1. Firm decentralization around the world
Nick Bloom (Stanford Economics)
John Van Reenen (Stanford GSB/LSE)
Lecture 8: February 2010 1

2. Summary of last lecture on hospitals, school and retail management
Huge variation in management practices seems common to every sector we have studied to date
The very same practices around monitoring, targets and incentives always associated with superior performance
Demonstrates there are basic management best practices, with the grid identifying a core set of these
Grid plus data useful tool for change management in that provides evidence base on key practices

3. What is decentralization?
Measurement and stylized facts
Determinants of decentralization
Firm Performance and decentralization

4. What is decentralization?
One aspect of firm organization - how power and decision making are distributed
Unlike basic management practices no typical “good” and “bad” – just different styles which tend to vary by country
Why should you care:
Key in designing your business or advising others
Changing over time, generating broader economic effects, like increased inequality and worker empowerment

5. Are things really changing?
"Globalization and the arrival of the information economy have rapidly demolished all the old precepts. The management of global companies, which must innovate simultaneously and speed information through horizontal globe-spanning networks, has become a daunting challenge. Old, rigid hierarchies are out ...."
Business Week The 21st Century Corporation, cover story August 21-28, 2000.

6. Evidence is limited, but appears yes, firms are decentralizing
Delayering – Number of positions reporting to CEO
Source: Rajan and Wulf, 2006, 300+ large US corporations

7. Formal vs. Real Authority
Formal authority in the organization chart (“organogram”)
E.g. Profit centres vs. cost/sales centres
Real authority
How things actually get done. Power
Can be different across different types of decisions
e.g. pricing, products and M&A

8. Real authority models: Centralized

9. Real authority models: Decentralized
e.g. Universities, but academics generally pretty inefficient

10. What is decentralization?
Measurement and stylized facts
Determinants of decentralization
Firm Performance and decentralization

11. Central HQ Plant Example A: Domestic Firm 2 Sites, Single Plant
(New York Site)
Plant
(Albany Site)
D, Decentralization

12. Central HQ Plant 1 Plant 2 Plant 3
Example B:
US Domestic Firm
Multi-Site, Multi-Plants
Central HQ
(New York Site) D3
Plant 1
(Buffalo Site) D1
Plant 2
(Albany Site) D2
Plant 3
(Scranton Site)

13. The Decentralization Survey Page

14. Our empirical decentralization measure
Main measure averages the z-score (scores normalized to mean 0, standard-deviation 1) of each variable:
Hiring senior employees (discrete, 1 to 5)
Maximum Capital expenditure (continuous, in $)
Introduction of new products (discrete, 1 to 5)
Sales and marketing (discrete, 1 to 5)

15. Decentralization measure
Decentralization varies heavily across countries
Most centralized
Asia
Southern Europe
Least centralized
Scandinavian countries
Anglo-Saxon countries
This is central graph in paper. Anglo-saxon/scandinavia vs asia and europe; ranking pretty independent of gdp pc or prod
Decentralization measure

16. Decentralization measure (higher number is more decentralized)
Decentralization also varies across firms
Note: Variation between countries by three digit SIC is 55%, 45% within country and industry
Decentralization measure (higher number is more decentralized)

17. External validation – decentralization over time
Only prior firm decentralization measure to cross-check against we are aware of is from Hofstede
Surveyed c.100,000 IBM employees across 50 countries during the 1970s & 1980s
Questions on management style (autocractic/paternalistic or consultative) and preferences for delegation
Combined into Power Distance index (1-100), low means limited (preference for) delegation

18. ‘Power distance’ measures from 1970s & 1980s matches our 2000s decentralization data quite well
This graph shows : a) cross country ranking coherent with similar measures of hierarchical organizations b) persistence over time c) striking given multi industry vs single firm comparison
Correlation= 0.80
Power distance

19. Fiscal decentralization (whether Governments are centrally run) looks similar – so something cultural?
Firm Decentralization
This graph shows : a) cross country ranking coherent with similar measures of hierarchical organizations b) persistence over time c) striking given multi industry vs single firm comparison
Correlation= 0.83
Fiscal Decentralization (Arzaghi and Henderson, 2005) 19

20. Decentralization measure
Decentralization also varies across ownership types – with founder and government firms centralized
This graph shows : a) cross country ranking coherent with similar measures of hierarchical organizations b) persistence over time c) striking given multi industry vs single firm comparison
Decentralization measure 20

21. What is decentralization?
Measurement and Stylized facts
Theories & Determinants of decentralization
Firm Performance and decentralization

22. What do you think are the benefits and costs of decentralization?

23. Some Benefits of Decentralization
Saves on costs of communication and information transfer
Improves swiftness of reacting to market changes [e.g. because of less specialization, more multi-tasking]
Increase in employee involvement, so possibly more job satisfaction

24. Some Costs of Decentralization
More “mistakes” (e.g. co-ordination in price setting)
Harder to exploit returns to scale/specialization – i.e. repetition of same activities across different divisions
More stress from control?

