http://theincidentaleconomist.com/
The search frictions paradox
August 23, 2011 at 7:00 am
Austin Frakt
I concluded yesterday’s post with a paradox. On the one hand, there is evidence that search frictions increase the rate at which individuals change plans. On the other, there is evidence that search frictions decrease switching or lead to status quo bias. How can both be true?
The answer is that status quo bias is expected when search frictions are very high or very low while increased plan turnover (switching) is expected when search frictions are at a moderate level. To understand why, it is important to recall what search frictions do. They are a source of plan market power, a means by which plans can charge premiums above marginal cost. Think of it as the power due to obfuscation that can be leveraged through marketing. This is something we all experience when confronting a complex product that we can’t fully evaluate.
The degree to which plans can take advantage of search frictions varies, which causes a distribution in premiums, even controlling for observable plan factors. That’s what Cebul at al. showed. Click through for the explanation. Here’s the chart, again.
So, back to the paradox. How do search frictions lead to status quo bias and higher turnover. By email, Jim Rebitzer, a coauthor on the paper that provides the chart above, explains,
If search frictions are very very small, the distribution of prices collapses around marginal costs (perfect competition is a special case of the model). With very very small frictions, there would be very little incentive to change plans if there were any switching costs at all.
If search frictions are very very large, the distribution of prices collapses around the consumer’s maximum willingness to pay for insurance. In this case there would be very little turnover because: (1) there is little price dispersion; and (2) the rate of arrival of information about outside offers would be very slow. This case matches your intuition that search frictions would lead to less rather than more turnover.
What we found in health insurance markets suggests moderate search frictions. Frictions large enough to support substantial price distribution and therefore large enough to support substantial turnover incentives. It turns out that moderate frictions are enough to distort markets in significant ways.
Translation: Perfect competition requires full information and, thus, has no search frictions. Moreover, products are homogeneous. There is no need to switch. When competition is not perfect and search frictions are very high (and also, in a trivial way, under monopoly), one should not expect much, if any, plan turnover. It’s just too hard to learn anything about other products (or, under monopoly, there are no other products). Status quo bias will reign. However, in a mid-range of degree of competition and search frictions, switching can exist.
Note that the degree of search frictions can vary by market sector (employers vs. Medicare beneficiaries) and can vary by type of searcher (high vs. low cognitive ability). Thus, status quo bias might predominate in some circumstances, plan turnover in others. Hence, the disparate results in the literature are reconciled. The paradox is resolved.
choice, search frictions, status quo bias
Status quo bias
August 22, 2011 at 7:00 am
Austin Frakt
In a Health Affairs paper published last week, Michael McWilliams, Christopher Afendulis, Thomas McGuire, and Bruce Landon examine how Medicare beneficiaries respond to the large number of health plan choices available to them. The perspective and results are consistent with other findings and my own intuition and experience observing my mother shop for Medicare plans. However, as I’ll return to at the end of the post, their results, combined with the search frictions work I described last week, lead to a paradox.
First, some background from McWilliams et al.:
[M]ore choice may be detrimental if there are too many or overly complex options, particularly in high stakes decisions that involve health or money.3,4 Consumers may choose inferior options or make no choice as a result of cognitive overload, anticipated regret, or bias toward the status quo.5–9
Therefore, providing seniors with more health plan choices could increase enrollment in Medcare Advantage or could decrease enrollment if beneficiaries become overwhelmed and choose traditional Medicare by default. Moreover, elderly Medicare beneficiaries with cognitive deficits may have particular difficulty identifying the most valuable option from a complex set of Medicare Advantage and traditional Medicare alternatives. 10,11 [...]
Previous research suggests that the many and complex plan offerings in the Medicare Part D prescription drug program contribute to suboptimal choices by elderly beneficiaries, thereby limiting the gains from competition.15–19
This is the standard view. Too much choice among complex options is paralyzing, giving rise to status quo bias. I’ve experienced it myself. In the context of McWilliams et al., the status quo is traditional Medicare, as opposed to Medicare Advantage. It’s the natural default option, the most popular choice by far (about three-quarters of beneficiaries choose it today), and relatively simple, if less generous in certain ways.
Still, people value choice. So, we want some, but not too much. If status quo bias increases with number of choices, too many options is self-defeating. What’s the right number of Medicare Advantage plans? Take a guess, then check out the chart below.
