Tim Harford The Undercover Economist

Articles published in October, 2009

Why don’t all waiters get their just desserts?

Dear Economist,
Do waiters in mid-priced restaurants work less than those at high-end ones? If not, shouldn’t their tips be the same (in absolute terms)?

Dear Manoj,

I can see where you’re coming from: a 10 per cent tip on a £20 meal is less than a 10 per cent tip on a £100 meal. If it’s the same waiter doing the same job, shouldn’t the tip be, say, £5 for each meal – a 25 per cent tip in one case and a 5 per cent tip in the other?

This is not what happens. According to a survey by the economist Ofer Azar, the absolute size of tips in the US is overwhelmingly dependent on the size of the bill. In Europe, formal service charges often replace tips and the FT’s restaurant insider, Nicholas Lander, tells me that such charges tend to be proportionate to the bill – or if anything, to be a higher percentage in the fanciest restaurants.

I am not sure the puzzle is quite as perplexing as you think, though. First, the connection between what the customer tips and what the waiter gets is far from straightforward. Waiters are not slaves: if tips are too low to attract them, then the restaurant owner will have to add a wage. And if a waiter can earn hundreds of pounds in tips at a top restaurant, the owner will be able to demand a share without running short of staff.

Second, high-priced restaurants tend to have fewer customers per waiter, to ensure attentive service. They receive higher tips, but fewer of them.

Despite these points, it is of course possible that waiters are paid more in better restaurants. But in a capitalist society, skilled workers expect to earn more. I suggest you sample the quality of service at El Bulli or the Fat Duck, and pop into Pizza Hut on the way home. Then tell me again that the waiters should earn the same at each place.

Also published at ft.com.

31st of October, 2009Dear EconomistComments off

Why feedback can be just so much noise

Should managers be giving more frequent performance appraisals? Do “customer feedback” questionnaires serve any useful purpose? The answers are not obvious. A feedback-free environment is not conducive to learning new skills, but then again, feedback itself can be confusing or demoralising.

I suffered from both too little and too much feedback in my last year of school. That was when I decided to stop going to piano lessons, having been coasting lazily at a mediocre level for years. My piano teacher, who had maintained a tactful silence, wistfully remarked that I had a beautiful touch on the keyboard – better than any of her hard-working, virtuosic prodigies. I was not impressed. Had she said that five years earlier, I might have worked harder. (Or so I told myself.)

It was also the year that I decided to spend less time with my A-level Further Maths exercises and more time with my girlfriend. I judged that my modest mathematical skills would not deliver a grade I needed to get into university, which would have to come from some other subject. Getting a C was no more useful than getting an E. So I stopped working, duly got the E, and did indeed get into university by other means.

Apart from revealing a deep inner laziness, these anecdotes point out that feedback can cut both ways. My ongoing maths grades proved clearly to me that I was facing an insurmountable obstacle, so I gave up. But had the gap between my ability and my target been smaller, that feedback might have spurred me into action instead of into the arms of my girlfriend.

Furthermore, as Gary Klein points out in a thoughtful and wide-ranging new book on decision-making, Streetlights and Shadows, much of what we might think of as feedback is useless. We may discover that something we did went well, but not why. For a simple, oft-repeated task, that may be enough. But much of life is not like that, and simple feedback about outcomes will do nothing. Too much feedback, too soon, can also be counter-productive in the long run. We may learn quickly while the feedback continues, but when the trainer or manager disappears, we discover that we have no idea how to continue teaching ourselves.

Obviously, feedback is often just what is needed. A new research paper from Oriana Bandiera and Valentino Larcinese of the London School of Economics, and Imran Rasul of University College London, studies the effect on student performance of an accidental experiment. Bandiera and her colleagues realised that the MSc courses at a leading university all had the same basic structure (exams followed by a thesis) but for historical reasons – and apparently at random – some revealed the exam grades before the thesis was written, while others withheld information about all grades until the course was complete.

The results were surprisingly clear: feedback about their exam performance seemed to spur everyone on. Struggling students decided to pull their socks up; high-fliers were inspired to yet greater heights. The effect was roughly comparable to that of shrinking class sizes – but presumably a great deal cheaper to achieve.

