Outliers, Inequality, and Injustice

Ezra Klein writes:

But since we justify income inequality by understanding success as an outcome of virtue, there's a tendency to ascribe achievement to diligent effort rather than the market's amoral decisions to attach high value to certain spheres of labor and low value to others. The important variable for success, however, does not seem to be hard work but profession. If you're in a high-value profession, hard work can do you a lot of good. If you're not, it may not do you much good at all. And though anyone can work hard, we're mostly able to admit that not everyone has the specific constellation of opportunities that lets you go to law school, or spend your time goofing off in amateur political punditry. Occupation is rather more useful for understanding why someone's hard work pays off than is their relative level of toil, but since occupation is more clearly contingent on circumstance, and high-value occupations have more obvious barriers to entry, they also raise questions of justice in outcome, and thus have fairly uncomfortable answers for those atop the pyramid. So hard work it is.

If “success” equals “high income,” then Ezra's right. Now, I deny that income inequality, per se, requires justification. There's no good reason to treat equality in money as some kind of moral baseline, deviations from which must be accounted for. If you understand that prices convey information about supply and demand, and that a wage is a price, then you understand that differences in wages for different kinds of labor convey information about the supply of different kinds of labor relative to demand. Wage inequalities are how people can know what's a “high-value profession” and what isn't. It guides our choices about the kinds of skills to seek. We need that guidance because effort isn't enough. You can work as hard as you like in a low-wage job, and you'll still be in a low-wage job. If Ezra and I have any disagreement here, it's probably in our sense of what does and does not “raise questions of justice in outcome.” But even here I think we may agree. As I said, the mere fact of inequality raises no question at all about justice. But if the extent of inequality is in part a function of barriers to entry, then injustice is likely. Still, the injustice is not the differences in outcome, but in the morally unjustified infringements of liberty that help explain them.

Gladwell makes a big deal out of being in the right place at the right time. The fact that he talks almost exclusively about Americans obscures just how crucial this point is. Simply being born an American or a Canadian or a Swede is a much bigger deal than having had a computer lab when nobody else had one, or having been a youth hockey player born in January. If you were born where there were computers at all in the 1980s, or where there are youth hockey leagues, then you're already an outlier of sorts.

Ezra's right to draw our attention to unjust barriers. Far and away, the greatest of all such barriers are restrictions on immigration. Wages also convey information about more than supply and demand; they convey unseful information about labor productivity. Skill level is hugely important to productivity and therefore wages, but two workers with identical skill levels may get paid very different wages because skills do not translate into the same level of productvity everywhere. Institutional structure and technology helps determine productivity and therefore wages. If you are barred from entering a political jurisdiction where technology and institutions will best complement your skills, you will earn less than someone with the same skills, who tries exactly as hard, but who lucked into the jurisdiction. Immigration law is often explicitly intended as opportunity hoarding, and the inequalities it creates reflect this injustice. 

Within countries, many professions have succeeded in creating barriers that brake growth in the supply of skill, ensuring unfairly high wages. That's unjust. That's also relatively trivial compared to the more fundamental failure of public education to deliver enough to millions of studens. Wage differences tell us which skills are most-highly valued in the labor market. But the ability to make the most of these signals, to acquire these economically-valued skills, depends on a foundation of basic, multipurpose skills that too few are given a chance to develop. If you grow up in inner-city America, the outrageous barriers thrown up by the medical cartel do not loom large among the injustices to which you are subject.

Success will always be contingent. Opportunities cannot be evenly spread. And lucky people should be encouraged to make the most of their luck. But the difference in opportunity between a typical American and a typical Mexican is more than a matter of luck. We can build a wall or create a common labor market. The difference in opportunity between a typical poor American and a typical middle-class American is more than a matter of luck. We can prop up our failed system of education or fundamentally reconfigure it. Justice forbids one and demands the other.