In the chart, “adjusted probability” is the raw probability, controlling for cognitive ability and other beneficiary socio-demographic characteristics, out-of-pocket costs for various options available, and prior traditional Medicare enrollment (details in the appendix). Enrollment into Medicare Advantage peaks at about 30 plans. (Interestingly, today there are also about 30 PDP options available to Medicare beneficiaries.) Does that mean 30 is the “right” number of plans. No. It is the number that maximizes enrollment. That is all. By itself, it does not imply beneficiaries are making good choices.
One explanation for low enrollment with low Medicare Advantage plan number is the poor matching between plans and beneficiaries. As choices grow, more beneficiaries find plans that suit them. But if choices grow too numerous (above 30), enrollment falls again, as shown in the chart above.
All other things being equal, Medicare beneficiaries should be more likely to enroll in Medicare Advantage as benefits of the program grow. However, to the extent that beneficiaries’ ability to comprehend choices relates to willingness to deviate from the default option (traditional Medicare), there should be different enrollment responses between those with low and high cognitive function. That is, enrollment among those with more difficulty processing complex information should be less responsive to plan generosity. That’s exactly what the authors found, as shown in the following chart.
Though all of the above is intuitive to me, and probably to you too, there is a paradox. Consider the search frictions findings of Randall Cebul, James Rebitzer, Lowell Taylor and Mark Votruba, about which I posted last week. They explained that search frictions are the cause of greater plan turnover, at least among fully insured firms. Search frictions, as you might recall, are anything that impedes the arrival or processing of information.
Wouldn’t more plans increase search frictions? That seems plausible to me. But if search frictions cause more plan turnover, what’s the source of status quo bias? My intuition is consistent with McWilliams et al., that search frictions impede switching (status quo bias) not encourage it.*
This paradox can be resolved. I’ll provide the explanation tomorrow.
* I’m purposefully generalizing. Cebul et al. studied firms. McWilliams et al. studied Medicare beneficiaries. I’m presuming, or assuming, the results of both generalize. I find that plausible.
References
3 Botti S, Iyengar SS. The dark side of choice: when choice impairs social welfare. J Public Policy Mark. 2006;25(1):24–38.
4 Kunreuther H, Meyer R, Zeckhauser R, Slovic P, Schwartz B, Schade C, et al. High stakes decision making: normative, descriptive, and prescriptive considerations. Marketing Letters. 2002;13(3):259–68.
5 Anderson CJ. The psychology of doing nothing: forms of decision avoidance result from reason and emotion. Psychol Bull. 2003; 129(1):139–67.
6 Dhar R. Consumer preference for a no-choice option. J Consum Res. 1997;24:215–31.
7 Gilovich T, Medvec VH. The experience of regret: what, when, and why. Psychol Rev. 1995;102(2):379–95.
8 Iyengar SS, Lepper MR.When choice is demotivating: can one desire too much of a good thing? J Pers Soc Psychol. 2000;79(6):995–1006.
9 Samuelson W, Zeckhauser R. Status quo bias in decision making. J Risk Ins. 1988;1:7–59.
10 Finucane ML, Slovic P, Hibbard JH, Peters E, Mertz CK, MacGregor DG. Aging and decision-making competence: an analysis of comprehension and consistency skills in older versus younger adults considering health plan options. J Behav Dec Making. 2002;15:141–64.
11 Hibbard JH, Slovic P, Peters E, Finucane ML, Tusler M. Is the informed-choice policy approach appropriate for Medicare beneficiaries? Health Aff (Millwood). 2001;20(3):199–203.
15 Abaluck JT, Gruber J. Choice inconsistencies among the elderly: evidence from plan choice in the Medicare Part D program. Cambridge (MA): National Bureau of Economic Research; 2009. (Working Paper No. 14759).
16 Hanoch Y, Rice T, Cummings J, Wood S. How much choice is too much? The case of the Medicare prescription drug benefit. Health Serv Res. 2009;44(4):1157–68.
17 Winter J, Balza R, Caro F, Heiss F, Jun BH, Matzkin R, et al. Medicare prescription drug coverage: consumer information and preferences. Proc Natl Acad Sci USA. 2006; 103(20):7929–34.
