Loading document…
Opening in Pages for Mac...
Your browser isn’t fully supported.
For the best Pages for iCloud experience, use a supported browser.
Learn More
Cancel
Continue
Beware Business Fads
:
Disruptive Innovations and Competition Policy
Joshua Gans
1
University of Toronto
December 2015
Business has its fads. Most of the time these involve changes in management practices — re-
engineering the corporation, total quality management and the like. But occasionally they move
beyond fi rm boundaries. A couple of decades ago, it was all about outsourcing and supply chain
management. But since the late 1990s, and especially alongside the evolution of the
commercial internet, the fad de jour has been
disruption
.
Disruption is seemingly everywhere and happening to everything. And like all such things it is
coin with two sides. On one side is a notion that would warm a competition regulator’s heart:
disruptive entrepreneurs are all about unleashing new technologies to bring down sleepy
incumbents. On the other side, are the incumbents themselves. Disruption holds that they have
never been more vulnerable; just a few ounces of complacency away from doom.
Combine these two sides and competition regulators and the laws that house them are
themselves argued to be disrupted. When entry is free, often and strong and incumbents are
petrifi ed, those promoting competition through regulation can go home. There is simply no more
than they can do. At least, that is what the direct implication of the disruption notion would imply.
However, like all fads, while there is a grain of truth at their heart, I will argue here that,
fundamentally, little has changed. Indeed, once we unpack what we know about disruption and
combine it with some hard-headed economics, we see that the role for competition regulators is
as strong as ever and, surprisingly, how they go about their business is as traditional as it has
ever been.
The plan for this paper is as follows. I outline the main theory of disruption as outlined by Clay
Christensen. I then relate it to what is traditionally known from economic theory before turning to
look at the evidence for disruption. I then explain how the evidence suggests that not much has
changed for competition authorities and they still have an important role. I conclude by
mentioning some issues in relation to the nascent sharing economy.
The original Christensen view
The disruption movement, if it can be called that, began with the work of Clay Christensen. The
Harvard Business School professor wrote a book entitled
The Innovator’s Dilemma
that tried to
1
Thanks to Thomas Ross and participants at the Canadian Bar Association Workshop in September 2015. All views
remain my own. Contact:
joshua.gans@gmail.com
.
2
look at why great fi rms failed.
2
Of course, it was well known that fi rms that had been great could
see their glory days over and then succumb to competitive pressures. But Christensen argued
that instead what could cause great fi rms to fail was when they followed precisely the principles
of good management that MBA professors would argue they should follow. Take, for instance,
the notion that you should listen to your customers when deciding which new products to pursue
and launch. Christensen argued that this was a good idea if the new product innovations built on
or sustained what your customers’ valued. But what if this was not the case? When Blackberry’s
customers saw the iPhone in 2007, they did not push the company to produce a similar product.
Why? Because it had no keyboard and if there was one thing that Blackberry users loved, at
least at the time, was their keyboards. As it turned out, that never changed but, in the process,
RIM (Blackberry’s maker) failed to move quickly enough on the real opportunity from iPhones
and then Android smart phones — an operating system and ecosystem that generated large
numbers of apps.
3
In many ways the story of Blackberry is a poster child for the disruption movement — now worth
less than a tenth of its market value from its peak in 2010. But there were others including
Encyclopedia Britannica and Blockbuster video. What was important was the message: even
well managed companies were not safe. There existed innovations that were disruptive in that
they came at established fi rms seemingly out of nowhere. To be sure, RIM, Britannica and
Blockbuster all knew, investigated and considered the innovations that were later held to be
their demise. Britannica, for instance, was a leader in online digital encyclopedias as early as
1996 while Blockbuster launch video on demand over the internet as early as 2000. In each
case, these moves undermined their existing business models and were quickly discarded. In
the meantime, new entrants took the charge absent these internal conflicts. The end result was
history.
The notion that incumbent fi rms might face weaknesses in the face of innovative entrants had
been considered before Christensen. Joseph Schumpeter had famously identifi ed the waves of
creative destruction that drove capitalism although he was ultimately pessimistic and believed
that powerful large fi rms would end up stifling that process and along with it, innovation. What
Christensen brought to the table was a more nuanced approach.
