Aftermath
Page 12
The nudges are just getting warmed up. At the commuter train station you are admonished to “Step back from the platform,” “Mind the gap,” and “Watch out for the sliding doors.” If you drive instead of taking the train, you’ll encounter a roadside laser-guided speed indicator that flashes your speed in red and displays “SLOW DOWN” if you’re one mile per hour over the speed limit, but turns green and offers a warm “THANK YOU” if you slow down.
There is more to the proliferation of nudges than mere annoyance. In one hospital, the digital prescription system sent so many nudges designed to reduce medication errors that the staff learned to ignore them. This resulted in more incorrect dosages than when the staff relied on their own training and judgment. In one case described by Dr. Bob Wachter of the University of California, San Francisco, a doctor entered certain incorrect information about a teenage patient into a medication robot. The robot issued a warning, “but the doctor clicked out of it because unnecessary warnings popped up on the screen all the time.” The robot then dispensed 38.5 times the intended dosage, which was handed to a nurse. A bar-coded backup system next issued a prompt to the nurse, telling her that she was administering the pills to the correct patient despite her misgivings about the dosage. The patient nearly died.15
Choice architecture is rapidly approaching the point of totalitarian overreach. In July 2018, Argentina enacted a law making all 44 million citizens of the country involuntary organ donors. As Thaler would no doubt commend, there is an opt-out provision, but it requires an affirmative act. Who knows if the government bureaucrat who receives the opt-out notice records it properly or if that record is readily available when needed? Declarations of death are often not as cut and dried as most imagine. The temptation to take viable organs in borderline cases is strong. Thanks to choice architecture, state power can now literally rip your lungs out without your consent. That’s quite the nudge.
The nudges never stop coming. Online or in real life, the audio-video-digital cavalcade of nanny-nudges inundates us like a never-ending word to the wise. We learn to ignore it, but the cost in cognitive dissonance is high; not quite the free lunch Thaler and Sunstein suppose.
Transcending these singular objections is choice architecture’s quintessential weakness—the idea that Thaler and Sunstein are smarter than you and have correctly identified the right choice in every design. Thaler, Sunstein, and their ilk base this belief on their higher IQ scores, educational attainment, and academic credentials. This presumed intellectual superiority empowers behavioral economists to prescribe various “correct” courses of conduct for the rest of us, and to implement those prescriptions with nudges embraced by institutions that interface with the public.
The evidence for this presumption is entirely absent. On binary outcomes, economists have worse forecasting records than random guesses produced by monkeys, due to deficient models, herding behavior, and fear of losing one’s professional standing with an out-of-consensus forecast (an incorrect yet in-consensus forecast is always acceptable because “everyone thought the same thing”).
There’s more to economists’ dismal analytical ability than herding and self-interest. Prevailing economic models bear no relationship to the real dynamics of capital markets or economic systems. Examples of this discordance include the Phillips Curve, the presumed inverse relationship between employment and inflation, which has no empirical support. As noted, the period from 1960 to 1965 exhibited low unemployment and low inflation. The period from 1965 to 1968 exhibited low unemployment and high inflation. The period from 1977 to 1981 exhibited high unemployment and high inflation. The period from 2008 to 2013 exhibited high unemployment and low inflation. In sum, there is no causal relationship between employment and inflation. Still, economists insist on using the Phillips Curve to guide monetary policy with grotesque results, including a 120:1 leverage ratio on the Federal Reserve’s balance sheet. Another myth maintained by economists is value at risk, or VaR. This is the primary risk-management model used by banks and regulators. It assumes that the degree distribution of risk events is normal (the bell curve), that there is a risk-free rate (short-term Treasuries), that markets are efficient, and that prices move continuously in response to news. None of these assumptions is true. The degree distribution of risk events is a power curve, not a bell curve. Treasuries are not risk-free, as recent U.S. government credit downgrades and debt-ceiling debates reveal. Markets behave irrationally, and prices gap violently up and down with no opportunity to buy or sell as the market reacts to shocks. VaR is junk science, yet it rules the roost of risk. The most egregious example of junk science in economics is the widespread use of dynamic stochastic general equilibrium models, DSGE, for economic forecasting and policymaking. The name gives it away. Capital markets are not “equilibrium” systems; they are complex systems. This explains why professional economic forecasters are not only frequently wrong, they are directionally wrong, persistently wrong, and wrong by orders of magnitude. With this modeling record as a calling card, why should everyday citizens accept the choice architecture of economists?
