The Teleological Paradox in Utilitarianism and Education

In my brief response to Community College Dean a few weeks back, I said something that I think is pretty obvious, but that is often ignored:

humanities advocates spend so much time fighting the instrumental approach to education [because] you’ve got to pretend like time doesn’t matter, or else the education won’t work.

Consider the classic paradox of hedonism articulated by Henry Sedgwick: across a whole range of domains, you cannot maximize utility if you take the maximization of pleasure as the motivation for engaging in activities that tend to be pleasurable. Your motivation matters. Even a hedonist has to have projects, and those projects are only reward to her if she takes them on for their own sake and ignores the utility she will gain.

For instance: being in love increases your utility, but if you approach a potential partner with an offer of mutual hedonism, he will rightly suspect that you are not seeking long-lasting love. “I really want to have a relationship with you, not because I think you’re awesome, but because I think it will make me happy and this seems like a good way to do it” is no way to fall in love; the potential partner will rightly say, “That’s weird, I don’t want to be loved selfishly, I want to be loved for myself.” You might think that you are then lying to your partner when you claim to love him for himself, but in fact we are impressively good at catching such lies. We have to really take that “for-himselfness” of loving seriously, to the extent that we’d even be miserable if our lover dies, in order to garner the hedonistic benefits of loving. The rational hedonist courts just this sacrifice.

You might also go to church to increase utility, but folks would think you were crazy if you went to church and said, “Hi! I’m here not because I have faith in the particular doctrines of this institution, but because I’ve been told that belonging to a community of religious inquiry will increase my utility.” Worse, like a bad Pascalian wager, you’d lose your Sabbath and you wouldn’t get the (mundane) benefits! So it is that the rational hedonist, motivated only by happiness, even courts irrationality!

Now, there’s a similar problem for education: if we instrumentalize education by treating it like a set of skills and practices, or even worse, as the acquisition of discrete knowledge, then the real benefits (especially of college education) will be lost. The real benefits of education are soft skills that are hard to “acquire” in that discrete sense. Habits of mind that enable analytic writing, close reading, critical thinking and problem solving skills cannot be learned unless the student takes a long detour through irrelevant material. So there’s a similar teleological or motivational problem to the one facing the hedonist: you have to read the Classics for their own sake in order to become a better advertising executive, even though reading the Classics isn’t directly relevant to advertising. Following Sedgwick, , T. M. Scanlon calls this the “teleological paradox” in a long footnote to What We Owe to Each Other, which can be described thus:

though the telos in question may depend on factors within an agent’s control, that does not mean that it is rational for the agent to target it and make it in that sense a matter of active demand.

In many areas, you must do things that are only instrumentally-related to your goal in order to accomplish your goal. The risk, however, is that this process of divergence from the goal might actually become counter-productive or self-deceptive: how will we know if the instrumental means supplants the true goal entirely?

Now, here’s where it gets interesting. In metaethics, there is a response to the teleological paradox and the problem of self-deception: R. M Hare’s “two-stage” or “two-level” utilitarianism. In the first stage, our ordinary lives, we take on ordinary justifications for our projects. But every so often we reflect on the value of particular methods and motivations, and during this reflective second-stage we tweak them from the perspective of the overarching goal. Maybe we only do this at the level of institutions or laws. So, instead of saying “You should have a family because it will bring you pleasure,” we normally just advocate family life for the normal, intuitive reasons that preserve love and loyalty as ultimate ends. But at the level of policy, we still ask questions like, “Should we incentivize large or small households?” or “Should we give tax breaks for children or not?” That second stage allows us to take on the utilitarian perspective for the purposes of improving our projects. So the teleological problem dissolves, so long as we’re able to willfully blind ourselves to our ultimate motivations: we get to be utilitarians some of the time, but when it matters we can be fathers and mothers, lovers and church-goers, citizens and consumers, etc.  Virtue and deontology are then sublimated under a utilitarian perspective, called to bear when they’re best suited to some basically utilitarian goal.

The same solution works for education. In the classroom, the library, and the lab, we can embrace wasteful irrelevance, detours into difficulty, and the rigors of basic research. Then later, at the level of syllabus-construction, course-design, academic policy, project funding, or tenure-line evaluations, we can ask: “What are the instrumental educational goals that we’re trying to accomplish? What is the best (most efficient, most effective) method for achieving those goals?” Yet because we are not fundamentally committed to any particular major or method of instruction, we can also ask: “How many Classics majors do we really need? Can we get the same benefits from Philosophy or Anthropology?” The close, careful reading of abstruse literature, abstract and irrelevant mathematical work, or the cultivation of the jargonistic language of High Theory all become tools, but tools we take up as if they are ends in themselves.

