Something Not Unlike Research

Alex Tabarrok on Economic Mobility

 post by Bill Gardner

Tabarrok is interested in the moral significance, if any, of economic mobility. He first contrasts two societies, "Stasis" and "Churn".

In society one there is no generational mobility, each generation has the same income as the previous generation. Let’s call this society Stasis.

Society Stasis
Family	Generation 1	Generation 2
A	100	100
B	50	50
C	25	25
Total	175	175

In the second society, some generations rise and some generations fall but there is no growth, let’s call this society Churn.

Society Churn
Family	Generation 1	Generation 2
A	100	25
B	50	50
C	25	100
Total	175	175

He argues that there is little to prefer between Stasis and Churn. But there is a reason to prefer Churn: Relative to Stasis, the deprived of Generation 1 have the compensation of knowing that their children will do well. A and C are symmetrical: A does well in Gen₁ and poor in Gen₂, while the reverse is true for C. If you view families as bearers of well-being across generations, then the utilities of A and C accumulated across generations are equal and B is not too far behind. In this sense, Churn is fairer than Stasis. 

Perhaps the idea of families as bearers of utility is problematic. Tabarrok seems to assume that utility = income in each generation. But this is not necessarily the case: Suppose that the utility function for a parent in Gen₁ is

Utility(Gen₁) = a * Income(Gen₁) + (1 - a) * Income(Gen₂), 0 < a < 1.

Then the average utility is unchanged but the range and variance of utilities in any generation is less in Churn than in Stasis. If that matters to you, Churn is fairer. It might not matter to you: A utilitarian could well argue that such utility functions are wrong. This is because the function gives higher weight to the well-being of my children, violating the principle of treating each other person as having equal moral worth.

Otherwise, Tabarrok is right: Mobility in and of itself is not important to justice. We value mobility because, as one of his commenters points out, societies without mobility almost certainly lack fair equality of opportunity or, as Tabarrok points out, liberty.

He then considers a third society, Growth.

Society Growth
Family	Generation 1	Generation 2
A	100	200
B	50	100
C	25	50
Total	175	350

Growth seems clearly better than Stasis, because everyone is better off in Gen₂ in Growth than they would be in Stasis. Would Growth be still better if we added some Churn? Just a bit, I would say, for the reasons advanced above. However, seeking to add random Churn would be crazy. It would not improve fair equality of opportunity and would likely compromise liberty.

The problem with Growth, from a liberal egalitarian point of view is not that it lacks mobility. The key problem is that the worst-off group gains the least from Gen₁ to Gen₂. Rather than add random Churn, a sane liberal policy would look for ways to increase the gains for C. I think that Taborrak would agree that this, rather than mobility, is the fundamental problem with Growth. Of course, the next and harder questions are whether attempts to do this would increase or decrease average utility, and if so, how much decrease we can accept to achieve a more just distribution of gains. 

John Broome on an Irony of Welfare Economics

post by Paul Kelleher

Bill and I are currently reading and studying John Broome's new book Climate Matters: Ethics in a Warming World. It is an excellent if deceptively challenging book (Broome's prose is so effortless that one can be led to think mastering the ideas will be, too). I am certain the book will shape the way I think about virtually all matters in public and population health. Indeed, I am currently writing a paper on its main themes that I hope to post here in a week or two.

Broome's chapter on discounting future goods is among the best in the book. He notes that discounting is a common practice in cost-benefit analyses of climate change, and the debate over the proper discount rate has been viewed by many as the ethical question raised by global warming (although Broome would not agree with that, I don't think). I have discussed the issue of discounting a couple of times on the blog, but if you have to choose, read what Broome has to say (naturally). 

Broome explains that cost-benefit analysis is a kind of shortcut: economists want to understand the impact that various policies will have on society, and the gold standard would be a way to assess the impacts they will have on individuals' well-being. But economists are notoriously pessimistic about being able to measure well-being directly, so they use a proxy: they measure what individuals are willing to pay for commodities, since these commodities can serve as a proxy for the human well-being they produce in those who enjoy them. The more someone is willing to pay for a given commodity, the more well-being it would produce for that person (if she were to acquire it). 

Broome points out what all economists already understand: other things equal, the same commodity will produce more well-being if enjoyed by a poor person than if enjoyed by a rich person. This phenomenon is known as diminishing marginal utility, and it can be present even when the rich person is willing to pay more than the poor person for commodity. So unless the economists' short-cut for measuring well-being takes this phenomenon into account, the shortcut will likely be biased in favor of policies that benefit the rich, since their higher willingness to pay will be interpreted as a higher capacity to benefit. Broome remarks, "Cost-benefit analysts have ways of correcting for [this bias in favor of the relatively rich], but in practice they rarely take the trouble to do so." I have discussed this fact about cost-benefit analysis at length elsewhere. 