18 Kling JR, Mullainathan S, Shafir E, Vermeulen L, Wrobel MV. Misperception in choosing Medicare drug plans [Internet]. New York (NY): City University of New York; 2009 [cited 2011 Jan 13]. Available from: http://web.gc.cuny.edu/economics/ SeminarPapers/Fall%202010/ Kling.pdf
19 Tanius BE, Wood S, Hanoch Y, Rice T. Aging and choice: applications to Medicare Part D. Judgm Decis Mak. 2009;4(1):92–101.
UPDATE: For clarity.
choice, Medicare, Medicare Advantage, search frictions, status quo bias
Can shopping ever be easy? Obfuscation in the information age
August 19, 2011 at 12:57 pm
Austin Frakt
By email, Jim Rebitzer, one of the authors of the search frictions paper I wrote about yesterday, throws some cold water on my idea of easy, computer-enhanced health plan shopping:
I saw your post today about the possibility of friction-less insurance markets. You make very good points that I very much agree with.
Glenn Ellisson has been looking at this issue from a theoretical IO perspective. Bounded rationality and information processing costs, it turns out, are an old idea in IO (who knew?) that have been revived by electronic search engines and shoppers on the internet. In a theory model he finds that when companies can create frictions by cleverly obfuscating, an exogenous technological change that reduces frictions might actually stimulate more obfuscation. Empirically he looks at this in the context of specialized electronic search engines for computer parts.
It’s an interesting set of ideas (but as a matter of policy and practicality sort of besides the point).
Here are the cites if you are interested.
Ellison, Glenn and Sara Fisher Ellison. 2009. “Search, Obfuscation, and Price Elasticities on the Internet.” Econometrica, 77(2), 427-52.
Ellison, Glenn and Alexander Wolitzky. 2009. “A Search Cost Model of Obfuscation,” National Bureau of Economic Research, Inc, NBER Working Papers: 15237.
Ellison, Glenn. 2006. “Bounded Rationality in Industrial Organization,” N. a. P. Blundell, Advances in Economics and Econometrics: Theory and Applications, Ninth World Congress. Cambridge Univesity Press.
However, if my disutility due to technology-encouraged obfuscation stays roughly constant, I’d still prefer a 10 minute computerized search and selection process to a multi-hour one that has no possibility of jarring me from the status quo (ignoring privacy issues).
Health insurance search frictions, part 2
August 18, 2011 at 12:00 pm
Austin Frakt
This is the second of two posts today on health insurance search frictions. I assume you’ve read the first one.
As promised, this post summarizes some of the content of Unhealthy Insurance Markets: Search Frictions and the Cost and Quality of Health Insurance, by Randall Cebul, James Rebitzer, Lowell Taylor and Mark Votruba.
The effect of search frictions on premiums is illustrated in the following chart from the paper, but it isn’t obvious. I explain below.
The chart is the probability density of “residual premiums” (the horizontal axis). Residual premium is the portion of premium that cannot be explained by characteristics of the plan and its potential enrollees. In other words, it’s the result of an attempt to control for the fact that plans in a market vary in their benefits and cost sharing as well as in the health risk of the population that enrolls in them. All these factors affect premiums and are what make comparing plans apples to oranges (or aardvarks).
In computing residual premiums, the authors controlled for “plan type (Preferred Provider Organization (PPO)/Point of Service(POS)), deductible level, copayment for typical office visit, the inclusion of prescription drug coverage, and such characteristics as firm and establishment size, percent of workers who are full time, percent female, age distribution of workers, and mean payroll.” This is, obviously, not a complete set of controls. What the authors assume is that (a) premiums in the SI market reflect variations in marginal costs governed by unobservable characteristics and (b) those unobservable characteristics have the same effect in the FI market as in the SI market. Consequently, the dispersion of SI residual premiums shown in the chart above serve as a counterfactural to that of the FI ones.
The FI distribution has greater dispersion than the SI one, reflecting greater search frictions. (If it isn’t clear to you why this should be, go back to the first post.) The authors use the differences between these two distributions to estimate the implications of those search frictions. They find that search frictions in the FI market led to a transfer of $34 billion in consumer surplus (consumer value) from consumers to insurers in 1997. They also find that frictions led to greater plan turnover, with 64% more FI groups and their members switching plans than would have otherwise (about 20% of policyholders switched plans in a year). That’s troubling because plan switching is a disincentive for plans to invest in preventative care, which is under provided in the US.
