Innovation, Technological Conditions and New Firm Survival

1
Innovation, Technological Conditions and New Firm
Survival
Paul Jensen, Beth Webster Hielke Buddelmeyer
Centre for Microeconometrics Workshop 22nd
November, 2006
2
OUTLINE

• Motivation and Research Objective
• Stylised Facts About Industrial Evolution
• Dataset Construction
• Descriptives
• Empirical Model
• Results and Conclusions

3
MOTIVATION

• International empirical evidence suggests that
• Firm survival has important effects on market
  structure, aggregate productivity growth and
  technological change
• Innovation, firm size (size-at-birth) and
  organisational structure are important
  determinants of firm survival
• However, almost nothing is known about either
  causes or effects of firm survival in Australia
• All that is known comes from 2 years of ABS data
  on the nature/extent of business exits in
  mid-1990s
• We address this lacuna in the literature by
• Mapping patterns of entry/exit using ASIC data on
  a large sample of company births/deaths
  1997-2003
• Linking these data with other firm, industry and
  macro data in order to analyse the causes/effects
  of survival

4
OBJECTIVE

• In this paper, we consider the following
  questions
• Firm Level How does innovation shape survival
  for new vis-a-vis incumbent firms?
• Industry Level How does the speed of
  technological change in an industry affect
  relative survival rates?
• Macro Level Are new firms more susceptible to
  business cycle effects than incumbent firms?
• To address these issues, we model firm survival
  using a piecewise-constant exponential hazard
  function
• Our dataset is an unbalanced panel of 260,000
  companies alive at some stage during 1997-2003
• Numerous cohorts of entrants
• Time-varying industry-level measure of tech
  conditions
• Firm-level measures of IP stocks and flows
• Some aggregate macroeconomic fluctuation

5
INDUSTRY EVOLUTION

• Stylised stages of development (variation in
  length/intensity of stage across product markets)
• Stage I product launch stimulates firm entry
• Stage II shakeout of firms, mainly due to tech
  change
• Stage III consolidation of oligopolistic
  structure
• How generalisable are these evolutionary stages?
• What do we know about firm survival?
• Age matters? No proxies for other unobserved
  factors
• Founder quality (skills, experience) affect
  survival
• Stage of industry evolution (formative/mature
  hi/low tech)?
• Size effects large firms are more likely to
  survive
• Subsidiaries (de alio births) outperform de novo
  births
• Macroeconomic conditions affects hazard rates
• Gross entry rate matters entry exit
• Innovation increases the likelihood of survival?

6
DATA (1)

• Our dataset consists of
• 261,262 companies alive during 1997-2003 as
  determined by ASIC registration/deregistration
  data
• Unit of analysis is Australian Company Number
  (ACN)
• The data were linked (by company name) to
• Yellow Pages in order to get ANZSIC codes
• IBISWorld data
• Parent/subsidiary concordance was constructed
  using ASIC ownership files
• IP Australia data to construct stocks/flows of
  patents and trade marks (measured at parent
  level)
• Companies that changed name/address were treated
  as continuing entities
• NB most mergers do not result in deregistration

7
DATA (2)

• 67 of ASIC records were matched to Yellow Pages
• Sample by industry looks representative of
  population (although some under-representation of
  restaurants due to the fact that company names ?
  trading names)
• Yellow Pages filters out non-trading companies
• The following ABS data also linked into the
  dataset
• Industry level profit margin
• GDP, interest rates and
• ASX stock market index
• Thus, our linked dataset has detailed
  time-varying firm-level, industry-level and
  macroeconomic variables

8
DESCRIPTIVES

• Death is defined as deregistration of an ACN or
  disappearance from the Yellow Pages
• Age profile companies vary from 0 to 124 yrs old
• Trends in birth/death rates
• Births are decreasing over the period
• Deaths are increasing over the period
• But net entry rate is positive overall

9
EMPIRICAL MODEL

• Piecewise exponential hazard function
• Company age (years) is the unit of time analysis
• Incumbents are defined as any company born prior
  to 1990 who we observe in 1997-2003
• New firms are defined as new ACNs 1997-2003
• Our set of explanatory variables xi consists of
• Patent/trade mark stocks (i.e. renewals) (log1)
  yrs
• Patent/trade mark flows (i.e. applications)
  lagged number of applications (log1) (Shadow of
  death)
• Size dummy (all IBIS firms are large)
• Parent and subsidiary dummies
• Private/public firm dummy
• 1-digit ANZSIC industry dummies

10
EMPIRICAL MODEL (2)

• Other explanatory variables are
• Gross entry rate number of entrants in 2-digit
  ANZSIC industry relative to number of incumbents
  (proxies intensity of competition or barriers to
  entry)
• Risk industry profit margin over the tangible
  capital-output ratio (proxies capital intensity)
• Industry innovativeness a (lagged) weighted
  1-digit industry index of RD expenditure and
  employment, IP applications and labour
  productivity (to proxy for the effects of process
  innovations). Variable measures the speed of
  technological change
• Macro conditions factor of ?GDP and ??GDP
• Interest rate 90-day bank bill rate
• Stock market ASX index

11
RESULTS (1)
12
RESULTS (2)

• Firm-level innovation results
• Patent applications actually increase the
  likelihood of exit for incumbents (since they
  embody market risk)
• Patent stocks lower hazard rates for incumbents
  and have no effect for new firms (market filter
  yet to take hold)
• TM stocks unambiguously improve survival rates
• Firm size (crudely measured) matters larger
  firms are much more likely to survive
• Entry begets exit, especially for new firms.
  Maybe low barriers to entry, but high barriers to
  survival
• BUT in industries characterised by rapid
  technological change, new firms are more likely
  to survive

13
RESULTS (3)

• AND new firms also are less likely to be
  affected by the riskiness of the industry
• Macroeconomic variables also play an important
  part in shaping firm survival
• All macro factors are significant, but the
  relative effect is greater for new firms
• Increase in interest rates increase hazard rate,
  but new firms are more vulnerable
• Increase in GDP aids all firms, but provides a
  greater boost for new firms
• New firms are more susceptible to a fall in the
  stock market

14
CONCLUSIONS

• No simple linear relationship between innovation
  and performance must take account of Knightian
  uncertainty (market or technological)
• Thus, it is important to separate innovation
  investments from innovation capital
• New firms play an important role in technological
  change in fast-moving industries, new firms
  drive the gale of creative destruction
• New firms are particularly sensitive to changes
  in macroeconomic conditions
• However, some puzzles remain why are spinoffs
  more likely to die? Better data from ASIC may
  help explain this