measuring and explaining management practices across firms and countries october 2007
DESCRIPTION
MEASURING AND EXPLAINING MANAGEMENT PRACTICES ACROSS FIRMS AND COUNTRIES October 2007 Nick Bloom Stanford & NBER John Van Reenen LSE & NBER. MOTIVATION. Large persistent productivity spread across firms and countries: people typically claim this is due to differences in “management” - PowerPoint PPT PresentationTRANSCRIPT
MEASURING AND EXPLAINING MANAGEMENT PRACTICES ACROSS FIRMS AND COUNTRIES
October 2007
Nick BloomStanford & NBER
John Van ReenenLSE & NBER
MOTIVATION
Large persistent productivity spread across firms and countries:
people typically claim this is due to differences in “management”
• But what is the role of management?
• And why does it vary so much across firms and countries?
SUMMARY OF THE PAPER (1 of 3)
(1) Measuring Management
•Develop a survey tool to “measure” management practices
• New data on 732 firms in US,UK, France & Germany.
•Management data:
• Appears consistently measured within firms
• Correlated with productivity, profits, Tobin’s Q, growth & survival
• Robust to measurement error and bias
SUMMARY OF THE PAPER (2 of 3)
(2) Explaining Management
•Observe big spread in management practices (Fig. 2 over)
• Wide cross firm spread (like profits & productivity)
• Significant differences across countries
• US 1st, Germany 2nd, France 3rd and UK 4th
•Demonstrate that two factors appear significant:
• Production market competition – positive effect
• Family managed firms – negative effect
• Family firm ownership but not management is fine
• Family ownership and management problematic, particularly under primo geniture CEO succession
0.2
.4.6
.81
1.2
Den
sity
1 2 3 4 5
0.2
.4.6
.81
1.2
Den
sity
1 2 3 4 5
0.2
.4.6
.81
1.2
Den
sity
1 2 3 4 5
0.2
.4.6
.81
1.2
Den
sity
1 2 3 4 5
FIRM LEVEL AVERAGE MANAGEMENT SCORES
France n=137 n=157
n=290n=154UK US
Germany
SUMMARY OF THE PAPER (3 of 3)
(3) Quantifying this Effect
•Competition and family-management important, explains about 50% of firm-level management tail; and between 1/3 to 2/3 of US-Europe management gap:
• Europe has lower levels of competition
• UK & France also many more primo geniture family firms due to Norman legal origin & tradition
1. Why should management practices vary?
2. “Measuring” management practices
3. Evaluating the reliability of this measure
4. Describing management across firms & countries
5. Explaining management across firms & countries
OUTLINE
Why Should Management Practices Vary?
Two models - not mutually exclusive
•“Optimal choice of management practices”
• Another factor of production (like advertising)
• No “better” or “worse” style of management – depends on firm’s circumstances
•Exogenous managerial inefficiency (Mundlak, 1961; Lucas 1978)