4
Not all innovations would be
the death knell of incumbents — only ones that were disruptive. And those innovations had two
characteristics. First, they tended to make design trade-offs that offered lower performance on
key metrics incumbents and their customers valued. Thus, they appealed to niche or under-
served consumers initially. Incumbents chose to ignore those because they tended to be at a
lower end of the market.
But this was only the fi rst element of a disruptive innovation. The second was that the
innovations had a trajectory of improvement on precisely the metrics that mainstream customers
valued. According to Christensen, incumbent fi rms who sensibly ignored those innovations
when they fi rst appeared found themselves facing entrants with more competitive products after
2
Christensen, Clayton M. (1997),
The Innovator’s Dilemma
,
Harper Business: New York
.
3
The precise story is a little more complex but is something I outline in
The Disruption Dilemma
, MIT
Press: Cambridge (MA), 2016 (forthcoming).
4
H
e was, in fact, not the only one. Rebecca Henderson and Kim Clark posited a supply-side theory that
focused on innovations that would be hard for incumbents to develop and sustain. See Gans (2016) for
more details.
3
just a short time. In his book, he argued that, by the time all that happened, it was too late for
the incumbent fi rms. They would be disrupted.
To support this theory, Christensen offered up numerous cases, the most famous of which was
the hard disk drive industry. That industry had its origins in the 1960s with pioneering efforts by
IBM for its mainframe computers before smaller disk drives took off for mini-computers in the
1970s. Christensen’s study began when Control Data Corporation (CDC) was the largest
independent maker of 14 inch drives for mini-computers. However, throughout the 1970s and
1980s there were several successive step changes in hard disk architecture. In each case, the
physical size of the disks fell (to 8 inch, 5.25 inch, 3.5 inch, 2.5 inch, 1.8 inch etc) but at the cost
of capacity. Not surprisingly, for incumbents at each stage, when they explored a smaller drive
with their customers, those customers claimed they were not interested. However, as ultimate
consumers moved from mini-computers to personal computers and then to laptops, smaller size
had some obvious benefi ts. Christensen showed that, in most cases, newer sized disk drives
were brought to market by new entrants rather than existing incumbent market leaders.
Christensen then went further and argued that those market leaders themselves failed as a
result of this competitive pressure. However, while that certainly did happen on one occasion as
Seagate managed to win against CDC, as we will see, the pattern of creative destruction did not
appear as strong as Christensen had maintained. This is important because what made the
book “scary” to people such as Intel’s Andy Grove was the notion that incumbents were more
vulnerable than they thought. Even with good management, Christensen had argued that they
could be felled by disruptive innovation. Moreover, from an antitrust perspective, innovation
dynamics were spurring a high degree of competition in such industries and hence, antitrust
authorities, so it was argued, should leave those industries be. However, if, instead, it was the
case that all the innovation-based entry did not lead to a rapid change in market leadership of
incumbents, the role of antitrust regulators could be much more important.
What does economic theory say?
At its essence, disruption theory involves a simple set of relationships. First, a disruptive event
occurs — which is usually the emergence of an innovation or technology that is, for want of a
better term, a bad fi t with what incumbents in the industry are doing. Second, incumbents pass
on developing that new technology due to internal conflicts while entrants, who do not face such
conflicts, take the charge. Third, the entrants do so well that they end up being a competitive
threat to incumbents. Fourth, the incumbents fail to catch the entrants and so lose leadership
and, ultimately, much more. From the perspective of the role of competition theory, it is this last
stage where all of the action is. If it is true that entrants can outpace incumbents to the latter’s
doom, then competition is working well without a regulatory hand. If that is not true, then we can
not presume that competition — which might translate technological leadership into market
leadership — is working as it should.
It is worth noting that the fi rst three stages of disruption theory have a long-standing basis in
economics. Beginning with the work of Kenneth Arrow,
5
economists have long tried to
understand the differing incentives of incumbents and entrants to innovate. Arrow noted
something interesting in this regard. If a new entrant were to enter a market on the basis of a
5
Kenneth J. Arrow, “Economic Welfare and the Allocation of Resources for Invention,” in
The Rate and
Direction of Inventive Activity
, NBER: Cambridge (MA), pp.609-626.
