Perverse incentives
One of the cool things about studying economics is that it teaches a new and exciting way of thinking about things. This means we can look at all sorts of topics in a whole new light, spot problems that we never knew were even there, and come up with efficient solutions. And THAT means every now and then we come across an everyday, routine situation and say, “Err, what?”
This has happened to me a few times recently, all related to the problem of incentives. One of the neat areas in the economist’s toolkit is contract theory, which came to public prominence last year when two of its champions were awarded the Nobel prize. Contracts and incentives govern most of the relationships in our everyday lives, both formal (such as with our employer, our bank, the government…) and informal (neighbours, strangers, and of course, partners). And it’s absolutely critical to get the incentives right. Normally, they’ve sorted themselves out over the years, either through social norms or legal obligations. But every now and then, we come across situations where the incentives don’t align with the situation.
One example from earlier this week was when I was investigating customs clearance when moving back to Australia. Shipped goods are inspected by customs officials, and – if they decide the goods don’t meet the test – they can charge you for treatment or destruction of your property. That is, the officials, and only the officials, get to decide whether or not you have to pay them more money. Now in this case, we could assume (probably correctly) that the officials are acting as unbiased members of a government organisation in which there’s no corruption going on, and so the inspection decisions are made on the merits alone. But part of designing good contracts is reducing the potential risks of corruption developing in the future, and here the incentives don’t really line up.
Not that this will affect my shipping decisions. But fast-forward to yesterday when I started investigating buyer’s agents, a whole industry I previously didn’t even know existed. Basically, real estate agents are employed by would-be home buyers to search for property on their behalf, negotiate the price and execute the sale. Sounds good in principle, especially for the busy worker who doesn’t have the time or know-how to search themselves. But the fee structure is truly baffling. Buyer’s agents charge a (quite sizeable) percentage of the sale price – say, 3% of the final amount you pay for the house. That means when they’re searching for property and negotiating with the seller, they make more money if they get a worse deal for you. The incentives are almost exactly aligned in the wrong direction. How does this possibly make sense?!
Granted, I’m new to this field, but the more I’ve searched online the surer I am that the incentive system in this industry is really perverse (to use the economic term). So then I started to think about what sort of contract would actually align the incentives of agent and buyer. There are a couple of better alternatives to the status quo, though it turns out this problem is not as simple as it appears.
A fixed-fee model is the easiest, and a minority of agents do use this. But this means there aren’t any monetary incentives for the agent to really work hard to get you the best deal. (There are some reputational benefits of course, but that’s another story.) For the buyer who’s risk-averse and just wants to avoid getting screwed over, however, let’s see if we can do better. First, the agent searches for properties within the buyer’s strict criteria (price range, location, size etc). Then, when the buyer has agreed on one she likes, she agrees that she will pay in total the listed property price. The agent then gets paid whatever amount lower than the asking price the agent can negotiate. E.g. imagine the listed price is $500,000 and the agent negotiates it down to $490,000. The buyer pays $490,000 to the seller and $10,000 to the agent. The point is that the buyer knows exactly what her total cost will be before the negotiation takes place, reducing the uncertainly, as well as reducing the risks for the agent to act against the buyer’s interests.
It’s not a perfect solution, though. This contract might influence the agent’s selection of properties to present to the buyer; maybe we’ll only get to see properties that the agent thinks are overpriced (and therefore easily negotiated downwards). Can we do better?
The solution I came up with is complicated, but we can outsource most of it to a computer (in theory, anyway). First, the buyer comes up with a list of strict criteria for the perfect property. E.g. up to $700,000, within 4km of the office, near to parks and schools, at least 3 bedrooms, living space of at least 150 square metres, blah blah. Then there’s a second list of “optional extras” that the buyer likes: a garden would be nice, a pool would be amazing, more than one bathroom or more than one car park sounds good, open-plan kitchen, close to cafes, possible to ride a bike to work, and more blah blah. We feed our strict and desired preferences into our fancy computer software, which spits out a score function. Then, every potential home, for a given price, is given a score. If a house just meets the bare minimum criteria with no optional extras, the score is 0, and the agent gets a minimal flat fee. For every point higher in the score, the agent gets paid more. E.g. if the price is negotiated down to $5,000 less than our maximum price, the score goes up, the agent gets paid an extra $2,000, and everybody’s happy. If the location is closer to the office, the score goes up, proportional to the distance or travel time. Extra bedroom? Score goes up. Outdoor pool? You beauty!
The nicer the property and the better the deal is to the buyer, the more the agent earns. The key point is: it’s in the agent’s interests to try harder for the buyer. Incentives are aligned. Everybody wins.
So… Anyone know a good programmer?