Re: [wg-c] non/for profit

[Jonathan Weinberg <weinberg@mail.msen.com>]

On market power and market share . . . 

	(An aside: though the reader might not know it from some of the language
in the message quoted below, I support the proposition that ICANN should
expand the number of gTLDs to the limit of the technically feasible and
operationally stable.  Further, I believe that one of the advantages to
having lots of gTLDs is the opportunity for market competition between
registries.)  The question on the table is whether, during a transitional
period in which we have introduced, say, 6-10 new gTLDs (some specialized,
some not), we will have sufficient competition to constrain initial
registration prices to competitive (that is, marginal cost) levels.
Lock-in is a separate matter.  I think there are useful points being made
here, and if we can make them without biting each others' heads off, that
would be all the better.

	I think the market power and market share analysis is slightly more
complicated than has been presented.  (Caveat: I'm not an economist, nor do
I play one on TV.  But I've hung out with economists, and I do the best I
can.)  Each gTLD is a monopoly, in the sense that the .xyz registry is the
only entity that can sell registry services in that TLD.  What provides the
competition is the fact that different TLDs can be substitutes, or
near-substitutes, for each other. 
The extent of a TLD registry's market power will depend on the extent to
which satisfactory substitutes are available.  That is, .biz seems like a
great substitute for .corp, but you may find it less desirable as a
substitute, if you're in the newspaper business, for .news.  In the
current, constrained, name space, there's room for registries aiming at
niche markets (say, .md) to exercise market power within those niche
markets — which consist of the people for whom other available TLD strings
are significantly less good.  The expected effect of their exercise of
market power is that they charge higher prices, generating fewer
registrations, but making more money.  In an unconstrained name space, that
wouldn't be so, b/c even niche markets could see their choice of TLDs.

	When it comes to general-purpose TLDs, it seems to me correct that we
don't have to worry about prices in the short term, no matter what the
registry's incentives. Given NSI's regulated prices, and assuming the
existence of 2 or 3 other registries operating general-purpose TLDs on the
CORE/Nominet model, then there should be ample market constraints on any
new for-profit registries.  I'm not sure that we'll see the same
constraints on specialized gTLDs.  This doesn't automatically mean that we
need to impose rules or structures designed to avoid situations in which
the operator of a specialized TLD seeks to charge inefficient monopoly
prices.  Milton, at least, would argue that any such provisions would
constitute a cure worse than the disease, and he may be right.  I'm
interested in what others have to say.

	On another matter: I think Kent and Eric make a good point that gTLD
registries will have an economic incentive to consolidate.  That's a
problem for those of us who believe that competition among registries is
good, since (as Milton pointed out a while back) the salient issue in
achieving competition is the number of registries, not the number of gTLD
strings.  Rather than responding to this by banning for-profit registries,
I think the best available response is a regulatory one, in the nature of
antitrust, limiting the number of gTLDs a single entity can control.
That's a complicated rule to draft, as Kent's pointed out, but it's doable. 