4
new innovative product, what they would receive as a reward were the profi ts from that product.
By contrast, suppose an incumbent, already in the market, where to think of launching a new
product. What they would receive would be the profi ts from that product but they would also lose
profi ts from the previous generation of the product. Thus, what fundamentally distinguishes an
incumbent and an entrant in thinking about whether to put forward effort to generate new and
better products is a difference in their (net) rewards from that activity. For the exact same
product, the incumbent’s reward is lower than the entrant’s precisely because the product will
replace what the incumbent already is earning. This replacement effect suggests that, to some
extent, all new innovations do not ‘fi t’ with incumbents in the same way they do for new entrants.
Thus, in the absence of other frictions, we might see entrants being more likely to bring
innovations to market than incumbents.
This example presumes quite a bit of symmetry in the opportunities facing the incumbents and
entrants. In reality, if you want to improve a product rather than, say, launch a new product
completely, an incumbent has an advantage. This is because to get the new improvement to
market, the incumbent already has an existing product to work with. By contrast, an entrant, to
be an effective competitor, has to supply a product — improvements and all. Thus, we could
presume that for many innovations, the costs to the incumbent of developing that product are
lower than those of entrants.
Disruption theory emphasizes this type of thing by focussing on the notion that only certain
types of innovations will be disruptive — in short, of the type that entrants do not have a big cost
advantage relative to incumbents. To Christensen, coming up with an entirely new disk drive
with a distinct physical size was an example of this. Similarly, coming up with a touch-based
phone, a digital encyclopedia or streamed video would be something neither RIM, Britannica or
Blockbuster had a particular ‘technical’ advantage in. Therefore, a more nuanced approach to
whether competition is working as it should, would be to consider whether the industry under
study was more or less prone to the types of innovations that were disruptive as opposed to
being merely sustaining. Importantly, two industries could see high levels of innovation but may
be very different in terms of whether practices might be of concern to antitrust authorities.
Which brings us to step 4 — the fi nal step. Suppose that, in fact, a new disruptive innovation
had emerged and it was brought to market by an entrant who then found themselves able to
compete with incumbents. Would that be the end for the incumbents? The picture painted by
Christensen and the business leaders who subsequently carried the disruption movement was
that the answer was yes. Incumbents would be unable to catch the entrants and would
subsequently lose their market position. However, economic theory by contrast suggested that
was not inevitable and, indeed, that incumbents had tools and incentives to prevent such
outcomes.
Faced with existential threat, incumbents have two broad options. The fi rst is, in fact, to meet
the threat. One of features of disruption theory is that entrants enter but take some time to
improve new products to be competitive with incumbents and compete for their primary
customers. In other words, built into the theory is time. Moreover, this is not just simply a matter
of waiting it out. Entrants have to invest to make their products better but incumbents can
similarly divert resources to meet a competitive threat. No better example of this exists than
when, in the mid-1990s, having ignored web browsers in a manner consistent with disruption
theory, Microsoft realised the threat Netscape posed to its operating system dominance. Bill
Gates penned an 8 page email that announced a new division staffed with thousands of
5
engineers that would be devoted to catching up and subsequently outperforming Netscape. It
did just that becoming the dominant in browsers with Internet Explorer — at least for a time.
Netscape, by contrast, having led the way, fell out of the market entirely.
Why did Microsoft double up its investment in this way? Of course, one reason is that it could.
An advantage incumbents have is the ability to marshall resources. But another reason is that it
had a preservation incentive. While prior to something like Netscape coming along, Microsoft’s
incentives may have been muted due to the replacement effect, when such entry becomes
inevitable and, by doubling up, the incumbent cannot just meet but neutralize a competitive
threat, its incentives switch. In this situation, while Netscape’s reward from continual innovation
is a foothold in competition with Microsoft’s, Microsoft’s reward is to continue to hold on to its
monopoly rather than face permanent competition. In this case, the difference in profi ts between
monopoly and competition is greater than the profi ts in competition itself and so Microsoft’s
reward from innovation is relatively higher. The end result is that when a threat becomes
existential while an entrant’s entry is not permanent, incumbents have a stronger incentive to
devote more resources to innovation to preserve their market dominance.
