Discussion of Global Absolute Intergenerational Mobility by Yonatan

Discussion of Global Absolute Intergenerational Mobility, by Yonatan Berman (Paris School of Economics, France Increasing Inequality in Joint Income and Wealth Distributions in the United States, 1995 to 2013, by Louis Chauvel, Eyal Bar-Haim, Anne Hartung (University of Luxembourg) & Philippe van Kerm (University of Luxembourg & LISER) By Tim Smeeding Lee Rainwater Distinguished Professor of Public Affairs and Economics Research | Training | Policy | Practice

How’s your math ? --better than mine • Both papers ‘magical’ in the sense that a few parameters, a few assumptions, some fancy math and we can tell you almost everything • I will try my math, then see if I am right ? • I will look at actual real data and add some intuition, I hope • I invite others to join in , please

‘Global Absolute Intergenerational Mobility ‘--Yonatan • Expands the estimation of absolute mobility trends to several developed and developing countries and to the global level---- WOW-HOW? • Combines the marginal income distributions for parents and children and their copula, i. e. the joint distribution of parent and child income ranks • Reduced-form statistical model captures the relationship between absolute mobility, income growth, income inequality and relative intergenerational mobility”.

More claims- • “. . offer a theoretical study of the relationship between the canonical measures of intergenerational income mobility and show empirically that a simple model of a bivariate log-normal income distribution can describe adequately the long run dynamics of absolute mobility” • Shows that absolute and relative mobility are inversely related

How do you do all this with so much missing data ? • Assume we have a can opener? -- Figure 2: --gives the relationship between relative mobility indices in empirical copulas – to argue that the marginal income distributions and a single observation of a relative mobility measure, such as the rank correlation, can provide robust estimates of absolute mobility within various countries. ? ? --and in so doing, he infers within country behavior from cross country evidence , as in the Gatsby curve ? ?

Gatsby: cross country = within country ? ? - World Bank style

Gimme more- • “assume that the rank correlation in each country is fixed in time --as long as the income growth and inequality among 30 -year-olds is similar to that of the entire adult population, absolute mobility estimates will be very close even if the entire adult population is considered” ? • “Overcoming data limitations only requires acquiring the marginal distributions for 30 -year-olds only, while also capturing well the top of the distribution” ? • ? ? Facts-- age income profiles by education are not constant over time , so lines will cross after age 30 and rank order is not preserved

More- • Although Chetty et al. (2014 a, p. 1574) argued that “the income distribution is not well approximated by a bivariate log-normal distribution”, we find that the decline in absolute mobility in the United States is ” ? • “While such a simplified model may not explain adequately some aspects of mobility and inequality, for the purpose of estimating long run absolute mobility trends, this model is satisfactory “ ? ? • Yonatan--“despite the inverse relationship described, in practice, as also observed in Section 2, the effect of changes in relative mobility on absolute mobility, is usually minor. ” • Is that true ?

Strong ending claims • “using widely available data on marginal income distributions and limited data on relative intergenerational mobility, it is possible to produce estimates of absolute intergenerational mobility without the need for high-quality panel data sets—by assuming a fixed copula in time – to provide meaningful and reliable estimates of absolute mobility. • We also find that a model as simple as a bivariate log-normal distribution is satisfactory for describing the dynamics of absolute mobility with high confidence.

And ta da- • --”absolute mobility decreases with both increasing inequality and decreasing growth, but increases when relative mobility is low ” • --‘absolute mobility is very sensitive to acrossthe-board economic growth’ • Ok--lets have an intuitive look at what Yonatan is talking about, just by looking at the data

Income Inequality Trends : 1978 -2014: USA, China, France; top 1% shares

Income Inequality Trends : 1978 -2014: USA, China, France; top 10% shares

Income Inequality Trends : 1978 -2014: USA, China, France; bottom 50% shares

Growth and inequality winners and losers : compare China, USA, France Source, WTID, http: //wid. world/

Bottom Line • I can find some intuitive sense of a tradeoff between absolute inter-generational mobility and inequality due to economic growth • But different points in the distribution experience different rates of growth-- does that mean relative mobility is inversely related to absolute mobility and that absolute mobility is unstable ? • If you are right– AER lead article- not JID ! • HELP-Markus, especially

