in spousal income than magnitude.

1. https://www.investopedia.com/terms/m/marriage-penalty.asp

Concretely, if my wife and I decide to file "married filing separately", we become two returns and income inequality is reduced in the analysis (but not at all in the real world).

(I now see how “per capita income” led you to the conclusion you reached. That’s my bad.)

If by eliminate, you mean equalize the inequality - then the remaining 99% will contribute more as they'll be making more money...

Sounds great. I’d rather have Nordic-style social democracy.

Suppose you have a well run society with good safety nets, high standard of living and 'investment in innovation' which leads to some great startups and a host of wealthy people. Well, you have 'inequality' ... but is that bad? Some might argue that, but I'm not so sure.

Now you have another state with bad social conditions, and the wealth generated is mostly through aggressive anti-union tactics, low wages, and power leveraging etc.. This is generally the kind of 'inequality' that nobody likes.

So in the former case, we want to see surpluses from such innovation 'help everyone' and they usually do: for example Netflix for $10/month is one of the most massive consumer surpluses imaginable, it's such a great deal for consumers. Innovators usually only capture a small chunk of the surpluses they create.

So while we can still argue that 'asset/income inequality' is a problem in the former case, it becomes more difficult.

But it gets worse. I hinted at 'consumer surpluses' which are the 'profits' individuals make when the exchange money for something more valuable than that money i.e. the thing they buy. For most people, Netflix is worth vastly more than $10/month. The delta between 'that upper range value' and $10, is called the 'consumer surplus' and it's never measured! Essentially, societies can get really rich, without any monetary measure of such wealth at all!

Suppose the founders of Netflix, and employees weren't making that much money ergo you maintain basic asset/income equality with the rest of the population. But regions that have access to Netflix are materially wealthier than regions that don't have access to Netflix - because they can watch 100's of movies on demand for dirt cheap - but again, that differential in surplus is not measured at all, it's nowhere to be seen in the GDP or on our books! (Caveat: we 'kind of measure' the increase in real value of a product with inflation. If tomatoes one year are 'riper, fresher, redder' than the previous year, they try to account for that. But for most products it's just impossible).

A more extreme case is vaccines. A nation with strong vaccination stays healthy, a nation without gets sick and that direct measure isn't really part of the GDP. In fact the sicker nation might require more healthcare spending ... which increases the GDP in a really perverse way. Due to the odd artefact that vaccines are priced irregularly (people will only pay a few dollars for a vaccine which could save their lives, over $100's of thousands for a cure once they have the disease), they don't work well on the free market: it makes more sense to make expensive cures than cheap preventative vaccines, even though in the later case the surpluses are much, much more vast.

Raw measures of inequality are only rough data points and they really don't say a whole lot.

We need to understand more qualitatively how that wealth is distributed, how fairly, and the kinds of surpluses being generated by those in higher income brackets: are they just capturing rent on labour, or are they really 'growing the pie'.