There is an irony here, however. For Broome also notes that economists standardly discount the importance of future benefits at some positive rate. This practice gives more weight to the enjoyment of a commodity here and now relative to the enjoyment of the same commodity in the future. And one key economic rationale for doing this is that people in the future are expected to be richer. Broome goes on:

If people are richer in the future, that means additional commodities bring less benefit on average to future people than the same commodities bring to present people. A kilo of rice in one hundred years will contribute less on average to the well-being of the people who eat it than a kilo contributes today. This is a good reason for discounting future commodities. Ironically, although cost-benefit analysts generally ignore the diminishing marginal benefit of money when they are aggregating value across people at a single date, their main case for discounting future commodities is founded on this diminishing marginal benefit.

Broome does not pull punches when he goes on to note that the effect of this irony is that it protects the current rich on two fronts: it downplays the case for transfers from the current rich to the current poor (since it ignores the greater capacity of the poor to benefit from the same expenditure), and it downplays the case for transfers from the current rich to future generations (since their well-being is discounted relative to well-being here and now). Broome's view is that if people in the future really will be better off than we are, then we really do have reason to discount the value of future commodities. But this same reason suggests that cost-benefit analysis should no longer ignore the fact that the same commodity brings more benefit to the current poor than it would if enjoyed by the current rich.  

Taking death panels seriously

post by Bill Gardner

One of the important decisions that will be made across the US in the next years is whether states will expand Medicaid coverage to include more uninsured patients. The Affordable Care Act provides increased federal funding to do this, but the decision in National Federation of Independent Business v. Sebelius permits states to refuse to expand Medicaid and cover these patients.

There are many things at stake here. Randy Barnett thinks that the ACA expands the power of the federal government. Tyler Cowen (@tylercowen) points out that spending more state revenue on Medicaid may require spending less on education or other desirable programs. Josh Barro (@jbarro) argues that taxation to expand public spending on health care creates significant dead weight losses of utility. 

What is also at stake are some of the lives of people who would be covered by Medicaid if coverage expanded. To see this directly, consider what happens when a state contracts Medicaid coverage, as is happening in Arizona. A surgeon, Randy Oppenheimer, has made a video in which some of his patients -- working people who lack insurance and do not qualify for AZ Medicaid -- talk about the consequences of being uninsured. I warn you that the video includes disturbing images of cancer.

 

It's not my view that Oppenheimer's video makes a definitive case for expanding Medicaid coverage. Expanding Medicaid might rescue these patients, but what else would it do? Are we sure that Medicaid expansion would, when all consequences are registered, save lives? I expect, for example, that Barro would argue that the economic gains from reduced taxation would extend lives; perhaps even more than would be saved through the expansion of Medicaid. So it is not my view that Barnett, Cowen, or Barro are apologists for killing.

What I want is for everyone to see that decisions about Medicaid coverage are not simply opportunities to satisfy norms about the relationship between federal and state power or to resolve state budgets. They are also life and death decisions for human beings. Arizona has decided that having dependent children entitles you to lifesaving care, but being single does not. Look no further for a death panel.

And who was on that panel? State voters determine the extent of Medicaid coverage as much as anyone does. Do not imagine that the health care rationers are distant experts, like the to-be-named (and god help them when those names become public) members of the Independent Payment Advisory Board. The US has 50 death panels, the state electorates. We are rationing health care, we have always rationed health care, and the forthcoming decisions by the states about whether to expand Medicaid will be decisions about rationing health care. We are all on death panels, and we need to appreciate the responsibilities that come with that role.

h/t Ross D. Silverman (@phlu)

Thinking about the cost of global warming: Getting beyond GDP

post by Bill Gardner

Preventing (or mitigating) global warming is going to be expensive. But how much should we spend on prevention, and when? To answer that question it seems natural to first estimate how much global warming will cost. But to do that, we have to define a measure of such costs. And that's not easy.

One way to calculate the cost of global warming is to project future trajectories of global gross domestic product (GDP) based on different assumptions about how the climate works and different regimes of energy consumption and greenhouse gas emissions. So one idea is that we should pick the mix of prevention/mitigation strategies that maximizes long term GDP. This is appealing because it gives us a single dimension to rank solutions on. Nordhaus and Stern (I've read only the former) are comprehensive analyses of the costs of warming and preventing warming that measure harm using GDP. For a brief, thoughtful (and conservative) discussion, see Manzi. 