The key to this process is that at the administrative or policy level we have to seriously believe that these projects matter: we have to actually commit ourselves to the claim that studying the Classics (or Philosophy or Anthropology) is more useful than studying Accounting, even if what we want are more accountants! That may seem odd or self-deceptive, but the evidence suggests we have no other choice: when we study the data and look at the Collegiate Learning Assessment, we end up concluding that the most useful education is the one that focuses on the least useful work.

So, even though we’re pure instrumentalists at the policy level, as instrumentalists, we become committed to the rejection of instrumental approaches. Not always, not if better evidence comes along, but for the time being, given our current knowledge, etc. We become instrumentally non-instrumental. When we are in front of the classroom or when we are advising students about majors, we should discourage an instrumental relationship to education. And that means that we have to discourage instrumentalism when we are deciding which programs to fund, too.

Season of Political Irrelevance Update

Weigel has predicated a lot on the conditional statement: “If you look at it right, then you’ll see serious policy.” But we don’t have any evidence for the antecedent, that the public or the media *will* “look at it right.” More to the point, I don’t agree that the waivers or tax loopholes are among the most important problems facing this or the next president. Effective tax rates are much too low on the rich, but the best solution isn’t better income tax laws, it’s a progressive consumption tax, which is not on the table.

The policy issues under discussion are partly unsubstantive because they’re subject to deep red meat divisions. They won’t be resolved; they’re designed to be perpetual. The stuff I want to talk about is stuff that a whole host of people from both sides could agree on, if they were allowed to spend the election talking about it. Elections are about drawing distinctions and offering a choice, but they’re also about parsing the electorate into roughly equal demographic slices. Policy is about taking the broad overlapping consensus and the best evidence and acting on that.

Over the next two years, there will be very little movement on the things we argue about during this election season. During the same time span, there will be a great deal of movement on the things we ignore this election season. That suggests irrelevance.

The Season of Political Irrelevance

It is my considered opinion that the next three months will involve no serious deliberations regarding substantive public policy. Though readership and viewership for such matters will be at its highest, none of the things discussed will be discussed in a way that comports with public reason or with anything like the goal of exchanging reasons and evidence in the search for truth-tracking beliefs. Our best analysts, pundits, and public intellectuals will be busy with horse-race coverage and fact-checking the candidates’ claims. Worse, few of the matters discussed in highly rhetorical fashion, upon which our fellow citizens will be asked to make their determinations, will even be relevant to the public policy matters that ought to concern us most.

Here are the things I suspect we will discuss most:

  • Taxes
  • Jobs and the economy
  • The distribution of income and wealth
  • Globalization and outsourcing

Since none of these are under the control of the presidency, it’s absurd to stage the debates on these matters around the presidential election. And yet we will.

Since this is a highly cynical claim for a democratic political philosopher to make (well, not contentious among professionals) here are the things we ought to be talking about:

  • Climate change
  • Mass incarceration and its causes
  • Immigration
  • Regulatory agency capture by the financial sector
  • The proper size and role of the US military

Notice that all of these policies are administered by agencies under the President’s control. But perhaps even this is undemocratic. Accounts of politics that focus on leaders and the vertical measures of “greatness” are at odds with the pervasive sense of horizontality that ought to guide us in a democracy. Presidents are not the only political actors, nor even the most important: they perch atop the bureaucratic state barely able to steer it, using the reins merely to hold on to their office a bit longer.

Here’s Arendt in Reflections on Violence:

These definitions coincide with the terms which, since Greek antiquity, have been used to define the forms of government as the rule of man over man—of one or the few in monarchy and oligarchy, of the best or the many in aristocracy and democracy, to which today we ought to add the latest and perhaps most formidable form of such dominion, bureaucracy, or the rule by an intricate system of bureaux in which no men, neither one nor the best, neither the few nor the many, can be held responsible, and which could be properly called the rule by Nobody. Indeed, if we identify tyranny as the government that is not held to give account of itself, rule by Nobody is clearly the most tyrannical of all, since there is no one left who could even be asked to answer for what is being done. It is this state of affairs which is among the most potent causes for the current world-wide rebellious unrest.

Still relevant, forty-three years later.

Another Badly-Aimed Attack on the Basic Income Guarantee from Crooked Timber

John Quiggin has been taking up the case against the basic income guarantee at Crooked Timber recently. See here and here. Unfortunately, he is attacking a weak man version of the policy.

