Category Archives: income inequality

23/6/21: Covid19 Deaths and Income Inequality


An interesting, although intuitively straight forward note on the determinants of Covid19 deaths:

As Youyang Gu @youyanggu states, "I believe income inequality is the single best predictor of total Covid deaths in the US. Not income, but income *inequality*. The R^2 is surprisingly high: 0.35."

There are some potentially important issues with this analysis (some are explored here:, but the conclusion seems to be qualitatively robust. 

23/7/20: Globalization and Populism: A Recent Study

I recently came across a fascinating paper by Dani Rodrik, an economist always worth reading. The paper, titled "Why Does Globalization Fuel Populism? Economics, Culture, and the Rise of Right-wing Populism" (NBER Working Paper No. 27526, July 2020) argues that "there is compelling evidence that globalization shocks, often working through culture and identity, have played an important role in driving up support for populist movements, particularly of the right-wing kind."

Rodrik carries out "an empirical analysis of the 2016 presidential election in the U.S. to show globalization-related attitudinal variables were important correlates of the switch to Trump."

  • "Trump voters were more likely to be white, older, and college-educated. 
  • "...they were significantly more hostile to racial equality and perceived themselves to be of higher social class. 
  • "The estimated coefficient on racial attitudes is particularly large: a one-point increase in the index of racial hostility – which theoretically ranges from 1 to 5 – is associated with a 0.28 percentage point increase in the probability of voting for Trump (Table below, column 1). 
  • "By contrast, economic insecurity does not seem to be associated with a propensity to vote for Trump.

"The finding that Trump voters thought of themselves as belonging to upper social classes ... largely reflects the role played by party identification in shaping voting preferences. When we control for Republican party identification (cols. 2 and 6), the estimated coefficient for social class drops sharply and ceases to be statistically significant."

"Note, however, that racial hostility remains significant, although its estimated coefficient becomes smaller (cols. 2 and 6)."

The other columns in the table above examine attitudes towards globalization (columns 2-5).

  • "All three of our measures enter statistically significantly: 
  • "Trump voters disliked trade agreements and immigration; 
  • "They were also against bank regulation (presumably in line with the general anti-regulation views of (cols. 2-5) the Republican party). 
  • "These indictors remain significant in the kitchen-sink version where they are all entered together (col. 6)."

"In none of these regressions does economic insecurity (financial worries) enter significantly. This
changes when we move from Trump voters in general to switchers from Obama to Trump (cols. 7-12). ... financial worries now becomes statistically significant, and switchers do not identify with the upper social classes. "

"Switchers are similar to Trump voters insofar as they too dislike trade agreements and immigration
(cols. 9-11). But they are dissimilar in that they view regulation of banks favorably. Hence switchers
appear to be against all aspects of globalization – trade, immigration, finance. the regression."

Rodrik postulates "a conceptual framework to clarify the various channels through which globalization can stimulate populism" on both "the demand and supply sides of politics". He also lists "the different causal pathways that link globalization shocks to political outcomes". 

Rodrik identifies "four mechanisms in particular, two each on the demand and supply sides:

  • (a) a direct effect from economic dislocation to demands for anti-elite, redistributive policies; 
  • (b) an indirect demand-side effect, through the amplification of cultural and identity divisions; 
  • (c) a supply-side effect through political candidates adopting more populist platforms in response to economic shocks; and 
  • (d) another supply-side effect through political candidates adopting platforms that deliberately inflame cultural and identity tensions in order to shift voters’ attention away from economic issues."

The full paper, accessible at is choke full of other insights and is absolutely worth reading.

13/6/2020: What Do Money Supply Numbers Tell Us About Social Economics?

What do money supply changes tell us about social economics? A lot. Take two key measures of U.S. money supply:

  • M1, which includes funds that are readily accessible for spending, primarily by households and non-financial companies, such as currency outside the U.S. Treasury, Federal Reserve Banks, and the vaults of depository institutions; traveler's checks; demand deposits; and other checkable deposits. 