Doubling up in this way is a highly competitive response but it is also costly to incumbent fi rms.
That leads to the second option they may have — to acquire the entrant. To an antitrust lawyer
that seems like a fairly obvious response. It is also obvious to an economist. After all, so long as
it is permitted, a merger that can diminish competitive pressure is in the interest of both the
incumbent and entrant concerned.
Why then was it not given weight by the disruption movement? For Christensen, he emphasized
that by the time the incumbent realised that acquiring the entrant was a good move, it would be
too late. The entrant would already have market leadership in its sights and so acquisition may
be too costly for the incumbent — indeed, it may be unaffordable. In addition, Christensen
argued that integration of the two fi rms would not get rid of the issues the incumbent had in
promoting and developing the new disruptive technology. He argued something similar in
relation to the incumbent’s ability to catch up by doubling up investment.
Thus, the question regarding the relevance of competition policy in the face of disruption can
move up one level. It hinges on the incumbent’s incentive and ability to respond to the entrant
but also on whether it can do so either by doubling up on investment in response or by
undertaking an acquisition. If these are too costly or too late, as Christensen suggests, then the
new entry will be successful in overturning incumbent leadership and competition authorities
can relax. On the other hand, if they are neither of these and are viable options to protect
incumbent leadership, competition authorities have a role to play. What role? That needs to be
discussed. However, before considering that let me turn to the evidence on whether disruption
really does leave incumbents flatfooted.
What does the evidence say?
While there are certainly examples where fi rms that seemed to have an unassailable market
position, fell from grace, the question is whether an industry can be prone to disruption over the
long-term so that we can relax about competitive forces operating in a socially benefi cial way.
As already noted, Christensen identifi ed that hard disk drive industry as an example of an
industry prone to disruption. For that reason, it has received much attention from economic
researchers over the last couple of decades. As I will highlight here, the picture painted is
6
somewhat different from that of Christensen although it is consistent with the incumbent
response to incumbent as highlighted by economic theory.
As a starting point, it bears repeating that, for the most part, Christensen was correct when he
showed that for large step-size changes in hard disk drive confi gurations, it was entrants rather
than incumbents who brought the new innovations to market fi rst.
6
A recent study by Mitsuru
Igami highlighted why.
7
He examined the move from 5.25 inch drives to 3.5 inch drives. The 3.5
inch drives would become the most popular drives ever for personal computers and laptops. But
actually this drive took some years to be introduced. For a few years, it was exclusively supplied
by new entrants before being successfully promoted by Conner Peripherals, another entrant,
who came to dominate the 3.5 inch segment in its early years. Igami examined why it was that
the market leader in 5.25 inch drives, Seagate, took so long to enter that segment. He found
that, consistent with disruption, Seagate were concerned about the replacement effect
explaining about two thirds of their delay. Interestingly, he also found that Seagate had a cost
advantage that translated into the new segment — something that could have accelerated its
entry but also gave it something more to protect in terms of existing margins. This serves to
reinforce the role of entrants in bringing new innovations to market.
That said, apart from one instance — the move from 8 inch to 5.25 inch drives — in general, the
incumbents ended up catching up by investing more heavily in the new designs when they
found themselves under competitive pressure. In other words, the displacement predicted by
step 4 of disruption theory did not come about. Instead, both doubling up on investment and
acquisition were demonstrable incumbent responses in this industry.
On doubling up, incumbents general caught up with entrants by investing more heavily than
them in new designs after they entered the market. Josh Lerner found that the late-comers to a
new segment (that is, the leaders in the previous segment) ended up being the market leaders
again after a short time.
8
Similarly, acquisition played an important role in the industry. Following its successful leadership
in the 3.5 inch segment, Conner Peripherals was acquired by Seagate in 1993 during the period
when Seagate had fi nally started to catch them in that segment. That process was part of an
ongoing consolidation that had seen Seagate purchase Control Data Corporation in 1989 (the
incumbent it did displace when it introduced the 5.25 inch drive) and over the next two decades,
acquisition was the main form of exit for new entrants in the industry. Later Seagate bought
Maxtor in 2006, Samsung
’
s drive division in 2011 and LaCie in 2012. Maxtor itself had been an
acquirer of competitors including MiniScribe in 1990 and Quantum in 2000. All told, Seagate
was responsible for the exits (directly or indirectly) of nine of its rivals by acquisition.