It doesn’t look like he actually opposes the BIG, in theory, but his objections all appear to demonstrate that a Basic Income is much too hard and expensive to implement, short of a revolution. As such, he ignores the experimental evidence out of Namibia, Manitoba, and the US. Quiggin starts by eliminating an unconditional and universal basic income guarantee and instead focuses on a means-tested guaranteed minimum. Then he assumes that this will reduce employment as workers leave the workforce (presumably to surf). He also assumes that it will be funded by income taxes, natural resource extraction taxes, or capital expropriation.

All of these assumptions are wrong, and caste the BIG in the worst possible light. The real appeal of the BIG is that it is easy to implement, can work piecemeal, will likely increase employment and productivity, and can ramp up slowly, and even unsteadily, to make the transition politically palatable and in accordance with the rule of law.

Cheap at Any Price (“Cost” versus “Dead-weight Loss”)

How much would a basic income cost? There are a few ways to ask this question:

  • “How much would we have to raise taxes in order to pay for this policy?”
  • “How much productivity would we lose under this policy?”
  • “What would the effective tax rate be under this policy?”

It’s important to note that these are separate questions. We would certainly have to raise tax rates to pay for a basic income: there is not a currently a revenue stream devoted to it. These higher taxes might reduce people’s propensity to work, and still more productivity might be lost because people choose not to work when their needs are met. However, depending on how those taxes are collected, such an increase might not increase the effective tax rate: the difference between the tax rate and the services supplied. The early Crooked Timber discussion totally confused effective rates with headline rates, in much the same way that people pretend that the rich actually paid 91% under Eisenhower. They didn’t.

The best way to ask this question would be:

  • “What is the dead-weight loss of a basic income guarantee and its associated taxation, compared to what we have now?”

The answer is that, on the most plausible accounts, a basic income guarantee is actually cheaper than the current system.

It’s important to note that the current cost of government in the US is approximately 21.41% of GDP. That’s the GDP divided by the Disposable Personal Income. Doing it this way adds back transfers payments and the positive benefits of government like firefighters and medical research. 21.41% is how much money we spend on all the stuff we don’t really want: invasions and aerial drones, pretextual traffic stops, cavity searches, TSA employees, drug warriors, and CIA black sites.

Government as a whole produces net benefits (a social surplus) because it makes transactions possible through the rule of law. Yet GDP probably overvalues government services and undervalues technological innovation over time, so it’s pretty controversial whether the number is anywhere close to this  21.41%  in reality.

In contrast, calculating the economic cost of a transfer payment is pretty easy: I take $100 from Bill Gates and give it to a poor person, there’s no lost disposable personal income in that transaction. Gates has less, but the poor person has more. Yet some ways of funding that transfer can be said to produce a dead-weight loss, an “excess burden” compared to other revenue-generating mechanisms. If it costs me $10 dollars to pay for the administrative costs of running the $100 transfer, the transfer tax costs 10% of disposable personal income. If Bill Gates produces $10 less income than he would otherwise because he has less of an incentive to work knowing that some of his money will be taxed, then the deadweight loss is 20%. ($10 from lost productivity, and $10 to pay for the transfer.) The current administrative costs of Social Security are 0.9%, which isn’t much, so the question is, how much less will Gates and the poor person work under the BIG?

It’s a trick question, of course: Bill Gates doesn’t work, and neither do the very poor! The capitalist allows his investments to work for him, and the poor can’t afford to work or they’ll lose their benefits. So some transfers are less distortionary than others.

Unconditional & Universal (Guarantees reduce Dead-weight Losses)

Here’s Quiggin:

The big question is whether current workers will respond by leaving the workforce and relying on the basic income. We’d expect and want this to happen to some extent – the whole idea is to free workers from absolute dependence on wage income. But if the shift is too large, the tax burden will become unsustainable.

How large is too large? Suppose the employment rate falls from 60 per cent to 50 per cent, and that capital income falls in line with labor income, so that a larger benefit cost is being supported by a smaller income.

The basic income guarantee solves a problem that Quiggin seems to think it creates: various parts of the government actively discourage remunerative work by creating a “poverty trap” consisting of various means-tested programs that are necessary for survival but are lost as incomes increase. Many of the poor literally can’t afford to work: they’re exiled from the (remunerative) labor pool. That’s a powerful coercive effect that more than trumps the lessened interest in employment felt by middle-class workers who conclude they have enough. This will also be a problem with a negative income tax, which is why a universal guaranteee is actually more efficient.