I wonder, though, if Bandiera and her colleagues have really figured out the causal mechanism. In their theoretical model, the results make sense only if overconfident students put in more effort when unpleasantly surprised while diffident students work harder when given a boost. A more plausible explanation, it seems to me, is that uncertainty is paralysing for everyone.

And I wonder whether the results will carry over to the workplace. Honest appraisal is to be expected, eventually, in education. In the office, it can be awfully inconvenient for all concerned.

Also published at ft.com.

Business Life: Pay what you want

First published in Business Life, April 2009

Deciding how much to charge customers is a crucial decision for most businesses, and many devote huge effort to complex pricing schemes. Not all, though: some businesses turn the whole task over to customers. “Pay what you want” seems to be a new business fashion – but is it taking customer sovereignty too far? And is the model sustainable?
Pay what you want has two clever features. The novelty attracts customers and publicity, too. And affluent or price-insensitive customers tend to pay more. Any well-run business will try to offer low prices only to customers who demand them, not to every customer; pay what you want might achieve the same result without the fuss.
The Achilles heel of pay what you want is, of course, blindingly obvious: what if people don’t want to pay anything at all? No wonder the model has caught on in two very specific contexts: digital goods such as music files, software or blogs; and cafés or restaurants. With digital goods, the cost of providing an extra copy is close to zero, and collecting real money is difficult thanks to piracy and a customer base that is used to getting what it wants without paying. Inviting contributions is better than nothing. In a café, customers are used to tipping staff and find it hard to accept service from a smiling waitress and then pay little or nothing. Honour – or guilt – can be a powerful motivator. So can social norms: Americans, well used to tipping, were reported to have paid much more than others when invited to pay what they wanted to download Radiohead’s album “In Rainbows”.
My concern is that when the novelty wears off, pay what you want will collapse as a model, at least for restaurants and cafés. Not only will the free publicity ebb away, but there is every reason to suppose that customers will start to exploit the offer. The economists John List and Uri Gneezy have carried out an experiment that sheds light on this tendency. They hired temps to perform various tasks, and paid some of them well above the hourly wage that had been agreed. The question was whether the workers would put in extra effort out of guilt or simple gratitude. The answer: yes, but the effect wore off in a matter of hours. Selfish exploitation of the generous wage became the norm by lunchtime on day one. I suppose if the same thing happens to restaurant owners, they can always install a cash register.

30th of October, 2009Other WritingComments off

A video interview on Dear Undercover Economist

Cass Talks, courtesy of Cass Business School, London:

30th of October, 2009Dear EconomistVideoComments off

Loving and losing – is the cost too high?

Dear Economist,
With the imminent passing of my pet rat I am faced with a lot of grief; he has been a great pet and so I will be more saddened by his passing than if he had been a bad one. My question is: is it possible for the cost (the grief from losing a friend/pet/family member) to outweigh the benefit (the joy gained through time spent with them) and so make the purchase of my pet not worth it, as the net benefit would be negative? Would there be a point where you would say: “I don’t want to get involved because I love X so much that I will be destroyed if I lose him?”

Dear Ilka,

Your intriguing problem has not, as far as I know, been explored by economists before, although it has been discussed by artists. Your ailing rat puts me in mind of a departed sparrow, mourned in verse by Catullus. Paul Simon expressed the trade-off more directly in his early song “I Am a Rock”: “If I never loved I never would have cried.”

But poetic speculation gets us nowhere. Let’s head straight to the data. Andrew Oswald, professor of economics at Warwick University, provides the following data points to ponder, based on surveys of life-satisfaction. Relative to never having been married, being married is worth 0.38 “points” of life satisfaction on a scale of 1-7. Being separated is worth -0.24, widowed -0.19 and divorced -0.09.

This is not much to go on, but it is better than nothing. If we incautiously interpret these numbers as causal – in fact they are merely correlations – then we could conclude that 20 years of marriage is compensation for up to 40 years of widowhood. Ten years of marriage more than justifies 40 years as a divorcee.

For human marriages, these odds seem pretty good. For a pet rat, less so: the little darlings hit puberty at six weeks and rarely live past three years. Perhaps you should buy a tortoise next time.

Also published at ft.com.

24th of October, 2009Dear EconomistComments off

Want to help? Then make life harder for the aid agencies

A club sandwich, a pair of trousers, a ticket to the movies – in a typical market transaction, I choose and pay for my own desires.