So, what’s to be done about search frictions and the trouble they cause in the market? The authors point to a government solution in the form of a subsidized public option that is simple to understand, though does not provide the flexibility that some segments of the market demand.
The socially optimal government policy is to price the public insurance backstop option below cost! Note that this policy will not completely crowd out private insurance, as privately provided insurance is more highly valued than government-provided insurance. Indeed, the size of the optimal subsidy is limited by the welfare costs of attracting clients away from superior private options. Rather, the intervention displaces insurance offerings on the far-right tail of the distribution, i.e., the highest-priced policies, and in so doing makes the market more efficient.
A subsidized government plan has ripple effects that alter incentives and prices throughout the market. A properly chosen backstop plan improves market efficiency by moderating the arms race in marketing. More precisely, [...] the subsidy reduces the payoff to marketing for all insurers; thus, each insurer can spend less without ceding an advantage to competitors. The government subsidized insurance also makes private insurance more attractive to consumers by compressing the distribution of prices toward marginal cost.
Opening up state insurance markets to provide more competition from out-of-state plans is, perhaps, the antithesis of a single, simple public option. About it, the authors write,
[I]n a search model, an increase in the number of insurers need not lead to lower prices if it does not reduce the cost to insurers of marketing and medical underwriting or reduce the cost to employer groups of evaluating offers. Indeed, this sort of proposal might exacerbate frictions resulting from information overload.
Finally, they conclude,
Health insurance reform is among the most pressing policy issues in the United States today. A better understanding of the causes and consequences of search frictions will be important for formulating better policy and improving the efficiency of insurance markets.
Even if one disagrees with the authors’ economic models, empirical technique, or conclusions, one cannot easily dismiss this conclusion. Those who view a public option less favorably and prefer a market with more options for consumers, for example, should be (and are) concerned with enhancing the ability of consumers to make good choices, thereby encouraging better functioning markets. Other than those who profit from them (no trifling interest group), we should all be concerned about search frictions in insurance markets.
choice, health insurance, search frictions
Health insurance search frictions, part 1
August 18, 2011 at 7:00 am
Austin Frakt
Not since my undergraduate study of physics have I thought so much about “friction” as I will today. This is the first of two posts on this topic. Come back at noon for the second.
Because products vary so much across many characteristics, health insurance is not easy to shop for. Comparing plans is an apples to oranges problem, or maybe it’s more like apples to aardvarks. The challenges of comparison create “search frictions,” inefficiencies in one’s ability to wisely choose, to be a savvy shopper. This motivates the recent development of (draft) standards for health plan labeling.
How much will the new health plan labels, required starting next March, help consumers in their search for plans? How much grease will they add to the otherwise highly frictional process? I don’t know. A place to start is an examination of those frictions. What are they and how much do they matter?
That’s the subject of Unhealthy Insurance Markets: Search Frictions and the Cost and Quality of Health Insurance, a new paper in the American Economic Review by Randall Cebul, James Rebitzer, Lowell Taylor and Mark Votruba. Theirs is an examination of search frictions in the employer-sponsored insurance (ESI) market, particularly those experienced by fully insured (FI) firms, in contrast to the self insured (SI). FI firms buy insurance from an insurance company. SI firms take on the insurance risk, only contracting out the management of their plans to third parties. Search frictions should be larger in the FI market than in the SI one because FI firms are searching for more: insurance plus administration versus just administration.
Those frictions are manifest in premiums. Even controlling for all the reasons why premiums should vary–plan characteristics, variations in the cost of care, the different mix of health risk of enrollees in plans–premiums will still vary due to search frictions and more so for FI firms. Think of it this way, when representatives of firms can’t easily assess a plan’s characteristics and compare it to other plans (search frictions) they will make errors in assessing and choosing plans. They know less about a plan than an insurer does.
This information asymmetry is a source of market power, permitting plans to charge above competitive prices, though to differing degrees perhaps. In other words, rather than one competitive price at marginal cost–what one would expect in a perfectly competitive market–prices are above marginal cost and in some distribution. Note that market power due to search frictions increases with the number of plans, while market power due to insufficient competition decreases. Therefore, nothing like a perfect market can ever exist for health insurance. Sorry!
Notice that I have not yet said anything substantial about the paper by Cebul et al. That’s the subject of the second post at noon.