I'm going to start by walking you through the following graph adapted from Reinhardt's article, Can Efficiency in Health Care Be Left to the Market?, a comment written in response to Kenneth Arrow's Uncertainty and the Welfare Economics of Medical Care. It is nearly identical to the one he provided in his recent NYT blog.Source: Uwe Reinhardt, Can Efficiency Be Left to the Market? In Kenneth Arrow and the Changing Economics of Health Care, Duke U Press, 2003.
The graph takes a form that is fairly common in economics. It is often used to depict a two good or two person world. The curve is often seen in intro econ textbooks as a production possibilities curve where it represents all possible combinations of the (two) goods that could be produced from a fixed set of resources. A similar graph, like that above, can be used to depict all possible combinations of (in this case, two people's) happiness (sometimes called "utility") that could be produced from available societal resources. As is frequently the case and without too much loss of generality, we have simplified by assuming a two-person world, Person A and Person B, among whom the available pool of happiness is to be divided. (Some of you will already be seeing at least one potential flaw with this, but bear with me.)
Now suppose that you are an economist and you believe that 1) happiness is difficult to measure and 2) it is difficult to compare the happiness of A to that of B. Let's therefore imagine that corresponding to each point on the curve is a certain distribution of national (in this case two person) resources. (Each possible distribution of resources could be represented by a straight line connecting W and Z with W representing 100% of national resources going to B (and 0% to A) and Z representing 100% of national resources going to A (with 0% going to B.) I think it would be uncontroversial under these conditions to assert that the share of national resources (e.g., output, education, health care) a person receives provides a rough indicator of relative happiness, i.e., the larger the share you receive, the happier we would expect you to be. We could call it an "indirect happiness function." (Some of you will be quick to point out that there is not a very good correspondence between income and happiness, but bear with me, please.)
At point W, B is a very happy person (or at least rich in terms of resources and the positive sentiments that may accompany this state), A not so much. At Z, A is a very happy, resource-rich person and B is the person left out in the cold.
The curvature of the happiness frontier that connects W and Z reflects the idea that the rate at which happiness increases (decreases) as one's share of resources increases (decreases), decreases (increases) as one holds more (fewer) resources. Think about someone with very low income, low access to education and health care, say person B at a point just shy of Z. The curvature is rather steep there, reflecting that a small increment in B's share of resources when B doesn't have very much will produce a much larger gain in B's happiness relative to the loss of happiness that A experiences. As B's happiness (and share of resources) gets closer to W, the curve flattens, reflecting that even large increases in resource-share produce relatively smaller increases in happiness for B relative to rather steep reductions in happiness for A. The converse is true for A who at W is experiencing very large losses in happiness (and presumably resources) with each of B's small gains in happiness and share of resources.
As Reinhardt has already pointed out, all points on the curve are Pareto efficient. If the distribution of happiness is at, say, point D (or any other point on the curve), we cannot move in any direction along the curve without making either A or B less happy. Only if we are at a point inside the curve is it possible to move to a place where at least one person is happier without the other person being made less happy. From point C, if we move to point G, we make B happier, while A remains as happy as she was before the move. (If you doubt this, print the above and sketch in lines from these points to the two axes and note how A and B's happiness changes.) If we move to point H, we make A happier, while B remains as happy as before. If we move to any point on the happiness possibilities curve that is bounded by points G and H, both A and B will be happier than they were at the Pareto inefficient point C.
Movement from C to anywhere between G and H seems pretty uncontroversial. After all, who wouldn't agree that if we can rearrange output or education or health care or anything else that contributes to happiness in ways that make no one less happy and make at least one other person happier, this is a Good Thing. The problem with Pareto efficiency arises if instead of being at, say, point E, where A and B have equal shares resources (and presumably equal happiness), we are instead at point D. Point D is Pareto efficient, but B has a near monopoly on both happiness and resources. Yet, among economists and some politicians, point D is preferred to point C because it is efficient.
As a thought experiment, imagine that at point D, A does not have enough income to attain adequate nutrition or health insurance, that A's life is at risk at point D. Surely that should matter, yes? Or imagine that point D was arrived at because B was an investment banker whose self-dealing behaviors precipitated an economy-wide downturn that also shifted inordinate amounts of wealth in B's direction. Does a rule based on not allowing A to be made better off at B's expense make sense here? Does a Kaldor-Hicks rule that requires A to compensate B for a move that improves A's welfare at B's expense make sense here? Should we be reassured to learn that competitive markets guarantee that trade and production will arrive at Pareto efficient points without consideration of how the initial or any subsequent distribution of resources has been achieved? How reassured should we be if markets aren't competitive? What if highly unequal distributions, such as D, cause the loser to implement involuntary income transfers, i.e., crime rises, trust is lost, A and B retreat into armed camps? Are the happiness of A and B, independent or are they interdependent?