Sometimes, however, I might buy something for someone else, and here trouble begins. If I am buying something – a goat, an HIV prevention course, a bit of paved road – for a complete stranger in a far-off land, the risks that something will go awry are far higher. How am I to know what is needed, where to send it, even whether it has been stolen en route?

This may be why we have aid agencies. Aid agencies are popular symbols of national generosity – witness the Tory commitment to ring-fence the Department for International Development’s budget, even as they speak of inevitable spending cuts elsewhere – and in principle should make better-informed decisions because they are in a position to put expert decision-makers on the ground.

In practice, things are not quite so simple. Aid agencies are government bureaucracies, of course. They are funded by governments and governments are also their typical beneficiaries. Even sympathetic critics tend to agree that aid agencies often spread themselves thinly across countries and sectors. Civil servants in poor countries are constantly tied up in meetings with aid agencies, while the agencies themselves fail to focus on what they do well.

Only a wild optimist would expect a market-style focus on innovation and value for money. So, in a new policy paper for the Center for Global Development, Owen Barder argues that it might be easier to change the rules of the game to encourage real competition than to change behaviour. He may be right.

There is hidden potential in the aid industry’s troubling fragmentation. The only reason industry watchers talk of “fragmentation” rather than “competition” is because different agencies don’t compete with each other in any useful way. High-level officials instead make high-level declarations about how the industry will “harmonise”. These declarations have little apparent effect – and if you imagine a Howard Schultz of Starbucks attempting to “harmonise” the world coffee-bar industry, you can see how idiosyncratic the harmonisation agenda actually is.

Could aid agencies be made to compete? India has already been streamlining aid, telling smaller donors to donate through multilateral agencies or not at all. But India is an unusually large and confident aid recipient. In most cases, countries will take all the aid they can get, regardless of its quality.

It might be more promising to apply competitive pressure from the donor end. Donors should start to ask whether their pet agency is actually well qualified to deliver a particular type of aid to a particular country – or whether the job could be contracted out to a charity, a rival aid agency, or indeed a private company. Aid agencies already hand out cash to charities, but a market mentality would demand more ruthless assessments of results and value for money. More radically, aid recipients could even be given aid “vouchers” redeemable for services provided by a range of charities and aid agencies. (Why give an unwanted present when you could give a voucher?)

All this may seem far-fetched. Perhaps, then, we could start by asking simple questions about where aid comes from, where it goes, how effective it is and how much is lost to administration – or worse. A few people – including Barder, and William Easterly and Tobias Pfutze of New York University – are tackling such questions. The aid industry needs to do a better job of providing the answers.

Also published at ft.com.

Buy my book or the hedgehog gets it

Here’s the hedgehog. Here’s the book. Thanks to NPR and Planet Money for inviting me on!

18th of October, 2009MarginaliaRadioComments off

Superfreakonomics reviewed

Superfreakonomics: Global Cooling, Patriotic Prostitutes and Why Suicide Bombers Should Buy Life Insurance
By Steven D. Levitt and Stephen J. Dubner
Allen Lane £20, 288 pages