Shorter Kenneally

The greatest gift Culture 11 has given the world is the dazzling thought of RIT political theorist Ivan Kenneally. I think I'll make Kenneally translation an occasional feature. On the subject of Thanksgiving

Thanksgiving is a holiday devoted to the virtue of gratitude which, one could argue, finds less than hospitable ground in the modern world. The Lockean position on nature, that it furnishes only worthless materials that gain value through an imposition of labor, could not be more inconsistent with gratitude; in fact, Locke specifically undermines any conception of nature that would inspire reverence for the evidence it gives us of God’s providence. Instead of gratitude for what nature provides Locke encourages pride in our mastery and possession of nature—we take pride in ourselves as the only part of nature that can refashion nature. It is hard for beings who fancy themselves as radically self-sufficient or autonomous to be grateful for anything.


In the modern world, there is so much more to be grateful for. Therefore, we are less grateful.


Technocracy vs. Liberal Democracy

Upon return from Singapore, Bryan Caplan writes:

Singaporean bureaucrats are less afraid to criticize their government than American bureaucrats are to criticize theirs.  Neither group would be afraid of legal punishment; but the Americans would be more worried that saying the wrong thing would hurt their careers.

Why is that? Commenter Devin Finnbar writes:

I believe it. Tocqueville noted that American democracy had less free speech than the European monarchies, because of the overwhelming social pressure to say the right thing.

I find this fascinating. Let me see if I can flesh out this idea a bit more.

In a democracy, policy shifts due to shifts in public opinion. Random changes in public opinion won't do. They have to be coordinated. Opinion is coordinated through signaling and sanctions, both subtle and unsubtle. So, in a democracy, stating an opinion is a move in a delicate coordination game. There is a lot of pressure to be on the team. Coalition democratic politics is largely about trying to sabotage the other side's attempts at solving the opinion-coordination problem. If you are a bureaucrat, you will be especially sensitive to the demands of conformity and solidarity in producing policy change, so you will tend to toe the line. You probably won't think about this strategically. People whose sincere opinions tend to track the needs of their political coalition will be trusted within political coalitions, and will tend to get assigned to desirable government jobs.

In a Singapore-style technocracy, public opinion is just one of many constraints to take into account in formulating policy. But then public opinion can't serve as the basis for a sense of the legitimacy of a policy or a policymaker. If the technocrat actually cares about legitimacy, then she probably cares a lot about effectiveness. The reason its okay to go over the heads of the people is that what you're doing actually works to make them better off. Additionally, if you're a bureaucrat, have a good idea, and can argue for it, it just might become policy. Especially if you are willing to let your boss take credit for it. In a well-functioning technocracy, status accrues to people who produce new ideas for effective policy.

So bureaucrats in a technocracy will be motivated to explore ideas, while bureaucrats in a democracy will be motivated to signal and recruit fidelity to the coalition's pre-assigned ideas. Free-thinking exploration could spell defeat! 

Some related thoughts:

  • The glacial nature of shifts in democratic public opinion are part of what kept the U.S. from adopting more heavily socialist policies mid-century. 
  • Implementing the policies best supported by the social-scientific consensus once meant “economic planning,” and that is bound to fail for familiar reasons. But the fact that those reasons are familiar explains why technocracy is now less likely to fail.
  • Milton Friedman claimed that capitalism and freedom are inextricably linked. If this is an empirical and not a conceptual claim, we could find that this is false if politically free people again and again choose against economic freedom, or if the rulers of politically unfree countries show some tendency to choose policies of economic freedom.
  • In principle, free-market technocracies seem dangerously unstable in ways liberal democracies do not. But that doesn't imply a free-market technocracy can't have a good run before becoming captured by malign forces. How much will it matter to people in democracies that their liberties are more secure in the long run if it comes to pass that a technocracy on a heater is actually producing 3x the democracy's per capita income?
  • In that scenrio, “Canadian citizen living in Singapore” will dominate “Canadian citizen living in Canada” or “Singapore citizen living in Singapore.” Why not live in the richest jurisdiction as long as you can always retreat to the freest if things go south? 
  • Singapore can't absorb a ton of Americans or Canadians, but China could. If China becomes a huge Singapore, do liberal democracies develop a brain drain problem? Could this push democracies toward freer market policies and vindicate Friedman in the end?

