Submarkets, Industry Dynamics, and the Evolution of the U'S' Laser Industry

1
Submarkets, Industry Dynamics, and the Evolution
of the U.S. Laser Industry

• Steven KlepperCarnegie Mellon University
• (joint with Ajay Bhaskarabhatla)

2
General Framework

• Diversity in buyer preferences
• Punctuated changenew submarkets
• Appeal to subset of buyers
• Latent substitutability
• If improved enough, new submarkets compete
  w/others
• Differential incentives opportunities to
  improve submarket offerings
• Evolution of dominant submarkets
• Can explain evolution of
• Market structure
• Character of innovation diversity of product
  offerings
• Turnover of industry leaders
• Apply to lasers
• Distinctive evolution of market structure

3
Organization of Talk

• Why lasers?
• Laser basics submarkets
• Model of industry evolution w/submarkets
• Explaining the first era of the laser industry
• Fusion of submarkets RD escalation
• Explaining the second era as well
• Testing the model on the laser industry
• General lessons general framework

4
Why Lasers?
Shakeouts in Innovative Industries Common
Sharp ? Entry Despite 20/yr Q growth
Entry, Exit, U.S. Automobile Producers,
1895-1966
5
My Theory of Shakeouts

• Growth costly
• Early entrants always larger (ceteris paribus)
• All entrants (firms) of limited size
• Returns to RD depend on size
• Process innovation, general product innovation
• Advantage of early entry
• New entrants eventually cant compete
• Later entrants forced out too

6
No Shakeout in High-Tech Lasers
Entry, Exit, U.S. Laser Producers, 1961-1994
Something wrong with my theory?
7
Laser Basics
First Ruby Laser
8
Lasers and Submarkets

• Laser types wavelengths
• Solid state crystals (Ruby)
• Semiconductor
• Chemical dye (tunable)
• Gases CO2, HeNe, Ion, Excimer, HeCd, Others
• Power, Size, Continuous/Pulsed
• Limited substitution, scope economies
• Wavelength dictates application
• Firms specialists
• 55 produced only 1/9 laser types
• 20 2 types
• 23 3-6 types
• Only 2 7-9 types
• Few commonalities in production across laser
  types
• Innovation created new laser types, submarkets

9
Submarket Theory of Industry Evolution

• K potential producers at start of industry
• Constant prob. ? new submarket created
• Appeals to new users
• Firms have same prob. ? submarket entry
• No scope economies
• Firm submarket sizedraw from F(Q)
• Total Q SSubmarket Q
• Constant prob. ? submarket destruction
• Industry evolution
• ? to steady-state submarkets, producers
• All entry, exit in period in 1 (different)
  submarket
• Prob firm exit in a period ? w/ submarkets
• E( firm submarkets) ? w/ firm age
• Prob firm exit ? w/ firm age ( size),
  independent of time of entry
• Firm growth ? w/ firm size ( age)

10
Cant Leave Well Enough Alone
U.S. Industry, 1961-2007
172
87
11
Key Technical Development (1988)
Diode Pumping
12
From a Ray to a Rainbow
ND Yag1064 nm
13
Submarket Model

• Creation of opportunity for new submarket
• Product RD rp required to develop
• Prob of developing s(rp) lt 1, s gt 0, s lt 0
• Imitation of developers after 1 period
• Developers play 1 period Cournot-Nash game
• Imitation ? p ? to AC c, Q developers ?, random
  entry smaller firms
• Substitutability across submarkets
• None initially
• Starts if p of submarket ? sufficiently ? as p
  ?
• Buyers same prob. switching
• Cost-reducing RD rc ?c g(rc)
• g(0) bounded, g gt 0, g lt 0
• 1 Period profit Qg(rc) rc, then imitated
• Q of developers too small for rc gt 0 to be
  profitable
• Tech ? ? quality Q demanded of 1 submarket
  (DPSS in 1988)
• (Only) developers expand to Q such that rc gt 0
  is profitable

