A Friction-free Future? Transaction Costs and Internet Markets
Bill Gates says:
"The information highway will extend the electronic marketplace
and make it the ultimate go-between, the universal middleman. Servers
worldwide will accept bids, resolve offers, control authentication and security
and handle all other aspects of the marketplace, including the transfer
of funds. This will carry us into a new world of low-friction, low-overhead
capitalism, in which market information will be plentiful and transaction
costs low. It will be a shopper's heaven." (The Road Ahead)
On the other hand, cybermaven Esther Dyson sees some
friction left in virtual space:
" But all the world's not a market. We'll be seeing friends in cyberspace,
shopping for things by
browsing around, asking for peers' advice, watching for what others buy
and how much they pay. There's more to shopping than efficient
And Mitch Kapor, erstwhile competitor of Gates, warns that 'friction-free'
is a stalking horse for:
"Control of key interfaces, for instance those of an operating system
such as Windows, creat[ing] so much leverage that a company can insulate
itself from normal market pressures, hardly an example of perfect
What's going on here? Since when did notions like 'friction' and 'transaction'
take on the supposed power to rearchitect markets and society? Are these
people using code or reciting some mantra?
In a way they are. The word 'transaction' itself is a likely source of
confusion - most folks may think of credit cards or ATMs, or maybe database
operations. The allusion in each of these passages is instead to a specific
economic concept, transaction costs. 'Friction' is an attempt at
an everyday metaphor for the same thing. My intent here is to unpack the
notion of transaction cost, and why some bright people see explosive
when it's combined with the Internet.
Transaction cost economics was originated by Ronald Coase, for which he
won a Nobel
Prize. 'Transaction' here is meant in a very broad sense. In addition
to credit cards and ATMs, it includes just about any exchange of value which
one can write into a contract or accomplish with a handshake (virtual or
real). Coase's contribution was to focus on the overhead inherent in finding,
negotiating, and governing exchanges of value. He reasoned that when the
overhead - the transaction cost - was low, that markets would form and find
efficient prices, but that high transaction costs would be a predictor of
market failure or value
exchanges that occur by non-market means.
A notion central to transaction cost is asset specificity. Specificity
relates to the scope of value of a good or service. For instance, a house
is location or site specific. So is cogeneration steam from an industrial
plant, or a T-1 link between two offices. An original artwork can be highly
specific in its value. A cargo of fresh fish, or an lot full of Christmas
trees, are highly time specific. From the technical realm, a program written
in assembly language to take advantage of a graphics accelerator is specific
to that hardware, but it would be less specific if written in high level
language to an abstract software interface.
The value of a specific asset is highly contingent or ambiguous. Advertising
or otherwise finding the buyer for which the value is maximized, and then
negotiating a fair exchange, is a relatively expensive proposition: the
transaction cost is high. Less specific assets are easier to take to market;
there are more potential buyers and they need less information to reach
a fair deal. Consider the difference between a part for a 70 year old antique
car on one hand, and a gallon of 87-octane gas on the other. Much of the
value of standardization (like octane ratings) is creating assets which
are less specific and more understandable. The transaction cost of matching
up a standard product up with a standards user is relatively lower.
Once the parties understand there's likely a useful exchange to be had,
then the terms have to be spelled out and a settlement made. And that's
part of transaction cost. Sometimes this is extemely simple: The ice cream
vendor hands me a chocolate bar, I hand over a dollar, and walk away never
to see him again. It can also go the other way: Hammering out a multiple
year contract with intermediate milestones, late delivery contingencies,
and specification change procedures can take months and hundreds of thousands
of dollars. That level of transaction cost may work for office buildings
or very large techical systems, but it would kill more mundane deals.
And the costs don't end when the handshake is made. Real people (and
have a characteristic called 'opportunism.' That means they cheat. Or,
more charitably, they interpret unanticipated events to their own favor.
Consider the 'dispute resolution' procedure built into using a credit card:
If you think the merchant cheated you, you can get the bank to withhold
payment until the matter is worked out. Having that contingency available
isn't free, it's built into the overhead of using a credit card; it's a
The longer the duration of the deal, and the more specific or contingent
the values being exchanged, the better the chance for something to go wrong,
or for parties to be opportunistic if a contingency wasn't spelled out.
Commercial contracts may have elaborate price adjustment and arbitration
clauses to maintain the relationship in spite of outside changes, and even
then may end up in court - a very high transaction cost. It should be obvious
that having a trustworthy trading partner, and a general
level of trust in the market and society, is worth real money in the
form of reduced overhead when creating and governing a deal.
OK, if you've lasted this far, let's reinterpret the statements by famous
"The Internet is great because it removes site specificity for many
transactions. All the goods of the world, at your fingertips. Forget driving
around to obscure bookstores, just order that odd-ball book from amazon.com
over the Web. Sure you may not know the merchant on the other end of the
wire, but you'll still be able to trust them. They have to price publicly
in an cybermarket you can browse, and they're running standard security
That's the key phrase, of course, standard software. Because this
friction-free market scheme only works if there's a standard way for all
the shoppers and all the merchants to meet and transact over the net.
the specificity - the problem in matching up buyer and seller - reappears
through the rear door and the market fails again. Standards make markets,
and market makers make money. In software, we call a standard a platform.
Guess who owns the platform? It will be heaven for shoppers, heaven for
Windows and NT, and heaven for Microsoft."
"I don't dispute that we need standards, Bill, but you've missing the
point with your bidding and authentication gabble. Geek toys don't build
trust, people do, and markets don't cause social cohesion, it's the other
way around. If you just think of the Internet as a big distribution chain
driven by better credit card standards, you're missing the point. Unless
the people can interact on the Internet in ways that promote trust and
they won't deal at a distance, and transaction costs will kill the market,
standards or no."
"Wake up and pay attention to what's going on here. All those Microsoft
standards may define a market, but since when did we let a market maker
trade in it as well? And isn't that what's happening when Microsoft not
only sells servers but content, and both operating systems and applications?
All the benefit derived from reducing specificity through standards may
go out the window if you let Windows win everywhere, unless you trust Mr.
Gates not to be a bit opportunistic somewhere down the road. And what
recourse will you have then, I might ask?"
Next Time: How to Make Platform War or Can a Guy Worth $20 Billion
be a Public Benefactor?
What strikes me as important about WebTV is that it is
a proof of concept for the notion that the
"client-side" can itself be implemented using a
client-server model and extremely inexpensive clients.
Or, another way of saying it, WebTV exploits the
three-tier model of distributed app deployment.
Microsoft (and others) have been promoting this model for
quite some time.
Join the conversation!
Most Active Topics:
Topic 30 The Global Brain
Topic 29 The Information Revolution - Paradigm Shift or Paranoid Shaft?
Topic 18 Knowledge Workers: Leading, Keeping
All Wealth of Networks Topics