Unfortunately, looking only at GDP is an inadequate way to think about the cost of global warming, for several reasons.

1. GDP sums up the value of the goods and services we produce and consume. It takes no account of the particular individuals who gain and lose. Every analysis I have seen suggests that the current and future global poor will be harmed by warming significantly more than the global rich; whereas the global rich will benefit substantially more from global energy consumption. Even if we were to conclude that the aggregate harm of global warming was acceptable, the uneven distribution of those harms is likely to be unjust.
2. GDP has an uncertain relationship to human well-being, and it is human well-being that truly matters. There are scenarios for global warming in which large areas of the world that are currently temperate will experience sustained summer peak temperatures in the range of 35-40°C (95-104°F), resulting in significant ecosystem loss. This might not, however, dramatically reduce GDP, because all the efforts we have to undertake to adapt to increased global temperature will contribute to GDP. Suppose an increase in temperature kept GDP constant but shifted our economical activity toward producing powerful air conditioners, and away from other things. We would be worse off, whatever GDP says.
3. More importantly, global warming may produce effects on well-being that are not captured by GDP. The increasing concentration of CO2 is rapidly increasing the acidity of the ocean. There is a point where the pH will be incompatible with the survival of most ocean species. If the oceans die, would the resulting loss of GDP be our only loss of well-being?

GDP calculations are important, but they are not a sufficient basis for decision making about global warming.

Silence on global warming

post by Bill Gardner

@baratunde yesterday:

So we've all just decided not to do anything about global warming, right? I just wanna make sure.

The discussion of global warming in the US seems to have disappeared. Why? There appears to be no short-term option for policy action. 1) You cannot pass a bill in the US Senate without 60 votes. 2) You have little possibility of obtaining a necessary Republican vote. 3) An attempt to change policy would certainly lose. 4) Democrats see no purpose in starting a fight they would certainly lose. 5) For better or for worse, the Democratic leadership does not talk about issues on which they will likely lose. 6) So no debate occurs, and  it looks like we have agreed not to do anything.

We have to talk about this. We have to take the long view and come to a full understanding of the scientific, economic, and ethical implications of climate change. If the globe is warming, the issue will be back, and we must be clear on what to do. In a certain way, having no policy options on the table is an advantage for a good discussion. Once the partisan battle is engaged, it is hard not to defend the consensus policy of your side.

Paul and I are reading John Broome's Climate Matters. He is White's Professor of Moral Philosophy at Oxford. Read it, and let's talk.

Making outcomes more equal can advance equality of opportunity

post by Bill Gardner

Paul and I have written recently about two different ways of understanding egalitarianism: equality of outcomes and equality of opportunity. Americans may be more sympathetic to the ideal of equal opportunity than equal outcomes. For example, Larry Summers:

Perhaps the debate and policy focus needs to shift from inequality in outcomes, where attitudes divide sharply and there are limits to what can be done, to inequalities in opportunity. It is hard to see who could disagree with the aspiration to equalize opportunity, or fail to recognize the manifest inequalities in opportunity today.

Policies that are targeted primarily at equality of opportunity, like No Child Left Behind, have bipartisan support in the US. Policies that equalize outcomes, like increasing the progressivity of the tax code, do not.

The thing is, it might be that making outcomes more equal for young families improves equality of opportunity for their children. Here's some evidence for this. Brendan Saloner wrote about a fascinating paper by Hoynes, Miller, and Simon about how increases in income for poor families improve birth outcomes for their children. These authors looked at the association between increases in the generosity of the Earned Income Tax Credit (EITC) -- which happened in 1986, 1990, and 1993 -- and changes in birthweights for infants.

We find that increased EITC income reduces the incidence of low birth weight and increases mean birth weight. For single low education (<= 12 years) mothers, a policy-induced treatment on the treated increase of $1000 in EITC income is associated with a 6.7 to 10.8 percent reduction in the low birth weight rate. Our results suggest that part of the mechanism for this improvement in birth outcomes is the result of more prenatal care, and less negative health behaviors (smoking). 

Being born more or less at term and at a reasonable weight is a marker for successful prenatal development. Many people believe that successful early development is critical for later development of life and workplace skills. In particular, James Heckman argues that 

interventions early in the life cycle of disadvantaged children have much higher economic returns than later interventions such as reduced pupil-teacher ratios, public job training, convict rehabilitation programs, adult literacy programs, tuition subsidies or expenditure on police. 

Heckman believes that the earlier the intervention, the greater the per-dollar lifetime benefit for the child. And an intervention that affects prenatal development is as early as you can get. So it is possible that among the most effective ways to increase equality of opportunity is to reduce inequality of outcomes among young families. 