14
Evolution

• Symmetric Nash equilibrium in rp
• Some firms develop new submarket
• Later submarket entrants capture smaller Q
• Before 1988, no cost-reducing RD
• Submarkets only add new possibilities
• Submarkets number of producers ? to steady
  state
• After 1988, SS Q ? SS entry occurs
• Then rc gt 0 for developers of SS lasers
• Their cost ? ? p of SS lasers ?, Q of SS lasers
  ( Q industry) ?
• Other firmslater SS entrants--exit SS lasers
• Cost gt p
• Eventually p of SS lasers? sufficiently that
• Some buyers of other lasers switch to SS lasers
• No new submarkets viable
• Producers of other lasers exit when lose all
  buyers
• Developers (earliest entrants) last to exit as
  are largest

15
8 Predictions

• laser producers industry Q
• Initially ? in tandem
• Continue to ? after DPSS in 1988
• Then shakeout Q continues to ?, producers ?
• SS laser producers SS Q
• Initially ? in tandem after DPSS in 1988
• Then SS shakeout Q ?, producers ?
• producers, Q in affected submkts
• ? then ? in tandem some time gt 1988
• SS innovation
• SS RD initially gt 0 (rp gt 0), then 0, then rc gt
  0
• ?SS patents/innovation w/DPSS after 1988
• Early SS entrants leading innovators

16
Predictions Continued

• Entry exit in sub-periods
• lt1988, in 1 laser type
• gt1988, in SS, then exit dispersed
• Industry hazard of exit
• lt1988, only via submkt destruction
• gt1988, ? due to exit in SS affected submkts
• Industry hazard, age time of entry
• lt1988, related to age, not time of entry (given
  age)
• gt1988, later entrants exit first
• Submarket hazard time of entry
• lt1988 unrelated
• gt1988, later entrants in SS affected submkts
  exit first

17
Industry Shakeout
18
SS Submarket
19
Other Submarkets
Helium Neon
Ion
Other Gas
CO2
CO2
Other Gas
Dye
20
SS Versus Gas Dye Patents
DPSS Lasers
Top 2 patenters industry leaders, long-time SS
producers
21
Entry by Laser Type Over Time
Lasers produced by entering firms in their first
year
Chi-squared for clustering 1970-89 significant at
0.001
22
Exit by Laser Type Over Time
Lasers produced by exiting firms in their last
year
Chi-squared for clustering 1970-89 significant at
0.02
23
Industry Hazard Rate
DPSS Lasers
24
Industry Hazard Time of Entry
Kaplan-Meier Survival Curves Control for Firm Age
1988 Cohort Significantly Different from other 2
Cohorts
25
Submarket Hazard Time of Entry
Solid State
CO2, Hene, Ion, Other Gas, Dye
1988 Cohort Significantly Different from Earlier
Cohort
26
Other Shakeout Theories

• Non-technological, better firms enter 1st
• DPSS lasers seem essential to story
• Hazard same for earlier entrants until gt 1988
• Dominant design/? MES
• Dominant submarket ? dominant design
• Proliferation of types of SS lasers after 1988
• New SS crystal lasers at natural wavelengths
• Vanadate crystal lasers
• Fiber lasers
• New SS lasers at shorter wavelengths
• But synthesizes two views of industry evolution
• Innovation ? after 1988, leaders longest SS
  survivors
• Innovation of diverse lasers, not scale econs
  seems key

27
Further Lessons

• Timing of shakeouts entry
• TV industry shakeout w/in 5 yrs vs autos lasers
• Absence of meaningful submkts in tvsno place to
  hide
• Why auto shakeout after 15 yrs?
• Model T in 1908 exploits vanadium steel
• Appeals to both rural and urban buyers fuses1
  large submkt
• Unleashes ? in process RD, Ford dominance, entry
  ? 0
• Turnover of leaders
• Can occur if new submarkets provide best
  incentives opportunities for improvement
• Hard disk drives
• Reversing the PLC
• Unequal submarkets militate toward dominant one
• But innovation can fissure submarketscameras
  after WW II
• Entry and ? product innovation at midlife