The Reality of Pareto Optimality

Earlier this week I discussed this question of the Kaldor-Hicks criterion versus Pareto optimality: two standards we use for government intervention from the welfare economics perspective. I argued that unless we constrain our assumptions about the world, a completely Pareto optimal government intervention is impossible to reach. After having thought about the issue some more, I realized that this is really a question about open versus closed systems. In a closed system, we can have a Pareto Optimal solution. A broad long-lasting insurance pool benefits everyone who is a part of that pool, but may raise costs for those outside of it. Foundation grants that fund projects in a single city may benefit everyone who lives in that jurisdiction, but it raises the opportunity costs of those living elsewhere.

Theodore Lowi approaches this idea of policy benefits and burdens from a political science perspective. If you remember from our reading on explaining policy choices, Lowi is associated with the idea that policy shapes politics. He argued that there were four types of policies: distributive, re-distributive, regulatory, and constituent (although his earlier work only acknowledged the first three of these categories) . Key to his argument is the idea that each of these types of policies have their own arenas of power, in other words, each of these types of policies produce different relationships between interest groups, elected officials, and agencies.

For this post, I will focus on distributive policies. The current form of distributive policy would be those earmarks and pork-barrel politics that we discussed during the first week of class. They distribute large benefits to a small group of people, but the burdens they place on others (mostly due to taxes and opportunity costs) are small and diffuse. While these may meet the Kaldor-Hicks criterion, they are certainly not Pareto Optimal. The original use of the term "distributive policy" was to describe 18th and 19th century American land policies. During that time, land in America was considered an almost unlimited resource. The American government was basically giving away land to anyone willing to travel west. From the perspective of 18th and 19th century Americans, I would argue that distributive policies, as defined in this way, were Pareto Optimal. There was enough land to go around, those willing to make the trip to the west could have it, freeing up space and resources for those remaining on the East Coast. Everyone had to travel to the west to benefit, so the opportunity costs were more or less the same for everyone.

I hope that by now you have all recognized the flaws in this argument. This assumes a very closed system. The definition of "American" in this example does not include those who were already living on the land in the West (some of whom had already been displaced from the East Coast), and who were brutalized, killed, and occupied so that American citizens could all benefit from free western land. (Of course, I am talking here about Native Americans/American Indians). Once we extend our system to include Native American/American Indian peoples, we no longer have a Pareto Optimal solution. These policies would not qualify as distributive policies in the modern sense of the term "concentrated benefits with small diffuse burdens", and they would fail the Kaldor-Hicks criterion, as well. The short-term and long-term harms  they imposed (most notably genocide, relocation, and rape) likely outweigh the benefits. This would be true for the colonialist actions of other nations, as well.

The interest in these two criteria for government action in class has allowed me to engage in a thought experiment about what they could look like in real-life. While grounded in economic and policy theory, these arguments are not based on research and should only be taken as thought experiments. I think one lesson here is that policy-makers and analysts need to be weary of solutions that seem to be Pareto Optimal. When burdens of any action seem nonexistent or even small and diffuse, we may be guilty of closed-systems thinking. We,  as responsible members of a global society, need to at least consider the harms and burdens that our actions will cause to those we consider outside our reference group, our society, and our nation. There may still be a good argument for action, but if we don't at least consider these possibilities we risk history's harsh judgement.

The Affordable Care Act and the Kaldor-Hicks Criterion

On Tuesday, I gave a lecture in class about different theories and models for examining the design of public policy, the policy process, and the effects of policy on society. I spent a lot of time on economic models of public policy, as this tends to be the dominant paradigm that policy scholars use to study all three aspects of policy. I discussed welfare economics in-depth including the two major justifications that we use for government action to maximize social utility: Pareto optimality and the Kaldor-Hicks criterion. As a reminder, the standard of Pareto optimality posits that government should intervene in the market if it can make at least one person better off without making anyone worse off. In the global, systemic world in which we live, this is basically an impossible task for public policy to achieve. As I will discuss here on Friday, we can think of some examples of Pareto optimal policies when we constrain our models of society in certain ways, but I have yet to come up with a public policy that does not harm a single person once opportunity costs and relative harms are taken into account. I invite you to help me brainstorm to come up with such a policy. In contrast, the Kaldor-Hicks criterion is easier to meet and is usually what we rely on as our standard for policy-making. The Kaldor-Hicks criterion states that if we can make at least one-person better off through government policy and the benefit to that person or group of people exceeds the harm imposed on others, thus creating an overall net-gain of utility for society, then government should act. Further, the person who is made better-off can, in theory, reimburse the person or people who are harmed, creating a theoretically Pareto optimal solution. Of course, this criterion may lead to extreme inequality even while it creates an overall net gain for society if the same people are always made better off and there is no requirement to compensate those made worse-off.