For fans of the multimillion-selling pop-economics book, Freakonomics, all that needs to be said is that the sequel’s title is an accurate description. This book is a lot like Freakonomics, but better.
The original, a runaway hit, had its genesis in Stephen Dubner’s masterful New York Times Magazine profile of “rogue economist” Steven Levitt. “Rogue” may be stretching it a bit, because Levitt is, in fact, a garlanded and hugely influential professor at the University of Chicago.
He has applied his statistical techniques, now much emulated, to unconventional topics such as the link between abortion laws and crime, or whether sumo wrestlers cheat (they do, he concludes). The 2005 book that resulted was wide-ranging, fascinating and above all, likeable – however, it showed signs of haste, and it was never clear whether it was supposed to be a book by Steven Levitt or about him.
Book cover of ‘SuperFreakonomics’ by Steven D Levitt and Stephen J DubnerSuperFreakonomics offers much the same range and amiability, but is more polished. The book’s chapters cover prostitution; data analysis in healthcare and counter-terrorism; altruism; innovation; and geo-engineering. The reader may not guess the central topic from the chapter titles or the opening pages, however, which betray a fondness for springing surprises and putting twists in the storytelling.
Detours are all part of the style; an afternoon reading SuperFreakonomics is like one of those thrilling and occasionally frustrating conversations where ideas tumble out so quickly that they keep interrupting each other. In short, the book’s organisation is deliberately on the freaky side, but if you simply resolve to read it from cover to cover you are guaranteed a good time.
My favourite chapter describes the research of John List, a colleague of Levitt’s, as he zaps some of the most famous results in behavioural economics. In the “dictator” game, well-known in economic circles, player A is given $10 by the experimenter and told they can keep it all. Alternatively they can give some to anonymous player B. Many players do, indeed, hand over money, a finding that troubles conventional economic theory.
List thinks many researchers have embraced this finding too easily, however. “What is puzzling”, he comments, “is that neither I nor any of my family of friends (or their families and friends) have ever received an anonymous envelope stuffed with cash”. The lab experiments, in which large numbers of students display a preference for sending cash to anonymous strangers, need to be questioned more closely. Yet List showed that with small modifications to the dictator game, experimental subjects could be persuaded not only to curb their generosity, but to confiscate cash from others.
There is much more here, and all is told with verve and care. Levitt and Dubner have a gift for explaining precisely how a researcher discovers something. Their epilogue, on Keith Chen’s attempts to introduce currency to a monkey society, is a model of how to tell a gripping story of scientific research without compromising on accuracy.
The most eye-catching chapters in the book are the first, on prostitution, and the last, on global warming. The chapter on prostitution flits from academic research into street prostitution, carried out by Levitt and sociologist Sudhir Venkatesh, to an engaging profile of a high-end escort and various digressions into the economics of gender and other topics.
One of these asides provides the book’s best moment: when the authors demonstrate that a prostitute gets more money through the use of a pimp than a homeowner gets through the use of a realtor, or estate agent. The financial impact of a pimp is greater than that of a realtor, “Or, for those who prefer their conclusions rendered mathematically, PIMPACT > RIMPACT.”
Those with a prurient curiosity (I am guilty), will find some of the descriptions of what prostitutes do all day rather coy. Those with an interest in the economic angle (guilty, again), will find some questions unanswered.
“The real puzzle isn’t why someone like Allie becomes a prostitute, but rather why more women don’t choose this career,” Levitt and Dubner write. Do tell, thinks the reader, but they don’t, even though literature on the puzzle does exist. The leading research on the question is written by two women, not two men, a fact that some people will find relevant.
The analysis of street prostitution is based on careful academic work. The account of high-end prostitution is merely a journalistic profile of a single successful and intelligent woman. But Levitt and Dubner seem to have decided that while data-driven discoveries are generally wonderful, you can have too much of a good thing.
The authors claim to prefer data to “individual anecdote”, but part of the secret of their success is that they like a good story more than anyone.
As for the final chapter on global warming, it is a striking discussion of geo-engineering, surveying various schemes for cooling down the planet rather than trying to prevent climate change by cutting carbon emissions. This is a strong story, but it is also one-sided, portraying the geo-engineers as brilliant iconoclasts, dismissing the objections to geo-engineering as the knee-jerk reaction of the unreflective, and failing to convey the views of a single credible geo-engineering sceptic. A well-deserved swipe at Al Gore does not really count.
According to this chapter, the only reason everyone is making so much fuss about carbon dioxide is that they’ve never heard of geo-engineering, or are the kind of stubborn Luddites who think technology never solved anything. I have some sympathy with that view but the section nevertheless needed more balance.
In the end, a book such as SuperFreakonomics stands or falls on its entertainment value. And on that count, there’s no doubt: it’s a page-turner.
More revealing, though, was that I’d folded over at least a dozen pages, resolving to go back, follow up the references, and find out more. This is a book with plenty of style; underneath the dazzle, there is substance too.

First published at ft.com.

17th of October, 2009Other WritingComments off

Fine wine or finer feelings?

Dear Economist,

After reading with interest your plan to start exercising last Christmas (by giving to charity if you didn’t meet your goals), I’d be obliged if you could offer an opinion on a similar scheme I have concocted. Wine is the problem. I drink too much of the glorious stuff, and am unable to convince myself that doing so is unhealthy.

Your idea of giving to charity would not work with me. I’m uncharitable, I’m afraid, and would probably rather lie than give my earnings away. Here is my alternative. Each month I will deposit the total amount I would spend on wine in the family joint bank account. If I want a bottle, it must come out of this account, but whatever is left at the end of the month is to be given to my wife and children.