For their part, the economists are unabashed. As noted, the most highly proclaimed success of choice architects like Thaler and Sunstein is their ability to manipulate employees into signing up for 401(k) tax-deferred savings plans. This is done by switching new employee choices from opt in to opt out.
Prior to choice architecture, a typical new employee document would include a question in the following form:
Our company offers a 401(k) savings plan. (Please see separate disclosure document for details.)
Please check here if you would like to participate: __________
This is an example of an opt-in clause, because the employee has to take an affirmative act to join the plan. Choice architects redesigned the form to read as follows:
Our company offers a 401(k) savings plan. (Please see separate disclosure document for details.) You have automatically been enrolled in this plan.
Please check here if you would not like to participate: __________
This is an example of an opt-out clause, because the employee has to take an affirmative act not to join the plan.
The changed wording is subtle yet powerful. In Nudge, Thaler and Sunstein give examples of employee savings plans where initial participation by employees increased from 20 percent to 90 percent once the switch from opt in to opt out was made. Thaler and Sunstein pride themselves on the fact that nudges still leave individuals free to choose. This is disingenuous because aggregate results are highly predictable even if individual cases technically allow choice. Thaler and Sunstein are preying upon personal biases to produce a result to their liking. They cannot easily evade responsibility for the outcome by hiding behind choice.
A real-world example shows why economists’ choices should not necessarily be trusted. Assume that two employees, Sue and Joe, both aged fifty-three, go to work for two companies with adjacent headquarters in a midsized city. Their salaries are identical at two hundred thousand dollars per year. Both are married and file joint tax returns with their spouses. Sue’s company and Joe’s company both offer 401(k) plans. Sue and Joe were each hired on the last business day of December 1999, and completed their new employee paperwork that day. Sue’s company uses an opt-in enrollment form; Joe’s company uses the opt-out form. Sue does not opt in and is not in her company’s plan; Joes does not opt out and is in his company’s plan. This is a predictable result according to behavioral psychologists, and a desirable result according to choice architects.
Sue and Joe both receive a discretionary $10,000 performance bonus on the first business day of each year; $10,000 is the maximum employee contribution that Joe can make to his plan. Each year Sue receives her $10,000 on the first business day of the year, pays her taxes on that amount at a 21 percent effective tax rate, and invests the $7,900 after-tax remainder. Joe also receives $10,000 on the first business day of the year, all of which goes into his tax-deferred 401(k) plan. No
taxes are due for Joe until he withdraws the money upon termination or retirement. For portfolio selection, Sue invests her after-tax bonus in physical gold bullion, which she keeps in safe non-bank storage. Joe invests in an equity index fund tied to the performance of the Dow Jones Industrial Average. In addition to normal market fluctuations, Joe’s stock portfolio pays 2 percent in dividends, which were reinvested and compounded tax deferred.
On the last business day of December 2017, Sue and Joe both retire from their jobs at age seventy-one, after eighteen years of service. Sue has no plan distribution since she never joined a plan. She sells her gold and pays a 28 percent capital gains tax at the special rate applied by the IRS to collectibles. Joe takes a lump-sum distribution from his plan, and pays his taxes at ordinary rates on the previously deferred income. With cash in hand, Sue and Joe look forward to their retirement years with one important difference. Sue has more money.
In this example, Sue’s cash balance at the end of 2017 is $265,725 and Joe’s cash balance is $265,713. It’s a small difference, but Sue wins.
