Saturday, May 6, 2017

Huck, Szech, and Wenner (2015) on Information Avoidance; or, Am I Getting Paid Enough to Finish this Outline?

Steffen Huck, Nora Szech, and Lukas M. Wenner, “More Effort with Less Pay: On Information Avoidance, Belief Design and Performance.” CESifo Working Paper No. 5542, October, 2015 (pdf).

• If information can influence your expectations and motivations, you might want to avoid acquiring information. For instance, knowing that there are high stakes riding on your performance might lead you to choke under the pressure. This paper reports on an experiment that tests whether people might choose to avoid information about their precise (piece-rate) wage. 

• The experiment is of the “real effort” variety, that is, subjects are asked to engage in a tedious, painstaking task: typing 60-character nonsensical alphanumeric strings, mixed uppercase and lowercase, backwards into a computer. For each correctly typed string, a piece rate is paid, and the subjects have one hour in which to type (or not, as they wish). In case they want to not work, they have a magazine at hand to help pass the hour. 

• The piece rate is random; with probability .5, it is quite low – .1 euro per string – and with probability .5 it is significant: 1 euro per string. The different conditions of the experiment turn on: whether the workers know their wage (Full Info); whether they choose to be or not to be informed of their wage until after the work hour is over (Info Choice); and whether they have no option but to remain uninformed (No Info). A fourth condition is where workers know that their wage is .55 euro per string, which is the weighted average of the high and low piece rates, and hence might serve as the expected wage for workers who do not know their wage. 

• Not surprisingly, if you know your wage, you work harder if the wage is high than if it is low (and you work at an in-between rate if you know you earn the medium wage). If you get to decide whether or not to be told your wage, well, 30 out of 95 subjects choose not to be told. These uninformed folks work like machines, even outpacing subjects who know they have a high piece rate.

• Perhaps those subjects who choose not to learn their wage rate happen to be really productive workers, that is, maybe the option of not knowing the wage within the Info Choice group selects for the best workers. This possibility is tested by the No Info treatment, where subjects don’t know their piece rate in advance, but not by choice: they are not given the option of learning their wage. But these subjects also perform at a very high level, so selection effects don’t seem to drive the result: not possessing wage information tends to bolster productivity. 

• When asked about their decisions, those subjects who choose to learn their piece rate in advance tend to explicitly note their interest in tailoring their effort to the rate. Those who choose not to learn their piece rate tend to mention the effect that a low rate would have on their motivation, and/or a fear that a high rate might lead to anxiety. 

• The authors adapt a model of Brunnermeier and Parker (2005; pdf here) to indicate that a high piece rate can both increase the revenue from effort, and also increase the costs of supplying effort: a high rate might induce anxiety, for instance. Further, agents receive utility not just from what happens today but also some utility (today) from anticipation of what is likely to happen next period. The anticipation utility provides a channel for utility-maximizing agents to hold biased beliefs, overoptimistic views of what the future will bring. With a high enough anticipatory utility, agents will prefer to avoid knowledge of their wage: they are information avoiders.

Bobadilla-Suarez, Sunstein, and Sharot (2016) on Under-Delegation

Sebastian Bobadilla-Suarez, Cass R. Sunstein, and Tali Sharot, “Are Choosers Losers? The Propensity to Under-Delegate in the Face of Potential Gains and Losses.” February 15, 2016 (pdf, perhaps somewhat updated, available here).

• People seem to value control over decisions, and are willing to sacrifice (in terms of expected returns) to make a decision themselves rather than to delegate. Perhaps for decisions involving losses, however, people might prefer to delegate, to insulate themselves a bit from painful choices. 

• This paper reports on two laboratory experiments that are conducted to look at delegation versus control. Sample sizes are rather small: 26 subjects for the first experiment, and 25 for the second. In some trials, participants can delegate the choice to an advisor, after being informed of the reliability and cost of the advisor. (In reality, the advisor is a computer player programmed with the appropriate reliabilities.) 

• Both experiments start with a learning phase. Participants are shown two geometrical shapes, and are asked to pick the better one. They are not told what criteria go into “better,” but if they pick correctly, they receive a higher payoff (or avoid a bigger loss) than if they choose incorrectly. Unbeknownst to the players, there is no underlying rule for “better,” the computer just designates, randomly, one of the two shapes as better. But humans are good at finding patterns in random events!

