INSIDE THE BLACKBOX: ECONOMIC PERFORMANCE AND TECHNOLOGY ADOPTION WHEN SPACE AND PRODUCT RELATIONSHI

1
INSIDE THE BLACKBOX ECONOMIC PERFORMANCE AND
TECHNOLOGY ADOPTION WHEN SPACE AND PRODUCT
RELATIONSHIPS MATTER

• Leonardo Becchetti
• University of Rome Tor Vergata

Annalisa Castelli University of Rome Tor Vergata
Monitoring Italy 2005 Productivity, Growth
Competitiveness Rome, ISAE, June 7th 2005
2
MOTIVATION RESULTS

• We highlight the relevance of studying the
  performance of productive units by considering
  space and product relationships.
• By applying the district/no district and
  independent/subcontractee status taxonomy, we
  find that district independent firms have higher
  value added and are more efficient than non
  district independent firms, but that ICT
  investment improve efficiency only for the
  latter.
• Our findings suggest that i) the quality of
  space relationships (district location)
  positively affects unmeasured forms of human and
  social capital ii) ICT adoption has a stronger
  impact on non district independent firms as the
  improvement of electronic links is more
  beneficial for productive units with poorer space
  relationships.

3
BACKGROUND LITERATURE (1)

• Limits of the ISTAT classification
• 1) Dichotomous indicators fail to take into
  account the intrinsic heterogeneity of industrial
  districts account (Tattara, 2002 and Iuzzolino,
  2004)
• 2) Focus on employment density in small and
  medium sized enterprises, which leads to
  neglecting the contribution of large firms and
  urbanization on local corporate performance
  (Iuzzolino, 2004).
• 3) Econometric estimates limiting the analysis to
  the effect of an ISTAT(-like) ID variable neglect
  the importance of the position of the productive
  units along the value chain.

4
BACKGROUND LITERATURE (2)

• Alternative solutions
• the relative density of manufacturing employment
  with respect to national average in the local
  labour system or in the municipality
• a global specialization index (Glaeser et al.,
  1992 Henderson, 1997 Henderson et al., 1995
  Costa et al., 1999) calculated on employees and
  on productive units in the local labour system or
  in the municipality
• industry specific specialization indexes (Glaeser
  et al., 1992 Bronzini, 2003 Becchetti et al.,
  2005)
• combination of ISTAT ID with value chain position
  indicators (this paper)

5
SPACE AND PRODUCT RELATIONSHIPS

• Space district classification
• We use the industrial district concept as a
  proxi for space relationships. Firms belonging to
  districts have been identified using the ISTAT
  quantitative indicator.
• Product position in the value chain
• Product relationships have been proxied using
  qualitative information on firms subcontracting
  activity. Firms having more (less) than 70 of
  their net sales subcontracted have been
  classified as subcontractee (independent).

Matching the two we get our fundamental
taxonomy district/no district -
independent/subcontractee
6
DATA

• Data come from the last two waves (1995-97 and
  1998-2000) of the Mediocredito Centrale Survey.
• We focus on a sample of 3903 firms selected from
  the 1998-2000 wave.
• In order to test if ICT investment carried out in
  the previous period has positive and significant
  impact on productive efficiency in the following
  period we match the last wave with the 1995-97
  one and obtain a sample of 1221 firms.
• We then classify the obtained samples by our
  proposed taxonomy based on geographical
  agglomeration and position in the value chain.

7
METHODOLOGY

• Descriptive evidence highlights differences among
  the 4 subgroups of firms. In order to disentangle
  the various determinants of these differences we
  follow 3 steps

• OLS we test the impact of traditional controls
  such as age, size etc. on a performance measure.
• Stochastic Frontier Approach we estimate a
  stochastic frontier production function to test
  whether the proposed taxonomy affects productive
  efficiency.
• Difference in distance estimate we follow this
  two step procedure to test the effect of ICT
  investment on productive efficiency.

8
1. OLS

• The highlighted heterogeneity among subgroups
  could depend on 2 classes of determinants (age,
  size, monopoly rents etc. or space and product
  relationships). The problem is that the two
  classes of variables are strictly correlated.
• To disentangle the two effects we regress value
  added per worker on age, size and other
  traditional controls in order to test whether the
  role of space and product relationship is
  significant.

