TCS Daily

Washington Still Doesn`t Get The New Economy

By Kevin Hassett - September 5, 2000 12:00 AM

The data in August were exactly what the old-economy gurus at the Fed were looking for. The economic mood of the month was captured concisely in the August employment report which showed that employment in the U.S. increased only slightly. Worries of an overheated economy driving us into an inflationary Armageddon have abated for a while, and you can expect markets to continue to rejoice. While the Fed has probably done some damage by tightening too much, they look to be ready to stop for a while, and the tech sector, especially, should rejoice. When the Fed slams the economy and the stock market, it is the tech sector---which relies so heavily on equity finance for new projects---that suffers the most.

That one month's bit of data could cause such a change in course for policy is fairly ridiculous. This is because the ability of the economic data to provide clues about how well we are doing has been undermined significantly by the new economy. The punchline is that there is a big difference between growth that comes from making more Chevettes and growth that comes from the introduction of totally new products, such as the wireless web. Existing methods make the benefits of technology seem much smaller than they really are. To see why, I need to provide a little background about how we measure improvements in well being.

Suppose we want to create a measure of how much "stuff" our society has produced and then compare it to how we did last year. Suppose that our whole economy has only one factory, which makes Chevettes. In 1999, our mythical factory has total sales of $100. In 2000, our mythical factory has total sales of $200. Output went up, right? Maybe not. Maybe the factory made one Chevette each year, and the price went from $100 to $200. What we really care about is how many Chevettes our economy is producing, not the number of paper bills that change hands. If the number of Chevettes is the same in each year, real GDP is the same in each year. You have to adjust your total sales observation for changes in the price of the goods sold.

That's how our economic agencies make the data that guide the Fed. Look at how much, in dollars, our firms have produced, then adjust that number for changes in prices to get a measure of how much "stuff" we have produced.

Which is all fairly easy to do if we keep making more and more Chevettes, and that is where our growth is coming from. That probably is even a good description of how we used to grow. But consider a simple but dastardly change to our example. In 1999 our economy has total sales of $100 (one Chevette). In 2000, our economy has total sales of $200, because in that year we make one Chevette and one immortality pill. How much better off are we? What price should we use to adjust our $200 to back out how much more "stuff" we made? Those are very tough questions. In 2000, we made a completely new good that didn't exist in 1999 (and presumably leads to quite a bounce in human happiness). No price exists in 1999 to compare the 2000 price to.

Now let's look at the real world. Our total current dollar output in 1998 was about $8.8 trillion. In 1999 it was about $9.2 trillion. How should we adjust those numbers to back out how much better off we were in 1999?

The government's statistical agency, the BEA takes a bundle of products that existed yesterday and today, looks at how the price of that bundle changed, and then uses that measure of price change to back out real GDP growth. In our mythical example, they would use the price of Chevettes to tell us how much better off we are because we now have an immortality pill! So when the BEA turned the 1999 number into one comparable to the 1998 $8.8 trillion, there wasn't much of a change, even though an enormous variety of new products---music clips, internet ready phones, satellite services---really took off that year.

What should they do? Superstar economist Jerry Hausman of MIT has solved that tricky theoretical problem. To accurately measure how much better off we are in 2000 we should go back and create virtual data in the past. Pretend that music clips, say, existed but the price was so high that nobody bought them. The best measure of inflation (and therefor real GDP growth) lets the price change for new products be the difference between today's price and the hypothetical price at which the demand for the product would have been zero yesterday. Think about how high that virtual price would be. At what price per month would nobody have bought wireless web services? If we made that correction, our price measure that we use to evaluate how well we are doing would show enormous price declines, or deflation. And our examples helps illustrate what that would mean. If the price of Chevettes goes from $100 to $1, then our real output goes from 1 to 200 in our two sample years.

In the new economy, the pace at which we are developing new products is unprecedented, and accelerating. Its difficult to pin the correction down easily, but back-of-the envelope calculations suggest that if the BEA correctly treated the introduction of new products, then the measure of GDP growth that would result would likely be at least double the reported number.

Which leads to two policy conclusions. First, our growth statistics are likely off by one hundred percent or so! Accordingly, allowing policy changes to depend a lot on little wiggles in the data is silly. Second, the economic benefits of the high tech sector are vastly understated, but this is not evident to our decision makers because they are looking at bad data. Our next president should take this measurement problem on, before bad data leads to bad policy.

TCS Daily Archives