Karelis says that persistent poverty-including behavior is the main cause of persistent poverty, that none of the other major theories can explain what’s going on, and that he can.  How?

He starts by attacking what initially seems like one of the most banal assumptions in all of economics: diminishing marginal utility.  If you could either have $500 for sure, or $1000 with 50% probability, you’d want the sure thing, right?  Economics aside, isn’t this just common sense?  Not according to Karelis.  The real story, he says, is that some goods (“pleasers”) have diminishing marginal utility, but other goods “(relievers”) have increasing marginal utility.  For pleasers, you prefer a sure thing over a gamble with the same expected value; for relievers, however, you prefer a gamble to a sure thing with the same expected value.

Relievers, such as salves, are goods that reduce pain, unhappiness, or misery. Contrary to the law of diminishing marginal utility, which purports to hold for goods generally, relievers are a major type of good that exhibit increasing (not diminishing) marginal benefit. The benefit of relievers—namely, relief—grows as consumption grows, but the benefit grows faster than the consumption. Equal increases in the consumption of relievers produce ever-bigger increases in the relief that is felt by the consumer.

The best evidence of this is common experience. Consider the following scenarios. In Scenario One, you are sitting quietly in a field, and suddenly you are stung by a bee on your hand. The spot that was stung hurts terribly, and it is virtually impossible not to pay close attention. In Scenario Two, likewise, you are stung on the hand, but at the same moment you also are stung six times elsewhere on your body. Surely the sting on the hand would be much more noticed in the first case, where it was the only sting, than in the second, where it was one of seven. The sting on the hand would be like a shout that is striking in a quiet street but hardly noticed in a riot.

Given all this, consider the impact of dabs of salve, each of which we may imagine relieves just one sting. Go back to Scenario Two, the scenario where you suffer seven stings. Suppose that in this situation a single dab of salve is applied to the sting on your hand one second after it happens. Bearing in mind that you still have six uncured stings elsewhere on your body, you would not expect applying this one dab to make much difference. It would be like quieting one shout in a riot. Most people would pay little for a dab of salve for their hand in this situation. But now change the supposition a bit more. What if, just before the dab is applied to your hand, six dabs of salve are applied to the six stings on your body? This surely makes the impact of the dab of salve on the hand much greater…


Moving now from physical to mental relief, we see the same principle in operation. The first scratch sustained by a new car distresses its owner. It is hard to look at the scratch. Each subsequent scratch causes new displeasure, but as the scratches mount up, the new ones add less and less new distress. The seventh may elicit nothing more than an irritated shrug. So the relief that comes from repairing the very first scratch that the car sustained will be greater if the other six are also being fixed than if the other six are not.

Turning to a different kind of mental discomfort, paying the first bill in a stack of overdue bills does little to relieve a guilty conscience. Having nine overdue bills is only a little less uncomfortable than having ten. But paying the last bill in the stack and moving from one to none is normally a big load off one’s mind.

How are “pleasers” different?

These are goods that cause positive experience, as distinct from removing negative experience. Examples might include a glass of wine along with a meal or a portion of ice-cream at the end. To lose or not to have a pleaser is not an evil but only the undoing or absence of a good. Interestingly, while the word “reliever” is a natural name for things that undo the effect of evils, there does not seem to be a name for things that obliterate the effect of pleasers, other than “killjoys,” perhaps.

Unlike relievers, pleasers do generally conform to the law of diminishing marginal utility. That is because what the limits of human attention are diminishing in this case is objects with a positive charge. Virtually all introductory economics textbooks remind students of the psychological impact of successive pleasers that are familiar from everyday experience, things like desserts and movies, to win support for the law of diminishing marginal utility. Students readily agree that the third helping of dessert brings less pleasure than the first, and so forth. But the textbooks do not recognize that significant classes of goods are not pleasers (or not pleasers at all levels of consumption—see below), and so they claim far more generality for the law of diminishing marginal utility than they should.

At first glance, none of this has anything to do with poverty.  But Karelis says otherwise, because many goods switch from relievers to pleasers as their quantity increases.