• Part of total-factor productivity
• Strictly “better” or “worse” styles of management
•Empirically we find some support for both
1. Why should management practices vary?
2. “Measuring” management practices
3. Evaluating the reliability of this measure
4. Describing management across firms & countries
5. Explaining management across firms & countries
SOME RELATED LITERATURE - EXAMPLESManagement, organisation & performance
• HRM / Management practices: Ichinowski, Shaw, and Prenushi (1997), Ichinowski and Shaw (1995), Black and Lynch (2001), and Lazear (2000); Cappelli and Neumark (2001), Bartel, Ichniowski and Shaw (2004),
• Organisational practices: Bresnahan, Brynjolfsson and Hitt (2002) and Caroli and Van Reenen (2001)
• Individual managers: Bertrand and Schoar (2003)
Competition and firm performance
• Empirics: Nickell (1996), Syverson (2004), and Aghion, Bloom, Blundell, Griffith, and Howitt (2005)
• Dynamic theory: Jovanovic (1982) and Hopenhayn (1992)
• Theory: Schmidt (1997), Raith (2003) and Vives (2004)
Productivity dispersion & dynamics
• Establishments: Baily, Hulten, and Campbell (1992), Bartelsman and Dhrymes (1998), and Jensen, McGuckin and Stiroh (2001), Foster, Haltiwanger and Syverson (2003)
• Countries: O’Mahony & Van Ark (2004), Caselli (2005)
Family firms
• Empirical: La Porta, Lopez-De-Silanes and Schleifer (1999), Bertrand et al (2004), Villalonga and Amit (2004), Bennedsen, Nielsen, Perez-Gonzales & Woflenzon (2005),
• Theory: Burkart, Panunzi and Schleifer (2003), Caselli and Gennaioli (2005)
• Economic History: Landes (1969), Chandler (1994), Nicholas (1999)
STEPS TO TRY TO MEASURE MANAGEMENT
1) Developing management practice scoring
• Scorecard for 18 monitoring, targets and incentives practices
• 45 minute phone interview of (manufacturing plant) managers
2) Obtaining unbiased responses
• “Double-blind”
•Interviewers do not know company performance
•Managers are not informed (in advance) they are scored
3) Getting firms to participate in the interview
• Introduced as “Lean-manufacturing” interview, no financials
• Endorsement of Bundesbank ,UK Treasury, Banque de France
• Run by 10 MBAs (loud, assertive & business experience)
Score (1): Measures tracked do not indicate directly if overall business objectives are being met. Certain processes aren’t tracked at all
(3): Most key performance indicators are tracked formally. Tracking is overseen by senior management
(5): Performance is continuously tracked and communicated, both formally and informally, to all staff using a range of visual management tools
MONITORING - i.e. “HOW IS PERFORMANCE TRACKED?”
Note: All 18 dimensions and over 50 examples in Bloom & VanReenen (2006).
ADDITIONAL CONTROLS FOR BIAS & NOISE
8 INTERVIEWEE CONTROLS
• Gender, seniority, tenure in post, tenure in firm, countries worked in, foreign, worked in US, plant location, reliability score
3 INTERVIEWER CONTROLS
• Set of analyst dummies, cumulative interviews run, prior firm contacts
5 TIME CONTROLS
• Day of the week, time of day (interviewer), time of the day (interviewee), duration of interview, days from project start
MANAGEMENT SURVEY SAMPLE
• US (290), UK, France and Germany (≈150 each)
• Medium sized manufacturers (100 - 10,000 employees, median ≈ 600)
•Medium sized because firm practices more homogeneous
•Manufacturing as easier to measure productivity
• Obtained 54% coverage rate from sampling frame
•Response rates uncorrelated with performance measures
ADDITIONAL MATCHED DATA WE COLLECTED
HR Survey
• Skills, demographics, hours, organisational characteristics, number of competitors etc.
Ownership & Family Survey
• Shareholders & managerial characteristics, family involvement, family progression rules etc.
Performance Data
• Separately match company accounts - so collect management and performance data from completely different sources
