Bryan Caplan  

What I Learned at the Tower Records Going Out of Business Sale

DeLong, Dasgupta, and Stern... Can 600 Economists Be Wrong Ab...

Big finding as the Titanic goes down: The greater the demand for a CD at its regular price, the greater the elasticity of demand for that CD. When everything is 50% off, CDs that normally sell 1 per week may sell 3 per week. But CDs that normally sell 100 per week may sell 1000 per week.

This rule seems to work in every genre. In classical, for example, almost every Wagner disk was gone, but there was still an ample selection of Bax. In punk, every Bad Religion and Dead Kennedys disc was gone, but there were lots of obscure compilations left.

Non-economists probably won't find this surprising, but it is. If one CD is 10 times as popular as another, you'd expect stores to stock 10 times as many copies of it. And if stores are already doing that, the fact that the popular disk still sells out more quickly is not trivial.

Does this law generalize to other markets? Is it a near-universal rule that products in higher demand at their regular price also have higher demand elasticity? Why or why not?

Comments and Sharing

COMMENTS (12 to date)
Bob Dobalina writes:

If one CD is 10 times as popular as another, you'd expect stores to stock 10 times as many copies of it.

Why don't you walk down the hall and ask your pals on the Operations faculty if that's indeed true. Because I think you have to assume a lot of stuff away to make it so.

Brandon Berg writes:

If one CD is 10 times as popular as another, you'd expect stores to stock 10 times as many copies of it.

I suspect that the explanation is this premise is wrong. I've never worked in retail, so I don't know for sure, but I suspect that infrequent sellers are overrepresented on the shelves.

Suppose a store receives shipments once per week. Now consider first a CD that avearges two copies sold per week. Maybe the store will order enough to bring their stock up to four after each delivery: the two that they expect to sell, plus a safety buffer, since two per week is just an average. So they average three copies in stock.

Now consider a CD that averages one sale per month. They'll probably keep just keep one copy, and then order another one whenever someone buys it. They average about 0.88 copies in stock (since it's out of stock for half a week per month).

So even though the first CD is eight times more popular, there are on average only 3.4 times more copies in stock.

chris writes:

What are the microfoundations here? Intuitively, I'd suppose the opposite were true. A big cut in the price of CDs would cause me to buy more CDs from artists I'd never heard of, in the hope of finding some great new music. But I wouldn't buy anything by J-Lo or Christina Aguilera, as I know and don't like their music.
This suggests price elasticity is inversely related to sales volume, for me. So why do others behave in the opposite way?

Brad Hutchings writes:

At the extreme, where discount = 100%, one could test this by looking at original Napster (or gnutella, ec.) data. I wonder what happens to the long tail curve as price goes to free. I'd bet the front gets taller and steeper, the tail lower and flatter as price goes to free. Which would suggest, contrary to popular wisdom,
that the Internet does more to unify us culturally than to break us apart. I.e. in the old world of content, we had several fairly strong "hits", but not a lot on the edges. In the new world of content, we have fewer but stronger hits, with a lot on the edges that don't get much attention. Do I win a Nobel Prize?

John Thacker writes:

The greater the demand for a CD at its regular price, the greater the elasticity of demand for that CD.

Well, consider that you can often find hit CDs for much cheaper at Best Buy, Target, Wal-Mart, etc. than you can the less popular CDs. So the assumption of a greater elasticity of demand for hits seems warranted.

Brandon Berg's point still seems quite relevant to me. It's not clear that your assumption of "If one CD is 10 times as popular as another, you'd expect stores to stock 10 times as many copies of it," should actually hold. Rare CDs may be ordered less often from the supplier than popular CDs, so stores don't need to stock as much of the popular CDs.

To be effective, a retail music store like Tower needs to have one copy of a bunch of different obscure discs in stock at all times. They may only rarely sell that disc, so they don't re-order it often. But they need it in stock so that customers can count on being able to walk into the shop and find what they're looking for. Otherwise customers will realize that special requests are best done on the Internet.