Christina Romer to Head the CEA

Christina Romer's appointment to the chair of the President's Council of Economic Advisors is excellent news. Her recent papers on taxation with David Romer represent a real advance in blending economic theory with rigorous and ingenious historical research. Of course, good economics isn't always good politics, which leaves high-ranking academic economists in a bit of a tight spot as they attempt to help the administration meet its aims while also attempting to maintain their professional reputation as social scientists. (This is, I imagine, rather harder for right-leaning economists such as Mankiw and Holtz-Eakins, already a distinct minority even in economics.) I'm interesting to see the extent to which Obama's excellent economic team will and won't be willing to diverge from their expert opinion in order to provide political cover for bad economic ideas. Goolsbee's truth-telling campaign gaffe about Nafta comes to mind. 

In honor of Romer's appoinment, I reprint the post I wrote about her and David Romer's recent work on taxes over a year ago for Free Exchange:

The unbearable lightness of being Martin Feldstein

Posted by: 
Free Exchange | Washington, DC

JONATHAN CHAIT apparently is unimpressed by citations to the work of personages such as Martin Feldstein, the president of the prestigious National Bureau of Economic Research and the George F. Baker Professor of Economics at Harvard University. Indeed, Mr Chait has a knack for drawing the bounds of intellectual respectability so tightly around himself that by late afternoon even his shadow falls outside the charmed circle. Even so, one must admit that Mr Feldstein's Clark medal and his endorsement by the New York Times for the job of Chairman of the Federal Reserve does leave one with a residue of suspicion. The Bank of Sweden has not bestowed upon him its coveted prize—though it's true he has been mentioned, specifically for his work on the theory of taxation. So let's not be too hasty to take him seriously. 

Because Mr Chait is a self-avowed empiricist, perhaps he will favour this new NBER paper (free version here) by Christina and David Romer of the Univesity of California, Berkeley (despite somewhat embarrassing credentials, even slightly more lackluster than Mr Feldstein's). It is a dazzling empirical investigation of the effects of tax cuts and increases on economic output in the United States since the end of the second World War—one that significantly improves on the methodology of earlier attempts to estimate the effects of tax changes. They find that tax increases appear to have a very large, sustained, and highly significant negative impact on output. Since most of our exogenous tax changes are in fact reductions, the more intuitive way to express this result is that tax cuts have very large and persistent positive output effects.

The economists Romer looked at every tax change legislated at the national level since WWII.  Impressively, they scoured “presidential speeches, executive-branch documents, and Congressional reports … to identify the size, timing, and principal motivation for all major postwar tax policy actions.” They then categorized each tax change based on whether or not it was intended as a forward-looking correction to the direction of the economy (they call these “endogenous” changes), or intended for other reasons, such as to reduce an inherited deficit or to boost long-run growth (the “exogenous” changes.) This allows them to tease out the output effects of tax cuts and tax increases set in place for different reasons.

Those interested in raising taxes, but unwilling to take seriously Martin Feldstein's estimate of the deadweight loss of tax increases, will need to grapple with Mr and Ms Romers' new findings. For example:

Our baseline specification suggests that an exogenous tax increase of one percent of GDP lowers real GDP by roughly three percent.

This is bad news for those with aspirations to higher levels of tax-financed spending. However, they find that not all tax hikes hurt the same. Tax increases specifically intended to offset budget deficits largely avoid the negative effects of other kinds of increases, in part by improving the climate of investor confidence.

In another fascinating new working paper on the “starve the beast” hypothesis (it is false, FYI), the Romer duo directly discuss the revenue effects of tax cuts. So, the question at the heart of Mr Chait's attack on supply side economics: do tax cuts pay for themselves? 

