ICT adoption in developing countries: firm-level evidence from Brazil and India Simon Commander (LBS) Rupert Harrison (IFS) 1 st June 2006.

ICT adoption in developing countries: firm-level evidence from Brazil and India

Simon Commander (LBS)

Rupert Harrison (IFS)

1st June 2006

Structure of the talk

New firm-level data from Brazil and India

Summary of results in two areas:

ICT, organisational change and productivity

Constraints on ICT adoption

Conclusions and other work

New firm-level data

500 face-to-face questionnaires in each country, April-May 2005

New firm-level data

500 face-to-face questionnaires in each country, April-May 2005

6 three-digit manufacturing industries, representing 30% of manufacturing employment in Brazil and 20% in India

New firm-level data

500 face-to-face questionnaires in each country, April-May 2005

6 three-digit manufacturing industries, representing 30% of manufacturing employment in Brazil and 20% in India

Sample stratified by region and firm size

New firm-level data

500 face-to-face questionnaires in each country, April-May 2005

6 three-digit manufacturing industries, representing 30% of manufacturing employment in Brazil and 20% in India

Sample stratified by region and firm size

Broad range of detailed information on:

Performance outcomes (employment, sales etc)

ICT adoption and organisational change

Occupation and skill shares of the workforce

Management practices

Reported constraints on ICT adoption

The sample

Brazil India

Median emp. N Median emp. N

Electronic Components 50 76 43 78

Plastic products 80 81 65 126

Soap and detergents 65 77 65 47

Auto components 187 78 120 110

Machine tools 69 81 70 52

Garments 65 99 89 63

Total 70 492 70 476

Summary measure of ICT (1 – 5)

1) ICT is not used at all

“How would you best describe the degree of ICT usage in your firm?”

Summary measure of ICT (1 – 5)

1) ICT is not used at all

2) ICT is used only for some office applications along with accessing the internet, emailing, etc.

“How would you best describe the degree of ICT usage in your firm?”

Summary measure of ICT (1 – 5)

1) ICT is not used at all

2) ICT is used only for some office applications along with accessing the internet, emailing, etc.

3) ICT is used for some advanced applications. Most processes are automated but there is no integration into a central system.

“How would you best describe the degree of ICT usage in your firm?”

Summary measure of ICT (1 – 5)

1) ICT is not used at all

2) ICT is used only for some office applications along with accessing the internet, emailing, etc.

3) ICT is used for some advanced applications. Most processes are automated but there is no integration into a central system.

4) Most processes are automated and some of them are integrated into a central system.

“How would you best describe the degree of ICT usage in your firm?”

Summary measure of ICT (1 – 5)

1) ICT is not used at all

2) ICT is used only for some office applications along with accessing the internet, emailing, etc.

3) ICT is used for some advanced applications. Most processes are automated but there is no integration into a central system.

4) Most processes are automated and some of them are integrated into a central system.

5) Almost all processes are automated and integrated into a central system

“How would you best describe the degree of ICT usage in your firm?”

Summary measure of ICT (1 – 5)

0.2

.4.6

1 2 3 4 5 1 2 3 4 5

Brazil India

Fraction

Fraction

Fra

ctio

n

Summary index of ICT adoption

Graphs by c

20032001

Other ICT measures

Brazil India

% of firms who have adopted PCs 98 98

% of firms who have adopted servers 90 63

Number of PCs per employee 0.28 0.22

Number of servers per employee 0.04 0.02

% of non-production workforce using PCs 69.6 53.9

% of prod workforce using ICT-controlled machinery 23.3 15.3

Other ICT measures

Brazil India

% of firms who have adopted PCs 98 98

% of firms who have adopted servers 90 63

Number of PCs per employee 0.28 0.22

Number of servers per employee 0.04 0.02

% of non-production workforce using PCs 69.6 53.9

% of prod workforce using ICT-controlled machinery 23.3 15.3

PCs per 1000 people 86 11

ICT expenditure as % of GDP 6.7 3.7

Other ICT measures

Brazil India

% of firms who have adopted PCs 98 98

% of firms who have adopted servers 90 63

Number of PCs per employee 0.28 0.22

Number of servers per employee 0.04 0.02

% of non-production workforce using PCs 69.6 53.9

% of prod workforce using ICT-controlled machinery 23.3 15.3

PCs per 1000 people 86 11

ICT expenditure as % of GDP 6.7 3.7

Date of adoption:PCs (both 98%)

0.0

5.1

.15

1975 1980 1985 1990 1995 2000 2005 1975 1980 1985 1990 1995 2000 2005

Brazil India

Density

kdensity v302a

De

nsity

Date of first introduction: PCs

Graphs by c

Date of adoption:Servers (90%, 63%)

0.0

5.1

.15

1975 1980 1985 1990 1995 2000 2005 1975 1980 1985 1990 1995 2000 2005

Brazil India

Density

kdensity v302b

De

nsity

Date of first introduction: Servers

Graphs by c

ICT adoption and productivity

Micro literature from developed countries has found very high estimated returns to ICT (Brynjolfsson and Hitt, 2000, 2003; Stiroh, 2002)

