LOCATION PATTERNS OF ISRAELI PHARMACEUTICAL AND ELRCTRONIC FIRMS

1
LOCATION PATTERNS OF ISRAELI PHARMACEUTICAL
AND ELRCTRONIC FIRMS

• ??? ????? ????? ????
• ??????? ?????? ?????? ??????, ???????

2
Presentation

• Introduction
• Agglomeration
• Literature review
• Hypotheses
• Research design
• Spatial regression model estimation
• Results
• Conclusions
• Study implications

3
Introduction

• Understanding and explaining spatial organization
  of firms is central in industrial location
  economics
• Pharmaceutical and electronic industries are
  among the most significant and prominent branches
  of Israeli economy, understanding their location
  pattern is essential in firms location in Israel.

4
Agglomeration

• Agglomeration Economies
• May be defined as cost reductions that occur when
  many economic activities are carried on in one
  place (Blair, 1995).
• May be derived from the geographical
  concentration of firms engaged in similar
  activities or within the industry, leading to
  further local clustering of related firms and
  accumulation of knowledge (Odile, 2002).
• Agglomeration economies are cost-reducing factors
  that diminish uncertainty and increase production
  efficiencies (Shefer and Frenkel, 1999).
• McCann and Shefer (2004) mention information
  spillovers, non-traded local inputs and skilled
  local labor pool.

5
Literature review

• Empirical studies
• Many innovative industries are characterized by
  the apparent tendency to cluster in particular
  locations (Shefer and Frenkel, 1999).
• Locally captured spillovers play an important
  role in driving US pharmaceutical firms
  productivity (Chacar and Lieberman, 2003)
• Geographical localization of spillovers of
  knowledge was established (Jaffe, 1989 Furman,
  Kyle, Cockburn, Henderson, 2004 )
• Patent citations tend to occur more frequently
  within the state in which they were patented than
  outside of that state.
• University research has a positive effect on the
  productivity of local firms

6
Hypotheses

• The major question examined was does
  autoregressive spatial models provide a better
  explanation then models that disregard the
  spatial effects.
• The models are examined on the pharmaceutical and
  electronic industries in Israel.

7
Models Examined

• Traditional Tobit Regression
• Models with spatial effects using LeSage Matlab
  Spatial econometric toolbox (1999)
• SAR-Spatial Autoregressive Model
• SDM-Spatial Autoregressive Durbin Model
• SEM- Spatial Autoregressive Error Model

8
Hypotheses

• Location patterns of pharmaceutical industry and
  electronics industry in Israel are similar.
• Both industries are characterized by
  agglomerative location patterns.
• The following would affect firms location
  decision
• Location of a university
• Location of a hospital, generating demand for
  drugs and enabling collaboration in research
• Size of population is a proxy for the pool of
  workers
• Presence of firms in other industries in the same
  region

9
Overview of Israeli Pharmaceutical industry

• There are about 30 manufacturing companies in
  Israel that are devoted to the production of
  pharmaceutical products. They produce mainly
  human drugs, veterinary products and active
  pharmaceutical ingredients. The five leading
  companies are Teva, Agis, Dexxon, Taro and Rakah.
• The major amount of the firms are RD
  small-middle size companies

10

• Dependent variables
• Number of pharmaceutical companies in each
  natural region
• Mostly small middle sized companies with a varied
  line of activity such as RD, manufacturing,
  marketing etc that are located within Israel.
• Joins a number of pharmaceutical related sectors
  such as biotechnological and biological companies
  that mainly involved in drug research and
  development activities and chemical companies
  that manufacture biochemical, organic and other
  chemicals

11
Phamaceutical Industry

• The database includes about 170 companies
• The database sources are Dun and Bradstreet guide
  and Matimop internet site (http//www2.matimop.org
  .il).
• Besides pharmaceutical firms, firms of related
  sectors were included such as biotechnological
  and biological companies that are mainly involved
  in drug research and development activities and
  chemical companies that manufacture biochemical,
  organic and other chemicals

12
Overview of Israeli Electronic industry

• Israel's fabless sector is third only to the USA
  and Taiwan.
• The basis of the Israeli semiconductor industry
  are the very strong microelectronic academic
  departments in Israel since the early 60s. They
  contributed to a skilled manpower that later
  often migrated to the Silicon Valley, gained
  experience there and returned to Israel.
• Three noticeable representatives of electronic
  industry - Motorola, Intel and Tower