• We find that district firms have significantly
  higher value added net of considered controls.
• Repeating the estimate on subgroups of
  independent and subcontractee firms we find that
  all the effect is due to the significantly higher
  value added of district with respect to non
  district independent firms.

9
(No Transcript)
10
2. STOCHASTIC FRONTIER APPROACH (a)

• OLS estimates do not discriminate between rent
  extraction and productive efficiency and do not
  simultaneously take into account distances from
  the efficiency frontier for a given production
  function.
• The SFA allows to distinguish between production
  inputs and efficiency/inefficiency factors and to
  disentangle distances from the efficient frontier
  between those due to systematic components and
  those due to noise.
• We estimate the model with Frontier 4.1 on both a
  three year panel than a cross-section in which
  all quantitative variables are included as three
  year averages in order to smooth potential, year
  specific, measurement inaccuracies.

11
2. STOCHASTIC FRONTIER APPROACH (b)

• The estimated model

(1)
(2)

• space relationships significantly and positively
  affect productive efficiency since district
  firms, whatever their position in the value
  chain, are significantly less distant from the
  efficient frontier.

12
3. DIFFERENCE IN DISTANCE (a)

• The evaluation of whether ICT investment in the
  previous wave (1995-97) has positive and
  significant effects on productive efficicency in
  the following wave (1998-00) requires 2 steps
• 1st step we estimate a common stochastic
  frontier model (the 2 equations above presented,
  the 2nd including only industry dummies).
• 2nd step we regress changes in the estimated
  distance from the efficient frontier in the two
  periods on a series of regressors traditionally
  used in the second equation of the stochastic
  frontier model.

13
3. DIFFERENCE IN DISTANCE (b)
The chosen specification is the following

• among the 4 subgroups of ICT investing firms,
  only non district independent firms exhibit a
  significant reduction in distance from the
  efficient frontier.

14
(No Transcript)
15
SFA RESULTS
space relationships significantly and positively
affect productive efficiency since district
firms, whatever their position in the value chain

• A likely interpretation of this result is that
  space relationships may help to accumulate hidden
  and unmeasured social and human capital
  components which significantly affect the
  productivity of visible and measurable labour and
  capital inputs. The result is a significantly
  higher productive efficiency of the two
  observable inputs.

16
DIFFERENCE IN DISTANCE RESULTS
among the 4 subgroups of ICT investing firms,
only non district independent firms exhibit a
significant reduction in distance from the
efficient frontier

• This result is consistent with the effects of
  geographical agglomeration. According to the
  literature, district location fosters hidden or
  not entirely measurable forms of (human and
  social) capital by reinforcing informal networks
  and technological spillovers. The underestimation
  of the stock of overall capital explains the
  observed difference in mean and variability of
  productive efficiency between the district/non
  district subgroups. By considering that
  electronic links are a substitute of space in
  fostering networks, we are not surprised that ICT
  adoption has a stronger and significant impact on
  non district independent firms and that ICT is
  more beneficial for those firms with poorer space
  relationship.

17
CONCLUSIONS

• Measuring firm intangible assets is becoming of
  increasing relevance in a framework of global
  integration in which innovation plays a decisive
  competitive role.
• In this paper we have argued that an important
  dimension of these intangible assets may be
  incorporated by space and product relationship of
  individual productive units.
• Our findings show that quality of space
  relationships affects firm productivity in both
  value added per worker and firm efficiency and
  that ICT adoption generates significant
  improvement in efficiency only for the subgroup
  of non district independent firms.
• We interpret our results by arguing that district
  location significantly stimulates accumulation of
  hidden and unmeasured forms of (human and social)
  capital and that ICT acts as a partial substitute
  for these factors in firms which do not benefit
  of local agglomeration economies.

18
FURTHER SUPPORT FOR THE ROBUSTNESS OF THE
DISTRICT EFFECT ...(1)
19
FURTHER SUPPORT FOR THE ROBUSTNESS OF THE
DISTRICT EFFECT ...(2)