“Increasing Inequality in Joint Income and Wealth Distributions in the United States, 1995 to 2013” by Louis and company • “ we propose a new method that is able to provide a thorough examination of tails: the isograph and the logitrank ”. • “—we find income inequality increased significantly; the wealth- to-income ratio measuring the importance of wealth relative to income, increased significantly; the association between high wealth and high incomes increased as well”

Some buts- • “Wealth may be buffering or delaying the repercussions of a shrinking middle (income) class. As inheritance plays a key role wealth inequality has also implications for intergenerational equity. ” • BUT Inter-vivos transfers are much more important than inheritance in determining future life chances and dynastic behavior

Another one- • “With the observed trends of the U. S. isograph, we detect now a distribution comparable to a Zipf curve (Gabaix, 2009), exceptional by the intensity of its inequality in the set of LWS countries”. • Of course--other LWS countries do not have the upper half—only the SCF ( why PSID is there in LWS too)

What do you do, Louis ? • Zipf curves; isographs; “leptokurticity of wealth “– and more that I have never heard before (– I am too simple, I admit ) • Throw way the bottom of the distribution as uninteresting ? • “At the very top, the increase in the W/I ratio seems unsignificant “ • Lets look and see what we get—again, JUST BY LOOKING

Luckily—I also use this very same data, to study the same phenomon “Inequality in 3 -D: Income, Consumption, and Wealth” • • Jonathan Fisher, Stanford University David Johnson, University of Michigan Timothy Smeeding, University of Wisconsin Jeffrey Thompson, Federal Reserve Board • https: //www. federalreserve. gov/econres/feds/ files/2018001 pap. pdf

Top 10 percent INCOME share ) (source http: //elsa. berkeley. edu/~saez/Tab. Fig 2015 prel. xls Note 2015 exceeds 1927 for all time high 45% 40% 35% 30% Including capital gains 2012 2007 2002 1997 1992 1987 1982 1977 1972 1967 1962 1957 1952 1947 1942 1937 1932 1927 1922 25% 1917 Top 10% Income Share 50%

The distribution of family WEALTH is growing far more unequal Source : SCF at http: //apps. urban. org/features/wealth-inequality-charts/

1 -D inequality: comparison of share held by top 5%

2 -D measures of inequality 1) Percent of households in the top 5% of: a) Income and Consumption b) Income and Wealth c) Wealth and Consumption 2) Share of Income by top 5% of: a) Consumption distribution b) Wealth distribution 3) Share of Consumption by top 5% of: a) Income distribution b) Wealth distribution 4) Share of Wealth by top 5% of: a) Income distribution b) Consumption distribution

2 -D inequality: Top 5% shares in two dimensions (1989=100)

2 -D measures of inequality 1) Percent of households in the top 5% of: a) Income and Consumption b) Income and Wealth c) Wealth and Consumption 2) Share of Income by top 5% of: a) Consumption distribution b) Wealth distribution 3) Share of Consumption by top 5% of: a) Income distribution b) Wealth distribution 4) Share of Wealth by top 5% of: a) Income distribution b) Consumption distribution

2 -D inequality: Top 5% shares in two dimensions – Share of wealth

3 -D inequality: Percent of households in top 5% of income, consumption, and wealth

Y= C +/- change NW 1989 -2016 • Measures of one-dimensional inequality understate the level of inequality today and the growth in inequality since 1989 • Inequality in income (Y), consumption (C )and wealth (or net worth, NW) all rising separately • Inequality in any two dimensions increased faster than in any one dimension • Inequality in all three dimensions together rose by the most • What fraction of all households that were in the top 5% of the income (Y) distribution , were also in the top 5% of the consumption (C) distribution and the top 5% of the wealth (NW) distribution year by year ? In 1989— 32 % ; by 2007 it was 49%-In 2016 , 44% ( but still growing rapidly since)

And, all this high and rising wealth inequality is understated ! • SCF ( best survey we have) shows wealth as of March 2016—since then, markets up 30+ % • Who wins ? --The “change in net worth” folks, and this change does not show up in household income data until it is realized • Most stocks and financial wealth including defined contribution pension plans are owned by the top decile ( about 75 %) • E. g, 1998 and 2017, two “very good years”

Bottom Line • Louis et al right about the USA • BUT do we need all the hoopla or can we just look at the data?

But also consider--global inequality —the elephant’s nose