Walgreens drug stores had a big forecast error that cost executives their jobs. The CFO had to cut their forecast of revenues from pharmaceuticals paid by Medicare plans by $1 billion. That's about a 12% miss. Don't worry. He got severance of $3.2 million and a performance bonus of about $1.2 million (or almost 1% of the miss. I wish I could miss a forecast and get a bonus like that. I've never messed up Walgreens forecasts; all I want is a chance.

One of the reasons for the severance payments is to prevent a former executive from working elsewhere. Maybe the money would be better spent getting competitors to hire him.

* * *

Skepticism about statistics is always a healthy thing. One of the ways to emphasize the point is to look at silly applications of statistical methods to illustrate the care needed in using them to solve real problems.

For those who suffer with baseball teams who have trouble winning, as I do with my beloved New York Mets, we can entertain ourselves by looking at “spurious correlations” of two independent data sources. These data have nothing to do with each other, but but appear to be statistically related.

Here are some examples from a Mets fan blog. There is a 62% correlation of hot dogs eaten by Joey Chestnut at the July 4 Nathan's Hot Dog Eating Contest and the percentage of hitters pitcher Bartolo Colon strikes out. Perhaps you prefer the 89% correlation of mortgage rates and the percentage of batters who swing and miss at strike three when Daisuke Matsuzaka is pitching.

There is a more serious site about statistics that generates spurious correlations. Did you know that there is a 96% correlation of per capita consumption of cheese and the number of people who die from being tangled in bedsheets?

The owner of the site is Tyler Vigen, a former intelligence officer and now a J.D. candidate at Harvard Law School. His book about misuse of statistics and some of his favorite spurious correlations will be published by Hachette in May 2015.

As Vigen notes, the purpose of these is to make sure that people understand that correlation and causation are not the same. What you are seeing are two data sets that happen to have similar patterns over time. If you hunt hard and long enough for data you can find one that fits what you need. One time I prepared a correlation for a WhatTheyThink April Fool's edition where we compared printing shipments to the number of home runs hit by Ken Griffey, Jr. Vigen's site lets you create thousands of crazy calculations. It should be part of every statistics and econometrics class.

The other lesson to learn for legitimate correlations is that all correlations relate to those data at that time. They are not “forever” statistical relationships. Back in grad school (the first time) a finance professor noted that she worked with some commodities traders who developed models that they found valuable when their model stopped working.Then they knew that a buying or selling arbitrage opportunity presented itself.

Models are supposed to be an abstract representation of reality. They are not perfect, and cannot be. Use them cautiously, but still use them creatively. Models are there to help us learn and explore things that are not easily seen.

* * *

Stock buybacks are still distorting the stock market averages. The Dow Jones Industrial Average, for example, does not take into account the total market capitalization of the stock. That is, when a company buys back its stock it does not make the value of the company go up, but makes the share price increase because there are fewer shares. A company can be doing poorly, borrowing money to buyback stock, making its stock price rise. Buybacks are supposedly “shareholder friendly.” Wouldn't you prefer a company that had reasons to reinvest this extra cash into new products, improving productivity, employee training and bonuses, and especially increasing its dividend? Instead, many executive bonuses include incentives to increase the stock prices or to increase earnings per share. Stock buybacks do that.

The Wall Street Journal reports that “In mid-August, about 25% of nonelectronic trades executed at Goldman Sachs Group Inc., excluding the small, automated, rapid-fire trades that have come to dominate the market, involved companies buying back shares. That is more than twice the long-run trend...” and continues “According to Barclays, companies in the second quarter spent 31% of their cash flow on buybacks, the most since 2008 and up from 14% at the end of 2009.”

Companies that prefer to just give the money to stockholders as dividends are penalized. Why? Because dividends do not increase the stock price, and therefore do not increase the stock market averages. If your company's stock price is flat, but you have a record of increasing and reliable dividends, you are making a mistake, and not working in the interest of stockholders.

Many companies seem to try to appease everyone: they try to have increasing dividends and also have buybacks. Big institutional investors and the company executives like the buybacks. I think I'd prefer to have them creating new products or building up their cash for special situations of opportunity.

* * *

Furniture retailer Ikea has been getting a lot of buzz about its online video promoting its print catalog. It's called the “Ikea BookBook” and its benefits are playfully described with funny skits and narration laced with computer jargon. It's well done and deserves its high viewership.

It reminds me of another video that's not promoting a product, but a skit called “Medieval Help Desk” where a new technology, a book, replaces the established and much-preferred scroll.

# # #