America’s unhappy middle

Branko Milanovic unpacks the malaise of the US middle class and its implications for Democratic strategy towards the presidential election.

The voting behavior of the American middle and lower-middle class was, according to many, one of the reasons for the (unexpected) victory of Donald Trump as Republican candidate in the 2016 presidential election.

Various explanations have been offered for the malaise of the American middle class: prolonged lack of income growth, combined with massive growth at the top, which has made the distance between the top 1 per cent and the middle class greater and at times seemingly unbridgeable; an education system with high and unaffordable tuition fees, limiting opportunities to children from modest parental backgrounds; outsourcing of ‘good’ jobs to China and other emerging economies; technological change that has made many ‘routine’ middle-class jobs redundant. A recent book by Anne Case and Angus Deaton condenses all these factors into the dismal health statistics of ‘deaths of despair’.    

One way to check whether there was indeed something unusual in the voting behaviour of the American middle class in 2016 is to compare growth rates of real, after-tax income of different income groups during the period after the Global Financial Crisis, which broadly corresponds with Barack Obama’s two presidencies, and the percentage of votes for Hillary Clinton, the Democratic candidate to be his successor.

Figure 1 shows real per capita income growth between 2007 and 2016 for various fractiles (parts) of the US income distribution. The poorest groups are on the left: decile one stands for the 10 per cent of the poorest Americans; decile two for the second poorest 10 per cent of Americans and so on, all the way up to the top 5 per cent and the top 1 per cent (at the extreme right).

The data demonstrate that, with the exception of the two bottom deciles, US growth was ‘pro-rich’: incomes grew at higher rates for the rich than for the middle class. The story of economic growth under Obama was thus not favourable to the middle or the lower-middle class; the real winners were those at the top of the income distribution, the top 5 per cent, the top decile and even the two highest deciles—the ‘dream hoarders’ described by Richard Reeves in his 2017 book.

Figure 1: cumulative rate of growth of real after-tax per capita income, 2007-16 (%), by income fractile

Figure 2, recently published by Thomas Piketty in his Capital and Ideology and derived from voting surveys, shows the pro-Clinton vote in 2016, arrayed against the same axis of income distribution as in Figure 1. The overall shape of the growth curve is similar: support for Clinton was relatively high and stable among the bottom 30 per cent of the population, dramatically declined throughout the middle of the income distribution to reach its trough around the seventh decile but steadily rose afterwards towards higher-income households. Indeed, for the very top 1 per cent, pro-Hillary support was slightly down, compared with the top 5 per cent, exactly in line with the evolution of the growth rate.

Figure 2: vote for the Democratic candidate (%) in 2016 by income fractile

An interesting difference however emerges on closer inspection. I have put the two graphs together to compare the patterns more precisely. I have normalised both figures by the results for the top 1 per cent (in terms of income and voting behaviour). The relative Clinton votes (blue) are shown on the left vertical axis; the relative growth rates (brown) on the right vertical axis.

What is immediately apparent is the anomalous behaviour of deciles four through seven: while their real income growth improves compared with the lower (poorer) deciles, the pro-Hillary vote declines. This cross-movement of growth and votes ends around the eighth decile, where we find a movement in the two variables that we expect—higher support for the Democratic candidate being associated with better economic outcomes during the previous eight years of the Democratic administration.

Figure 3. Real income growth and pro-Clinton vote

Thus even at such an aggregate level we do retrieve evidence of unexpected behaviour among the middle deciles—some 30 to 40 per cent of the US electorate who, when compared with those poorer than themselves, experienced better economic outcomes but voted much less for the Democratic candidate.

Now, whether people who are in the middle are exactly those mentioned at the outset as having been adversely affected by the financial crisis, globalisation, outsourcing and/or technological change, we do not know. It seems however reasonable to link the two trends.

For individuals in these strata to become more Democrat-leaning, then, it may not be just economic growth (under a Democratic administration) that matters but something more. Whether it is less job uncertainty, more accessible schools for their children or healthcare for themselves, or something else, a rough analysis like this cannot tell.

But it seems that the reductionist adage ‘It’s the economy, stupid’, associated with the 1990s presidential electoral victories of Clinton’s husband, Bill, cannot explain all—at least not any more: wider social-policy considerations need to come into play. Which perhaps Joe Biden, should he receive the Democratic nomination as expected, might take into account as the 2020 election approaches.

This article is a joint publication by Social Europe and IPS-Journal. See our series of articles on the US Election 2020