Industry and Trade Data
• OECD
1. Why should management practices vary?
2. “Measuring” management practices
3. Evaluating the reliability of this measure
a) Internal/External validation
b) Contingency
c) Measurement error/bias
4. Describing management across firms & countries
5. Explaining management across firms & countries
INTERVAL VALIDATION OF THE SCORING
1
2
3
4
5
1 2 3 4 5
`
1st interview
2nd i
nte
rvie
w
• Re-interviewed 64 firms with different interviewers and managers
Firm average scores (over 18 question)
• Firm-level average correlation of 0.759
EXTERNAL VALIDATION OF THE SCORINGPerformance measure
cit
cit
ccitm
ccitk
ccitl
cci
cit uxmklMNGy '
ln(capital)
ln(materials)
management(average z-scores) ln(labor)
other controls
• Use up to 11 years of accounting data for 1994-2004
country c
• Note – not a causal estimation, only an association
Dependentvariable
Sales(in Ln)
Sales(in Ln)
Sales(in Ln)
ROCE Tobin Q (in Ln)
Sales growth
Exit
Estimation1 OLS OLS OLS OLS OLS OLS Probit
Firms All All All All Quoted All All
Managementi
0.085(0.025)
0.034(0.011)
0.042(0.012)
2.469(0.688)
0.250(0.075)-
0.018
(0.006)
-0.200[0.026]
Ln(Labor) it
0.999(0.014)
0.539(0.021)
0.540(0.021)
2.172(1.202)
0.209(0.109)
-0.022
(0.011)
0.233[0.045]
Ln(Capital) it
0.103(0.013)
0.104(0.013)
-0.148(0.899)
-0.029(0.086)
0.024
(0.008)
-0.158[0.045]
Ln(Materials) it
0.362(0.020)
0.354(0.020)
-0.439(0.723)
0.130(0.050)
-0.010
(0.007)
-0.084[0.231]
Controls1 No Yes Yes Yes Yes Yes Yes
Noise controls No No Yes Yes Yes Yes Yes
Observations 6,267 5,350 5,350 5,089 2,635 4,777 709
Firms 732 709 709 690 374 702 709
EXTERNAL VALIDATION: PRODUCTIVITY & PROFIT
1 Includes country, year, SIC3 industry, skills, hours, firm-age, and public/privateRobust S.E.s in ( ) below. For probit p-values in [ ] below
EXTERNAL VALIDATION – ROBUSTNESS
Productivity correlations robust to type of TFP estimation
• OLS, Olley-Pakes, GMM & Within-Groups
Results also significant in most recent cross-section (2003/04)
Results significant in both Anglo-Saxon (US and UK) and
European (France and Germany) country subsets
CONTINGENT MANAGEMENT PRACTICES
Dependent VarHC
Management
FC Manage
ment
HC-FC Manage
ment
HC-FC Manage
ment
HC-FC Manage
ment
Level Firm Firm Firm Firm Industry
Ln (% degrees)i
firm level
0.220
(0.039)
0.100(0.043)
0.120(0.043)
Ln (ave wage)i
firm level0.337
(0.122)
Ln (% degrees)j
Industry level (US)
0.281(0.169)
Standard Errors Robust Robust Robust Robust Clustered
Firms 732 732 732 424 732Note: “HC management” average z-score of the 3 most human capital focused questions (questions 13, 17 and 18). “FC management” average z-score of the 3 most fixed capital focused questions (1, 2 and 4). “HC-PC management” is the difference of these two measures.
CONCERNS WITH OUR MANAGEMENT MEASURE?
Three potential issues:
1) Measurement error (classical), but
• Attenuation downwardly biases our results
• We try to control for this with “Noise” controls (management & interview characteristics)
CONCERNS WITH OUR MANAGEMENT MEASURE?
(2) Firm performance-related measurement bias in management score (i.e. the “happy manager” problem), but
• Surveying methodology using examples tries to minimize this
• Competition and management positively linked (later)
• Management-performance link is as important in France & Germany (where managers less likely to “talk up” Anglo-Saxon practices) as it is in UK & US
• No link between past productivity growth & management
• Not all questions significant (and not linked to “subjectivity”)
• Other subjective questions insignificant – i.e. “feel-good” work-life balance questions, organisational devolvement questions
So potential problem – but no evidence that major phenomenon
CONCERNS WITH OUR MANAGEMENT RESULTS?
(3) Reverse causality (management correctly measured but better firm performance causes better management),
• Yes – but main point of performance estimations is external validity of the measure
• Also note that if interpretation is effect of management on productivity note that the bias is ambiguous
1. “Measuring” management practices
2. Evaluating the reliability of this measure
3. Describing management across firms & countries
4. Explaining management across firms & countries:- competition- family managed firms
OUTLINE
0.2
.4.6
.81
1.2
Den
sity
1 2 3 4 5
0.2
.4.6
.81
1.2
Den
sity
1 2 3 4 5
0.2
.4.6
.81
1.2
Den
sity
1 2 3 4 5
0.2
.4.6
.81
1.2
Den
sity
1 2 3 4 5
FIRM LEVEL AVERAGE MANAGEMENT SCORES
France n=137 n=157
n=290n=154UK US
Germany
COUNTRY LEVEL MANAGEMENT SCORES*
3.07
3.14
3.31
3.35US
Germany
UKTypical UK managers?