  • MZM, which is M2 less small-denomination time deposits plus institutional money funds, or in more simple terms, institutional money and funds available for investment and financial trading.
Here we go, folks:

Does this help explain why Trumpism is not an idiosyncratic phenomena? It does. But it also helps explain why the waves of social unrest and protests are also not idiosyncratic phenomena. More interesting is that this helps to explain why both of these phenomena are tightly linked to each other: one and the other are both co-caused by the same drivers. If you spend a good part of 20 years pumping money into the Wall Street while largely ignoring the Main Street, pitchforks will come out. 

The *will* bit in the sentence above is now here.

Assortative Mating and Math on "The Bachelor"

Scandal erupted on the Australian edition of The Bachelor when the Bachelor, 32-year old astrophysicist Matt Agnew, asked 28-year old chemical engineer Chelsea McLeod to solve math problems in order to find the combination numbers of a safe on their third and final date before the show's conclusion.

We're not making this up. Here's a video excerpt from the show:

The task greatly upset the show's fan base. Here's the story from the Daily Mail:

Since day one, Bachelor Matt Agnew has embraced the fact Chelsie McLeod shares his same passion for science.

But fans were left confused during the pair's final date on Wednesday's semi-final, as the astrophysicist, 32, tasked the chemical engineer, 28, with math equations.

'This would literally be my worst nightmare!' one fan Tweeted, after Matt described the maths problem as a 'fun activity' for Chelsie to complete.

'I cannot believe you're making me do this right now!' Chelsie said to Matt, which viewers appeared to agree with.

Matt handed Chelsie a pen and paper to solve a math problem to open a safe, which housed a present for the star.

At least it ended well. Chelsie correctly solved the math problems, determined the combination to the safe, and revealed the prize:

After cracking equations, Chelsie typed in the code which opened to a box with a necklace engraved with the chemical formula for oxytocin.

The formula was a reference to their first meeting on the red carpet where she gave him a temporary tattooed of the same chemical.

'So you got to put some oxytocin on my chest, to my heart, and this way, the necklace, it'll be close, some oxytocin near your heart as well.

It's probably no surprise that the two ended up together in the aftermath of the show's suitably romantic final rose ceremony.

But more than just a triumph of math and love, the episode demonstrates assortative mating in action, where couples come together because of their shared potential in addition to their shared attraction. Which in the case of their shared potential, more often than not also reflects their shared earning potential.

But since most couples don't have the accelerated contrivances of a televised dating show to utilize during their courtships, how does that tend to work in the real world?

Alparslan Tuncay described how couples made up of individuals with similar earnings potential come often come together in a paper Marginal Revolution's Tyler Cowen described as "the best results on assortative mating and inequality I have seen". Here's the abstract:

This paper studies the evolution of assortative mating in the permanent wage (the individual-specific component of wage) in the U.S., its role in the increase in family wage inequality, and the factors behind this evolution. I first document a substantial trend in assortative mating, as measured by the permanent wage correlation of couples, from 0.3 for families formed in the late 1960s to 0.52 for families formed in the late 1980s. I show that this trend accounts for more than one-third of the increase in family wage inequality across these cohorts of families. I then argue that the increase in marriage age across these cohorts contributed to the assortative mating and thus to the rising inequality. Individuals face a large degree of uncertainty about their permanent wages early in their careers. If they marry early, as most individuals in the late 1960s did, this uncertainty leads to weak marital sorting along permanent wage. But when marriage is delayed, as in the late 1980s, the sorting becomes stronger due to the quick resolution of this uncertainty with work experience. After providing reduced-form evidence on the impact of marriage age, I build and estimate a marriage model with wage uncertainty and show that the increase in marriage age can explain almost 80% of the increase in assortative mating.

Marrying later contributes to this outcome because it allows the each partner in a couple the time to demonstrate their earning capacities, which would either reinforce the pairing or lead to a break up if the two are incompatibly mismatched. Tuncay's findings confirm what we've seen in the data for income inequality in the United States over that period of time, where there has been no change in the income inequality level for individuals, but a rising trend for both households and families as a greater emphasis on college education and early career establishment before marriage has become common in American society since the 1960s.

Assortative mating provides a very reasonable explanation for these outcomes. And we've just seen it at work, in all its oxytocin-enhanced sweetness, on The Bachelor Australia.

