Tuesday, November 08, 2016

High-end inequality colloquium, week 3, Ilyana Kuziemko

Yesterday Ilyana Kuziemko came by to discuss her co-authored paper, Support for Redistribution in an Age of Rising Inequality. As background, here is the paper's abstract:

"Despite the large increases in economic inequality since 1970, American survey respondents exhibit no increase in support for redistribution, in contrast to the predictions from standard theories of redistributive preferences. We replicate these results but further demonstrate substantial heterogeneity by demographic groups. In particular, the two groups who have most moved against income redistribution are the elderly and African-Americans, two groups relatively more reliant on it. We find little evidence that these subgroup trends are explained by relative economic gains or growing cultural conservatism, two common explanations. We further show that the elderly trend is uniquely American, at least relative to other developed countries with comparable survey data. One story consistent with the data on elderly trends is that they worry that redistribution will come at their expense, in particular via cuts to Medicare. We find that the elderly have grown increasingly opposed to government provision of health insurance and that controlling for this tendency explains roughly half of their declining relative support of redistribution. For blacks, controlling for their declining support of race-targeted aid explains a large portion of their differential decline in redistributive preferences (raising the question of why support for race-targeted aid has fallen during a period when black income catch-up to whites has stalled)."

These are interesting and credible findings. Herewith an expanded version of my thoughts regarding the paper:

Two discussion topics.  First, as per the paper, why didn’t rising U.S. inequality from 1978-2006 prompt an increased demand for redistribution?  Second, what are the lessons learned regarding the future?

1.         Why no increased demand for redistribution?
a) Is there a paradox?
The paper notes that, under the “workhorse political economy model” in which voters simply follow their own narrow economic self-interest, it would be paradoxical to find that rising inequality didn’t trigger increased support for redistribution.  The model – which, happily, the authors, no less than I, regard as a useful strawman rather than something that is actually credible – posits that each voter’s support or opposition for addressing income inequality is simply a function of mean income minus own income, as this would determine whether symmetrically compressing income inequality would yield one a gain or a loss.

This is not a convincing model for numerous reasons.  Let’s even posit that people didn’t care about anything other than the effect on own income under the above setup.  The model would still founder on the paradox of voting – i.e., the fact that, since I cannot have any statistically significant effect on the outcome, my voting (and even informing myself about the economic stakes to me) are a total waste of time when modeled in such a framework.  People “shouldn’t” vote, and if rational “wouldn’t” vote, unless something else was going on.

(As an aside, I spent close to an hour at the polling place this morning.  Suppose that I disvalue a Trump victory at $1 billion – if I had no conscience, I might in a sci fi hypothetical accept that much from the gods, after-tax, in exchange for his winning –  and that I very generously rate the chance that my vote will affect the outcome at 1/500 million.  Then my behavior would suggest that I value my time at only $2/hour, surely below how I usually treat it.

What’s a better basic account of how voters generally make choices?  I rather like this quote from the newly published Christopher Achen and Larry Bartels, Democracy for Realists:  

“[M]ost residents of democratic countries have little interest in politics and do not follow news of public affairs beyond browsing the headlines. They do not know the details of even salient policy debates, they do not have a firm understanding of what the political parties stand for, and they often vote for parties whose longstanding issue positions are at odds with their own.  Mostly, they identify with ethnic, racial, occupational, or other sorts of groups and often – whether through group ties or hereditary loyalties – with a political party. Even the more attentive citizens mostly adopt the political positions of the parties as their own: they are mirrors of the parties, not masters. For most citizens most of the time, party and group loyalty are the primary drivers of vote choices.”

Unfortunately for purposes of explaining how rising inequality might be expected to affect voters’ political preferences regarding policies to address it, this leaves us with something of a black box.  On Election Day 2016, for example, we are seeing identity issues more than usually cashed out in straightforward terms of race and ethnicity.

b) High-end inequality vs. low-end inequality
The Kuziemko paper, reflecting the data (in the form of survey questions) on which it must rely, does not tease apart high-end and low-end inequality.  Yet I feel this issue is central to people’s attitudes, wholly apart from my arguing that the issues posed are analytically quite distinct.

“Inequality” is an abstraction.  Suppose I am in between the 20th and 99th percentile, whether ranked by income or wealth, and thus regard myself as “middle-class.”  What I mean by this is that I see people above me and people below me, so – aggregate statistics be damned – I place myself in the middle.  I may think something about the “rich,” and something about the “poor.”  But these are two different groups to me, and I may have particular sentiments about each.