Like most Americans, I have a strong preference of policies that favor equality of opportunity. But a preference for such measures should also lead to support for certain policies that equalize outcomes for parents of young children, such as the EITC and paid maternal leave.

 

Medicaid as if the people in the system mattered

post by Bill Gardner

Tyler Cowen wrote about the future of Medicaid in last Sunday's New York Times.¹ He pointed out several problems -- albeit some are arguable -- with Medicaid. He warned that Medicaid is likely to be heading for trouble because

as the Medicaid expansion [built into the Affordable Care Act] receives more publicity, voters may realize that they want to spend less money on it.

Liberals should look past Cowen's attempts to bait them.² There is a vital message here.

The progressive elements of the Affordable Care Act can be undone by legislation or regulations interpreting that legislation.

The left has comforted itself with the idea that once the legislation is in place, the ACA will become more popular, like Social Security and Medicare did. Time is supposedly on our side. 

But it is not. I am confident that many elements of the ACA will become popular when they are in place, such as the prohibition on insurance companies considering pre-existing conditions in issuing policies. However, Medicaid is fundamentally redistributive: it is a program for low income children, low income people in long term care, and people with disabilities, all of whom are unable to finance their own care. These funds come from taxes on more affluent people. There are significant groups in the US that, for reasons that seem good by their lights, do not want to pay for the care of the indigent. They have defeated many previous attempts to achieve a more egalitarian health insurance system, and they are not leaving the field after one defeat. Cowen's piece games out several paths future health care conflicts might take. Liberals need to think this way.

Time is not on our side because the continuing increase in the elderly population, the continuing increase in per-capital health spending, and the increasing fiscal problems of US states will make the redistributive components of the ACA harder and harder to finance. Therefore, we have to make the health care system more popular by making it better, rather than just waiting for the rest of American to see the light. We must urgently solve the problem of health care growth, and we have to improve every aspect of service delivery in Medicaid and elsewhere.

Nevertheless, I come not to praise Tyler Cowen, but to bury him. He discusses future changes in Medicaid without considering the outcomes for the population it cares for. He supports mandated insurance for catastrophic care. If so, which persons and what services, if any, will no longer be covered? If some of the population currently covered will be excluded, does he anticipate that they will receive care financed through some other mechanism? If another financing scheme will cover them, will any money actually be saved? And if no other scheme comes into place...?

In addition, Cowen expects that

The greater likelihood is that, over time, American voters will rebel against Medicaid and dismantle the subsidies that keep the states locked in, and will prefer instead to spend the money on other programs.

This frames the moral structure of the decision to pay for Medicaid like a consumer choice: do I want Medicaid, or other programs like roads, or education? Those are the (seemingly) benign choices. Shouldn't we also consider some of the others things states want to do with their finances, such as reduce taxes altogether? Many states have looming responsibilities for the pensions and health care of retired public sector workers. Should Medicaid be reduced to finance benefits for those more affluent persons? Finally, the Medicaid resistors are cast as rebels against the Federal jailer. However, when voters make this choice, they are usually not consumers making choices for themselves. Relatively few Medicaid families vote, and many patients cannot vote because they are infants in NICUs, demented persons in long term care, or persons with cognitive handicaps in residential care. 

Cowen may have views on the likely outcomes of these political struggles for the people served by Medicaid. He may believe -- and I am open to persuasion -- that this population will be better off under his catastrophic care scheme. But let's discuss Medicaid as if the people in the system mattered. 

---

¹If Cowen's views interest you, see the exchange between Aaron Carroll and Cowen here, here, and here.

²Cowen is an equal-opportunity flame-baiter. Marginal Revolution has lots of posts profaning the sacred cows of libertarians and conservatives.

Two Quick Points About Equality of Opportunity

post by Paul Kelleher

As a coda to Bill's post on equality of outcomes vs. equality of opportunity, I want to make two points about the ideal of equality of opportunity.

First, strictly speaking, a principle of equality of opportunity says only that opportunities must be equal; it does not and cannot evaluate the outcomes among which opportunity is to be equal. But many of us would not tolerate an "opportunity society" wherein every child had the same chances of either ending up in an obscenely affluent minority or in a very poorly off majority. If it is right to prefer greater equality of outcomes than this, then equality of opportunity cannot be the sole or the primary principle of distributive justice.