[A]lthough a tax cut leads to a sharp fall in revenues in the short run, it does not have any clear impact on revenues at horizons beyond about two years. Second, roughly half of the tax cut is offset by legislated tax increases over the next several years. Taken together, these findings suggest that a substantial fraction of the rebound in revenues is the result of non-legislated changes. The key source of the non-legislated changes in revenues is almost certainly the effect of the tax cut on economic activity. In Romer and Romer (2007b), we find that a tax cut of one percent of GDP increases real output by approximately three percent over the next three years. Since revenues are a function of income, this growth undoubtedly raises revenues. 

So there you have it. Tax cuts don't exactly “pay for themselves”, but they also don't diminish revenue after about two years. That is, after about two years, the government receives revenues equal to what it would have received at the higher rate, but taxpayers enjoy a lower burden. It is an important advance to discover that because cuts do lead to an immediate dip in revenue, they often inspire offsetting tax increases that retard the growth effect of the original cut. Nevertheless, the effect of cuts on output is generally strong enough to bring revenue back to where it would have been otherwise.

They do, however, add “an important caveat” to their “finding that tax cuts partially pay for themselves through more rapid growth”: the output effects of a tax cut may not last forever, in which case the cut could lead to shortfalls over the long run “in the absence of other legislative changes.”

The lesson, then, is that it is indeed irresponsible to think of a tax cut as a free lunch. If citizens wish to enjoy lower taxes, and do not wish to foist their debt on the next generation, they cannot avoid the responsibility of also cutting spending. This is how it should be. Truly “free” tax cuts make big government too much of a bargain. If you care about nothing more than getting out from under high tax rates, then self-funding tax cuts are a political dream come true: you don't have to ask anyone to give anything up. But if you worry about the intrusions, abuses, and injustices other than the confiscatory tax rates enabled by a massive state, you should not want too much slack in the relationship between the spending levels and tax rates. 

For their part, empirically-minded champions of the welfare state like Mr Chait should be encouraged to discover that taxpayers seeking relief are not in fact the mortal enemy of government spending. The real nemesis of both government revenue and individual well-being is a slowed rate of economic growth, which, research shows, tax increases generally deliver.

Andrew Gelman on Voting and So Much More

In this week's episode of Free Will, I chat with Columbia political scientist and statistician Andrew Gelman about his book Red State, Blue State, Rich State Poor State, the election, why rich people are increasingly voting Democratic, and even a bit about happiness. Andrew is a man of exceeding epistemic virtue and was worried that he riffed too much on topics outside of his expertise. Of course, I have no such scruples, and found the chat fascinating, because Andrew is.  

Getting the Numbers Right

My farseeing sister points me to this AP report which helps clarify the matter of accounting for GM's labor costs:

GM, which negotiated the four-year deal that serves as a template for UAW deals with Chrysler and Ford, says its total hourly labor costs dropped 6 percent this year from pre-contract levels, from $73.26 in 2006 to around $69 per hour. The new cost includes laborers' wages of $29.78 per hour, plus benefits, pensions and the cost of providing health care to more than 432,000 GM retirees, GM spokesman Tony Sapienza said.

The total cost will drop to $62 per hour in 2010 when the linchpin of the contract — a UAW administered trust fund — starts paying retiree health care costs.

But that's still $9 more than the $53 per hour that GM estimated Toyota now pays in the U.S., and the gap could be even wider. Toyota spokesman Mike Goss said the company's total labor costs at its older U.S. plants are around $48, with about $30 per hour in wages.

The remaining difference largely is due to “legacy” costs, the cost of a 100-year-old company paying its retiree pensions, Sapienza said.