25. Factors affecting Decentralization
Technological
Economic
Cultural/Legal

26. Technological Factors affecting Decentralization
Complexity: Size, Product and Multinational status
Learning: Age and Proximity to technological frontier
ICT: Information costs and Communication costs

27. Complexity: Larger Firms are more decentralized
Number of employees in the firm

28. Complexity: Multinationals (even controlling for size) are more decentralized
Dependent variable: Decentralization of plants
Foreign Multinational
0.157***
(0.058)
Firm employment (in logs)
0.052**
(0.022)
Plant employment (share of firm)
0.089***
(0.030)
Plant Skills (% employees with a degree)
0.085***
(0.015)
Observations
3,660
Industry dummies (112)
yes
Country dummies (12)
Other controls (60)

29. Learning: Younger firms are more decentralized29

30. Learning: More technologically advanced firms (i. e
Learning: More technologically advanced firms (i.e. close to frontier) more likely to be decentralized

31. ICT: Decentralization and the use of ICT
Enterprise Resource Planning
0.104*
0.116**
Information technology
(0.054)
NETWORK
-0.098*
-0.110**
Communication technology
(0.053)
Computers/Employee
-0.041
-0.021
-0.031
(0.031)
Information technology (IT) increase decentralization
Communications technology (CT) decreaseas decentralization
In last column ERP assoc with a 11.5% of a s.d. more autonomy (Networks 11% less)
Notes: Controls are SIC3 dummies, country dummies, noise controls (interviewer dummies Interviewee tenure and seniority, etc.), public listing, CEO onsite, plant size,

32. Factors affecting Decentralization
Technological
Economic
Cultural/Legal

33. Economic Factors affecting Decentralization
Competition
information: Improves the value of timely responses to local conditions
incentives: Lower risk of manager abusing autonomy as incentives more aligned with firm
Skill: more able workers better at coping with responsibility

34. Competition: Firms facing more competition are more decentralized
Competition proxies
Dependent variable: Decentralization
Imports/production (three digit industry, )
0.184***
(0.073)
Lerner Index of competition (three digit industry except firm)
2.265***
(1.081)
Number of competitors
0.094**
(0.034)
Observations
2,497
3,587
Notes: Other controls are SIC3 dummies, 12 country dummies, noise controls (interviewer dummies Interviewee tenure and seniority, etc.), public listing, CEO onsite, plant size,

35. Skill: Firms with more skilled workers are more decentralized
Proportion of employees with a college degree

36. Legal and Cultural Factors affecting Decentralization
Trust
In high trust areas (e.g. Sweden) managers likely to be trusted to carry out more activities
Rule of Law
In good rule of law areas (e.g. US) top management less worried about theft by middle management
Religion
Evidence that firms from areas with more centralized religions tend to be more centralized (causal?)

37. Trust and decentralization
How to measure trust? World Value Survey has question:
“Generally speaking, would you say that most people can be trusted or that you can’t be too careful in dealing with people?”
Trust by region of the country defined as % of people answering “yes” to first part of the trust question
Experiments show this question linked with trust/trusting behavior (Glaeser et al, 2000, Sapienza et al, 2007)

38. Trust by country (means and regional spreads)
The graph shows median level of trust. The vertical bars denote minimum and maximum levels.

39. Trust, rule of law and decentralization
Trust (region)
1.196***
0.825***
0.732**
(0.429)
(0.290)
(0.298)
Rule of law (country)
0.473***
(0.102)
Observations
3549
Country dummies no
yes
Other controls
Notes: Other controls are SIC3 dummies, noise controls (interviewer dummies
Interviewee tenure and seniority, etc.), public listing, CEO onsite, plant size, competition

40. Also look at multinationals as another test of trust
Could worry about bias due to trust proxying for other country/regional variables
So look at affiliates of foreign multinationals and investigate whether trust in their home country also matters
Bilateral trust: (on average) how much do people in the multinational’s home country trust people in the country where the subsidiary is located (Eurobarometer Survey)

41. Central HQ Plant Example A: Domestic Firm 2 Sites, Single Plant
(New York Site)
Plant
(Phoenix Site)
D, Decentralization

42. French CHQ Sweden CHQ Plant 1 Plant 2
Example D
Japanese MNE
Global HQ
(Tokyo Site)
Do not observe D
French CHQ
(Paris Site)
Sweden CHQ
(Stockholm Site)
Observe D
Observe D
Plant 1
(Lund Site)
Plant 2
(Lyon Site)

43. Decentralization and trust in multinational firms
Trust (region of location)
0.609
0.563
0.446
(0.592)
(0.843)
(1.908)
Trust (country of origin)
0.749***
0.152
(0.301)
(0.152)
Trust (bilateral from origin cty to location cty)
1.809***
(0.768)
Full set of controls
Yes
Regional dummies No
Country origin dummies
Clustering
Region
Origin country
Observations
867
280

44. What is decentralization?
Measurement and Stylized facts
Determinants of decentralization
Firm Performance and decentralization

45. Does decentralization matter for performance?
Short run: generally depends on firms circumstances (e.g. country, industry and technology)
Short run: one recent factor that has been rapidly affecting this is IT. Evidence that impact of IT on productivity much higher in better managed and more decentralized firms
Long run: to be large you need some decentralization, so key for growth (Penrose 1959 and Chandler 1962)

46. Productivity growth and IT
“High Org” = more decentralized
IT and decentralization appear complementary

47. Wrap-up Decentralization varies heavily by country and ownership
Northern European and US firms decentralized, and Southern European and Asian being centralized
Founder, family and government very centralized
Factors driving this appear to be both economic – competition, technology and skills – but also cultural around trust, rule-of-law and even religion
In short-run decentralization is a choice variable depending on circumstances, but necessary for growth in long-run