I remember exactly when I started going to the warehouse club. It was to buy an air-conditioner. Sure enough, the machine was cheap and well-made (in Korea), and soon I was hooked on Costco.

For a time, I told myself that I liked the enormous store -- a so-called "big box" outlet in a giant shopping center -- because of its essential simplicity. The local supermarket confronted me with a bewildering array of products, confusingly priced. There were just two choices for each product at Costco, a fancy brand (meaning an advertised one) and the house brand.

Gradually I became aware of the store's grocery department -- and the price differentials there. The extent of the bargain was difficult to gauge, because seldom had I bought such quantities of things before: 24 rolls of paper towels, a pound of vitamins or two large pork tenderloins. But a UBS Investment Research survey last year found the cost of a representative basket of groceries was 8 percent to 27 percent lower at Wal-Mart than at the large supermarket chains.

About the same time, I started going to Trader Joe's or Whole Foods for cereal, cheese and wine (the famous "Two-Buck Chuck" from Charles Shaw, whose cost has climbed; it now exceeds $36 a case); and to nearby Wilson's Farm for vegetables and fruit. And of course if I needed something I had forgotten, a quart of milk or a box of brown sugar, I would pick it up at the convenience store at the corner.

Before long, I was hardly ever going to the supermarket. When I did, I found narrowed aisles, loaded with ever more slightly-differentiated products, priced as confusingly as possible. It dawned on me, as it already had dawned on the managers of the supermarket, that we were not in Kansas any more.

It was a couple of years later, in January, 2005, that Gillette sold itself to consumer products giant Proctor and Gamble.

But when I went to look for a good description of the various aggregate effects of the changes that were occurring, I couldn't find one.

The first thing that caught my eye was an influential paper that has been making the rounds recently, by Jerry Hausman, of the Massachusetts Institute of Technology, and Ephraim Leibtag, of the Economic Research Service of the US Department of Agriculture.

If you studied only the Consumer Price Index, said Hausman and Leibtag, you would have to wonder whether the index-maker, the Bureau of Labor Statistics, even knows that Wal-Mart exists.

The changes themselves have been extensive. Wal-Mart began selling food in 1988, and today it is the largest grocery chain in the US, 50 percent bigger than second-biggest, Kroger, bigger than numbers three and four (Albertson's and Safeway) put together. Then there are the warehouse clubs (Costco, Sam's Club and BJ's) that require annual dues in exchange for low prices.

These "non-traditional shopping formats," as they are known (not only Wal-Mart, but Kmart, Meijer and other "super-centers") now account for something like 25 percent of all US food expenditures. They are "high-spend" destinations, as opposed to "low-" and "medium-spend" destinations, which are mainly convenience stores. Caught in the middle are the supermarkets.

It turns out, however, that that the Consumer Prices Index reflects none of these dramatic changes. That's because the Bureau of Labor Statistics employs a "linking procedure" that assumes prices at Wal-Mart are exactly the same as those at Kroger after quality adjustments. The items may be physically identical. Their prices may differ by 25 percent. But they go into the index exactly the same, thanks to an assumed "quality differential." Between the goods obtained.

Such linking procedures are a standard problem in the economics of new goods. Econometrician Hausman wrote a primer on them fifteen years ago in the The Economics of New Goods, the volume that included William Nordhaus' classic study of the price of illuminating a room at night.

(The cost of lighting had dropped a thousand-fold further than traditional measurements suggested, Nordhaus reported, thanks to faulty linking procedures, and the treatment of most other major inventions showed similar discrepancies.)

Omitting the effect of new goods in the CPI was surprising, Hausman wrote, at least from the standpoint of using it as an instrument to detect changes in the world at large, because most modern business strategies were of two basic varieties -- aimed either at becoming the low-cost producer of a basic no-frills good (house-brand toothpaste, for example, or a plain black telephone), or else learning to differentiate an ordinary product from its competitors by adding a little something extra to it so as to command a higher price (a secret ingredient for the toothpaste, portability for the telephone).

New-good bias is only one source of distortion in price indices, Hausman notes; "outlet substitution," as in the case of the rise of Big Box outlets, is another. And the arrival of a super-center or a warehouse club in a particular geographic market amounts to the same things as the introduction of a new good, creating both a direct effect (by offering a lower price option to consumers) and an indirect effect (by causing traditional supermarkets to lower their prices because of the increased competition).