The big rise in inequality in the U.S. since the mid-1970s has been concentrated on the high-end.  So survey questions that emphasize the low end certainly run the risk of missing respondents’ feelings about the piece that has actually moved a lot.

Consider 2 questions from the surveys that the paper analyzes.  First, what I’ll call “Question 1,” which shows up in the paper’s Figure 2:

“Some people think that the government in Washington ought to reduce the income differences between the rich and the poor, perhaps by raising the taxes of wealthy families or by giving income assistance to the poor. Others think that the government should not concern itself with reducing this income difference between the rich and the poor.” [Respondents then pick a number between 1 and 7 to reflect where they fall on the spectrum between these 2 view.]

While this question is a bit of mishmash, it does notably discuss BOTH the rich and the poor.  Kuziemko et al find little to no change in sentiment as between the polar choices in the period from 1978-2006

Then there’s “Question 2,” which shows up in the paper’s Figure 3:

“Some people think that the government in Washington should do everything possible to improve the standard of living of all poor Americans....Other people think it is not the government’s responsibility, and that each person should take care of himself.”  Once again, people are asked to locate themselves along a spectrum.

Here we find a significant decrease in pro-redistributive sentiment from 1978-2006, despite rising inequality during this period.  Why the difference between this and the flat, no-change finding for Question 1?

Unfortunately, this is not a clean comparison given the two questions’ multiple moving parts – e.g., what the government should concern itself with in Question 1, versus the role of individual responsibility in Question 2.  But the differences include mentioning both the rich and the poor in Question 1, and just the poor in Question 2.  So it’s consistent with positing, or at least speculating, that increased concern about the rich, which only Question 1 evoked, might have offset reduced concern about the poor so as to keep the overall trend in Question 1 flat rather than negative.

c) American exceptionalism
Here as so often, we have a paper dealing with American exceptionalism.  The paper’s headline finding is that US seniors have moved sharply against redistribution during this period.  Might it be a byproduct of their rising life expectancies?  The paper says no, because seniors in peer countries have not so moved despite having similar life expectancy increases.  The paper suggests that “Medicare threat” is the answer to the riddle.  U.S. seniors are worried that increased redistribution to younger Americans will come at THEIR expense, via Medicare cuts.  In effect, they’ve already got theirs, so it’s time to kick away the ladder that lifted them to safety, and make sure that others cannot benefit as they have.

To anyone who has followed, say, “Obamacare” politics from 2009 through to the present (although this falls outside the period that the paper surveys), this is wholly  plausible and credible.  We do indeed have a unique structure for government-funded healthcare that we have observed affecting seniors’ political sentiments in this way. 

But there are also other ways in which the US is unique. For example, as my colleague David Garland’s book regarding the death penalty points out, in many way the U.S. looked a lot more like Western Europe 50 years ago than it does today, and we have diverged in multiple dimensions, partly because our political elite (pre-Trumpism) split into liberal and conservative wings.  We have a far less generous social safety net, we’ve retained the death penalty, religion plays a larger role in politics than in many peer countries, libertarian and pro-market ideology are stronger here, we have racial and ethnic issues that predate contemporary immigration issues, etcetera.  So it might be interesting also to evaluate other possible explanations for American exceptionalism here.

d) Seniors
Again, the “Medicare threat” explanation for seniors’ rising hostility to redistribution in the US makes a lot of sense.  I also note that it would not be entirely irrational for seniors to be concerned that opening up the budget more to help younger people would potentially crowd out expenditures on their behalf, especially with rising healthcare costs that help put Medicare on at least the long-run chopping block.

But just how rational is it?  (And again, recall that voters generally are not “rationally self-interested” as the “workhorse political economy model” irrationally posits, because it’s not worth the effort to figure out where one’s interest lies, given the voting paradox.)  And what about another threat that they have chosen to ignore?  This is the clear fact that the anti-redistributive Republicans to whom they have rushed as a voting bloc actually thirst to sharply cut both Medicare and Social Security – whereas the (relatively) pro-redistributive Democrats whom they now shun are eager to protect both programs.

True, Republicans in the Paul Ryan vein have not only tried to soft-pedal their Social Security and Medicare retrenchment plans, but also have promised that any changes would only affect younger age cohorts.  But then again, mightn’t the same thing be true of Obamacare expansions et al, especially since the Democrats like Social Security and Medicare and thus don’t want to cut them?