Bad manufacturing management - a UK tradition?
“Efficient management is the single most significant factor in the American productivity advantage”[Marshall Plan Anglo-American productivity mission, 1947]
France
3.58
3.25
3.13
US FIRMS ARE ALSO BETTER IN EUROPE
Average management score by firm type in UK, France and Germany*
Domestic
Non-US multinational subsidiary
US multinational subsidiary
* Controls for any sample selection on size (direct and group) and listing
# in sample
379
44
20
1. “Measuring” management practices
2. Evaluating the reliability of this measure
3. Describing management across firms & countries
4. Explaining management across firms & countries:- competition- family managed firms
OUTLINE
Factors we did not find a significant relationship for
Unions: negative but not significant•But: (i) sample ≈ 450 firms; and (ii) issues over causation•Was negative and significant for two individual practices:• Fixing/firing bad performers,• Rewarding good performers
CEO Pay: no link in levels – but issues over causation
Ownership/Governance: positive but insignificant for ownershipconcentration and board indepedence measures:
•But sample only UK/US quoted firms (≈ 350)
Leverage: nothing with debt/equity – but issues over causation
Competition & Models of Management Practices
“Exogenous managerial inefficiency” – positive impact
•Selection models Hopenhayn (1992) or Syverson (2004)
“Optimal choice model” – ambiguous impact
•In contracting models balance between opposing profit and market-size effects (Raith 2003, Vives 2004).
Competition proxies Dependent variable: Management
Import penetration (SIC-3 industry,1995-1999)
0.144(0.040)
0.156(0.084)
1 - Lerner index1
(SIC-3 industry except firm itself, 1995-1999)
1.515(0.683)
1.318(0.637)
# of competitors(Firm level,2004)
0.142(0.051)
0.145(0.049)
Full controls2,3 No Yes No Yes No Yes
COMPETITION AND MANAGEMENT PRACTICES (TABLE 4)
1 Lerner index = (operating profit – capital costs)/sales ≈ rents2 Includes 108 SIC-3 industry, country, firm-size, public and interview noise
(analyst, time, date, and manager characteristic) controls, = 732 obs3 S.E.s in ( ) below, robust to heteroskedasticity, clustered by country-industry
3 competition proxies from Nickell (1996) & Aghion et al. (2005)
FAMILY FIRMS & MANAGEMENT – AN OLD TOPIC
Alfred Chandler1 and David Landes2 both claimed UK & French industrial decline relative to US & Germany linked to family firms
“The Britain of the late 19th Century basked complacently in the sunset of economic hegemony. Now it was the turn of the 3rd generation…and the weakness of British enterprise reflected their combination of amateurism and complacency”
“French enterprise was family-owned and operated, security-orientated rather than risk-taking, technologically conservative and economically inefficient”
1 Alfred Chandler, “Scale and Scope: The Dynamics of Industrial Capitalism”, (1994)2 David Landes, “The Unbound Prometheus: Technological Change and Industrial Development in Western Europe from 1750 to the Present”, (1969)
WE DO FIND GREATER UK & FRENCH FAMILY MANAGEMENT IN OUR DATA (100 YEARS ON),
% UK Fra Ger US
Family1 largest shareholder 30 32 30 10
Family1 largest shareholder and family CEO
23 22 12 7
Family1 largest shareholder, family CEO & primo geniture2 15 14 3 3
1 Family defined as 2nd generation or beyond (so not the founder). Shareholdings combined across all family members. 2 Based on question: “How was management of the firm passed down: was it to the eldest son or by some other way?”. Non primo geniture alternatives in frequency order: other sons, son in-laws, daughters, brothers, wives, nephews and cousins.