Reliever/pleasers do not fall completely into either of the preceding categories. Rather they are relievers at low levels of consumption and pleasers at high levels of consumption. Examples include many “basic” goods—things that benefit virtually all consumers: food, shelter, clothing, transportation, leisure, and opportunities to take part in community life. Characteristic of these goods, besides being generally valued, is that they can be used or consumed at three levels: insufficient levels, where shortfalls make for misery and more consumption makes for relief; sufficient levels, which cause neither misery nor positive pleasure; and above-sufficient levels, where the consumer takes a positive enjoyment or satisfaction from consuming them.

Again, the set of goods that serve as both pleasers and relievers includes many basic goods…

The “dual citizenship” of goods of this third type shows itself in their marginal benefit. They act just like pure relievers when insufficient amounts are being consumed, which is to say they yield increasing marginal benefit. But they act just like pure pleasers when it is more-than-sufficient amounts that are being consumed, which is to say they yield diminishing marginal benefit… For this class of goods, in short, the applicable generalization is not the law of diminishing marginal utility but a law of diminishing marginal impact of deviations from sufficiency.

Why then do the poor persistently make impoverishing choices?  Because they’re way down in the “reliever” segment of the reliever/pleaser continuum.  The marginal benefit of doing better is trivial; their only hope of getting to a tolerable position is to make big risky bets.  If they lose, they’re only slightly worse off (because they’re already near maximum misery); if they win, they can have an OK life for a while.

To be more specific about non-work, the postulate of increasing marginal utility makes sense of the fact that very poor people are less likely to exert themselves for money than others. After all, they stand to benefit less from a given addition to their income than not-so-poor people do—not more, as conventional wisdom would have it. So at any given wage, they stand to gain less from an hour of paid work. Recall the person with seven bee-stings who would not sacrifice much to relieve the sting on his hand, seeing that the pain of it was drowned out by the pain of the six stings on his body. This would seem to be the position of very poor people, for whom work, school-work, and (in a much different way) moderation in alcohol use constitute sacrifices that would buy them too little felt relief to be worth making, so many are their troubles.

To be more specific about consumption smoothing, the postulate of increasing marginal utility amid poverty can also rationalize the fact that poor people are less likely than others to save for a rainy (or rainier) day, because, according to that postulate, leveling income down wastes some of the relief that could have been wrung from the sum of the available resources over time. The poor person whose income varies for reasons beyond his control is like someone who awakes every morning with two stings that will hurt all day if not treated with dabs of salve, and who awakes every second day with two dabs of salve on his bedside table. Seeing that two stings are less than twice as painful as one, relieving one is less than half as beneficial as relieving two. So it would be a waste of potential relief to lay aside one dab from the pair he gets on alternate days in order to smooth consumption at the level of one dab per day. At the end of a month, in other words, the same number of dabs will be consumed whether the pattern is 2-0-2-0 . . . or 1-1-1-1 . . . , but the total amount of relief is greater with the first pattern than with the second.

What, however, determines when a good switches from reliever to pleaser?  For Karelis, it mostly comes down to what people in your society take for granted.  Thus, we should not be surprised to see poor immigrants striving for success while seemingly similar natives shrug:

No one doubts that different cultural groups within the United States have different histories, and that these different histories create different economic norms and expectations. For instance, having come from much poorer countries, Asian immigrants to the United States often have relatively low norms and expectations. By contrast, African-Americans, who are closely acquainted with the lifestyles of middle-class whites, and who have long been exposed to “the American dream” and all it implies, often have relatively high norms, if not exactly expectations. As a result, Asian immigrants with low incomes are probably less unhappy with their economic situation than African-Americans in similar straits. When taken together with our hypothesis that the marginal benefit of income is increasing for poor people, this difference in norms and expectations predicts that the felt relief of the marginal dollar will be greater for poor Asian immigrants than the felt relief of the marginal dollar for similarly poor African-Americans. So we should expect poor Asian immigrants to be more willing to do low-wage jobs than equally poor African-Americans. This predicted situation is precisely what we see in fact.

Karelis provides many more details, but we’re basically done.  Yes, poor people could improve their condition emulating the behavior of the middle-class.  But they reasonably spurn this genuine opportunity, because even if they did everything right, their gain would be smaller than their sacrifice.  Being poor is so awful that solving half your problems is barely better than solving none of your problems.

Read Caplan’s next Karelis post here.