ICT adoption and productivity

Micro literature from developed countries has found very high estimated returns to ICT (Brynjolfsson and Hitt, 2000, 2003; Stiroh, 2002)

Possible explanations include: unobserved heterogeneity, unmeasured complementary investments in e.g. human capital, software, workplace organisation

ICT adoption and productivity

Micro literature from developed countries has found very high estimated returns to ICT (Brynjolfsson and Hitt, 2000, 2003; Stiroh, 2002)

Possible explanations include: unobserved heterogeneity, unmeasured complementary investments in e.g. human capital, software, workplace organisation

Some evidence that unmeasured complementary organisational changes may help to explain high measured returns to ICT (Bresnahan, Brynjolfsson and Hitt, 2002; Basu, Fernald, Oulton and Srinivasan, 2003)

ICT adoption and productivity

Micro literature from developed countries has found very high estimated returns to ICT (Brynjolfsson and Hitt, 2000, 2003; Stiroh, 2002)

Possible explanations include: unobserved heterogeneity, unmeasured complementary investments in e.g. human capital, software, workplace organisation

Some evidence that unmeasured complementary organisational changes may help to explain high measured returns to ICT (Bresnahan, Brynjolfsson and Hitt, 2002; Basu, Fernald, Oulton and Srinivasan, 2003)

No micro-econometric evidence in developing countries

ICT adoption and productivity

Dep var: Log sales Brazil India

(1)

Levels

(2)

2 year difference

(3)

Levels

(4)

2 year difference

Log employment 0.392***(0.109)

0.337***(0.066)

Log materials 0.359***(0.062)

0.434***(0.051)

Log fixed capital 0.214**(0.083)

0.188***(0.038)

Log ICT capital 0.173***(0.042)

0.115***(0.034)

Observations 172 312

R-squared 0.84 0.87

ICT adoption and productivity

Dep var: Log sales Brazil India

(1)

Levels

(2)

2 year difference

(3)

Levels

(4)

2 year difference

Log employment 0.392***(0.109)

0.190**(0.096)

0.337***(0.066)

0.180**(0.078)

Log materials 0.359***(0.062)

0.369***(0.124)

0.434***(0.051)

0.343***(0.104)

Log fixed capital 0.214**(0.083)

0.159(0.132)

0.188***(0.038)

0.154(0.099)

Log ICT capital 0.173***(0.042)

0.193***(0.070)

0.115***(0.034)

0.101**(0.039)

Observations 172 130 312 248

R-squared 0.84 0.59 0.87 0.38

ICT adoption and productivity

High returns persist after controlling for occupation, skills, software, management practices and other controls

ICT adoption and productivity

High returns persist after controlling for occupation, skills, software, management practices and other controls

Measure two forms of organisational changes associated with ICT adoption in the literature:

“Over the last three years which of these changes related to working practices for the production workforce have been introduced in your firm?”

1. Removed a level of hierarchy, or reduced no. of reporting levels

2. Improved monitoring of individual workers or groups of workers

ICT adoption and productivity

High returns persist after controlling for occupation, skills, software, management practices and other controls

Measure two forms of organisational changes associated with ICT adoption in the literature:

“Over the last three years which of these changes related to working practices for the production workforce have been introduced in your firm?”

1. Removed a level of hierarchy, or reduced no. of reporting levels

2. Improved monitoring of individual workers or groups of workers

Similar questions for Admin & Clerical workforce and Managers

ICT adoption and productivity

No impact of organisational changes in the full sample

ICT adoption and productivity

No impact of organisational changes in the full sample

If we drop low ICT adopters (i.e. only desktop applications)

Strong evidence of complementarity between flattening hierarchies and ICT capital investment in Brazil

Strong direct effect of flattening hierarchies on productivity growth in India

No effect of improved monitoring in either country

ICT adoption and productivity

No impact of organisational changes in the full sample

If we drop low ICT adopters (i.e. only desktop applications)

Strong evidence of complementarity between flattening hierarchies and ICT capital investment in Brazil

Strong direct effect of flattening hierarchies on productivity growth in India

No effect of improved monitoring in either country

So there is evidence for complementarities between some types of organisational change and ICT investment…

…but not in the tail of low ICT adopters (c. 30% of the sample in Brazil, 40% in India)

Constraints on ICT adoption

High measured returns to ICT investment

Important policy question: what aspects of policy / economic environment constrain firms from adopting ICT?

Constraints on ICT adoption

High measured returns to ICT investment

Important policy question: what aspects of policy / economic environment constrain firms from adopting ICT?

We use regional variation in infrastructure and other reported constraints to identify effects on ICT adoption and returns to ICT

India: State mean number of days with power-related problems (power cuts or surges)

Both countries: region/state means of other reported constraints on ICT adoption

Constraints on ICT adoption

Region/state mean proportion of firms reporting as a constraint to ICT adoption:

Brazil India

Mean s.d. Mean s.d.