13
Electronic Industry

• Includes Semiconductors design, Semiconductor
  Manufacturing, Passive Components, Industrial
  Equipment, General Electronics Equipment and
  Electronics Assembly (EMS)
• The database includes about 250 companies
• The source of data is Israel Science and
  Technology website (www.science.co.il)
• Contains firms engaged with activities such as
  RD, manufacturing and marketing

14
Independent Variables

• Population (Pop)the number of persons
  distributed by natural regions (in thouthands).
• Plants (Plants) the number of establishments
  with more than 5 employees distributed by natural
  regions.
• Control variables
• University (UNIV) Distribution of Israeli
  universities. The variable receives value 1 in
  case university is present in the region, and 0
  otherwise.
• Hospital (HOSP) since this work focuses on
  pharma industry, hospitals represent one of the
  customers for the pharma firms and also research
  partner. The variable receives 1 in case of
  presence of hospital in a given region.

15
Traditional Regression

• PharmaceuticalTobit Regression

16
Variable significant at plt0.05
17
.

• Weight Matrix
• The proximity term in this work bases on natural
  regions distribution.
• ISR CBS (Central Bureau of Statistics) provides a
  map of Israel divided by regions, sub-regions and
  natural regions
• Rook method was used to examine spatial
  contiguity

18
Spatial Autoregressive Durbin Model

• SDM Tobit Pharma Regression
• Variable significant at plt0.05
• Variable significant at plt0.1

• SDM Tobit Electronic Regression
• Variable significant at plt0.05
• Variable significant at plt0.1

• Tobit method was used due to distribution of
  observations since this method allows dealing
  with cases where sample observations are censored
  or truncated. Given that the number of companies
  cannot be less than zero, the dependent variable
  is bounded on the lower end by zero and justifies
  the use of Tobit specifications for the models
  used.

19
Results

• Spatial lag parameter ?
• Pharmaceutical industry
• Positive and significant coefficients
• Presence of spatial agglomeration economies at
  work for the dependent variable (the number of
  companies in a certain region depends on the
  number in neighboring regions).
• Electronic industry
• Positive but insignificant at the 10 percent
  level
• Suggesting that the number of electronic
  companies in one region was not influenced by
  number of companies of nearby regions.
• .

20
Results

• Population
• Pharmaceutical industry
• Negative and insignificant
• W-population is positive and significant at 10
  level
• Positive influence of population presence in
  nearby regions on firms decision to locate in a
  certain area
• Electronic industry
• Negative and insignificant

21
Results

• University
• Insignificant, however positive in both
  industries
• Contradicts a common hypothesis in spatial
  literature that proximity to universities or
  research institutions is beneficial for plants
  (Jaffe, 1989, Feldman, 1994, Jaffe, Trajtenberg
  and Henderson, 1993)
• Contradicts thesis hypothesis
• Consistent with previous research that was
  conducted in Israel (Frenkel, 2001, Felstnstein,
  1996)

22
Results

• Plants
• Represents the presence of other firms in the
  area
• Positive and significant at 5 level in both
  industries
• Consistent with study hypothesis that presence of
  other firms in the same region will have a
  positive influence on the firm.
• In the pharmaceutical industry WPlants is
  significant but negative, it measures the impact
  of remote plants.

23
Results

• Hospital
• Represents proximity to the client and also
  proximity to the source of innovation (similar to
  influence of universities).
• Positive and significant in pharmaceutical
  industry
• Aligned with study hypothesis of positive
  correlation between hospitals and pharmaceuticals
  firms

24
Main Conclusions

• The presence of spatial agglomeration economies
  between the different natural regions
• Confirmed in the pharmaceutical industry
• Not confirmed in electronic industry

25
Conclusions cont.

• Autoregressive models are more appropriate to
  examine agglomeration effects.
• Though the traditional Tobit model results in the
  pharma industry in R-squared of 0.69, SDM results
  in R-squared of 0.7326. The spatial regression
  results in better goodness of fit, i.e. higher
  R-squared and in more significant results
  regarding the spatial effect.

26
Main Conclusions cont.

• Result per industry
• Pharmaceutical RD process may require usage of
  specific resources such as special equipment,
  clean rooms or unique skills. Significance and
  positive sign of hospital variable strengthen
  this claim, since many hospitals are engaged in
  research activities and can represent valuable
  source of resources such as specific equipment
  and access to laboratories for the firms.
• Most of the small middle sized electronic firms
  do not have need for special resources due to the
  branch nature.

27
Plants Variable

• The variable was significant in both industries
• Represents an existence of industrial
  establishments in the area, indicating that firms
  prefer to locate near other firms. However the
  type of activity of those firms activity is less
  important.
• The presence of industrial establishments in the
  area indicates existence of infrastructure such
  as communication lines, different facilities,
  roads, accessibility etc.
• Electronic firms need access to general
  infrastructure, and they are less dependent on
  other electronic firms.