There was a request from one of my students to apply the Kaldor-Hicks criterion to healthcare reform, specifically the Patient Protection and Affordable Care Act, often referred to as the Affordable Care Act (ACA). In order to do that, we need to take a step back to talk about the concept of insurance in general. Why do we have insurance for some things (car insurance, unemployment insurance, home-owners insurance, social security insurance etc.) and not others? For the basic answer to that question, turn back to my blog post on why we have health insurance, here. In general, insurance is a solution to the problem of imperfect information about our future health that allows us to pool risk across many different people. At a single point in time, we essentially have a classic Kaldor-Hicks situation. The healthy in our (hopefully large) risk pool pay a little more in premiums than they would on medical care for themselves, while the sick pay much less and are not left bankrupt after the sudden on-set of a serious illness. However, rarely is anyone healthy over the entire course of their lifetime. Given enough time in an insurance pool, those who paid in extra when they were healthy will be compensated or reimbursed by paying less than they otherwise would when they are sick. 

So how do people decide whether or not to purchase insurance? Those who are unhealthy, who believe that they will spend less in premiums than they will for their own care will of course purchase insurance, if they can. What about the healthy? They are choosing between spending their money on insurance premiums and other goods and services. Remember, we are talking about rationally self-interested utility-maximizers here so they will each have their own individual utility curves that map this trade-off. This curve is a function of their disposable income, their knowledge about their current and past health, their knowledge about family health history, their degree of risk aversion, and the price of health insurance. Those people who are healthy, young, and risk-neutral or risk-seeking likely will choose to spend their money on other goods and services. Those people with very little disposable income may be forced to spend their money on other goods and services. Those people with an extremely high amount of disposable income who are risk-neutral may decide that they have enough money to cover their individual health costs regardless of what happens to them and avoid purchasing insurance. 

Like any good public policy from the economics perspective, the ACA adjusts this utility curve. The provision that young people under 26 can be added to their parent's health insurance plan reduces the cost of insurance for that young person (who is more likely to be healthy and less risk-averse then the general population). This should incentivize young people (or their parents) to pay for health insurance. By creating state health insurance exchanges, there is greater risk pooling which should lower the costs of health insurance relative to the individual market, incentivizing more healthy uninsured people to purchase insurance. By providing tax-credits for the purchase of health insurance to low- and moderate- income individuals the act increases the income available to spend on health insurance and increases the opportunity cost of purchasing other goods and services, incentivizing more people who would not have otherwise been able to afford insurance to purchase it. Similarly, by imposing a tax penalty on those who do not purchase insurance, the individual mandate changes the opportunity costs of not buying insurance (for those making enough to afford insurance, see more about this here). All of these changes should increase the pool of healthy people who purchase insurance, making those already in the pool better off, but possibly penalizing those who decide not to purchase insurance. 

Once again, this is not the end of the story. We also have provisions that will increase the short-term burden on the healthy individuals in the pool. Health insurers can no longer exclude the sick from health care in the same way that they previously could. They also cannot rescind coverage from individuals when they become sick. The influx of expensive to treat patients into insurance coverage may mean that premiums increase and the Kaldor-Hicks criterion is violated in the short-run. However, in the long-run when these healthy people themselves become sick, these regulations ensure that they will then benefit and be compensated for their previous over-payments. In some ways we can think of an insurance system with a large risk pool as Pareto optimal for those inside the pool. As I will mention on Friday, those  outside the pool will likely still be harmed, making the solution technically non-pareto optimal. Now, when we attach insurance to employment and the labor-market shifts from long-term employment in a single firm to high turn-over positions, how does that affect the ability to remain in an insurance pool long-term? How does that affect our Kaldor-Hicks criterion? Of course there's a lot more depth we can go into here, but I will save that for another time.