This Is the End


Markets, Risk and Human Interaction

October 23, 2014

Ex Ante versus Ex Post Social Policy

I have written various posts on social policy related to the question of whether and how we redistribute income. One argument of equal opportunity and social egalitarianism is that it is fair to control for the randomness of life. For example, take the luck of the draw in who your parents are. If your best career choice was to be born into a wealthy family, if you got ahead because your father could pull strings to get you an internship at a big investment bank; because you built up your high school resume by spending a summer building houses in Tibet while a classmate had to spend the summer working at the 7-11; or because you could afford two years of one-on-one tutoring for the SAT, the result might be to score higher in an objective standard of selection and have a thriving career and high income as a result. But is that objective standard a fair standard, is it really merit based?

I think of income redistribution as an ex post policy. Another approach is to make ex ante adjustments to level the playing field, and then step away and let the chips fall where they may. When properly executed the ex ante approach is consistent with a meritocracy, and indeed creates a better, deeper and more successful meritocracy than ignoring the differences in essential endowments.

Assume that there is an objective standard for merit, and a test that correctly ranks the subjects in terms of that standard. (For the record, though basing merit on a testing regime is common in many societies, I do not advocate it). Also assume that we can identify the factors that govern success on the test that are within the control of those taking the test, such as how hard they work, as well identify as the factors that are beyond their control. Given these two assumptions, one scheme for the redistribution, suggested by John Roemer (and in this short post I cannot do justice to his argument and stray from it in various respect), is first to define what constitutes the endowment of important characteristics that are outside a person's control, and then assign people to cohorts based on their levels of this endowment. For example, if the endowment is parents' wealth and parents’ education, we place people into cohorts based on the level of these two factors, with the cohorts made narrow enough so that we can take all those in each cohort as being the same with respect to the endowment.

The social policy is to reward the top one percent of those in each cohort equally, and redistribute as need be in order to do so, and do so for the next percentile across each cohort, and so on all the way down. The idea is that the cohorts take into account differences due to what is beyond a person’s control – the endowments on which the cohorts were based – whereas the person’s place within the cohort is based on those things that are within their control, such as how much effort they expend. That is, by definition a person is not responsible for his cohort, and should not be rewarded or disadvantaged by it, but he is responsible for where he sits in that cohort. So when we make decisions in terms of social policy, whether it be redistribution of income or equal opportunity, we adjust percentile by percentile across the cohorts.

Redistributing income or other rewards percentile by percentile across the cohorts is an ex post approach to social egalitarianism. It is ex post because we look at results versus the endowment and ask how we can make things fairer given the outcomes, where we have determined that fairness means people should be accountable and rewarded only for those things that are within their control.

We can use this methodology on an ex ante basis, where we adjust to create equal opportunity rather than equal ex post reward. Applying the cohort approach ex ante is consistent with a merit-based policy. Doing so requires one more assumption, namely that we only put in the endowment things that are both outside the person’s control and also are improvable if targeted with resources. The aspects of the endowment that we adjust for are those that only make a difference over time and where any reallocation occurs before that difference is significant. This means we do not adjust for innate ability or genetically-based advantage, even though these are things that are outside of one’s control. We do not define the cohorts based on these things, we do not equilibrate these through matching percentile by percentile across the cohorts.

To give a sense of the ex ante social program which melds social egalitarianism with the objectives of a meritocracy, consider the following, which makes a good story and also happens to relate to an actual situation. A number of ten year-olds are trying out for a spot on the school's tennis team. There is one spot available, and at the end of the day it comes down to a match between two boys. One has taken tennis lessons for years and practiced many times a week, the other has only played casually every now and then. The well-trained boy wins the match, but only barely. Given his lower endowment of lessons and practice, the loser is a remarkable talent. On the basis of an ex post meritocracy where the established metric was who won the match, the comparatively mediocre boy got the spot. Yet what is a meritocracy ex post is a mediocrity ex ante. Absent the resources for training and competition the inexperienced boy would receive from being on the team, he goes back to other activities, and the world must make due with one less exceptional tennis player. We will all survive; but the point repeats itself many times over in fields that are weightier.

If we take the amount of tennis training as the endowment that is outside one’s control, the boy without the past training might be at the top one percent of the bottom cohort, while the boy who beat him might be in the middle range of the top cohort. A merit-based social policy will reallocate resources for training to the inexperienced boy; perhaps a level of resources equal to that which the top one percent of the top cohort is receiving. If our measures are correct, and if, as assumed above, the reallocation is given early enough in the child’s formative period, we will have someone who is exceptional at tennis who otherwise would have fallen by the wayside. And we will have used those resources more efficiently than if we kept them with the other boy.

The approach I am describing is not the one we are take in the U.S. when we attempt to level the playing field by making adjustments to create equal opportunity. The social program we follow in the U.S. fails in part because it does not distinguish between that which is within and outside of one’s control. It does not even attempt to define cohorts and pick across the various percentiles to reflect effort. And when it moves into ex ante sphere, it throws resources at differences that are not improvable. (There is one program that might appear to move in this direction: The top ten percent of students from every Texas high school are admitted into the university system. We can think of each school as representing a cohort, with the program giving the same opportunity to those in the top decile of each. But it is pointed in the direction of ex post rather than ex ante social egalitarianism because by the time of high school graduation a student who has had poor preparation is likely to be too far gone for the equal opportunity to yield equal results).

In addition to fostering a fairer society, the ex ante approach described above will yield better results than running people through the merit-based process absent this social policy because there will be a larger group that is equivalent to the top decile of the top cohort. But no matter how well we execute ex ante social egalitarianism, we cannot get away from addressing it ex post as well. Not everything can be measured, and some things that can be measured cannot be mitigated. At some point we have to say we have done enough; we do not want the end result to be physicians equally drawn from the top of each cohort if, at the end of the day, all of the ameliorative actions have failed to close the gap. The question that remains then is at that point, ex post, do we enact a social policy to redistribute income, or do we consider the social task to be completed?