Wednesday, May 02, 2007

Giving birth 

I've been giving a talk around St. Cloud to several audiences over the last two months on the number of business births in town. When the Wall Street Journal ripped the Bureau of Labor Statistics last week for poor employment data (subscribers link), I had wanted to write back to the editorialists that this thing is a lot harder than it seems. The talk has focused on the fact that our employment data was revised upward quite sharply in late February as the result of reconciling the survey data in the Current Employment Survey with the Quarterly Census of Employment and Wages. (Here's the text of the revision or re-benchmarking of the CES.)

My talk (I'll have a link here when I assemble it from the PowerPoint later) focused on the errors in the birth-death model used to build the CES estimate. To oversimplify, each month BLS has to guess how many new jobs are created from firms they could not possibly have surveyed because they either are new firms or firms that have closed their doors and no longer answer the phone. 300,000 jobs were missed, BLS thinks, because they underestimated the number of business births. We're not sure why that happened; at a talk over lunch today, one audience member commented to me that he was finding many small business owners forming LLCs and LLPs and that there are many others that nobody probably knows about. I don't think I have good data on this at all for the local area. So perhaps many more small business owners are out there that we don't know about.

Another example (h/t, Craig Newmark) is in the restaurant business.

"Do you know me?" asked Rocco DiSpirito in a 2003 TV spot for American Express. "I'm a chef who already runs two restaurants in New York. Now I'm opening a third on national television in a time when nine out of 10 restaurants fail in the first year."

Like many viewers, H.G. Parsa did know DiSpirito from his NBC reality show The Restaurant. The nine-out-of-10 figure was familiar, too. As an associate professor in Ohio State University's Hospitality Management program, Parsa had heard it many times before. But based on his 13 years of restaurant-industry experience, he still didn't buy it.

...His research�consistent with similar studies�found that about one in four restaurants close or change ownership within their first year of business. Over three years, that number rises to three in five.

While a 60% failure rate may still sound high, that's on par with the cross-industry average for new businesses, according to statistics from the Small Business Administration and the Bureau of Labor Statistics.

It's an excellent article with explanations for business deaths as well -- many may be planned, where one operated a firm to meet an objective and then closed it because the firm no longer made sense, not that it had failed. I don't think you could have even written this sentence fifty years ago without someone thinking you daft.

Economies are dynamic. The changes we are witnessing mess with our data in ways we have trouble predicting. So when I read pieces like this one on the age distribution of income -- which is really interesting -- I have to pause and wonder, is the data right?

One of my friends at BLS replied to me last week when I sent the WSJ article that the only way to get better data from the government is to demand it. (Or find ways to get it privately, though the data that comes from the government relies often on information collected as a by-product of tax collection, so it's hard to replace privately.) Rather than that, I prefer to question the data at every step. Most people don't understand the process of collection, and that leads to people saying silly things.