Second---and here I am just repeating Richard Arneson's insight---a principle of equality of opportunity presupposes a hierarchy of outcomes, access to which is to be equalized by giving everyone the same chances of achieving them. But if equality of opportunity presupposes a hierarchy of outcomes, what justifies allowing the hierarchy in the first place? Arneson's answer is that a hierarchy of outcomes is permitted on efficiency grounds: people will work hard to end up on top, and this makes everyone better off by improving productivity. Assume this is true. Arneson then asks: if the hierarchy is permitted for reasons of efficiency, what should we do if equalizing opportunity has costs in terms of efficiency? This could occur if equalizing opportunity required greater equality of outcomes (since unequal outcomes among parents can lead to unequal opportunities among children).

Arneson doubts that equality of opportunity has intrinsic value, since implementing it might well require policies that make everyone's lives worse (by reducing effort and productivity). An alternative position is that equality of opportunity has intrinsic value, but that its value must be traded off against the value of making the worst off in society as well off as possible. I have expressed related doubts about equality of opportunity and social mobility here. 

On equality of opportunities and equality of outcomes

post by Bill Gardner

This terrific graph was posted by Dylan Matthews (the graph is from Miles Corak). Matthews used it to make this point:

the distinction between equality of opportunity (usually phrased in terms of upward income mobility) and equality of outcomes (the raw distribution of income or wealth in an economy) is not as big as it sometimes appears. More specifically, countries with high inequality of outcomes (as measured by the Gini index of economic inequality) tend to have low social mobility (as measured by the association between parents’ and childrens’ incomes) as well.

He concludes that

The distinction between equality of outcomes and opportunity has some theoretical appeal, but in practice, you get both or neither.

However, the distinction between equality of outcomes and equality of opportunity is of more than theoretical importance. These phrases pick out answers to the question of why inequality is unfair, and different motivations for action to reduce inequality. (You can, of course, care about both kinds of inequality. Or neither.)

Even more importantly, the phrases point to different social policies. If you care about the inequality of outcomes, the natural course of action is to redistribute income or wealth from the haves to the have-nots. If you care about the inequality of opportunities, a natural course of action is seek the equalization of health and education for children, as does Larry Summers in the target article that prompted Matthews' post. Or, as James Heckman argues, you should intervene before public school begins.

These policies are very different. On the one hand, their direct effects benefit different people. Most inequality of outcomes policies benefit adults, whereas inequality of opportunities policies generally benefit children (i.e., the next generation of adults). There is also a difference in time frame. A policy targeting inequality of outcomes can take effect as soon as the beneficiary cashes a check. A policy targeting inequality of opportunity takes effect over a generation, if it ever does.

So if the policies are very different, why do we get either both equality of opportunity and equality of outcomes (Scandinavia, to a lesser degree, Canada), or neither form of equality (South America, to a lesser degree, the United States)? I don't know, but there is an obvious possibility. If your government and culture consistently supports child health and education across generations, this should reduce inequality of opportunity, and equalizing opportunity will reduce inequality of outcomes. 

Should the US have 50 Medicares and Medicaids?

post by Bill Gardner

Brendan Saloner has a great post about federalism and social policy. He discusses whether all Americans should have similar protection, or whether states should have increased discretion about devising a local safety net. The Republican Governors resisting Obamacare would, presumably prefer the latter. The director of Health and Human Services for South Carolina argues that 

South Carolina is not Massachusetts or Vermont no matter how desperately the think tank crowd would want it to be.  We have wildly different rates of poverty, educational attainment, racial mix and economic bases which are the primary drivers of health. Massachusetts was 93% insured prior to their reforms because of their wealth; they weren’t wealthy because they were 93% insured.

Letting states evolve their own solutions to social welfare is attractive, just because that's a more democratic way to proceed. A future world in which big countries devolve into smaller, loosely affiliated countries might be a better place than a future of increased centralization (i.e., East Asia, Eurasia, and Oceana). In addition, decentralization of socialized health insurance is feasible: Each Canadian province has its own health insurance plan, supported by provincial taxes. 

There are, however, several practical problems, all related to scale. First, insurance works better when you have a larger pool of beneficiaries. Second, information technology systems have high fixed capital costs. It's not that expensive to purchase a system that just cuts checks, but if you plan to measure health care quality, monitor population health, or in any way manage care, you are into serious money, and small states will not be able to afford it. Finally, many people suggest that having the individual states experiment with social programs will allow us to learn what works. But having 50 states self-select health care policies is not an optimal experimental design. You'd learn much more from experiments reached across the whole country and randomized by county, or by whether a Medicare participant's social security number was odd or even. In addition, having 50 individual health insurance systems means you largely lose the ability to pool national data to study a problem -- this is a major limitation of the Canadian system.

So there is no clear answer to whether the US should have 50 independent health insurance systems. But the question will be coming up repeatedly in the next few decades.