28
Population Variable

• The population variable was insignificant in both
  industries
• Israel is relatively a small country
• About 40 of employees work one or more natural
  regions far from home, therefore the dependence
  on proximity to population center is limited
  (based on CBS data)
• University variable was also insignificant in
  both branches
• Consistent with previous research that was
  conducted in Israel (Frenkel, 2001, Felstnstein,
  1996)

29
Study implications

• Pharmaceutical firms
• Would find it beneficial to collocate with firms
  that engage in the same type of activity
• Creating specialized industrial parks or centers
  would be valuable for this industry.
• Electronics firms
• Would not see in location with other electronic
  firms a valuable advantage
• Existence of better infrastructure would attract
  electronic firms to the region.

30
QA
31
Agglomeration - cont.

• Types of agglomeration economies
• Internal agglomeration economies
• External economies
• Linkages between two activities
• Vertical relationships
• Backward linkage
• Forward linkage
• Complementary relationships

32
Literature review

• Webber (1929) and Lösch (1944) models are
  cornerstone of the location theory.
• Weber location theory (1929)
• The model highlights the relationships between
  the input-output structure of the firms
  production function and the influence of input
  and output transportation costs on the firms
  optimum location.
• The central place theory of Christaller (1933)
  and Lösch (1944)
• Scale economies and transportation costs will
  lead to the creation of lattice of central
  places, each serving the surrounding area.

33
Literature review cont.

• Location of Innovativeness
• The Diamond of Competitive Advantage Theory
• Porters diamond of competitive advantage
  theory identifies four major conditions that
  shape a cluster that include
• Factor conditions such as costs, infrastructure,
  resources, and scientific and technical knowledge
  in the region
• Demand conditions which refer to the strength of
  local and export demand Related supporting
  industries which provide local sourcing
• Firm strategy and rivalry which refers to
  cooperative and competitive relationships among
  firms.
• Supporting industries

34
Literature review cont.

• Location of Innovativeness
• The Diamond of Competitive Advantage Theory
  -cont.
• The success of an individual firm may be
  partially traced to the size, depth, and nature
  of the cluster of related and supporting both
  public and private enterprises. Clusters provide
  constituent firms a competitive advantage not
  afforded by dispersed firms. Location advantages
  accrued from being in a cluster and provide less
  costly access to specialized inputs like
  components, machinery, business services, and
  skilled personnel in comparison to dispersed
  participants (Porter, 2000).

35
Spatial Econometrics Spatial Autoregressive
Regressions

• The standard linier regression has the following
  form
• y Xß e
• e N(0,?2)
• The most general statement of spatial auto
  regression is
• y ?W1y Xß u
• u ?W2ue
• e N(0,s2In)

36
Spatial Econometrics Spatial Autoregressive
Regressions cont.

• Model with restrictions
• SAR Model - Spatial Autoregressive Model
• y ?W1y Xß e
• e N(0, s2In)
• SEM model - Spatial Error Model
• y Xß u
• u ?Wue
• e N(0,s2In).
• SDM model - Spatial Durbin Model
• y ?W1y Xß1 W1Xß2 e
• e N(0, s2In)

37
Spatial Econometrics cont.Weight matrix

• Contiguity matrix
• NxN symmetric matrix where wij 1 when i and j
  are neighbors and 0 when they are not
• Makes for a fairly sparse matrix
• W matrix is usually standardized so all columns
  sum to 1
• wsij wij / Sj wij
• Makes operations with the W matrix as an average
  of neighboring values

38
Spatial Econometrics Weight matrix - cont.

• Types of contiguity matrix
• Rook contiguity - define Wij 1 for regions that
  share a common side with the region of interest.
• Bishop contiguity - define Wij 1 for entities
  that share a common vertex with the region of
  interest.
• Queen contiguity - for entities that share a
  common side or vertex with the region of interest
  define Wij 1.

39
The Diamond of Competitive Advantage Theory
40
Pharmaceutical Firms Data Distribution
Axis X -Range of number of companies Axis Y
-Number of natural regions
41
Electronic Firms' Data Distribution
Axis X -Range of number of companies Axis Y
-Number of natural regions
42
Population Distribution
Axis X -Range of population number (in
thousands) Axis Y -Number of natural regions
43
Plants Distribution
Axis X -Range of plants number Axis Y -Number
of natural regions
44
Maps of Israel natural region
45
Weight Matrix
46
Employed Persons by District