Computer numbers don’t always equal fact

By John Cunniff The Associated Press
Saturday March 10, 2001

Housing sales are viewed by economists as an indicator of the economy’s good or poor health, so when a real estate group reported a big sales decline in January, it deepened an already gloomy outlook. 

And it should have. Homes are catalysts. When people change addresses they tend also to spend on furniture, appliances, landscaping and various supplementary items. They may renovate, and maybe even buy a second car. 

It explains the consternation that greeted the National Association of Realtors’ announcement of a sudden, 6.6 percent drop in January sales of existing homes. It tipped the indicator arrows toward recession. 

That was on February 26. On March 7, the arrows snapped back to a more  

neutral position. 

We erred, said a Realtor spokesman, blaming computer software. Instead of a 6.6 percent decline in sales of existing homes, that sector rose a healthy 3.8 per cent, adding 480,000 sales to the original estimate. 

Because this is a big economy producing big numbers, small glitches can become magnified into popular misinterpretations. But it isn’t glitches alone that cause problems. The best numbers can do it, too. 

It has become that way, ironically, as more people with more money react to government economic numbers that once concerned only a circle of professionals – academics, economists, executives, government officials. 

Now, more people than ever are directly invested in the economy via home ownership, stocks, mutual funds, 401(k)s and other individual retirement plans, corporate pension plans, credit cards and the like. They are better educated, too, and the media serves them more and more information. 

The reaction to computer numbers can be immediate rather than delayed, as it once was. And magnified, too. And, unlike the old days when the pros knew the limitations of numbers, worsened by imperfect understanding. 

It is not fully understood, for example, that numbers may be subject not only to errors, but to incomplete data. That they are often revised. That they may be seasonally adjusted, perhaps imperfectly. That some are meaningful, some not. 

But as reliance on computer-driven numbers grows, the pros are having their problems and those problems can cause economic landslides. 

The evidence is played out daily: semiconductor and other industries overestimating their markets’ growth rates; bad guidance from Wall Street; corporate chiefs lowering earnings estimates made just a month earlier. 

There may be no substitute for running an economy by the numbers, but a bit of old fashioned intuition can occasionally help lessen the impact. 

It is difficult, though, and can be embarrassing to attempt defending instinct and intuition, while the worst of decisions can find an alibi and security in blaming the computer numbers. Ask weather forecasters. 

Numbers in a modern economy, though off the mark and misinterpreted to boot, seem chiseled out of the hard rock of truth. They dictate. 

John Cunniff is a business analyst for The Associated Press