• In the delegation phase of experiment one, participants are given the option, prior to being shown the shapes, to delegate their decision to an advisor (actually, a chooser) whose probability of choosing correctly is revealed in advance, along with the cost of using the advisor. After the delegate-or-not decision, the two shapes are revealed, a choice is made (although, if made by an advisor, not revealed), and onto the next round (60 in total). Subjects do not learn the outcome of individual trials, but are compensated at the end on a random subset of ten of the trials. 

• Given that the subjects themselves have a 50% chance of choosing correctly, it is easy to determine when the expected monetary payoff to the subject is higher by delegating the decision to the advisor (thanks to those known probabilities and costs; incidentally, the price of using the advisor is only imposed if the advisor is used, and if the advisor chooses correctly). 

• The advisors are configured (costs and accuracy) in such a way that the subjects would do best (in expected monetary terms) by delegating on half of the trials. Instead, subjects choose to delegate only about 30% of the time, whether they are choosing for gains or to avoid losses. The departures from “optimal” play are almost all in the direction of failing to delegate the decision when expected returns would have been higher with delegation. 

• Subjects’ perceptions of their own abilities to choose are canvassed, and the aversion to delegation is not the result of excessive optimism with respect to the accuracy of their own choices. 

• The second experiment is nearly identical to the first one, except that the decision to choose or to delegate is now made each round after the shapes themselves have been revealed. Still, there is substantial, even slightly worse, under-delegation. 

• The fact that under-delegation also is present in the “loss” trials indicates that people, at least in this setting, don’t seem to want to shield themselves from painful decisions through delegation.

Thursday, May 4, 2017

Gilovich and Kumar (2015) on Experiences versus Material Goods

Thomas Gilovich and Amit Kumar, “We’ll Always Have Paris: The Hedonic Payoff from Experiential and Material Investments.” Advances in Experimental Social Psychology 51: 147-187, 2015. [The version outlined below has different page numbers (and perhaps other discrepancies) from the version available here.]

• Experiences provide long-lasting memories, at least relative to material goods; experiences are gifts that keep on giving. Perhaps subjective well-being (SWB) and income would be more closely related if people shifted their spending from material goods to experiential goods. 

• Vacations, concerts, dining out: these are examples of experiential goods. Electronic gadgetry and clothing fall into the material goods category. People consistently report higher happiness from experiential goods relative to material goods. 

• Experiences are better for SWB for a variety of reasons. First, they are less subject to the adaptation which, over times, erodes the pleasure boost given by a material acquisition. Experiences are more pleasurable (and interesting) to talk about, and can become embedded in our identity, our story. 

• Adaptation is helpful in allowing people to come to terms with negative events, so if bad experiences are not subject to adaptation, we could suffer long-term from unpleasant moments. But experiences tend to be malleable, so that bad experiences – the vacation where it rained the whole time – tend to be looked at more favorably, through rose-tinted glasses, over time. Indeed, the negative features of experiences can become, in retrospect, their chief virtues. 

• Even in anticipation of consumption, experiences bring more pleasure than material goods. Waiting in line is less unpleasant when it is to see a play than to buy a gadget. And after the fact, people tend to think of money spent on experiences as having been better spent than that expended on material objects. 

• The social element that is attached to many experiences is one of the sources of their SWB advantages. And talking about experiences after the fact is one of their social elements. Talking about material possessions is a bit of a snooze fest. 

• The endowment effect is connected to an unwillingness to part with something that we “own”; people tend to show larger endowment effects, greater attachments, to potential experiences than to material goods. 

• Material possessions like cars are relatively easy to compare along quality dimensions – your neighbor’s Ferrari is probably better than your Ford. But who had a better vacation, or a better time at the concert? These are harder to compare, and hence, experiences to some extent protect us against unfavorable social comparisons. And if we enjoy our concert, we are not much concerned about opportunity costs, those potentially better concerts we sacrificed to go to “ours”. 

• If you buy a bad car, you regret it, you have buyer’s remorse. But the more frequent regret in the experiential realm concerns foregone opportunities, not buyer’s remorse. And it is easy to find the good in a bad experience, but not so easy to find the upside of a malfunctioning computer. “Surprises” tend to be positive when they arise during experiences, but negative when they are connected to material goods! 

