As a young child growing up in suburban Kansas City, my parents paid me a penny for every dandelion I picked from our yard. Their intention, of course, was to reduce the number of dandelions. I suspect, however, that my parents never realized that I would sometimes intentionally blow dandelion seeds back into our yard. Doing so was irresistibly fun, but it also meant more dandelions for me to harvest in the future. I eventually learned that I probably shouldn’t have blown those dandelion seeds throughout my yard – and somewhere along the way, I became fascinated with the idea that good intentions could potentially lead to unintended consequences.
Unintended consequences arise because we live in an interconnected and complex world. The interconnectedness and complexity of the world imply that a small change in one part of a system can potentially generate far-reaching effects in another part of the system; moreover, policymakers may be unable to anticipate or understand how their policies affect the incentives of individuals within the system.
I attended Williams College, a small liberal arts college in western Massachusetts. I initially planned to major in biology, studying the complexity of biochemical systems; however, somewhere between my time reading Shakespeare and throwing Frisbees on the quad, I grew increasingly interested in the complexity of social and economic systems, and so I eventually graduated with a double major in biology and economics. After college, I enrolled in the economics Ph.D. program at the University of Wisconsin-Madison, where I decided to begin formal study of the unintended consequences of public policies.
Examples of the unintended consequences of public policies are manifold. For instance, the dandelion example above is sometimes referred to as the “cobra effect,” in reference to a British policy in colonial India. In an attempt to reduce the number of venomous snakes, the British government offered a reward for dead cobras. While the policy was initially deemed effective, some individuals eventually realized they could increase their income by breeding cobras, which ultimately increased the size of the cobra population. Another historical example, common in 18th- and 19th-century northern Europe, was the so-called “window tax,” a property tax whose amount depended upon the number of windows in a house. (An explicit “income tax” was politically unpopular, so the intention was to create an income tax without explicitly taxing on the basis of income.) Houses with more than a certain number of windows were subject to taxation; therefore, to avoid taxation, many households bricked up windows to fall below this threshold. (The tax was eventually repealed.)
In both of these examples, the public policy attempted to address a social problem without understanding or taking into account the policy’s effects on individuals’ incentives – and, in particular, incentives that ultimately affected the outcome of interest. This is why careful and systematic consideration of a policy’s intended and unintended effects is imperative in the design and evaluation of a public policy; however, just because a policy could theoretically affect individuals’ incentives doesn’t mean that the policy actually will: The extent to which individuals respond to incentives is an empirical question, which requires an empirical answer. For example, consider a proposed increase in the minimum wage, which is intended to benefit low-wage workers. Opponents of such an increase may argue that an increase in the minimum wage will create an incentive for employers to hire fewer minimum wage workers, thus making low-wage workers worse off; however, the extent to which employers reduce their hiring of low-wage workers in response to a minimum wage increase is an empirical question, not a theoretical one. (This particular question is still hotly debated within economics.)
In my own research, I have studied the unintended consequences of policies in a variety of contexts, including federal income taxation, Social Security and higher education. Most recently, a research collaborator and I have examined the effects of expanding eligibility for Medicaid, a public health insurance program. In particular, we have focused on parental Medicaid, which provides public health insurance primarily to parents of Medicaid-eligible children.
According the U.S. Census Bureau, in 2010, approximately 256 million Americans had some form of health insurance, including approximately 49 million Medicaid enrollees; however, approximately 50 million Americans had no form of health insurance – an uninsured rate of about 16 percent.
Unlike Medicare, a well-known federal program that provides health insurance for primarily older Americans, Medicaid is a jointly funded federal-state program that provides health insurance for primarily low-income families. Since individual states manage their own Medicaid programs, they have a good deal of flexibility in implementation, including determining program eligibility. One of the main eligibility criteria that states use, in part, to define Medicaid eligibility is family income; consequently, in an attempt to reduce the number of uninsured individuals, many states have increased income eligibility thresholds.
The effectiveness of expanding public insurance eligibility thresholds to reduce the number of uninsured individuals rests primarily on two assumptions: (1) uninsured individuals will enroll in public insurance if they are eligible, and (2) expanding public insurance eligibility will not create an incentive for privately insured individuals to drop their private coverage in favor of public insurance – a phenomenon known as “crowd-out.” In this case, crowd-out is considered an unintended consequence of an eligibility expansion since large amounts of crowd-out will imply small increases in overall insurance coverage (and consequently small reductions in the number of uninsured).
Regarding assumption (1), my collaborator and I found that parental Medicaid eligibility expansions resulted in moderate, but statistically meaningful, increases in Medicaid participation. This finding is consistent with previous research, which estimated that approximately 20 to 40 percent of eligible adults do not enroll in Medicaid. Since a large number of eligible individuals do not enroll in Medicaid, we should not necessarily expect large increases in Medicaid enrollment when income eligibility limits are increased.
Regarding assumption (2), we found no evidence that parental Medicaid eligibility expansions result in departures from private insurance; in other words, we found no evidence of crowd-out. This finding implies that any increases in Medicaid enrollment generate approximately one-for-one reductions in the number of uninsured individuals. One might expect the degree of crowd-out to depend on a state’s initial income threshold – states with higher initial thresholds may experience more crowd-out as the income threshold rises to meet families whose incomes may afford them private insurance; however, even when we allowed for differential effects along the income threshold distribution, we found no evidence of crowd-out. Therefore, whereas crowd-out is accepted as a theoretical possibility resulting from a Medicaid eligibility expansion, the empirical evidence suggests that the magnitude of crowd-out is, at most, negligible.
Put together, our findings imply that states’ parental Medicaid expansions have resulted in moderate, but meaningful, reductions in the number of uninsured individuals; however, since a large number of already-eligible individuals (many of whom are uninsured) do not enroll in Medicaid, it means that simply expanding Medicaid eligibility will not completely eliminate the number of uninsured individuals; therefore, additional research is needed to determine why Medicaid-eligible individuals do not (or cannot) enroll in Medicaid.
The world is so interconnected and complex that we cannot hope to completely understand it; however, to strike blindly and indiscriminately is the incorrect response to such long odds; at the very least, we should try to account for as many intended and unintended effects as possible. In doing so, we do the best we can to translate good intentions into good policy.
From Exemplars, a publication of the Grants and Research Office.