WHY DOES FAMILY INVOLVEMENT VARY ACROSS COUNTRIES?• Historical differences
• UK & French tradition of Primo Geniture:
[Oxford English Dictionary, 2005] “Feudal rule of inheritance introduced into England by the Norman Conquest. Replaced Teutonic gavelkind. Obligatory until the Statute of Wills [1540]. Still common in many places”
• US and German tradition of equal division (Menchik, 1980)
• Estate tax headline rates1: on family firms
• US ≈ 50% France ≈ 25%
• UK = 0% Germany ≈ 15%1 Rate on a $25m firm. In practice these taxes are often reduced/avoided by advanced tax planning, although this involves foresight, financial costs and some control loss.
FAMILY FIRMS AND MODELS OF MANAGEMENT PRACTICES
Likely family impact depends on involvement
•Ownership but not management probably positive
• Concentrated ownership so better monitoring
•Management probably negative
• Smaller pool to select CEO from
• Possible “Carnegie” effect on future CEO’s
• Both effects will be worse with primo geniture (succession of eldest son to CEO position)
FAMILY OWNERSHIP AND FAMILY MANAGEMENT (TABLE 5)
% Dependent variable: Management
Family1 largest shareholder-0.029(0.094)
0.304
(0.166)
Family1 largest shareholder & family CEO
-0.100(0.078)
-0.175(0.188)
Family1 largest shareholder, family CEO & primo geniture
-0.281(0.097)
-0.382(0.128
Observations2 732 732 732 732
1 Family defined as 2nd generation or later2 Note includes SIC-3 digit, country, skills, firm size, firm age & public controls
QUANTIFYING THESE EFFECTS:
• ACROSS FIRMS
• ACROSS COUNTRIES
0.2
.4
.6
.8
11.2
Density
1 2 3 4 5Average management score across questions and interviews - note dropping lean3
0.2
.4
.6
.8
11.2
Density
1 2 3 4 5Average management score across questions and interviews - note dropping lean3
MANY COMPETITORS AND NO (PG) FAMILY CEO
FEW COMPETITORS AND/OR (PG) FAMILY CEO
2.7% firms in tail1
9.0% firms in tail1
1 Tail defined as a score ≤ 2. In the whole sample 6.9% of firms are in the tail.Sample splits significantly different at 5%, but not if exclude firms with score ≤ 2
N=415
N=317
Dependent variable Management
Country is US Baseline Baseline Baseline Baseline Baseline
Country is Germany-0.045(0.064)
-0.081(0.075)
-0.090(0.075)
-0.051(0.074)
0.010(0.076)
Country is France-0.202(0.086)
-0.183(0.104)
-0. 131(0.103)
-0.075(0.102)
-0.028(0.102)
Country is UK-0.276(0.078)
-0.276(0.093)
-0.227(0.091)
-0.199(0.091)
-0.126(0.079)
Family owned, family CEO & primo geniture
-0.638(0.101)
-0.628(0.100)
-0.584(0.098)
# of competitors0.142
(0.052)0.161
(0.051)
Ln (% employees with a degree)
0.145(0.037)
Public & size controls No Yes Yes Yes Yes
Observations 732 732 732 732 732
ACCOUNTING FOR THE CROSS-COUNTRY SCORES
1 OLS on 732 observations. S.E.s in ( ) robust to arbitrary heteroskedasticity
• Original methodology for measuring management
• Product market competition & family management important
• Explain 50% of tail of badly managed firms
• Explain 2/3 of US-France gap & 1/3 of US-UK gap
• Last summer ran 3500 firm survey on firms in Europe, US and Asia covering management and organisational structure
Research design very flexible so any suggestions welcome
Quotes:
TO SUMMARIZE
BACK-UP
MY FAVOURITE QUOTES:
[Male manager speaking to an Australian female interviewer]
Production Manager: “Your accent is really cute and I love the way you talk. Do you fancy meeting up near the factory?”
Interviewer “Sorry, but I’m washing my hair every night for the next month….”