• Some material goods are purchased with the intention of using them to enhance experiences, such as watching movies with your friends and family in a home video center. But perhaps the actual consumption of these goods is less social and less experiential and even less common than anticipated. If people have the experiences in mind when they make the purchases, they may be systematically fooled, as it were, into overspending on what turn out to be merely material goods. 

• Perhaps connoisseurs are able to convert material goods (collectibles, say) into experiential goods, by harnessing the same social, identity, and even narrative advantages that other types of experiences present. 

• Can we design cities, and our lives, to maximize experience?

Tuesday, May 2, 2017

Clark (2014) on Adaptation and Easterlin

Andrew E. Clark, “Adaptation and the Easterlin Paradox.” Paris School of Economics, October, 2014 [pdf].

• One explanation of the Easterlin Paradox is that people care about relative, not absolute, incomes. A second explanation, addressed here, is that people adapt their reference point to reflect a higher income and so receive no extra subjective well-being (SWB) when the adaptation is complete. 

• The existing empirical evidence shows a good deal of adaptation to permanently higher incomes, but often, not full adaptation. Increasing income inequality, where richer people have recent higher-than-normal gains, can lead to a positive association in cross-sectional data between income and SWB. 

• But perhaps other contributors to well-being – employment life, health, family life, for example – are not as subject to adaptation as is income. Unemployment seems to be fairly resistant to adaptation – many people are as worse off (in terms of happiness) a few years into unemployment as they are when they first become unemployed. The same lack of adaptation applies to falls into poverty. This is the case only on average, of course, and perhaps we could invest in resilience so that adaptation to negative events would be more common. 

• On average, the happiness of marriage is subject to adaptation, so that the initial happiness boost (in many countries) is eliminated over time. Divorce is associated with a notable fall in subjective well-being that precedes the divorce for a few years, but by and large, this decline is eliminated a few years post-divorce through adaptation. Having children has no impact, on average, in long-term SWB. 

• People somewhat adapt to bad health events, but that adaptation tends to be far from complete. (Cosmetic surgery brings lasting SWB benefits (page 11)!)

Sunday, April 30, 2017

Easterlin (2016) Relocates the Seemingly Lost Paradox

Richard A. Easterlin, “Paradox Lost?” USC Dornsife Institute for New Economic Thinking, Working Paper No. 16-02, 2016.

• Easterlin’s version of the Paradox: At a given point in time, within a country, happiness is positively correlated with income, and furthermore, at a given point in time, richer countries are happier than poorer countries. That is, cross-section evidence suggests a positive association between income and happiness. Over time, however, happiness is not positively correlated with income. Paradox! But the time period in which the relationship between income and happiness collapses needs to be substantial: it is long trends in happiness and income that seem to be uncorrelated. 

• Over nearly 70 years, happiness trends in the US have been zero or slightly negative, despite per-capita income tripling. 

• Easterlin looks at countries with at least 1 million people, and that possess data from at least three Subjective Well-Being surveys, conducted over a period of at least ten years and one GDP cycle: 43 countries make the cut. He finds no significant relationship between growth and happiness in this panel data. 

• Some other researchers generate different answers because they look over shorter timespans. Transition countries, for example, tend to be included with only one phase of their transition cycle in the data, biasing results towards a positive connection between GDP and happiness.

Links to outlines of a few closely-related papers:

Betsey Stevenson and Justin Wolfers, “Economic Growth and Subjective Well-Being: Reassessing the Easterlin Paradox.Brookings Papers on Economic Activity, pages 1-87, Spring 2008. 

Daniel W. Sacks, Betsey Stevenson and Justin Wolfers, “The New Stylized Facts About Income and Subjective Well-Being.Emotion 12(6): 1181- 1187, 2012.

Richard A. Easterlin, “Happiness, Growth, and Public Policy.” Economic Inquiry 51(1): 1–15, January 2013. 

Richard A. Easterlin, “Happiness and Economic Growth: The Evidence.” USC Dornsife Institute for New Economic Thinking, Working Paper No. 14-03, November 6, 2014.

Wednesday, April 5, 2017

Samek (2016) Nudges Schoolchildren to Drink Healthier Milk

Anya Samek, “Gifts and Goals: Behavioral Nudges to Improve Child Food Choice at School.” CESR-Schaeffer Working Paper Series, Paper No: 2016-007, January, 2016; available at SSRN.

• The author arranges for a field experiment to take place at eight public elementary schools in Chicago; the 1400 or so child subjects do not know that they are participating in an experiment. 

• The only lunch menu choice that is available to these schoolkids every day is whether to get the (low-fat) white milk, the (low-fat) chocolate milk, or do without milk altogether. On the first day of the experiment, students are nonchalantly observed when they make their milk choices. Chocolate milk is the big favorite: more than 85% of kids take the chocy, 11% take white, and 3.4% go milkless.

• In the second and final day of the experiment, a control and two treatment goups are implemented. In the control condition, teachers very briefly explain the relative merits of white milk (chocy milk has added sugar, and that is less healthy, you see) shortly before lunch. In the “Gift” condition, the explanation is followed up by a gift of a sticker to all students, no questions asked and nothing required in return; and in the third condition, “Goal,” students are asked to make an unenforceable goal, a written pledge, as it were, to choose the white milk (though they can pledge themselves to chocolate). 

• The control condition induces a huge increase in white milk purchasing, to over 47%. The Gift condition brings a slightly larger shift to white, with the Goal condition in between. The goal-setting “works” better for younger kids, but the sticker “works” about the same for everyone. 

• The unconditional gift tries (and seemingly succeeds) to connect the milk choice to reciprocity; one advantage of an unconditional gift is that no enforcement (of conditions) is needed. 

• The choices in the Goal group suggest a good deal of time inconsistency – lots of kids who pledge to get the white milk change their mind when the decision itself is at hand – even though the elapsed time between pledge and decision is only about 15 minutes. 

• The fact that the nudge and its recorded effects are of the one-day-only variety is a pretty severe restriction on drawing policy advice from the experiment. My main takeaway, as it were, concerns the large shift towards choosing white milk in the control condition: a very short piece of pro-white-milk propaganda delivered by a teacher alters lots of elementary student milk choices. Use this power wisely!

Tuesday, March 28, 2017

Bratanova et al. (2015) Find that Ethical Food Tastes Better

Boyka Bratanova, Christin-Melanie Vauclair, Nicolas Kervyn, et al., “Savouring Morality. Moral Satisfaction Renders Food of Ethical Origin Subjectively Tastier.” Appetite 91(1): 137-149, August 2015 [pdf here].

• The taste of food isn’t just about the chemical and physical properties of the food; rather, it also depends on the environment in which the food is consumed and the expectations of the consumer: Zinberg’s Drug, Set, and Setting applies to food, too. 

• It is the connection between consumer expectations and taste that is at the heart of the analysis in this article. Taste expectations might be enhanced if you are supportive of the perceived underlying ethics of the food; for example, if the food carries a Fair Trade certification, you might expect it to taste better – and therefore it might actually taste better to you. Further, this mechanism can also augment your willingness-to-pay for the food, and bolster your intention to purchase the food in the future. The ethical origin creates a sort of halo effect whereby other dimensions of the food are perceived more favorably. 

• The first study involves a 2005 survey of some 4000 adult grocery shoppers spread throughout eight European nations; the relevant questions concern tomato sauce. First, beliefs about the environmental benefits of organic tomato sauce (relative to conventional, non-organic sauce) are gauged; then, whether buying organic is viewed as doing the right thing, or making the consumer a better person. 

• In the survey – which was not accompanied by actually tasting any food – beliefs about environmental benefits were positively correlated with beliefs about better taste; the beliefs about better taste increased future buying intentions. 

• Follow-up experiments added some actual tasting of food into the mix. Besides environmental benefits, beliefs about fair trade and local sourcing also were examined. 

• A biscuit (cookie) company was described as either environmentally friendly or unfriendly, in two treatments applied to a total of 112 undergraduate students. Students thought that the environmentally friendly company would produce a higher quality biscuit (though whether the perceived quality difference is due to the halo effect of being environmentally friendly, or to the alternative biscuit being produced by an environmentally unfriendly company, is unclear). Upon eating the biscuits, however, the reported actual taste of the two types of biscuits did not differ in a statistically significant way. 

• The final experiment looked at fair trade chocolate and locally produced apple juice, in comparison with conventional products (that is, not as opposed to environmentally unfriendly products). One hypothesis is that people who endorse altruistic values will find the fair-trade chocolate to be taste-enhanced, while people who endorse environmental values will like the taste of the local apple juice. The data from 50 undergraduates tends to support the fair-trade hypothesis, but for the apple juice, taste experience and willingness-to-pay are reduced for the locally sourced juice. Nonetheless, if we look only at that subset of the participants who endorse environmental values, the better taste and higher willingness-to-pay results are restored.

Wednesday, January 18, 2017

Bühren and Pleßner (2014) on “The Trophy Effect”

Christoph Bühren and Marco Pleßner, “The Trophy Effect.” Journal of Behavioral Decision Making 27: 363-377, 2014 [pre-publication version pdf here].

• For everyday (convenience) goods, the endowment effect is reflected in about a 2-to-1 ratio of willingness-to-accept (payment to relinquish a good) to willingness-to-pay (to acquire the good). But for environmental goods or tickets to a basketball game, the ratio could be much higher. The gap is smaller when folks are in a good mood and larger when folks are in a bad mood, though sadness (as opposed to a bad mood) tends to reduce the gap. 

• The authors conduct a series of experiments involving endowing half of the subjects with a pen, under four treatment conditions. All the subjects know that the pen, though a nice one, could be purchased at a nearby shop for 2.10 euro. 

• In the Baseline treatment, half the subjects are given pens, then, willingness-to-accept (wta) and willingness-to-pay (wtp) are assessed. In the Trophy treatment, there is a 15-minute math quiz. Those who score above the median receive a pen, and become potential sellers in the exchange game to follow. 

• In the Work treatment, sellers are chosen randomly, but then they must take the math quiz before being given their pen. (They worked for their pen, they are told, though their performance on the quiz is immaterial.) In the Lottery treatment, everyone is informed that there is a lottery for pens, and the half who win the lottery are given pens. 

• The baseline treatment results in the usual (but weird!) endowment effect, with wta about twice as high as wtp. The Lottery treatment results in a slightly higher, but statistically equivalent, ratio. (Losers in the lottery don’t have a wtp that is lower than in the baseline treatment.) The Work treatment gives similar results, though the average wta (for those diligent workers!) is enhanced. 

• The broader “winning” condition (Trophy plus Lottery Treatments) yields a higher wta (compared to the other two treatments). The broader “work” condition (Trophy plus Work) yields a jump in wta, too. The Trophy treatment has a huge impact on wta, which went to 4.40 euro – and also dropped wtp from 1 euro to .5 euro; the wta/wtp ratio assessed at the medians is 9.6! It looks like the work element (taking the test) is slightly more important than the winning element in driving the trophy effect. And math quiz “losers” don’t like having the pen around to remind them of their failure. 

• The authors replicate a standard finding, that people do not anticipate the endowment effect. In the Work condition, they expect their wta to be less than 2 euro, but it is over 3.5 euro after the 15-minute quiz. 

• In one study, participants are asked how likely they think it is that their math test had been mis-graded. Trophy winners seem to decrease their wta when doubt of their deserving creeps into their mind. In yet another study, there’s (1) a two-hour delay or (2) a one-week delay before the wta and wtp valuations are assessed. With a one-week delay, we are back at the baseline endowment effect: the trophy effect evaporates. And the two-hour delay is statistically equivalent to the one-week delay.

Monday, January 16, 2017

Morewedge and Giblin (2015) Look to Explain the Endowment Effect

Carey K. Morewedge and Colleen E. Giblin, “Explanations of the Endowment Effect: An Integrative Review.” Trends in Cognitive Sciences 19(6): 339-348, June, 2015 [pdf].

• Endowment effects tend to be demonstrated in one of two ways: (1) the “exchange paradigm,” where people who are endowed with a good are less willing to trade it for another good than pre-endowment preferences would suggest; and, (2) differences between the willingness to pay (wtp) for a good by a non-owner and the willingness to accept payment (wta) by an owner – differences which seem to materialize about the time the non-owner becomes an owner. 

• Loss aversion is a standard underlying explanation for the endowment effect; once you own a good (or even expect to), your ownership becomes your reference point, and situations where you no longer own it are evaluated as losses relative to the reference point. 

• The authors look at five underlying mechanisms that are claimed to give rise to loss aversion and endowment effects, and offer a more general framework: “attribute sampling bias.” 

• Buyers and sellers pay attention to reference prices (low for buyers, high for sellers) that will increase their transactional utility. Prices are just one example of how buying and selling activate different cognitive frames that direct attention and recall to different pieces of information. Buyers and sellers do not have the same information readily in mind, so the endowment effect does not arise because they value identical attributes differently. People even have trouble recalling frame-inconsistent information. 

• Psychological ownership raises wta for two reasons: (1) ownership creates a connection between your identity and the good; the better that you feel about yourself, the better you will feel about the good. Losing the good can seem like a threat to your identity. (2) ownership helps you remember good feelings about the good, which otherwise might be harder to access. 

• Their attribute sampling bias explanation resembles the well-known phenomenon where people seek out information that confirms their current beliefs. 

• Sadness tends to reduce or reverse the endowment effect – it creates an implicit interest in changing your circumstances, so non-owners are more willing to acquire a good and owners are more willing to sell a good. But most frames in the exchange setting (other than sadness) tend to bias people towards maintaining the status quo. 

• People are eager to trade away a “bad”, which can’t be explained by some other approaches to endowment effects, where whatever you own you value more highly. Further, endowment effects are higher for those goods (like environmental goods) that have vague attributes, giving more room for attention to a biased sample of attributes to take effect.

Viscusi and Gayer (2015) on Reasons to Distrust Government Nudges

W. Kip Viscusi and Ted Gayer, “Behavioral Public Choice: The Behavioral Paradox of Government Policy.” Harvard Journal of Law & Public Policy 38(3): 973-1007, 2015 [pdf].

• Behavioral departures from full rationality are often used as justifications for government interventions in markets. But the resulting government policies can institutionalize, not rectify, the rationality shortfalls.

• Regulators themselves possess rationality shortfalls, and there are features of the environment in which regulators operate that push them away from pursuing first-best policies.

• If the median voter possesses biases, then a democratic government is likely to share those biases.

• Less-than-rational individuals tend to bear the costs of their errors themselves – not so with regulators. Further, individuals might have enhanced incentives (relative to bureaucrats) to acquire relevant information. Regulators, like all of us, can be overconfident in their abilities.

• Can we trust the claims that people are less-than-rational, or at least the universality of the claims? People are heterogeneous, and what looks like a mistake might reflect specific, reasonable preferences.

• Many consumer durables now have energy efficiency mandates, justified by internalities. But the energy-efficiency misjudgment is far from a proven problem. The claims of energy savings tend to be engineering-based, unavailable in practice, while individual circumstances can rationalize seemingly myopic choices.

• We could be more confident in an even-handed use of behavioral economics if a lot of pre-existing mandates were being replaced by non-coercive nudges. But instead, we see behavioral economics being used to tighten regulation, not to loosen it. Notice that the less-than-complete (narrow) self-interest that behavioralists often emphasize implies that even externalities might be internalized without government policy.

• The EPA instituted a fuel economy labelling requirement that seems intended to remedy all the problems that their later efficiency mandates also targets – don’t they trust their own labelling regulation?

• In dealing with health and safety risks, government does not seem to be more rational than individuals. In practice, worst-case scenarios are over-weighted in regulatory policy; EPA methodology leads to cascading of conservative estimates so that extremely low-likelihood problems can dominate policy. Further, the number of people exposed to a risk, which should be central in formulating policy, is ignored.

• Increases in risk are a sort of loss, and loss aversion kicks in – in the form of alarmist regulatory responses to ebola, terrorism, etc.

• The FDA fears errors of commission much more than errors of omission. [Nonetheless, the evidence base is weak – look how often risks are revealed only after approval; then there are off-label uses, which are quite legal, despite not having been tested in the usual sorts of controlled trials.]

• If organic veggies carry less risk than non-organic vegetables, but they cost more, it could still be better health-wise for people to eat more, non-organic veggies.

• For people to accept a small increase in risk often involves a payment some six times larger than they would pay for the same reduction in risk – a reflection of loss aversion. The “first, do no harm” principle leads to a similar effect in policy. It is often hard to identify victims of the FDA’s failure to approve a useful drug.

• Agencies can have tunnel vision, treating their issue in isolation. In OSHA, for example, regulators are not even allowed to look at the costs of fixing a hazard. One result: costs per life saved vary widely across domains, in ways that seem to be far from optimal.