However, all that shelf space costs money; it's cheaper for an Internet retailer to consolidate rarely bought CDs in a fewer number of warehouses and ship them out. That's essentially why Tower (and every other specialty music/video retailer) is going out of business or declaring bankruptcy-- the big box stores stock mostly hits in large amounts and keep them cheap, taking advantage of elasticity. The Internet stores hit the specialty retailers on the other end, by aggregating supply for rare discs more effectively.

Bob Knaus writes:

First law of retailing: Variety Creates Demand.

I had full price control over my father's vegetable stand at the age of 15, and I knew it then. The more choices you give people, the more money they will spend. The function of "oddball" items, whether CDs or vegetables, is to create demand for the "mainstream" items. Any oddballs you sell are gravy.

Dunno why it works this way, but it's a basic element of human nature.

Mike writes:

I would bet that Tower tried (or should have tried) to run something close to just-in-time intentory; getting popular cds in medium sized batches as needed. The producers would also be shipping in those quantities as cds were pressed I imagine, keeping just a small buffer stock (with the exception of a new release). The low selling cds still came in similar quantities, just less often. Therefore Tower probably doesn't have a large backstock on top sellers but would have a relatively high quantity of low and medium sellers left. Plans the company has to continue an online store are also going to impact what is sold off cheap in their retail stores and what might be held back.

I think it would be interesting to contrast bookstores to music stores. There are similarities in the markets. Both stores seem to have a relatively narrow band of prices, and the suppliers in both (record labels and book publishers) share similarities with respect to their sources of inputs and the type of value added. There is a very high markup on books at the retail level, higher than cds (even the ones at the old, full price, Tower). Low selling book titles are sent back to the publisher in something like a consignment arrangement. Clearly bookstores have a better deal with their suppliers, why is that?


Dain writes:

I spent the last month working at Tower Records Emeryville (near Berkeley) as it was nearing emptying its inventory.

I took note of the rather obscure CDs that I wanted to pick up before I left. I knew that on the one hand their obscurity would more or less secure them a spot as I waited for the CDs to be price reduced even more. On the other hand, on the off chance someone else spotted a CD I wanted, that'd be it, as the store would not order another one.

And another somewhat off topic remark: the quality of the labor force was very poor. (There was actually a frenzy of hiring due to the increased business in the final days.) Says something about the quality of character and time horizons. More economic wisdom confirmed.

John F. Opie writes:

Hi -

Oddly enough, I have worked in retail, many many years ago.

The most popular items, i.e. those that sold 10x as much as any other similiar item, were stocked in more, often 15-20x more, as the items in question were seasonable and there was a single delivery time period. After that, you were either out of luck or you scrambled to find alternative stock.

However, the suppliers also knew that it was a hot seller, so virtually no discounts were made available and in many cases (this was back in a time period of price controls!) the margins for the retailer were reduced.

As the less popular items had significantly higher margins, we were told more often than not to keep only a minimum of the popular items on the shelves and to push the less popular.

Just my anecdotal evidence. :-)

Oh, and there was at least three cases were my boss made significant profits by overstocking (ordering three times his anticipated volume) and then basically auctioning off his overstock at the peak of the season to those who had ordered poorly. He was a pyschologist and understood his fellows fairly well. Rotten bastard to work for, though. :-)


sourcreamus writes:

Could it be that the sales of the more popular items were coming at the expense of other stores. If you want a really popular item you could get it at a discounter, a record store could be reserved for more obscure items since their variety is likely to be greater. If the popular record is on sale, people would buy it at Tower rather than waiting to buy it at Walmart .

Greg D writes:

I think sourcreamus is right. The sale draws in lots of people who wouldn't have otherwise come. And the people it draws in are people who are less likely to be looking for "specialty" items, and more likely to be looking for something that they otherwise would have bought at a discount store.

thus the more "popular" things disappear at a higher rate than normal, because the custumer distribution has changed.

Deiah Haddock writes:

Obviously the reason a CD will sell more at 50% off is simply because the oppurtunity of buying ONE CD is high. For the same price of one CD, one can buy the equalivent on itunes and instead pick and choose the track that he or she desires from more than one artist. On a CD we are forced to pay for all the tracks, even the ones we don't enjoy!!

Comments for this entry have been closed
Return to top