In certain cases, more extreme than the one described here, Thaler and Sunstein describe a decision not to join a company savings plan as “foolish beyond a doubt,” and ask with regard to a more typical case, “How can we nudge these people to join more quickly?” What they do not ask is whether choice architecture produces better real-world outcomes.
There’s a lot to parse in this example. The dollar price of gold was near twenty-five year lows in 1999, which gave Sue an attractive entry point and a head start on Joe. Still, gold suffered its worst bear market in history from 2011 to 2015, in which 50 percent of the gains from 1999 to 2011 were lost. Sue invested after-tax dollars while Joe invested more pre-tax. Joe had a 2 percent tax-deferred dividend on his portfolio, while Sue’s gold had no yield. And Joe enjoyed one of the longest stock bull markets in history from 2009 to 2017, including blockbuster years 2013 with a 26.50 percent gain, and 2017 with a 25.08 percent gain. On the whole, Sue faced enormous headwinds, one reason Thaler and Sunstein nudge employees into 401(k)s.
Still, Sue won the race.
Some of the difference between Sue and Joe is due to taxes. Prior to sale, Sue’s gold was worth $313,762, while Joe’s 401(k) had a predistribution balance of $408,957. Yet distributions are mandatory for Joe and taxes are inevitable. Joe did not have to take a lump-sum distribution; he could have taken smaller annual distributions over time. That’s true, but Sue did not have to sell her gold either. She could have held it indefinitely and paid no taxes, while Joe faced mandatory distributions and tax bills every year. The comparison is also artificially constrained by putting Sue 100 percent in gold and Joe 100 percent in stocks. If Sue had invested 50 percent in gold and 50 percent in stocks, her returns would have far surpassed Joe’s even without the benefit of 401(k) tax deferral. In a fifty-fifty portfolio, Sue could have created her own tax deferral simply by not selling her stocks until the end of 2017 and paying the lower capital gains tax rate, versus Joe’s ordinary income tax rate. Likewise, if Joe had invested 60 percent in stocks and 40 percent in a money market fund, his returns would have been greatly diminished, and Sue would have dominated the contest again.
There is an infinitely large set of Sue versus Joe examples that could be devised with variations in start dates, end dates, tax rates, and rates of return. Sue wins in some, Joe wins in others. That’s not the point. The point is that the choice architect’s a priori assumption that 401(k)s produce superior results and that employees should be pushed in the direction of joining a plan is false. Thaler and Sunstein would never even have constructed an example along the lines just described because their training and mind-set do not admit the consideration of gold as an investible asset. Yet Sue is not “foolish beyond a doubt,” as the Nudge authors would have it, even if Thaler and Sunstein appear to be arrogant in the same measure.
Apart from these obvious deficiencies of choice architecture (a bias against learning, dissipating impact, hidden costs of excessive nudging, adverse adaptive behavior in response to nudging, and false assumptions about superior choices), the nudge project contains a huge hidden irony. Thaler and Sunstein begin their book by mocking the model of humans making efficient decisions guided by rational expectations, so-called homo economicus. They go through a litany of behavioral biases confirmed by experimentation, including risk aversion, confirmation bias, and recency bias, that show human decision making is “irrational.” Then they present the field of choice architecture, which creates traps devised to push behavior toward efficient decisions guided by rational expectations, exactly what they mocked. The Thaler-Sunstein affection for homo economicus is palpable. If homo economicus does not exist in the real world, they will nudge him into existence like latter-day Frankensteins. They come not to bury homo economicus, but to raise him from the dead.