Reinhardt makes the point in his paper that inefficient, but more equitable, point C may well be viewed as superior to efficient, but highly inequitable point D. A society's decisions about this may well hinge on the stories it tells itself about A and B. If B is productive and A is slothful, then D may seem right and fair. If A is the producer and B is the slacker, then D will seem even more unfair than it might if both were equally productive or equally slothful. Given that there is likely to be some underlying heterogeneity in productive capacity between A and B, how are we to make such judgments? Would we want to base it only on productive capacity? How would we choose the "right" fair but inefficient point that is preferable to an efficient, but possibly unfair point, like D?
But there's more. It's
here that the correspondence between resource share and happiness
becomes difficult. Assume that C-E is a 45 degree line through the
origin that represents equality of resource shares. Then any movement
from C to the arc bounded by G-E requires that more resources go to B.
Similarly, movement from C to any point on the arc bounded by E-H
requires that more resources go to A. Yet both movements result in
higher happiness for both individuals. What should guide a decision to move from C to the G-H arc and where on it should we be?
One of the things about the above diagram that puzzled me from the very first time I saw one like it was: why that shape? Surely over some range, particularly when we're graphing the happiness of two individuals, it must be possible that the happiness of another delights us, yes? Clearly, yes, when the other is a near and dear friend or relative, but surely also for those more distal in our affections, too. Is happiness always a zero sum game at the frontier?
Here's Adam Smith, very first sentence of Theory of Moral Sentiments:
How selfish soever man may be supposed, there are evidently some principles in his nature, which interest him in the fortune of others, and render their happiness necessary to him, though he derives nothing from it, except the pleasure of seeing it.
There are economic models of altruism, but for me most are very unsatisfying. For one thing, I prefer to live in a world where I don't have to step over bodies or live behind gate houses and fences. I'm pretty sure that requires me to care about the welfare of people other than my immediate family and descendants and probably on levels that are not purely instrumental, reciprocal, or utilitarian. For another, it seems to me to be highly inefficient to create a society in which the absence of social and health insurance allows uncertainty to rob the able and talented of both entrepreneurship, innovation, and their worldly goods when bad luck, bad choices or bad bankers result in individual or society-wide downturns. These sentiments are not selfless, as altruism is often viewed. I derive something beneficial from living in an economy and a society in which we all share in some to be determined distribution of resources upon which depends our joint and individual happiness. I may give up some resources or output, but I may gain trust, safety, community, some free time, and possibly more innovation depending on how free time is used.
On the other hand, I might not benefit at all from improvements to others' well-being. Sometimes it's just the right, fair, just thing to do.
So what would the above diagram look like where over some range, A delights in or benefits from improvements to B's lot in life and vice versa? Here's something from John Rawls' Justice as Fairness: A Restatement (p. 62) (with apologies for the black background). (See also Atkinson & Stiglitz, Lectures on Public Economics, 1980, McGraw-Hill, page 338 and accompanying discussion.)
Source: Rawls, J, Justice as Fairness: A Restatement, Harvard U Press, 2003, p. 62-63.
In this diagram, person B (vertical axis) is now a representative person from the Least Advantaged Group (denoted LAG) and person A (horizontal axis) is now a representative person from the Most Advantaged Group (denoted MAG).This is a social welfare function or happiness possibilities curve with both axes representing a share of primary goods allocated to either MAG or LAG. The index of primary goods represents "a person's prospects of income and wealth over a complete life" and includes 1) basic rights and liberties such as freedom of thought, conscience, and the rest; 2) freedom of movement and free choice of occupation in a climate of multiple diverse opportunities; 3) powers and prerogatives of offices and positions of authority and responsibility; 4) income and wealth in an economy or social context where both have exchange value; and 5) social institutions that support self-respect, the worth of all persons, and the ability of individuals to advance their own ends with self-confidence. Note that this is a much richer list than simply assuming that a person's share of national resources adequately indexes their utility. It captures political and institutional aspects of society that also influence individual well-being. For this reason, it is richer than the Pareto efficiency criterion that dominates economic assessments of welfare. The curve lies below the 45 degree line from the origin because the initial distribution of resources (or primary goods) favors MAG.
Rawls provides one possible framework for choosing an inefficient point C, that lies somewhere within the happiness frontier in the first graph and not necessarily on the 45 degree equality line.
The main difference between this and Reinhardt's graph is that from O to D, the shares of primary goods and the happiness of both persons or groups (since each person is representative) are increasing together. Point D is the point at which happiness is (efficiently) maximized subject to an equal justice constraint denoted by the horizontal lines. (Think of the lines as the highest attainable jointly produced happiness that also provides the highest level of primary goods to LAG.) The justice constraint derives from Rawls' difference principle, which tolerates inequalities as long as they benefit LAG.