When UBS studied the situation, for instance, supermarket prices dropped dramatically in the two years after a new Wal-Mart opened -- from a high of as much as 40 percent more than Wal-Mart prices (and a low of 25 percent above) to a high of 27 percent.more (and a low of only 8 percent above its aggressive new competitor) But you can't tell any of this from the CPI. It's as if the BLS doesn't know that Wal-Mart exists. It was fifteen years after their introduction before the CPI began to take note of cellular telephones.

But changes in the array of stores scattered across the landscape are only part of the story of Wal-Mart's growth -- and not necessarily the most important part. Big changes have been taking place as well in the goods those stores are selling -- specifically in their countries of origin.

Here's where a new paper by Emek Basker.and Pham Hoang Van, both of the University of Missouri comes in handy. In "Putting a Smiley Face on the Dragon: Wal-Mart as a Catalyst to U.S. China Trade," describe a process by which scale economies in retailing combine with scale economies in manufacturing to powerfully amplify the effects of technological change and trade liberalization.

What happens in the Basker-Van story is this. Wal-Mart begins by making dramatic improvements in the technology of operating a chain store. They learn to site their stores in the country, at points equidistant from rapidly-growing cities; to manage their logistical chain better than their competitors, with new technologies such as bar-coding and radio-frequency identification (RFID) of inventory.

Better logistics and distribution gives the American chain lower prices than its stand-alone competitors, which allows Wal-Mart buyers to turn their attention overseas -- to China, in particular, where manufactures are looking for outlets for their wares. Wal-Mart buys goods in such volume that the Chinese firms are enabled to take advantages if a second set of technological advantages in manufacturing: computer-assisted design, new robots, more flexible materials, and so on.

The two types of scale economies interact: developments in the retail sector influence the location of manufacturing jobs and the rise of Chinese manufacturing give Wal-Mart a powerful tailwind, since the retail giant can effortlessly meet the fixed cost of doing business in thee People's Republic.

The idea that market size influences production patterns goes back to Adam Smith, note Easker and Van. But only such a model of specialization both within and without the big box wall can explain the alacrity with which Wal-Mart abandoned its "Buy American" campaign, announced with much fanfare in 1985. (The Arkansas retailer promised to pay as much as 5 percent more for products made in America.)

But in expanding from 859 stores in 1985 to 1,880 stores a few years later, Wall-Mart may have passed a threshold at which the But American campaign was no longer the most profitable way to run the business, the economists conjecture. In any even, the "Buy American" campaign was quietly sidelined and Chinese imports soared. No big reduction in governmental trade restrictions is required to explain the surge, they say. The effect of market-broadening is enough.

Many insights arise from this mental habit of systematically linking changes in distribution patterns with developments in global manufacturing and trade. Gilllete's decision to sell itself to Proctor and Gamble, for instance, presumably reflected a the managers' conviction that manufacturers of heavily-advertised products would need more power in their negotiations with Wal-Mart buyers who were determined to give their customers the best possible prices. P&G can be expected to move more and more jobs overseas as well.

Likewise, the rise of intermediate grocery outlets, such as Trader Joe's and Whole Foods, may also be traced to the rise of Wal-Mart and Costco's grocery business, too. Their business too is coming at the expense of traditional supermarket chains, and presumably reflects an improvement a better deal for consumers.

In the end, the ;anguaged used to describe these changes is probably the biggest impediment to understanding what the Wal-Mart story is all about. Econometrician Hausman speaks of "channel-blurring" in the retail sector. But to non-economists, biological analogies come far more easily.

For example, .New York Times reporter Damon Darlin wrote the other day about the the changes in the music business that have been sparked by the appearance of thr iPod, "An entire ecosystem has emerged around the music player, introduced by Apple in October 2001. Other manufacturers had produced MP3 players earlier,. But the simple design of the iPod, plus Apple's iTunes store, quickly helped Apple to dominate the market. And that simple design -- some might even call it bland -- encouraged people to personalize the machine." The sale of accessories -- cases, car re-chargers, docking stations -- topped $1 billion last year, Darlin reported, and is growing at a brisk clip.

The development of language and analytic tools to describe our evolutionary economy is many years away -- probably decades. But while we wait, it is useful to remind ourselves that concepts such as ecosystems and niche space may have more power to explain the rapidly changing patterns of specialization that affect our world at least intuitively, than even highly refined models of supply and demand.