All of the experiments designed to test the employment effect found that work effort increased under a BIG: in Namibia, workforce participation rose from 44% to 55%! The poor can work for each other if they become the middle class. In Manitoba, hours worked by new mothers and teenagers decreased, but only because the teenagers focused on education and the new mothers on child-rearing. In that sense, remunerative work declined slightly, but productive work effort increased: this isn’t about a “right to surf.” Each marginal dollar earned still increases a poor family’s well-being, but now they can afford to make longer-term decisions by investing in human capital.

Note: the US experiments with negative income tax do show a 13% decrease in hours worked for wages, but similar increases in educational effort to Manitoba’s Mincome, and didn’t test overall employment effects from new labor market entrants because the unit of experimentation was the household. This is a famous moment in social scientific research, when we noticed that the “unit of analysis” matters.

Populations with a basic income guarantee work more, not less!

Tax Consumption, Not Income (If you want more of it, why tax it?)

Quiggin worries that marginal income tax rates needed to fund a basic income proposal might become unsustainable:

Depending on the design of the tax scales and the mix between income and other taxes, the marginal rate for the average worker would probably be around 40 per cent, and with a moderately progressive tax scale, lots of workers would be paying marginal rates above 50 per cent.

For a lot of people, the 50% marginal tax rate mark has psychological significance, since it seems like you’re working for more for the government than for yourself. But that’s precisely why we should focus our taxing attentions on consumption, a belief shared by the majority of economists on both the left and right.

A Basic Income Guarantee causes some to receive payments who don’t actually “need” it. But that’s not such a problem if we fund the basic income with a sales tax or value added tax. (A value-added tax, or VAT, is like a sales tax that is collected all along the supply chain rather than only at the point of sale: it’s cheaper to administer and harder to cheat than a sales tax.) That way, we can recapture the basic income grant paid to the very rich, who will spend more of their earned and basic incomes on consumption and thus repay the basic grant. I’ve written about this before, herehere, and here.

An ideal level for both the VAT and the basic income might be to cause the bottom two quintiles of the population to receive a subsidy paid by the top three quintiles. In the US in 2003, this would have been about $34,738: anyone earning less than this would be subsidized by anyone earning more. Other proposals might aim to redistribute around the median income of $44,000, such that a household earning the median income would receive exactly the same amount in basic income that they pay out in VAT (if they spent every penny they earned.) Since the basic income is designed to supply the poorest with the means of subsistence, the VAT might be as high as 30%: in the median proposal, each household would receive 30% of the median income (about $13,200) as a guaranteed payment, but the median income household would pay out the same amount over the year in sales tax, thus breaking even. This is the best way to widen the tax base while enhancing progressivity.

Transition is as Easy as a Tax Rebate (A BIG can start small!)

Quiggin claims that that the BIG+VAT would require a “big bang” rather than a slow transition:

The simplest way to get to a universal basic income would be to pay it to everyone, then recoup the cost, through the tax system, from everyone above the basic level. While conceptually simple, this way of doing things would be almost impossible to implement except as a ‘big bang’, and is also too hard for me to evaluate.

Along with Chris Bertram, Quiggin has been saying that the path to a BIG is totally unimaginable:

So, any serious approach must, as Chris suggests involve transferring a large proportion of existing wealth from its private owners to the public. Leaving aside questions of justice, I can’t see an obvious transition path here. The problem, as illustrated by the existing wealth-based funds, is that it takes a lot of capital to generate a return that would finance a UBI at or above the poverty line. In the US context, you’d need something much larger than the Social Security Trust Fund (2.7 trillion) to get near the target. On the other hand, it’s hard to see how a universal payment at a level far below the poverty line could mobilise the kind of popular support needed for a policy of radical redistribution.

Hardly! You can start with a partial BIG+VAT and grow it over time. The current proposals for a carbon tax or a national sales tax would both suffice. Make these important Pigovian taxes palatable by giving them back to the people using classic Republican tax rhetoric. Just don’t give them back in the same distributional pattern as they were collected: take a 5% tax on all carbon emissions, and use the revenue to pay every American a small annual income. Call it a “rebate.” A partial BIG wouldn’t replace the current welfare state, but it could help ween us off means-testing, reduce the size of the poverty trap, and help us iron out the kinks in this plan while supplying lots more evidence for pundits, wonks, and social scientists to study.

Take your time, the poor will still be here: we can do 5% now, and another 5% in a decade.