Point B, which represents a higher level of total happiness or utility than point D in Rawls' graph, is the utilitarian or Bentham bliss point. B is the point that standard neoclassical economics and economic welfare analytics would identify as optimal because total happiness is at it's highest point. N is the Nash equilibrium, which in this case can be thought of as the happiness or utility maximizing solution you would get (in this world in which output and utility depend on cooperation between LAG and MAG) if you did not have the justice constraint (that happiness/utility be maximized subject to making LAG was well of as is possible). Point F is what Rawls calls the "feudal point" where all the utility (and presumably resources) are concentrated with MAG. The important thing to note about F is that MAG is extremely well-off, LAG is extremely disadvantaged, and societal happiness (and output) could be higher if primary goods were redistributed in such a way that LAG and MAG could move back to B. This last is important. It must be so if B is a true maximum of total utility.
I occasionally hear rhetoric that seems to suggest that the more resources are concentrated at the top (among the MAG), the better off we will all be (since it will trickle down upon us all). When Adam Smith was writing about the nature and causes of the wealth of nations, he was comparing point F to a point that was approaching point B from F, i.e., national wealth (and presumably total utility/happiness) was growing as commercial exchange (when not impeded by mercantile interests) and the division of labor reshaped the distribution of national income to favor LAG a bit more.
Also note that the Rawlsian bliss point, D, is not a point at which both individuals share equally in primary goods (or presumably in equal utility or happiness). That point would have to fall on the 45 degree line from the origin. Equality would result in both very low happiness for both and also very low national output.
For Rawls, the rising portion of the utility/happiness curve (which I'm assuming tracks a similarly shaped national output curve) results from cooperation in production and in each other's happiness. This might be because LAG delights in improvements to MAG's welfare and vice versa or it might be because both MAG and LAG benefit in instrumental ways from improvements to each others' welfare. For example, in the kind of world that libertarians and free marketers believe we live in, MAG may be gifted with entrepreneurial skill or inventive talent. MAG creates jobs and invents cures for cancer (MAG does not capture monopoly rents while doing these things). LAG benefits from this so LAG's happiness increases with MAG's contribution to jointly produced output.
There are other reasons why we might observe interdependent happiness and productivity. MAG and LAG might be related or they might be friends. They might be neighbors and LAG is about to have her house foreclosed. The deterioration of lot and house as it stands empty will affect the price of MAG's house. This is an externality of LAG's misfortune that negatively impacts MAG. Externalities are important sources of interdependent utilities that are often ignored, underestimated, or assumed away in economic analysis. At a minimum, cooperation would build and reinforce trust and reduce transaction costs. It almost certainly reduces crime and the costs associated with it as well as societal (and political) conflict, which benefits everyone.
Let me just make a final point for anyone with the fortitude to have read this far. Uwe Reinhardt has already described the ways in which "efficiency," a term with precise economic and scientific meaning, has taken on or perhaps has always implicitly contained value-laden meaning. The fact that it is almost always viewed as a Good Thing should prompt us to look at it more critically, particularly when we suspect we are moving over time to something like point D in the first graph above or point F in the second graph. But notice also that how we conceptualize productive effort and the way we draw our curves can shape the way we think about these problems. The first curve above requires that, unless one is in the interior of the curve, happiness is a zero-sum game, that I can only gain more happiness at your expense and you at mine. This problem will be amplified if we rather mindlessly assume that whatever point we happen find ourselves at is on that Pareto efficient frontier.
The second graph suggests that there is some range over which we can all gain in happiness and output if only we will cooperate.
Much of what passes for political discourse is really about where we as a society want to be on one of these curves and whether or not the curve accurately captures the utility/happiness "production process" of a society. I'm almost certain that it is cooperative over some range, much like that of any other production process. (Otherwise, there would be no competitive advantage to large corporations; all businesses would be small individual firms.)
The fact that over the last 30 years we seem to have been approaching feudal point F, where resources and whatever happiness it may afford go disproportionately to the most advantaged, should cause all of us concern. That the transfer has made the least advantaged even worse off should be reason to question the entire underlying Paretian and Kaldor-Hicks welfare framework, at least as currently conceptualized and taught.As to the incomparability of individual utilities, I may not be able to compare MAG and LAG's utility directly, but I'm pretty sure that when MAG has most of whatever happiness money can buy (and the resources with which to buy it), we as a society have or will have a problem. If the resources were obtained by bankrupting LAG or suckering him into buying a house that LAG couldn't afford, then the problem is (or will be) even worse. If MAG has been in effect rewarded for that behavior or even held harmless, I suspect the problems may become far reaching and less readily remedied.
But my most concrete concern right now is that, just as we have elevated prudence and her handmaiden, "efficiency," above primary virtues like beneficence and her bedfellows, sympathy and fellow-feeling, so have we economists shaped our students' thinking and public discourse in ways that have contributed to growing income inequality in the US. I believe we have done this in part by how we draw and teach our curves.