1/4/19: Hollowing out of the American Middle Class: the High Earners, and the 1-percenters

An interesting and insightful 2016 paper from John Komlos of CESIfo, titled "Growth of Income and Welfare in the U.S. 1979-2011" (CESifo Working Paper No. 5880), paints the pretty dire picture of the post-1980s dynamics in the U.S. labor markets, that laid the foundations of the current acceleration and deepening of political populism and opportunism not only in the U.S., but also in Western Europe.

Kolmos estimated growth rates in real incomes in the U.S. from the Congressional Budget Office’s (CBO) post-tax, post-transfer data. Kolmos also adjusts the real income data to improve the accuracy of the measures. The result is striking: "... the major consistent findings include what in the colloquial is referred to as the “hollowing out” of the middle class. According to these estimates, the income of the middle class 2nd and 3rd quintiles increased at a rate of between 0.1% and 0.7% per annum, i.e., barely distinguishable from zero. Even that meager rate was achieved only through substantial transfer payments." Of course, given that we have experienced positive growth in the aggregate economy in excess of these figures and well above the demographic change, this "hollowing out" of the middle class had to be accompanied by the "fattening up" of some other income classes, either the rich or the poor or both. Per Kolmos, it was the former one: "the income of the top 1% grew at an astronomical rate of between 3.4% and 3.9% per annum during the 32-year period, reaching an average annual value of $918,000, up from $281,000 in 1979 (in 2011 dollars)." Predictably, "...the post-tax, post-transfer income of the 1% relative to the 1st quintile increased from a factor of 21 in 1979 to a factor of 51 in 2011."

But what about the poor? Again, per Kolmos, "...income of no other group increased substantially relative to that of the lowest quintile. Oddly, the income of even those in the 96-99 percentiles increased only from a multiple of 8.1 to a multiple of 11.3."

Kolmos id this exercise for 'high' and 'low' ranges of income (depending on specific assumptions that were less and more conservative ratline to the CBO's raw data.

The results of the two calculations are shown in the chart below

Source: Kolmos (2016: 14)

In simple terms, this chart shows two interesting things:
1) The dramatic growth differential between income estimates for all quintiles compared to the top quintile is fully accounted for by the massive growth in income of the top 5% of the populations and especially by the growth in income of the top 1%.
2) The gap between high and low estimates for income growth are massive for the second and third quintiles (the middle class), and are relatively comparable for the first (low income earners) and 4th quintile (upper middle class). The gap becomes much smaller for the 5th quintile (high earners) and turns negligible for top 1%.

Kolmos attempts to convert income into more meaningful 9albeit harder to pin down) measure of well-being. To do this, he estimates the logarithmic utility function for the quintiles (logarithmic utility function preserves the property of the diminishing marginal utility - the idea that as our incomes continue to increase, each percent increase in our income results in progressively smaller gains in satisfaction/utility). Here is what he finds: "A logarithmic utility function yields a growth in welfare for the middle class of roughly 0.01% to 0.07% per annum, which is indistinguishable from zero. With interdependent utility functions only the welfare of the 5th quintile experienced meaningful growth while those of the first four quintiles tend to be either negligible or even negative." Chart below shows these estimates.

Source: Kolmos (2016: 15)

Focusing on the Percentiles section, markers 6-9 disaggregate the last 5th quintile into the ranges of top 81-90%, 91-95%, 96-99% and top 1%. It is quite evident that only top 5% (segments 8 and 9) experienced welfare gains of more than the 4th quantile cohorts.

This strongly implies that, contrary to some left-leaning policymakers' proposals and preferences, the problem of 'hollowing out' of the American middle class is not driven by the incomes of the top 81-90th percentiles, nor even by those in 91st-95th percentiles. The real source of the problem starts somewhere within the 96-99th percentile and most certainly extends to the top 1%.

The same is confirmed by looking at each cohort income relative to that of the top quintile, shown in the chart below

Source: Kolmos (2016:27)

In summary, thus, the problem with the 'hollowing out' of the middle class is not within theta 20% earners, nor within the top 10% earners. It starts much higher than that.