The British Chat-Up
MY FAVOURITE QUOTES:
Interviewer: “How many production sites do you have abroad?
Manager in Indiana, US: “Well…we have one in Texas…”
Americans on geography
Production Manager: “We’re owned by the Mafia”
Interviewer: “I think that’s the “Other” category……..although I guess I could put you down as an “Italian multinational” ?”
The difficulties of defining ownership in Europe
MY FAVOURITE QUOTES:
The bizarre
Interviewer: “[long silence]……hello, hello….are you still there….hello”
Production Manager: “…….I’m sorry, I just got distracted by a submarine surfacing in front of my window”
The unbelievable
[Male manager speaking to a female interviewer]
Production Manager: “I would like you to call me “Daddy” when we talk”
[End of interview…]
Score (1) People are promoted primarily upon the basis of tenure
(3) People are promoted upon the basis of performance
(5) We actively identify, develop and promote our top performers
INCENTIVES - i.e. “HOW DOES THE PROMOTION SYSTEM WORK?”
Note: All 18 dimensions and over 50 examples in Bloom & VanReenen (2006).
Score (1) Goals are either too easy or impossible to achieve; managers low-ball estimates to ensure easy goals
(3) In most areas, top management pushes for aggressive goals based on solid economic rationale. There are a few "sacred cows" not held to the same rigorous standard
(5) Goals are genuinely demanding for all divisions. They are grounded in solid, solid economic rational
TARGETS - i.e. “HOW TOUGH ARE TARGETS?”
Note: All 18 dimensions and over 50 examples in Bloom & VanReenen (2006).
Dependent Var
Ln (Sales)
Ln (Sales)
Ln (Sales)
Ln (Sales)
Ln (Sales)
Estimation1 Reduced form, OLS Full, OLS Full, IV
Management0.042
(0.012)0.216
(0.097)
Competition (Import penetr.)
0.089(0.032)
0.088(0.032)
Family CEO & primo geniture
-0.060(0.030)
-0.058(0.030)
Instruments(F-test)
Imports,Family(20.79)
Over-identifying restriction (p-val)
0.520
% 75:25 TFP gap accounted for
12% 63%
I.V. MANAGEMENT IN PRODUCTION FUNCTION
1 Other variables include log(Labor), log(Capital), log(Materials), country, year, SIC3 industry, skills, hours, firm-age, and public/private. All 709 observationsS.E.s in ( ) below, robust to arbitrary heteroskedasticity
-.2-.1
0.1
.2ep
hi_o
rig/e
phi_
p5/e
phi_
p95
1 2 3 4 5Log firm age
ephi_orig ephi_p5
ephi_p95
AGE AND MANAGEMENT PRACTICES (KERNEL1)
Firm age (in logs)
Man
agem
ent
scor
e
10 years
1 Point-wise confidence intervals (in feint) generated from 1000 bootstraps
75 years
34
56
7S
ales
per
em
ploy
ee
-2 -1 0 1 2Management Score
bandwidth = .8
Lowess smoother
FAMILY OWNERSHIP PROBIT
Dependent variableFamily owned, family CEO & primo geniture1
Country = UK0.109
[0.015]
Country = France0. 096[0.042]
Country = Germany0.058
[0.303]
Log (employees)-0.022[0.012]
Log (firm-age)0.052
[0.017]
Industry controls Yes
Observations 718
1 Marginal effects, p-values in [ ] brackets underneath
SOME LIMITED EVIDENCE FOR EFFORT EFFECTS?
*Includes 108 SIC-3 digit dummies, country dummies, firm size and type
S.E.s robust to arbitrary heteroskedasticity, clustered by country-industry
Dependent variable
Managerial Hours Worked
Lerner index(5-yr lagged)
6.660(4.129)
1.809(5.869)
Import penetration (5-yr lagged)
-0.230(0.444)
1.082(0.948)
Number of competitors
1.155(0.509)
0.935(0.623)
Firms 727 727 733 733 733 733
Observations 727 727 733 733 733 733
Full controls* No Yes No Yes No Yes