Photographer: Victor J. Blue/Bloomberg.
Photographer: Victor J. Blue/Bloomberg.

Well, my piece on Costco Wholesale Corp., Trader Joe’s and Wal-Mart Stores Inc. has attracted a lot of heated argument. Why did I use worldwide figures instead of U.S. figures for revenue and profit per employee? Why didn’t I point out that Costco’s profit margin is much lower than Wal-Mart’s? (Aside from the graph in which I pointed that out, anyway.) And if Costco’s business model is so much better than Wal-Mart’s, maybe Wal-Mart should start putting stuff on pallets, huh?

First things first: I used worldwide figures because international profit figures aren't broken out separately on its financial statements. But how much difference that distinction would have made is unclear. Both companies get about two-thirds of their revenue from their U.S. operations, which is also approximately the percentage of their employees who work in the U.S.

Now about that profit margin. In fact, I did address this, in the earlier post that I linked. To wit: the fact that one firm operates at a profit margin of 1.7 percent doesn’t mean that all firms should operate at a profit of 1.7 percent. Amazon has an effective profit margin of 0 percent, but that doesn’t mean that every other firm in the country should follow suit. For that matter, why not compare Costco’s returns with Apple’s 22 percent?

You know the answer, right? Running a tech company is much riskier than running a grocery store. You have to spend a lot of money developing products that might not pan out. And then some moron can invent a product that eliminates the need for what you sell, as happened to Kodak and Polaroid and about a zillion other technology firms. The survivors tend to have high returns, but a whole lot of money got burned on the way to producing those few winning companies. Grocery store margins, on the other hand, are razor thin, but it’s unlikely that anyone is going to come along and invent something to replace what you sell.

An industry might need higher margins for all sorts of reasons: Revenue might be seasonal, or cyclical, or the industry might rely heavily on leverage, or long-term leases, so that you need some padding to cover your debt payments. This is why the returns for an industry group tend to cluster together, even in competitive industries: Groceries have lower margins than department stores, which have lower margins than insurance, which has lower margins than pharma.

If you were a prospective shareholder in one of those firms, or industries, you wouldn’t just look at the margin on sales; you’d want to look at how risky that margin was. You’d also want to know how much capital was needed to produce that margin, a statistic that is known, appropriately enough, as “return on invested capital.”

And if you were a shareholder who had already invested your money in one of these firms, you’d scream bloody murder if your high-margin tech company suddenly decided to cut profit margins to grocery store levels. There goes the retirement fund!

Wal-Mart’s profit margin is about twice what Costco’s is. But its ROIC is only marginally higher: 13.77 percent vs. 12.88 percent. The company needs to put a lot more money into warehouses, trucks, whizzy computer systems, and cinderblock stores in order to generate those profits -- not surprising given the complexity of its supply chain, and the number of products it offers. Owners of capital generally ask to be compensated for using it to build stuff, rather than spending it. Wal-Mart is no exception. If it targeted Costco’s ROIC, rather than its own, Wal-Mart could free up a bit of money -- about a billion dollars. If it gave two-thirds of that billion to its 1.4 million U.S. workers, each worker would get about $470, or $9 a week.

Of course, that leaves aside the issue of whether this is the “right” ROIC for Wal-Mart, or Costco, a question that has only metaphysical answers. I’d guess, however, that it’s considerably less than most of the people making these sort of comparisons would be expecting.

If Wal-Mart’s business model is so bad, then why shouldn’t it just do what Costco does -- put the stuff on pallets and raise labor productivity? I thought that was the very point I’d addressed, but here goes: A store with 100,000 products in its aisles cannot just put everything on pallets, because the stores would have to be 10 or 20 times bigger than they are now. Remember what I said about the ways that Costco’s small number of SKUs drive their business strategy? This is one of them. Having a small number of SKUs means that you can fit them inside your giant warehouse store on pallets, while still making them easily accessible to buyers. If you have 20 different sizes of aspirin, each requiring its own pallet-width display (and no stacking!), then the store would stretch for nearly a mile, and anyone over 50 would need a golf cart to get from beauty supplies to housewares. Hope you didn’t forget anything, because backtracking’s a literal pain in the butt.

These sorts of queries are distressingly common, and they never make any sense. Say Wal-Mart decides to move to a high-value, high-throughput labor model like Costco. It moves away from stocking, and puts the stuff on pallets. That means that it has to reduce its number of goods in order to fit in a store. That means doing away with comprehensiveness and variety: one brand of each thing. The means you probably need to put a lot of time and effort into picking that one thing, making sure it’s got the broadest possible market appeal. Okay, so now Wal-Mart is Costco … except that there’s already a Costco, and it has had decades of experience being Costco. Which is why Sam’s Club underperforms Costco so drastically.

Meanwhile, all the folks in smaller towns who depend on Wal-Mart for a wide array of goods are seriously mad that they can no longer buy a tennis racket, and not much mollified when you tell them to look at their fantastic savings on 10-gallon buckets of mayonnaise.

It’s as absurd to suggest that all stores should turn themselves into Costco as it is to suggest that all cheese should be made by Kraft or all restaurants should be Olive Garden, just because these companies are extremely successful. The existence of one Kraft Foods does not prove that there is room in the market for six more. Moreover, even if there were, a world where each broad category supported exactly one business model would be depressing indeed. What if it wasn’t Costco that won? What if every retailer in the country had to be a Wal-Mart?

But let’s imagine that we somehow, against all odds, managed to transform Wal-Mart into some sort of Costco clone -- maybe not in every detail, but in the one that most of these commentators care about, which is to say, a retailer with higher labor productivity and higher wages. How much better off would Wal-Mart's workers be?

Well, for starters, about half of them would probably have been fired. Remember, this is what a high-productivity model means: You need fewer workers. So half the workers will be making more, but the other half will be making whatever their state offers in the way of unemployment benefits.

Of course, that assumes that Wal-Mart wants to keep the other half. With a higher wage, it might be able to get better workers. So many of the rest of those Wal-Mart workers might eventually join their former colleagues on the unemployment line. Over the long term, one hopes, they’ll get other jobs. But remember that they’ve accepted these Wal-Mart jobs voluntarily; no one forced them into the store at gunpoint. Which means that their employment alternatives are probably worse than working for Wal-Mart.

Over the very long term, a very-high-productivity retail sector implies a more productive economy, which we hope has more highly paid jobs for unskilled workers. But that long term can be very long. While progress makes humanity better off in the long run, that doesn’t preclude the possibility that some people will be permanently worse off for the rest of their lifespan.

All of which is to reiterate the point of the original post: Companies are complicated. Markets are complicated. You can’t just change the bits you don’t like and have the rest tick along exactly as before.