So there you go. I completely agree that it is deceptive to present the numbers in the way GM does in the document I link below. That GM would try to make its health and pension obligations look crushing makes sense from a driving-a-hard-bargain-with-the-UAW pespective. But if you want to get bailed out, it makes no sense to make yourself look like a complete basketcase, which exactly what the $73 number does. I would expect GM to be doing what its spokesman is doing here: trying to make the company look like it is in fighting trim, just one giant infusion of taxpayer money away from not losing massive quantities money. For that reason alone, I'd be surprised if the remaining gap between GM and Toyota, even after legacy costs are taken into account, isn't “even wider,” as the Toyota spokesman suggests.

Now, it bears emphasizing that Casey Mulligan's point, that a bailout would benefit mostly relatively wealthy workers and shareholders is true, even if we go with the the $30/hour wage figure. According to the BLS, the median hourly wage in 2007 was about $15 and the mean was about $20 (and the averages in assembly and production work are lower than that). But GM (and as far as I know, all the other car companties) pays a lot for overtime and so forth, which is where you get GM's estimate that the “average annual cash compensation for hourly employees in 2006 was $39.68 per hour.” Autoworkers who end up jobless because their company and union are run by doofuses deserve our sympathy. The fact that they're so much richer than the average American worker doesn't change that. The point is that subisidies to relatively rich shareholders and workers are basically the opposite of distributive justice, and that's true whether the right number is $70, $40, or $30.

New Deal not Prozac to the Great Depression

Tyler Cowen's NY Times Economic Scene column on the Great Depression and the New Deal is a welcome dose of measured judgment. Bottom line:

In short, expansionary monetary policy and wartime orders from Europe, not the well-known policies of the New Deal, did the most to make the American economy climb out of the Depression. Our current downturn will end as well someday, and, as in the ’30s, the recovery will probably come for reasons that have little to do with most policy initiatives.

You know one of the many things I love about Tyler? I'd say I know him fairly well, and have a pretty good sense of his overall intellectual framework (including his not-as-elusive-as-it-seems moral and political theory). But I cannot predict his opinions on most topics with any precision, yet always find them consistent with what I already knew about his commitments. What do you suppose that means?

Yup: Over Seventy Buck per Hour

I caught some crap for linking to Mark Perry on GM's labor costs, and to Casey Mulligan's linking to Mark Perry on GM's labor costs. I said I wasn't sure whether or not they were making a mistake, because I wasn't sure. 

Well, it looks like there's no mistake.  Here what General Motors says, via Mark Perry:

Let me quote directly from General Motors Manufacturing and Human Resources website (click on “Other Benefits,” or go here directly):

The total of both cash compensation and benefits provided to GM hourly workers in 2006 amounted to approximately $73.26 per active hour worked. This total is made of two main components: cash compensation ($39.68) and benefit/government required programs ($33.58).

The average annual cash compensation for hourly employees in 2006 was $39.68 per hour. Included in average earnings are straight-time pay, Cost of Living Allowance (COLA), night-shift premiums, overtime premiums, holiday and vacation pay. In 2003, GM workers logged 41,363 (hours in 000's) in overtime hours for an average of 371 hours per worker; in 2004, 39,409 overtime hours for an average of 374 hours per worker; in 2005, 33,555 overtime hours for an average of 337 hours per worker; and in 2006, 27,265 overtime hours for an average of 315 hours per worker. 

Benefit/government required programs in 2006 added an additional $33.58 for each active hour worked. These costs include: group life insurance, disability benefits, and Supplemental Unemployment Benefits (SUB), Job Security (JOBS), pensions, unemployment compensation, Social Security taxes, and hospital, surgical, prescription drug, dental, and vision care benefits.

… Note that the $73.26 per hour was for 2006, and it's probably higher now, so if the $73.26 per hour labor cost was incorrect, it was probably too low, not too high. 

Here are some links for additional sources for the $73.26 per hour cost for GM at Business Week and USA Today, both of which I assume do strict and careful fact-checking.

So, yes. Few workers are getting over $70 per hour in wages. Average cash wages in 2006 were just shy of $40 per hour. At least that's what GM says.  The value of benefits per hour was north of $33 per hour. It adds up to average total compensation per active hour worked of over $73. At least that's what GM says. Maybe they're wrong and making some kind of fundamental mistake adding things up this way. If so, maybe all the people angrily linking here, or demanding a retraction and apology, should send a fax to Detroit or something.