This is well-known in antitrust circles. It is only a few years ago that the industry went from 5 to 3
players in a short period of time due to the Seagate-Samsung and
Toshiba
-Hitachi set of
mergers. In those cases, antitrust authorities were concerned about the reduction in competition
6
The exception was the 2.5 inch drive.
7
Mitsuru Igami, “Estimating the Innovator’s Dilemma: Structural Analysis of Creative Destruction the Hard
Disk Drive Industry,”
Journal of Political Economy
, (forthcoming, 2015).
8
Josh Lerner,
“
An Empirical Exploration of a Technology Race,
”
The RAND Journal of Economics
, 28, no.
2(1997), 228.
7
but also on a potential reduction in R&D expenditures and so placed conditions on the mergers
to ensure those reductions did not take place.
But our understanding of this industry has now been aided by a 2015 study conducted by
Mitsuru Igami and Kosuke Uetake.
9
They took historical data from the industry to develop a
model to see if permitting those fi nal two mergers was a good idea or not. On the static side,
what they found is that compared to mergers in the past, these mergers had relatively large
effects. In particular, they likely led to a large reduction in consumer welfare while at the same
time also generating substantial realized effi ciencies. In the past, both of these effects had been
dampened by smaller scale. Nonetheless, even though the effects became large, they balanced
each other out.
What was more interesting was what the likely impact of a long term merger policy would have
been on the industry. For instance, suppose that antitrust authorities blocked mergers that
reduced the number of competitors below 5. If this had been the policy 15 years ago, it would
have reduced the rate of R&D because it would actually encourage some fi rms to exit the
industry. Specifi cally, fi rms that might otherwise have stayed in longer to fi nd a merger partner,
leave and with them goes any innovations they may have produced. The end result of this is
that while the R&D rate did not vary much when the industry moved from 5 to 3, had a 5
threshold been the policy, it would have slowed R&D earlier in the industry lifecycle.
How should competition authorities approach disruption?
What has been demonstrated is that disruption theory does not imply that competition
authorities can be relaxed and presume that industries will have a natural matching of
technological leadership with market leadership. In fact, the two can be divorced and, in some
cases, the instrument of that divorce can be practices that are often assessed by competition
authorities. That said, like all innovation, because there are dynamic issues associated with an
industry, the analysis of those practices requires care.
For example, w
hen we look at mergers we tend to consider them one case at a time. However,
when dynamics and innovation play a role, the case by case approach may not be
9
Mitsuru Igami and Kosuke Uetake, “Mergers, Innovation and Entry-Exit Dynamics: The Consolidation of
the Hard Disk Drive Industry,”
mimeo
., Yale, July 2 2015.
8
appropriate.
10
This is because the strength or tenor of the merger policy will have an impact not
just on the present case at hand but also on the prospects for future mergers.
11
To see why this matters, suppose that in
an
industry two fi rms wish to merge. Using static
analysis, we can assess the likely impact on prices and hence, consumer welfare. We can also
examine whether there may be any effi ciencies from the merger. But the impact on innovation is
more subtle. To be sure, competitive pressure to innovate will disappear between the merging
parties but may also change for others from that.
That, however, is not all that will happen. This is because the prospects for future mergers being
permitted or not will also have changed. That will impact on their likelihood and also have an
impact on what determines innovation prizes into the future. The hard issue is: in what way?
As it turns out there are competing effects and no amount of introspection can resolve them.
A
more permissive merger policy will make mergers more likely. On the one hand, when mergers
are more likely, that may reduce innovation competition and so cause innovation rates to fall. On
the other hand, mergers may themselves be part of the prize —
for instance, you are going to
be a more attractive merger partner if you have innovated more and so you can expect to get
more of the share of gains from mergers. This effect may mean that more permissive merger
policy may spur innovation. Which effect dominates is hard to say.