What Thaler and Sunstein don’t comprehend is that what they deem irrational behavior is a highly evolved adaptation well suited to survival in stressful circumstances. As in the classic experimental model noted earlier, in which a subject is offered the choice of $3.00 with a 100 percent probability of receiving it (expected value $3.00), or $4.00 with an 80 percent probability of receiving it (expected value $3.20), subjects overwhelmingly take the $3.00 even though the expected value is less than the $4.00 option. This is considered irrational by the behaviorists. Yet it’s completely rational. In an Ice-Age survival situation (where some of these biases evolved), the cost of not receiving a “payout” could be death by starvation or hyperthermia. Taking the lower expected value with certainty was a small price to pay to avoid an uncertain outcome that could be deadly. The difference between the higher and lower expected values is, in effect, an insurance premium the subject pays to avoid a catastrophic outcome. Homo economicus does not exist, but neither do bundles of bias in need of the ministrations of choice architects. People know what’s best for them without nudges. Addictions and bad behavior will always exist, no matter how many forms Thaler and Sunstein design. The keys to improved outcomes are better education and more information, not manipulation and behavioral modification.
The difficulty is that Thaler and Sunstein bring their own biases and irrational beliefs to the table. They are biased in favor of employer savings plans despite high taxes on distributions. They are biased in favor of stock portfolios despite regular market collapses of 30 percent or more. They are biased against gold despite centuries of wealth-preservation performance. They are biased against tobacco despite musical geniuses, ballet dancers, and painters who smoked two packs a day and created lasting beauty in their art. They are biased in their belief that the future will resemble the recent past. They believe that deep-seated survival skills are anachronistic despite evidence of civilizational collapse from the Bronze Age and ancient Rome to Watts and Katrina. Their greatest bias is they think they’re smarter than you, me, and everyone else. They’re not. Choice architects should confine themselves to academia and leave the rest of us in peace.
Bad Ragaz
Switzerland appears on a map as an irregular rectangle, the horizontal longer than the vertical. In the lower left-hand corner is Geneva, the portal to France. In the lower right-hand corner is Lugano, the pathway to Italy. In the upper left-hand corner is Zurich, a window on Germany. In the upper right-hand corner is—nothing, except the high peaks of the Swiss Alps and villages snug among them. It’s here you’ll find Davos, home of the uber-elite World Economic Forum. Not far away is St. Moritz, the classic and still exclusive ski resort. Least known, and farthest from big cities and elite venues, is the quintessential alpine town of Bad Ragaz in the Swiss canton of St. Gallen.
“Bad” means spring or bath in German, a reference to the natural thermal springs that have made Bad Ragaz a destination for centuries. Bad Ragaz is tiny, six thousand souls in a mere ten square miles, nestled among the high peaks of the Graubünden Alps, just three miles from the border with Liechtenstein.
I was in Bad Ragaz on Marc
h 10, 2016, as a panelist at a private investment conclave for a small group of the largest asset managers in the world. The two-day conclave took place at the ultraluxurious Grand Resort, which offers a private thermal spring, a wellness spa, and a shuttle to nearby skiing for those inclined to the outdoors. The conference theme was “How Will the Euro End?” This conference took place in the wake of the 2010–2015 European sovereign debt crisis, when elite opinion was certain the euro was heading for a crack-up. Greece, Spain, Ireland, and Portugal were urged to quit the euro or be kicked out. Italy was not far behind. My remarks to the assembled money mavens said that the euro was not breaking up, it was strong and getting stronger, and that new nations would be added to the euro while none would exit. That was the same analysis I had been presenting since 2010, mostly to elite ridicule. Yet my analysis proved valid in 2010, 2016, and is still true today.
One of the perks of public speaking is hopping from one exclusive venue to the next. The Bad Ragaz Grand Resort held up its end in that regard. Another perk is the chance to spend time with the other keynote speakers. Downtime at meals or offstage offers the chance to chat informally with renowned experts in diverse fields. While at Bad Ragaz I spoke privately with Dan Ariely, the leading behavioral psychologist in the world after Nobelist Daniel Kahneman. Our encounter was an opportunity for me to explore aspects of the Thaler-Sunstein art of behavioral manipulation that had troubled me since I first encountered them.