Wednesday, May 20, 2009

Google Runs a Regression

A former Kellogg student forwards this cool article about Google using an "algorithm" to try to figure out which of its employees are likely to quit.

Google Searches for Staffing Answers

While Google is being tight-lipped about what they're doing, I am here to tell you that it's not so complicated. I doubt they're doing very much more than running a simple regression, not that much more advanced than what MBAs learn in b-school. Linear regression is, as MBAs know, is the best linear unbiased estimator of an underlying relationship. So it's exactly the tool for picking out the relationship between "quits" and various factors affecting the work environment.

Here's how you do it:

The simplest way is to estimate what's called a "linear probability model." Let the dependent variable in your regression be a zero if the employee doesn't quit, and one if he does. Let the independent variables be all the data you have pertaining to the employee's personal characteristics and work environment. Run a linear regression, and look for factors that strongly predict quit behavior. In a linear probability model, you can interpret the coefficients on the explanatory variables as how much a one unit change in that explanatory variable increases the probability of a quit.

Now the linear probability model isn't exactly the best thing to do, because of some funny characteristics of probabilities. Specifically, because probabilities have to be between zero or a one but the linear regression model doesn't account for that, you can get situations where the predicted quit probability for a given individual is less than zero or greater than one. So keep that in mind. But the linear probability model is quick and dirty, and it will give you insight into the relationships in your data.

To do something a bit more rigorous, you can do a logit or probit regression. Most of your standard statistics packages can handle this kind of regression pretty easily, and these methods always yield sensible quit probabilities. You have to do some more work to figure out the strength of the relationship using these methods, but it's not too tough.

After you've identified factors that are associated with quits, you can think about doing something about those factors --- and perhaps reducing your turnover.

One danger in this kind of analysis, though, is misinterpreting what causes what. People who know they're likely to quit are probably people who aren't going to undertake long-term projects at work. So the data might well tell you that people who work on a succession of short-term projects are more likely to quit than people who work on long-term projects. But this doesn't mean that shifting everyone to long-term projects will reduce turnover, because it's the employee's turnover intentions that drive project choice, not vice versa. So you need to think hard about what your data means before making wholesale changes in your organization.

Now you're as smart as Google. But unfortunately not as cool.

Tuesday, May 19, 2009

News Flash: Demand Curves Shift

Back to Accounting shortly....

But I'm teaching Managerial Econ for the 2010 EMBA class this summer, so I'll of course be thinking about some managerial econ ideas over the next ten weeks. And here's one thing that strikes me every time I cover the material about where demand curves come from: Journalists love --- love! --- to write about shifting demand curves.

Remember last summer when gas prices were so high? Every time you opened a newspaper (yes, I'm old-fashioned), there was a story about how people were changing their behavior because of high gas prices. People increased their demand for hybrids, bus passes, and city-center apartments, and reduced their demand for hummers and houses in exurbia.

And here's the latest from the NYT:

Yes, I Look Fabulous, But Inside I'm Saving

Even Hollywood A-listers are reducing demand for certain goods and services now that everyone's 401k is 40% down from last summer. Why, Alice Cooper's bass player parks his own car rather than paying $10 for the valet.

For most goods, we reduce our consumption when our income falls. But falling income leads to hard choices, and these are the sort of choices that attract journalists. This article focuses on how Hollywood-types are reducing spending on status goods only a little, but are cutting back more on goods that don't convey status. One exec canceled her vacation but still splurged in an Audi A5. Apparently nobody can tell whether you went to Fiji or Fresno over spring break, but tongues wag over your sporty new ride.

This article was in the Times The Arts section, which just goes to show that you can find cool economics pretty much anywhere you look.

Friday, May 15, 2009

Economics for Accountants (And Entrepreneurs) (Part 3)

In Part I of this post, I explained why the Market for Lemons results in unrealized gains from trade. This means that there are buyers and sellers out there who could both be made better off by trading... But they can't get the deal done. In the specific case of the Market for Lemons, trade fails because the buyer cannot be certain he is dealing with a plum seller.

Here's an important, related notion: Every business plan ever written starts with the observation that there are unrealized gains from trade. This sounds like an outrageously bombastic proposition, but I argue that it is literally true. When VCs and entrepreneurs ask "What's the value proposition?" (which they do), they are really asking about where the unrealized gains from trade are.

If there were no unrealized gains from trade, then you could never start a new business. Why? Because what businesses do is buy inputs (capital, labor, and intermediate goods) from suppliers, convert them into output (think of goods and services), and sell the output to customers. If the trade between suppliers and customers were already perfectly efficient --- in the sense that there was no way to rearrange things to offer both suppliers AND customers a better deal than they've already got --- then you'd never be able to get the new business off the ground.

Unrealized gains from trade can arise in a lot of ways. As one example, consider search costs. Amazon.com's business is built (partly) around reducing search costs, and hence allowing buyers of books and sellers of books to find each other more easily. That is, before Amazon existed, some buyers and sellers were prevented from realizing gains from trade because of the difficulty of finding each other. A buyer who really wanted quite a specific book might not be able to find it at his or her local B. Dalton Bookseller, and hence the trade wouldn't happen. Amazon makes it easy for buyers to find just what they want. Think of all the trades that didn't happen before Amazon existed, and think of Amazon taking a just a small piece of the value in each trade. Those pennies add up.

As we've been discussing, unrealized gains from trade can also arise from informational problems, as in the Market for Lemons. Because there are unrealized gains from trade, there's money to be made from figuring out how to get the trade to happen.

This explains Carmax. It also explains Executive Search Firms (aka headhunters).

If you think about Carmax's business model, they are an intermediary in the used car market. That is, they buy AND sell used cars. And if they're going to make a profit doing that, then there must be some reason why they're able to do better than buyers and sellers would be able to do on their own. That is, at least some sellers have to prefer dealing with Carmax to dealing with buyers directly. And at least some buyers have to prefer dealing with Carmax to dealing with sellers directly.

Let's start on the buyer side. What buyers are afraid about in the Market for Lemons is buying a Lemon car at a Plum price. This is why buyers shy away from cars with plum prices. But what if Carmax is selling a used car at a plum price? Should buyers believe that the car is actually a plum? There's reason to think that Carmax can do a better job (compared to individuals) of committing to sell ONLY plums at plum prices. Carmax can more easily offer warranties. And it can more easily develop a reputation for honest dealing, at least compared to an individual who sells a car once every five years.

On the seller side, in the Market for Lemons the plum sellers can't sell at all, because buyers fear buying a lemon car at a plum price. But suppose Carmax employs skilled mechanics who can, after inspecting a car for a while, do a good job of distinguishing plums from lemons. Then Carmax can be confident that when it pays a plum price it isn't getting a lemon, and plum sellers will be better off selling to Carmax than selling directly to a buyer who is afraid of buying a lemon.

Now, this isn't everything about the Carmax business model, but it's part of their edge. They facilitate trade in the used car market and, like Amazon, take a small chunk.

Headhunters work on much the same principle. The Market for Lemons affects the labor market because no firm wants to pay a plum price only to find they've hired a lemon executive. Headhunter firms connect buyers (firms looking to hire) with sellers (executives looking for new opportunities) and reduce search costs like Amazon. But they also check references and try to develop a reputation for successful placements. This reduces information asymmetry, and is what allows the headhunters to charge high fees.

So, an important point about the Market for Lemons is this: If markets fail to realize gains from trade, then there's a profit opportunity. So firms (and other economic institutions) will try to arise to get that trade to happen. Accounting is one economic response to the Market for Lemons. Entrepreneurship is another.

So entrepreneurs and bean counters have more in common than they think.