I've just added Mark's blog to my feed reader. You should too. It's terrific.

Postmaterialism and Cohen's Maxim

It is a commonplace on the left that “programs for the poor are poor programs,” the second 'poor' meaning “poorly funded.” Call this “Cohen's Maxim” after Wilbur Cohen, a chief architect of Social Security. Cohen's Maxim is likely true when wealth transfer programs targeted to the poor are very unpopular relative to “social insurance” that is heavily marketed as “universal.” Yet if the idea is to secure a certain level of benefits for the genuinely needy, the universal social insurance scheme (which will waste a lot of money by taking it from the middle and upper classes and then giving most of it back to them later) will tend to be much more expensive than the targeted transfer scheme. Other things equal, one should prefer the means-tested program, since it frees up resources that can be used to (a) make the targeted benefits larger, (b) sent back to taxpayers while leaving the poor no worse off, or (c) spent on other desirable social programs. The only reason to prefer the social insurance scheme (if the point is to help the poor) is if there will otherwise be insufficient political support to keep benefits for the poor at a decent level.

Under what conditions would we expect Cohen's Maxim to be true? Conditions under which the middle and upper classes tend to resist financing welfare transfers. They might resist for ideological reasons, in which case Cohen's Maxim will rise and fall with trends in ideology. But they might also resist for reasons of perceived economic self-interested, which in turn might to some degree drive trends in ideology. What if past a certain threshold in income and wealth, voters became more concerned with questions of fairness and justice and less concerned with their own perceived economic interest? That's Ronald Inglehart's “postmaterialism” thesis in a nutshell, and it appears to be well-supported by evidence. We may be seeing it in the move of wealthy voters toward the Democratic Party. Obama's win over McCain among the wealthiest voters might be because he promised to tax them more and spread the wealth around. 

Suppose this trend continues and, as the median income rises, an ever larger portion of voters above the median comes to prioritize social justice over tax rates. Under those conditions, why think Cohen's Maxim would hold? 

By the way, I've always rejected Cohen's Maxim. Unemployment benefits are targeted, but generous and popular. My favorite argument against making Social Security into a means-tested program is that benefits would likely to be too generous, generating serious moral hazard while being unjust to boot. The core of this argument is the disproportionate heft of retirees as a voting bloc, and the relative unity of their interests. Forced intrapersonal transfers are preferable, my argument goes, to exploitative, class-based (younger to older) interpersonal transfers. But if economic growth makes us ever less fixated on our narrow our economic interest in the voting booth, this argument could break down, too.

So could postmaterialization someday put us in a position where it becomes feasible to get rid of elaborate schemes like social security and medicare–and even the personal account alternatives to these–and go with the most direct, efficient, and transparent safety net policy? Provide government assistance to people who fall beneath a certain minimum of resources. Otherwise, don't. 

Related: Here's my colleague Jagadeesh Gokhale smacking down some bad arguments for the Social Security status quo.

How Much Are GM's Labor Costs?

Felix Salmon says it's not 70$ an hour per worker. 

The average GM assembly-line worker makes about $28 per hour in wages, and I can assure you that GM is not paying $42 an hour in health insurance and pension plan contributions. Rather, the $70 per hour figure (or $73 an hour, or whatever) is a ridiculous number obtained by adding up GM's total labor, health, and pension costs, and then dividing by the total number of hours worked. In other words, it includes all the healthcare and retirement costs of retired workers.

A number of Felix's commentators argue that he's missing the point. I don't know whether he is or not. I put up the number because it came from thinkers I trust, and if I was wrong to do that, I'm happy to admit it. Anyway, it looks like we all agree that GM has extraordinary liabilities, even if some of them are being shifted onto the UAW, and that it continues to waste a ton of money that could be put to better use elsewhere.