Jon


Jonathan Weinberg
weinberg@msen.com



At 06:19 PM 11/16/99 -0500, Milton Mueller wrote:
>
>
>Jonathan Weinberg wrote:
>
>>         This is an interesting question; the larger issue is how much
competition we
>> have to have before the market will operate properly.  I'm not sure I
know the
>> answer.  There's some evidence, though, that NSI and the ccTLDs today
may not be
>> providing significant market constraint: Milton's noted in another
context that Matt
>> is seeking to sell .md domains for $299/year.  So
>
>This argument is utterly fallacious.
>
>Pray, what is the market share of .MD registrations? How many are being
sold at the
>$299 price? Of these, how many are speculative, based entirely on
anticipation of
>continued artificial scarcity? .MD's share of worldwide registrations is
less than one
>percent, if it is even measurable. In other words, there are severe
constraints on
>price -- .MD does not have anything near the amount of registrations it
could have if
>it were charging closer to the true market equilibrium.
>
>I can offer to sell my home for a million dollars. In fact, consider it
done -- anyone
>want to pay me a million dollars for a 4-bedroom, 10-year old home in
Syracuse? I love
>my home, but I'd sell at that price. That doesn't make it the market
price, and it
>doesn't prove a damn thing about the level of competition in the real
estate market
>here.
>
>And if, given many cheaper options, some people value .MD registrations
enough to pay
>$299 for them, who are we to say this should be prohibited or regulated?
If someone
>takes up the offer to buy my home for a million dollars, would you send in
the
>regulators to stop them, Jon?
>
>> at least some people think the current marketplace leaves room for
>> outrageously high, non-cost-based prices.
>
>Outrageous to who? No one has to pay that price.
>
>> It doesn't strike me as wholly
>> implausible that a registry annointed as one of the first "real,"
>> "official" new commercial gTLDs might be able to exploit pent-up demand
— and the
>> fact that its SLDs, in contrast to those in .com, are actually available
— to
>> extract significant rents.
>
>This is a more plausible argument. But such problems are inherent in the
policy of
>artificial scarcity, which the 6-10 limitation on the initial new gTLDs
perpetuates.
>So thank you for another strong argument for an initial declaration that a
large
>number should be added over a fixed period of time. It's not too late to
add your name
>to Position Paper B.
>
>>         [2] If there is a problem, is the non-profit/for-profit
distinction the
>> right way to approach it?
>>
>>         Milton urges that, in any event, non-profits are as likely to
set prices at
>> profit-maximizing levels as for-profits are.
>
>No, I did not. I argued that non-profits >might< engage in opportunistic
behavior, and
>that some demonstrably >do< act to maximize their revenues. Whether they are
>statistically *just as likely* to do so as for-profits is unknown, and is not
>relevant. The simple fact is that we have no idea which non-profits will
behave in
>which manner. Therefore, whatever policy is applied to new registries cannot
>discriminate between profits and non-profits. Or, if it attempts to so
discriminate,
>ICANN will have to engage in highly intrusive regulation of the
distribution of
>surplus revenue. This cure strikes me as worse than the disease, given the
existing
>options available to consumers.
>
>Indeed, your own argument below ends up conceding the irrelevance of the
distinction.
>You ask whether a ceiling price is necessary regardless of whether we are
talking
>about profit or non-profit registries. I view this as another step forward
in the
>consensus-building process. It seems that no one is now prepared to base
policy upon
>that distinction.
>
>> it seems possible to construct particular non-profit organizations in
forms that
>> will make them likely to set prices at cost- recovery levels.  As Kent
has pointed
>> out, setting up a registry as a cooperative controlled by its customers
is a pretty
>> effective way.  (The customer-owned cooperative, in the real world, is
often poorly
>> managed, and may have other flaws.  But it tends to set its prices at
cost-recovery
>> levels.)
>
>Once again, you've overlooked the most salient part of my argument. Yes,
>profit-maximizing registrars may have an effective incentive to impose
cost-based
>pricing on a registry. If that is true, the creation of two or three of these
>registries is sufficient to give customers another competitively-priced
option in the
>marketplace.  Such an option acts as a real constraint on the other
registries using
>other models.
>
>If the advocates of this model are sincere, they should be content to
accept the
>existence of two or three of them. that, and the regulation of NSI on
cost-plus
>models, more than compensates for any concerns about market power.
>
>I am just a bit more modest than others about making assumptions about the
optimality
>of a specific model. The cooperatively owned rgeistry may work fine. OTOH,
we may very
>well find out that in a deregulated market for-profit registries charge
less than
>non-profits, perhaps because they bundle the service with other, more
profitable
>services. There are lots of models, some we haven't even thought of yet. DNS
>registration is a tiny part of the overall Internet services picture. I
think it is
>idiotic for ICANN to be in the business of imposing business models on
these services.
>
>>         Nonetheless — as more than one person has pointed out — it may be
>> distracting and unhelpful to focus on the blanket "for-profit" and
"non-profit"
>> labels.  As Roeland notes, the mere fact that an entity is classed as
non-profit
>> does not magically endow it with "enlightened benevolence."  Rather, we
need to
>> consider more specific structures or rules that will lead registries,
>> notwithstanding market power, to set prices at or near competitive levels.
>
>You don't need "specific structures and rules." You need to let
competition happen.
>Right now, ICANN, the DoC, and this working group are the biggest obstacle
to setting
>prices "at or near competitive levels," because we are actively preventing
the entry
>of new competitors. Competition in monopoly industries is always driven
primarily by
>entry, not by regulation. At best, regulation eliminates certain barriers
to entry.
>But the regulatory mindset that bars hundreds of potential entrants in
order to
>satisfy some purely theoretical notions of how structure and conduct are
related is
>truly perverse.
>
>> This gets us back to the question raised before: To the extent that we
think
>> registry market power may be a problem (for either for-profit or non-profit
>> registries), b/c the market for registry services is not yet
competitive, what
>> structures -- or provisions in the registry contracts -- would be
appropriate to
>> ensure that the registries don't inappropriately exploit that market
power?  A few
>> days ago, I suggested the "ceiling price."  Does anybody else have other
>> suggestions?
>
>I have made a suggestion. There is no need for "specific structures and
rules" to
>regulate rates.
>
>You have attempted to answer that by arguing that the existence of
above-market prices
>in an obscure corner of the DNS universe shows that market constraints are
not
>effective. This was refuted by showing that the entity charging those
prices has no
>appreciable share of the market. I assume that you are familiar enough
with the
>literature on market power to understand that an entity with no market
share cannot
>have market power. It was further refuted by showing that even within the
artificial
>constraints of the 6-10 consensus, the creation of 2-3 shared cooperative
registries
>would, in addition to the regulation of NSI rates, provide the bulk of
customers with
>an option that, if your reasoning is correct, would protect them from
market power.
>
>--MM