In my podcast, The Answer is Transaction Costs, I have been working on the problem of transactions and what it means to change aspects of purchase or sale without otherwise changing the product. One of the great insights of economist Yoram Barzel was that it makes no sense to talk about commodities in the abstract, because it is the precise features of accessing, delivering, and using the product that affect the cost of use.
In fact, though, the insight is older. I had always heard of John R. Commons, one of the founders of the “Institutionalist School” of economics, but I had not read him until recently. In his 1931 American Economic Review paper, Commons said this amazing thing:
Individual actions are really trans-actions instead of either individual behavior or the ‘exchange’ of commodities… The shift is a change in the ultimate unit of economic investigation. … The smallest unit of the classic economists was a commodity produced by labor. The smallest unit of the hedonic economists was the same or similar commodity enjoyed by ultimate consumers. … The outcome, in either case, was the materialistic metaphor of an automatic equilibrium, analogous to the waves of the ocean, but personified as ‘seeking their level.’ But the smallest unit of the institutional economists is a unit of activity – a transaction, with its participants. (emphasis added)
Commons then goes on to point out that it is really the “trans-action costs” that determine the relevant prices that market participants are reacting to: “[agreements] must therefore be negotiated between the parties concerned before labor can produce, or consumers can consume, or commodities be physically exchanged”.
We need to take a step back, and think about a world without transaction costs. That’s hard to imagine; it’s something like a world without friction. We do a lot of things—streamline airplanes, and boats, use oil in engines, lubricate the hinges in squeaky doors—to reduce friction, of course, but if there were no friction in the first place, what then? You couldn’t stand up, brakes wouldn’t work, clutches would just slip.
In fact, you would be able to travel from Great Barrington to San Francisco on just a small push. It would take a long time, but you could do it. Of course, you couldn’t stop until you hit something. Friction is a property of the universe. Most of the things we do, and the way we do them, assume that friction is substantial.
Well, so is transaction cost, which is the “friction” of market processes. A while back I wrote a short piece on a “Boss who wore bunny slippers,” thinking that he could just make all the transactions that once justified a firm through spot contracts. But transaction costs ate them up, and the dumb boss who didn’t understand transaction costs got fired.
Remember, transaction costs are triangulation, transfer, and trust. If you sell something, you receive only the transferred payment, and then subtract transaction costs to see if you came out ahead. If it turns out that the costs of discovery, negotiation, delivery, and handling returns and shoplifting exceed your revenues, you stop selling. If you can find a way to reduce your transaction costs, you have a better chance of staying in business.
But for the buyer, it’s all transaction costs. I either buy the thing and pay the costs of the transaction, or I don’t and pay nothing.
An Example
All this jargon can be clarified with a simple example:
Suppose a widget has a sticker on it that says “$20”.
Suppose, further, that it costs the buyer $5 to shop, $3 to travel to the store, $1 to wait in line and pay, $3 to travel back home, and $4 in “risk” to trust that the product will actually work (1 in 5 widgets fail, and there is no warrantee).
So, what is the cost to the buyer? $31! $20 for the widget, $6 for roundtrip travel, $4 for risk of failure, and $1 for standing in line.
If the consumer does not go to buy the widget, he pays nothing. If he does, it costs $31, and a reduction in any of those aspects of the cost of the transaction will increase quantity demanded, because demand curves slope downward. If you cut the sticker price, you sell more. If you make the transaction faster, less burdensome in terms of travel or delivery, or just less inconvenient, you also sell more. To the consumer, all the costs of the transaction are considered together, as a package “buy or don’t buy.”
But what about the seller? The seller receives the $20 transfer, that’s it. The buyer pays $31, and the seller receives only $20, minus the transaction costs of paying for a cashier to take the money and a security guard to make sure people don’t steal.
Now, assume the seller offers “free” shipping. Let’s say it costs the seller $4 to deliver the widget. That saves the consumer $7 (he still has to shop, but online, but now has no travel cost to get to the store and no longer has to stand in line).
The question is what price to charge. If the seller charges $24 and offers “free shipping,” is the buyer being duped? After all, the seller is just adding the costs of shipping into the price.
The answer is “no,” the buyer is not being duped, for two reasons. First, since all costs are transaction costs to the buyer, online shopping with free delivery is a net cost reduction of $3 (he saves $7 by not traveling or queuing, but pays $4 for delivery). Before, the consumer was paying costs that were not being received by the seller. This transaction costs “wedge,” the difference between the total costs the buyer pays and the seller receives, has been a main focus of Pareto-improving entrepreneurial innovation in the past two decades.
The point is that to the buyer all costs are transaction costs, including the purchase price. You can raise the purchase price, if you cut the transaction costs by even more, and the consumer will be happy. Because to the consumer, all costs are transaction costs. By turning the deadweight losses of queuing and inconvenience into monetizable costs that can be charged, both the buyer and the seller are better off. Thinking in terms of transaction costs is central to understanding how real markets work.