These sorts of issues tax competition authorities and make analysis diffi cult. This is especially
the case when industries are undergoing disruptive change. In that situation, regulators may be
concerned that inaction today may, rightfully, lead to problems later on. Hence, increasingly,
there is earlier investigation and advice to government in general coming from competition
authorities.
Conclusion
A good example of this is in relation to what is currently termed “the sharing economy.” These
are the new entry into industries such as hotels and taxi/limo services that have been facilitated
by digital technology that can match under-utilized resources with consumers. Indeed, this
highlights a two-element defi nition of the sharing economy: (i) that there are individuals who
own key assets (such as cars or dwellings) and (ii) that there exists
a market platform to match
those individuals with consumers
. Element (i) isn’t something that is new but element (ii) is
which is what makes all this currently relevant. Basically, mobile technologies allowed temporal
10
See
Segal, I. and M. Whinston (2007), “Antitrust in Innovative Industries,”
American Economic Review
,
97 (5): pp.1703-1730
;
Gans, Joshua S. (2010), “When is Static Analysis a Suffi cient Proxy for Dynamic
Considerations? Reconsidering Innovation and Antitrust,” in J. Lerner and S. Stern (eds),
Innovation
Policy and the Economy
, Vol.11
;
and
Gans, Joshua S. and Lars Persson (2013), “Entrepreneurial
Commercialization Choices and the Interaction between IPR and Competition Policy,”
Industrial and
Corporate Change
, Vol. 22, No. 1, 131-151
for more discussion.
11
For an interesting perspective on disruption and competition policy enforcement see
de Streel,
Alexandre and Larouche, Pierre, Disruptive Innovation and Competition Policy Enforcement (October 20,
2015). TILEC Discussion Paper No. 2015-021. Available at SSRN: http://ssrn.com/abstract=2678890 or
http://dx.doi.org/10.2139/ssrn.2678890
. Their paper views mergers as a potential hinderance to disruptive
processes and argues that competition authorities need to be vigilant. As will be argued here, the
dynamic trade-offs are somewhat more subtle.
9
agglomeration issues for suppliers to be overcome so that, for instance,
a supplier could signal
their availability and location in real time.
In many respect
s
, this is a business fad all on its own.
Regulators, competition and beyond, are concerned about these new developments. The fi rst
concern is consumer safety. There are
existing regulations
concerning
the ability of individuals
to make available their assets due to concerns about consumer safety (at least that is how they
are
posited these days
).
Those
concerns have not
gone away. But the very fact that new
markets have arisen without such regulations gives us pause to wonder whether they are
necessary. Uber, AirBNB all should have failed if the regulations were making transactions safe.
They did not fail because those platforms substituted public regulation for private regulation.
Uber and AirBNB are some of the most regulated eco-systems in the world
.
The problem we
have is compatibility between the public and private regulations not any fundamental
disagreement that they should exist
for their intended purpose
.
The second concern is with respect to market power or dominance, sh
ould the private platforms
emerge into a dominant platform in the future. To be sure, that is exactly what happened under
the system of public regulation. Because of that only large scale entry could overturn the
existing system. Like
Orwell’s
Animal Farm
the danger is that we turn one monopoly into
another. If the new platforms write the public regulatory rules, there is a concern that we could
have that situation.
In this situation to foretell a danger competition authorities need the equivalent of canaries in a
coal mine.
One reason to be optimistic is that a certain form of competition is baked into the
system. For instance,
if
Uber and Lyft drivers are
not
employees, they cannot be compelled to
work. A feature of Uber is that drivers are free to come in and out of the system. Alongside that,
they are currently free to come into and out of the Uber platform. Their ability to 'platform shop'
disciplines the power of platforms.
There are also risks.
Consider a situation where drivers must be licensed but that Uber, for
example, takes on the costs of licensing the drivers and ensuring the cars are serviced. In
return, they require exclusivity to Uber. Then we potentially have the seeds of a problem.
Instead, we want to ensure that drivers can fulfi ll these requirements in an independent way to
avoid such tying. It should not matter as, one way or another, the market will compensate them
for the costs. The sharing economy is important. It could re-write how we, for instance, deal with
transportation. But it needs a competitive foundation.