Availability of skills 0.54 0.13 0.44 0.23

Unions 0.20 0.09 0.14 0.16

Labour regulations 0.40 0.10 0.30 0.27

Internet availability / price 0.45 0.08 0.49 0.28

Low internet usage 0.47 0.11 0.51 0.16

Lack of government support 0.52 0.10 0.39 0.25

Constraints on ICT adoption

Region/state mean proportion of firms reporting as a constraint to ICT adoption:

Brazil India

Mean s.d. Mean s.d.

Availability of skills 0.54 0.13 0.44 0.23

Unions 0.20 0.09 0.14 0.16

Labour regulations 0.40 0.10 0.30 0.27

Internet availability / price 0.45 0.08 0.49 0.28

Low internet usage 0.47 0.11 0.51 0.16

Lack of government support 0.52 0.10 0.39 0.25

Number of days disrupted - - 21.8 7.9

Constraints on ICT adoption

In India, firms in states with more power disruptions use less ICT and get lower returns from ICT investment

Constraints on ICT adoption

Andhra PradeshAndhra PradeshAndhra PradeshAndhra PradeshAndhra PradeshAndhra PradeshAndhra PradeshAndhra PradeshAndhra PradeshAndhra PradeshAndhra PradeshAndhra PradeshAndhra Pradesh

DelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiDelhiGujaratGujaratGujaratGujaratGujaratGujaratGujaratGujaratGujaratGujaratGujarat

HaryanaHaryanaHaryanaHaryana

KarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnatakaKarnataka

MaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastraMaharastra

Tamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil NaduTamil Nadu

UPUPUPUPUPUPUP

West BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest BengalWest Bengal

-.0

50

.05

.1.1

5.2

Sta

te-s

peci

fic e

last

icity

10 15 20 25 30 35State mean number of days in 2001 with power-related problems

R2 = 0.80

Coefficient (s.e.) = -0.009*** (0.001)

Constraints on ICT adoption

In India, firms in states with more power disruptions use less ICT and get lower returns from ICT investment

And power disruptions are very highly correlated across states with other reported constraints on ICT adoption

Constraints on ICT adoption

In India, firms in states with more power disruptions use less ICT and get lower returns from ICT investment

And power disruptions are very highly correlated across states with other reported constraints on ICT adoption

Suggests that a cluster of ‘poor institutions’ and infrastructure are associated with low ICT adoption and low returns to ICT

Constraints on ICT adoption

In India, firms in states with more power disruptions use less ICT and get lower returns from ICT investment

And power disruptions are very highly correlated across states with other reported constraints on ICT adoption

Suggests that a cluster of ‘poor institutions’ and infrastructure are associated with low ICT adoption and low returns to ICT

Much less regional variation in Brazil (very concentrated around Sao Paulo and the South) so can’t identify similar effects

Constraints on ICT adoption

In India, firms in states with more power disruptions use less ICT and get lower returns from ICT investment

And power disruptions are very highly correlated across states with other reported constraints on ICT adoption

Suggests that a cluster of ‘poor institutions’ and infrastructure are associated with low ICT adoption and low returns to ICT

Much less regional variation in Brazil (very concentrated around Sao Paulo and the South) so can’t identify similar effects

Interestingly, the least constrained Indian states look similar to Brazil in terms of ICT adoption and returns to ICT investment

Conclusions and other work

Very high estimated returns to ICT at the firm level

Conclusions and other work

Very high estimated returns to ICT at the firm level

Some evidence for effects of complementary organisational changes once drop tail of firms with low levels of ICT

Conclusions and other work

Very high estimated returns to ICT at the firm level

Some evidence for effects of complementary organisational changes once drop tail of firms with low levels of ICT

Poor infrastructure and policy environment are associated with lower ICT adoption and returns to ICT across Indian states

Conclusions and other work

Very high estimated returns to ICT at the firm level

Some evidence for effects of complementary organisational changes once drop tail of firms with low levels of ICT

Poor infrastructure and policy environment are associated with lower ICT adoption and returns to ICT across Indian states

Other work: ICT and the demand for skills (Harrison, 2006)

Strong evidence for skill-bias of ICT adoption in Brazil and India

Larger effect in Brazil than in India

Robust to controlling for endogeneity of ICT adoption

Conclusions and other work

Very high estimated returns to ICT at the firm level

Some evidence for effects of complementary organisational changes once drop tail of firms with low levels of ICT

Poor infrastructure and policy environment are associated with lower ICT adoption and returns to ICT across Indian states

Other work: ICT and the demand for skills (Harrison, 2006)

Strong evidence for skill-bias of ICT adoption in Brazil and India

Larger effect in Brazil than in India

Robust to controlling for endogeneity of ICT adoption

One day conference at London Business School, 19th June