As it grows, the BIG can replace the expensive or particularly paternalistic parts of the welfare state, things like Social Security (which isn’t actually egalitarian) and food stamps. The BIG is tailor-made for these kinds of gradual tradeoffs; it enlists the middle-class in poverty-alleviation. Just think of all the ways in which conservative economists and progressive activists can ally to accomplish economically efficient policy changes: a minimum wage becomes a clear obstacle to employment (with no distributional upside) if everyone has a BIG, and that means even greater workforce participation, as would decreased unemployment insurance, which current employers must factor in to the cost of hiring decisions.

What’s more, we needn’t transfer any wealth at all! For a consumption tax, the current economy is all we need to get started: the household part of the US economy is worth $57 trillion or so. We only need to redirect the income stream from that enormous wealth. If a rich person lives as frugally as Warren Buffet, she won’t pay much in consumption taxes, but that’s okay. We don’t want to discourage investment and productivity increases! We want more of that, not less. We want less conspicuous consumption, right? We want less class difference! You may think that this requires wealth redistribution, but even the Scandinavian social democracies actually have quite large wealth inequality. What matters most is how people live.

The Fallacy Fallacy [sic] of Mood Affiliation (Workplace Domination Part Two)

In his initial response to the the Crooked Timber bloggers, Cowen also suggests that he doesn’t like the “mood affiliation” of the CT bloggers:

I am not comfortable with the mood affiliation of the piece.  How about a simple mention of the massive magnitude of employee theft in the United States, perhaps in the context of a boss wishing to search an employee?

Cowens’ “fallacy of mood affiliation” is an interesting and useful attempt to describe a kind of sophisticated motivated skepticism that occurs when we evaluate evidence that counters our basically optimistic or pessimistic views of the world. When he first introduced it, Cowen described mood affiliations that caused people to misrecognize particular evidence regarding innovations or environmental effects because the particular evidence fails to confirm their preferences for optimistic accounts of future growth and environmental improvements.

But to those clear examples of the optimism bias, he added two other examples that are only indirectly related:

3. People who see a political war against the interests of the poor and thus who are reluctant to present or digest analyses which blame some of the problems of the poor on…the poor themselves. (Try bringing up “predatory borrowing” in any discussion of “predatory lending” and see what happens.) There’s simply an urgent feeling that any negative or pessimistic or undeserving view of the poor needs to be countered.

4. People who see raising or lowering the relative status of Republicans (or some other group) as the main purpose of analysis, and thus who judge the dispassionate analysis of others, or for that matter the partisan analysis of others, by this standard. There’s simply an urgent feeling that any positive or optimistic or deserving view of the Republicans needs to be countered.

#4 is also clearly a bias where in-group solidarity blinds us to evidence, and Cowen has written about this well in the past. It is not, however, an obvious “mood affiliation” except by analogy, and it serves a pragmatic purpose: you can only call your friends our for being biased so often before they stop being your friends.

#3, though, is neither a mood affiliation nor an optimism bias. We might call it an “unjust-world fallacy,” if we really need a name for it. However, I’d suggest we might want to avoid prejudicing discussions of what makes people poor with attributions of fallacies and congitive biases until we’ve evaluated the evidence.

Since “what makes people poor” is a hotly debated academic question, there’s a lot of evidence, and it pushes in multiple directions. (My own money is on some version of Buddy Karelis’s book, The Persistence of Poverty (pdf) though there’s plenty of room for poverty traps and marginal tax rate arguments.) People affiliate around these positions in many of the same ways that they affiliate around political parties. But there’s a serious dispute in the literature and the question really, really matters, so let’s not glibly reduce our opponents to fallacy-mongers here.

This is relevant to blogging about the workplace only because, by analogy, we’re supposed to believe that employees might be partly to blame for their domination in the same way that poor people are partly to blame for their poverty. But note, there are particular actions the poor engage in that make them poor: failing to finish high school, committing crimes, and getting pregnant out of wedlock are individual actions that primarily harm the individual who enacts them by reducing lifetime wages. In the workplace example, there just aren’t particular actions that workers engage in that justify their being searched or filmed while going to the bathroom (except maybe being unwilling to quit, fight, or unionize). Invading my privacy because somebody else has been stealing doesn’t really fit the kind of personal responsibility motif that Cowen was pushing in the original discussion of poverty. Plus, employee theft costs our economy about $15 billion, which is 0.1% of GDP, and that’s including serious embezzlement in addition to retail “shrink,” so it’s not really so big a deal as Cowen makes it out to be.

Mood affiliation concerns don’t appear to be relevant to workplace domination issues, they threaten to resolve into ad hominem and fallacy fallacy [sic] issues, so let’s drop them and look at the data and the arguments.