My point here is not that the seniors “ought” to support the Democrats, rather than the Republicans, on all this.  There are arguments of group self-interest that go both ways.  Rather, my point is that it’s plausible that they have made their choice based on group identity.  Today’s seniors may well be a group that looks younger Americans who might get government aid as different from themselves – in particular, as less white.  In short, our demographic transition from a white-dominated to a diverse multi-ethnic society may have particularly strong effects on current seniors who can see the transition happening.

So is it about group identity after all?  More on this question under Topic 2.

e) African-Americans
The paper finds that, while African-Americans are still more pro-redistributive than the overall population average, they have moved towards the average (and away from being ideological outliers) during the period from 1978-2006 – this despite lagging economic catchup with whites.  I wonder if this might have something to do with the demographic shift, which I gather William Julius Wilson and others have written about, under which middle-class and upper-middle-class African-Americans increasingly live in neighborhoods (e.g., in the suburbs) that were previously restricted to whites, but among whom these individuals can now find economic peers.

2.         Lessons learned
Here are some thoughts about what the paper’s analysis might suggest, looking back a bit and maybe forward.

a) Was Medicare a political fiasco for progressives?
Medicare proponents in 1965 were motivated in large part by the thought: Seniors first.  They wanted comprehensive national healthcare like that in peer countries, but decided to start with the politically easy part.  But then, rather than this being the leading wedge for a broader change, it turned into a game of “pull up the ladder,” with seniors organizing to ensure that no one else could get what they have.

Medicare’s deliberately adopted optics may have worsened the problem.  It’s structured to create an impression that one is paying for one’s own benefits, via the payroll taxes one pays up front and then the annual premium for Medicare Part B.  But of course it results in huge net transfers to current seniors who, in many cases, deny that it is even a government program.  So from a progressive standpoint the whole thing appears to have backfired to a degree.

b) Cognitive dissonance looking forward
The paper evaluates but rejects a cognitive dissonance hypothesis, under which, if one becomes (say) more conservative on Issue 1, one will seek to restore consistency (defined by political alliances) by also becoming more conservative on Issue 2.  So, if one moved to the right by reason of opposition to gay rights, one might then adopt one’s new bedfellows’ (so to speak) on tax progressivity as well.  The paper finds, however, that such an explanation does not account for the shift against favoring progressivity (such as among seniors) in the data set.

Might it work the other way going forward?  Suppose future seniors dislike the conservative positions on so-called social issues?  Might this influence them to adopt more progressive positions on distributional issues as well?  Obviously we don’t know, but let’s know consider the same question phrased more generally.

 c) Seniors: life cycle effects vs. age cohort effects
As the authors recognize, it’s hard in their data set to tell apart life cycle effects and age cohort effects.  The former are at work if, say, once one reaches retirement age one starts to see public policy through the lens of “Medicare threat” to one’s own benefits.  The latter are at work if, say, people who grew up under Johnson & Nixon differ from those who grew up under Reagan, who differ in turn from those who grew up under Bill Clinton, and so on moving forward.  (I use the presidents’ name to evoke the particular times, whether or not this mainly depended on their particular political adventures.)

If one were to posit the hope (or fear, depending on one’s political preferences) that seniors going forward will not be as anti-redistributive as the current cohort, even if the structural issues around Medicare remain the same, one might base that judgment on the possibility that America’s racial and ethnic transition, from white-dominated to genuinely multi-ethnic, is playing a central role.  Is it crucial to today’s seniors that they see younger age cohorts as different from themselves, given the ongoing transition, but that this perception once the transition has passed a critical stage?  As the favored cliché of news features would have it, Only Time Will Tell.

d) African-Americans
Again, the paper finds that they remain more progressive than the median voter, but have moved closer to the median position.  One question might be: How is this likely to affect political outcomes?  Under a median voter model – although admittedly this old warhorse, no less than the workhorse political economy model of narrowly conceived economic self-interest, has lost ground in recent years – it actually might not matter a whole lot.  Obviously, however, not a question that this particular paper aims to address.

e) Relevance (or not) of voter sentiments
More generally, while this is outside the paper’s scope, one should keep in mind that it is far from clear that general voter sentiments regarding redistribution have any discernible effect on policy outcomes.  Martin Gilens, for example, finds that the sentiments of the bottom 99% generally have negligible effects on policy outcomes.  While this would not eliminate one’s interest in better understanding the link (if any) between distributional changes in the economy and public political sentiments, it would indicate that the grounds for interest are mainly sociological and based on the value of knowledge for its own sake (not as an input into making political forecasts). 

No comments: