Over the weekend, I went with my wife to the grocery.  On the list was string cheese, which our chickens (don’t ask) get as a treat.

She reached for a package that had a prominent sign declaring “SALE!” and placed the package in the shopping cart.  When I asked her why she was buying that particular cheese, she said that she always bought it because it was always on sale.  I pointed out that if the cheese is really always on sale, then the “sale” price is in any practical sense the usual price.

I then tried to compare the cost of the “sale” cheese with the other brands.  As behavioral economists predict, the competing brands were packaged in different sizes, making direct cost comparisons difficult. 

A careful examination of the cost per ounce confirmed that the “sale” cheese was more expensive.  When I pointed this out to my wife, she shrugged and said she didn’t want to buy a larger amount of cheese — quite possibly exhibiting the endowment effect (the tendency for people to value what they already possess).  Either way, a good spouse knows that you have to choose your battles carefully, so I relented with a nod.

How does this apply to the pharmacy benefit?  When we talk about member savings with Home Delivery, we should (and do) highlight the cost comparison for comparable days’ supply rather than cost per prescription.  And we should frame the offer as one that focuses on getting the same medicine rather than stepping away from something one has already chosen.

Healthcare is rife with situations in which costs and benefits are separated over time.  For example, staying adherent to preventive therapies (e.g., antihypertensive medications) means bearing costs today to gain future benefits.

Researchers have repeatedly demonstrated that a significant proportion of people have very high discount rates; that is, outcomes that occur in the near term loom very large relative to those that occur later.  Even more interesting is the finding that a large element of this preference has nothing to do with how much later in time downstream outcomes occur; they’re simply discounted because they don’t happen now.

Much of this research has involved posing questions such as the following:

Which do you prefer?

A) $5 today

B) $6.20 in 26 days

By fiddling with the relative amounts and the delay, one can get a measure of time preference.

Three researchers at Stanford — Eran Magen, Carol S. Dweck, and James J. Gross — realized that in both options A and B above, there’s a “hidden zero.”  That is, if you choose option A ($5 today), you are also implicitly choosing to get nothing later; if you choose option B ($6.20 in 26 days), you are also implicitly choosing to get nothing today.

We know that framing can have big effects in decisions such as these.  It’s also been shown that people prefer increasing income over time, even if there is no economic difference between the income streams.  Thus, it’s quite plausible that making the zeroes explicit might change subjects’ choices.

To investigate, the Stanford team randomized subjects to receive the implicit version (as described above) or an explicit version.  An example of the explicit version is:

Which do you prefer?

A) $5 today and $0 in 26 days

B) $0 today and $6.20 in 26 days

Each subject faced 15 such decisions, and the researchers kept track of the fraction of times the subject chose the near-term option.  They referred to this percentage as the “impulsivity” score.

In the first study, the researchers simply asked the subjects which they would choose.  The subjects could answer however they wished but didn’t have to live with the results of their decisions.  In a second study, they informed the subjects that one of their choice pairs would be selected at random, and the subjects would receive the option they’d selected.  Here are the main results:

Two important points jump off this chart.  First, moving from a hypothetical decision to one in which there’s a chance that you might have to live with the results makes a big difference; impulsivity scores drop by about a third.  Second — and this is pretty interesting — making the zero explicit (that is, laying out both what you’re getting and giving up at the relevant points in time) has an effect of comparable magnitude.  Full results of the study were published in the July issue of Psychological Science.

These findings suggest that when communicating choices to patients, we need to do our best to make the comparison of options as “apples to apples” as we can.  It will be interesting to see how best to apply this insight to decisions like therapy adherence.

This month, the Mississippi breached levees that protect small towns in Illinois, Iowa, and Missouri.  I recently read an article about one town’s decision-making process regarding rebuilding levees after the 1993 floods.  The gist of the article was that there was a lot of talk immediately after the flood about making the levees stronger than before.  Eventually, those talks evaporated like the flood waters that instigated them.

In this case, the recency bias — in which we overemphasize the importance of recent events in decision-making — might have played a role in the town’s process.  But I wonder to what extent anchoring overrode that process.

Anchoring is the human habit of relying on one piece of data or information to make judgments.  With flooding, I am concerned about the phrase “100-year flood.”  A 100-year flood is a flood that has a 1% chance of occurring in any year.  If I interpreted this to mean “in all years,” a flood of that magnitude would occur only 1% of the time. 

If you thought that there was a 1% chance of a levee-breaching flood, would you reinforce the levees?  If you knew that over 20 years, there was an 18% chance of a levee breach (1% odds compounded over 20 years), would you reinforce the levee?

I also wonder what assumptions underlie that 1% chance.  Does that assume random distribution across all years?  Or does it take into account climactic patterns that would cluster floods?  If the latter is true, and the 1% chance over a long period of time is really series of 0.2% chances and 3% chances, the odds of recurrence within 20 years would be closer to 50% (assuming a 3% chance of occurrence).  Would you reinforce the levee then?

I hope the convention “X-year flood” does not lead decision-makers to incorrect decisions because of human bias.

Last week, Bob explained how hyperbolic discounting skews our judgment in decisions about the future.  Today, I’m going to look at one way to use this knowledge to improve decisions.

First, let’s briefly revisit the concept of hyperbolic discounting.  This is a theory that says that when people are making decisions, they first determine whether something happens now or later.  Things that happen now get full credit; things that happen later are discounted by about one-third to one-half.

I know “guaranteed employee wellness” is a boastful title, and I can’t ensure the wellness of your employees.  But if you implement the following, I suspect you’ll see a vast improvement in their eating habits.  The goal is to do this in a way that does not remove real or perceived choice.  (By the way, you must have a company cafeteria to make this work.)

Now, the problem with cafeterias is that when people go there, they are hungry (the “duh” statement of the year).  This is not a problem of the function of the cafeteria, but a problem of the mindset of the people.  When were a dealing with primal feelings, such as hunger, we tend to think short-term.  In the cafeteria, then, we discount costs and benefits that occur even 20 minutes hence.  When surveying the options for food, future health benefits get halved (mentally) while taste is getting full value.  So do you go for the salad bar, or get the burger and fries (just this one time).  How can we make good choices in the face of this hard wiring?

One idea:  Require employees to make their lunch choices 10 a.m.  Put all the regular menu items on the list, including the burger and fries, so no one can complain about choice constraints.  But before 10 am, both the benefits (taste, satiation, nutritional value) and costs (fat, guilt) are considered “future.”  So all choices are back on equal footing because all costs and benefits are discounted by half. 

To amp up the effect, have employees pay for the meals before 10 a.m. as well.  This will provide extra incentive when the wafting aromas from the grill are tempting.  Why not also give a 10% discount to use the early-order system?  A cost of $0.50 or so a day to remove temptation but create a healthier workforce might be a good investment.

Which would you prefer: $100 today or $100 a year from today?  Most of us would take the money today so we could invest it, spend it, or in some other way benefit from it immediately.  When it comes to benefits, it’s pretty obvious that sooner is better than later.

Economists, accountants, and other financial types typically handle this issue of time value using the net present value approach.  Put as simply as possible, they assume a constant discount rate over time; the further out the future event, the lower the net present value.  Formally, the equation they use is:

NPV = C / (1 + r)^N

Where C = the future financial outcome (e.g., $100), r = the periodic discount rate (the higher this number, the less you value events in the future), and N = the number of periods into the future the outcome occurs.  (This is the discrete time version; there’s also a continuous version of the formula as well.) 

If, for example, your discount rate is 5% per year, $100 one year from now has a NPV of $95.20 = $100 / (1 + 0.05)^1.  This means if your discount rate is 5%, you should be just indifferent between $95.20 today and $100 a year from today.  If the $100 was two years out, the NPV would be lower: $90.70 = $100 / (1 + 0.05)^2.  The rational choice of the discount rate should depend on what else you could do with the money in the intervening time as well as inflation.

Behavioral economists and psychologists are less interested in how people ought to behave and more interested in how they actually behave.  And what they’ve found is that when it comes to comparing things that occur at different points in time, people don’t adhere to constant discounting… not even close.  Instead, they first determine whether something happens now versus later.  Things that happen now get full credit; things that happen later are discounted by about one-third to one-half.  Importantly, how much later doesn’t seem to matter that much; something that happens a year from now isn’t that much different than something that happens two or five years from now. 

For those of us trained in the NPV approach, this seems counterintuitive.  But try out this example: Would you rather have $10 today or $15 a week from today?  A surprisingly large fraction of people take the $10.  Now change it slightly: Would you rather have $10 a year from now or $15 a year and a week from now?  Most people choose the latter.  This preference reversal cannot be explained by the NPV model, but is easily explained by the behavioral economic model.  (I leave the math to the reader to show that this is the case.)

As we’ll show in an upcoming entry, the behavioral economics approach also explains why it’s so natural to make resolutions about exercising and so difficult to execute them.

On a recent weekend, my family had an unusual number of social events.  After casually observing and acting upon social norms all weekend, I used the last event, my son’s fourth birthday party, to more carefully observe what was happening. 

For my son’s fourth birthday we invited three families, each with a child who is a good friend of my son.  We had a BBQ and made it a full dinner.  I observed the following norms:

• Price of entry — When we have friends over for dinner, everyone wants to bring something.  Even if there is nothing else to bring for the formal meal, guests will usually bring a bottle of wine or an extra dessert.  For the party, everyone brought a gift for my son, but none offered or brought food-related items. 
• Kid’s gift — The social norm among the families is that $10 to $15 is the limit.  No one wants to send a signal that operating outside that range is acceptable.  Given the dearth of gifts available in that range, you get a LOT of repeats.  So many, in fact, that one of the attendees was embarrassed because she intended to give my son the same gift that we had given her son JUST THE DAY BEFORE at her son’s party.  She bought a different gift.
• Shared parenting — With eight kids running around the house, skirmishes are bound to happen.  There was an unspoken sharing of peacemaking responsibilities when it was needed, regardless of whose children were involved.  So I felt it was my duty to remove the teeth of one child from the arm of another, even though neither was my child.  And no one had to tell me to do it!

There are many other social norms in our lives, most so assumed that even with keen awareness they go unnoticed.

We all know the promise of consumer-directed health plans (CDHP) – give people money and great information, and they will be able to make prudent decisions about how they allocate their healthcare dollars.  But if we examine the typical structure of CDHPs through the lenses of behavioral economics, some potential flaws emerge.

• Paradox of choice – Too many healthcare options, or the complexity of deciding among those options, may actually lead people to make no choice whatsoever, be it for treatment options, providers, etc.
• Present me vs. future me – Humans put a huge discount rate on dollars in the future; thus, we have a hard time rationing current spending – bye-bye HRA in April.
• Financial incentives are not enough – Often social norms are more powerful motivators than the market norms used by CDHPs to influence behaviors.

 Have we yet seen evidence of sustained behavior change through high-deductible plan designs?

I had the honor of working with Jack Wennberg when I was on the faculty at Dartmouth.  Jack pioneered the epidemiology of health delivery; that is, subjecting health delivery to the same techniques used to understand diseases in populations.  (He also gave me my first academic job at Dartmouth’s Center for Evaluative Clinical Sciences.)

One of Jack’s most interesting findings was that of supply-induced demand.  For example, the greater the number of hospital beds per capita, the greater the admission rate.  Jack explained that physicians making admission decisions are in essence sorting through patients.  The most severe or risky cases have to be admitted.  The clearly benign cases should definitely not be admitted.  But many cases are in the middle — these are patients with significant disease that can be effectively managed either in the hospital or at home.  Because there are a large number of patients in that middle zone, the supply of beds drives overall admissions: You always make sure you have a few beds in case serious cases show up but otherwise fill the beds with those patients who would probably be fine without admitting them to the hospital.

A few days ago, I was shopping at a higher-end cooking store for a cast-iron griddle to use on the outdoor grill — something called a piastra.  I picked up the heavy item with both hands, and as I headed for the register one of the employees pleasantly took the piastra from me and said he would hold it for me at the register.  I thanked him, turned around, and spent a little time looking at more grilling gear.  As I was mulling over how much trouble I’d be in if I came home with a new set of heavy-duty tongs (which I didn’t need), it struck me that I’d been nudged.  By freeing my hands, the clerk had created supply (i.e., room to carry more items), quite possibly in the hopes that it would induce demand.  It didn’t work that time, but I wouldn’t be at all surprised that it affects their bottom line.

A recent issue of New Scientist describes current research in the science of persuasion.  Here are the eight things you can do to be more persuasive:

1. Subtly mimic your counterpart’s mannerisms.  Be careful though. Being overt will backfire here.
2. Reframe issues to suit your needs.  Are estate taxes “inheritance taxes” or “death taxes”?
3. Limit your arguments in a discussion to no more than three.  The more easily we can come up with arguments for a position, the more confidence we place in that position — but more than three arguments are too many to recall.
4. Mentally exhaust your counterpart.  Either wait them out or catch them at the right moment.  Self control takes brain resources.  When those are depleted, persuasion is easier.
5. Persuade women face-to-face.  Persuade men by e-mail.  This effect may occur because women form communal bonds to get to agreement while men use face-to-face interactions to establish competence, which does not lead to persuasion but conflict.
6. Don’t use hesitations in your speech, such as “I mean” and “ummm.”  Hesitations tend to draw attention to the speech rather than the content.
7. Get people angry, but for your cause (or against your “common” enemy).  This effect is amped up when the angered person believes there is a solution to the problem.
8. Slowly move people off of their positions, especially if those positions are entrenched.  Strong arguments aimed at obstinate people may just reinforce those beliefs you want to change.

These items could also be labeled the eight things you should be aware of if someone is trying to persuade you.

After eating lunch in Express Scripts’ cafeteria recently, I started to walk back upstairs to my office when I was blocked by a large group of people.  To clarify, the large group of people was composed entirely of women.

These women happened to be congregated around an inconveniently placed table to solicit participation for the Susan G. Komen Race for the Cure, an excellent cause for which Express Scripts is a major local sponsor.

My question was, “Why are there only women at this table?”  So I stuck around for a few minutes to make sure my sample wasn’t too biased.  Not one man stopped at the table (actually a few seemed to purposefully avoid it), while women were inclined to at least take a peek, if not stop.  I can understand that breast cancer has special interest for women, but why did ZERO men stop by the table?

My hypothesis is that no men were working the table.  There were four enthusiastic women attempting to drum up support – no men.  So to what extent is the liking principle at work here?  We know that we like people who are similar to us.  By having men represented at the table, maybe we could generate even broader support.