This appears to be an excellent solution; in my view, the guilt of taking something away from my beloved wife and children is far greater than taking from myself. Do you agree?

Dear DW,

In classical economic theory, your scheme would be useless. Every pound spent on the demon drink is always a pound unavailable to your wife and children, and it should make no difference which bank account you put it in.

But Richard Thaler, a leading behavioural economist, has a theory of “mental accounting” that supports your plan. We do attach different labels to different pounds: this one is for my pension, that one is slush money. And Thaler has discovered that those labels make a difference to the way we behave.

Your scheme may well work, then. But like all commitment strategies, there is a risk that it will backfire, and you end up with the worst of all worlds. You may find yourself unable to stop drinking, feel more guilty than ever, and demonstrate unambiguously to your family that you love booze more than you love them. You evidently like to live dangerously: good luck.

Also published at ft.com.

17th of October, 2009Dear EconomistComments off

A brilliant (and doomed) template for healthcare reform

As the debate on healthcare drones on in the US, I have been struck by a heretical thought: the differences between the British National Health Service and the US healthcare system are not nearly as important as their shared weaknesses.

The difference between the two systems has been exaggerated of late. The uninsured in America are not barred from emergency rooms by security guards. The NHS has not assembled a death panel to do away with Stephen Hawking.

I’ve had experience of both systems. My wife’s life has been saved once by American doctors and once by British ones. One of my daughters was born in Washington, DC, the other in London. And I’ll admit that the systems feel very different. The outcomes are different, the bureaucracy works in a different way, the waiting times are different and the rules of access are different.

Yet in one vital way, the systems are exactly the same: at no point during my interactions with either system did I ever have to wonder about whether a procedure was worth the price. Large sums were spent on me and my family, but I never had to ask myself whether my doctors and I were treading the path of cost-effectiveness, straying off into wasteful indulgence, or indulging in dangerous penny-pinching. Someone else always picked up the bill.

There is an obvious alternative. We could pay for our medical treatment the same way that we pay for our cars or our food or a roof over our heads: out of our own pockets. Before rejecting the idea out of hand, at least acknowledge that it would encourage us to ask a very different set of questions, including: “is there a cheaper way that would work?”, “can I get better value treatment elsewhere?”, and even “would I save money if I drank less and exercised more?” The effect on cost and quality would be bracing.

Think about medical technology. Why does its price keep rising while the price of other technology keeps falling? Perhaps it is just bad luck, but I doubt it. As long as patients have no way to demand better value instead of simply better quality, cost inflation seems inescapable.

The obvious objections to this modest proposal are that some medical procedures are very expensive and need to be paid for by the state or an insurance company; that some people are poor and can’t afford as much treatment; and that patients would find it hard to make sensible choices.

The first two objections are valid, but they can be overcome without the necessity of insurance for everything. It is perfectly possible to design a system where redistribution, forced saving and “real” insurance – that is, against unexpected and very costly events – address these concerns without whisking away every bill before the patient sees it. Singapore has such a system. David Gratzer (a libertarian Canadian psychiatrist) has proposed a US version in his superb book, The Cure.

As for the third objection, it is true that patients do not today have the information they need to make sensible decisions about buying their own healthcare. But then, why would they, given the current systems? I recall the local press in the US being full of articles along the lines of “the city’s 50 best dermatologists”. Value for money was never mentioned, but ask patients to buy their own treatment and you can be sure that such articles would soon be supplemented by the medical equivalent of “cheap eats” reviews.

I understand that the whole idea is a political non-starter. But it’s a shame. Not only is it colossally wasteful to outsource medical decisions to bureaucrats, public or private, it is also infantilising for us as independent human beings. We can do better.

Also published at ft.com.



  • 1 Twitter
  • 3 RSS
  • 5 Podcasts
  • 6 Facebook


  • Fifty Inventions That Shaped the Modern Economy
  • Messy
  • The Undercover Economist Strikes Back
  • Adapt
  • Dear Undercover Economist
  • The Logic of Life
  • The Undercover Economist

Free Email Updates

Enter your email address to receive notifications of new articles by email (you can unsubscribe at any time).

Join 188,297 other subscribers.