Measuring & explaining management practices
Nick Bloom (Stanford & NBER)
based on work with Raffaella Sadun (HBS) & John Van Reenen (LSE)
MIT/Harvard Org Econ Lecture 1 (February 2010)
1
Two part lecture course
Lecture 1: Measuring and explaining management practices
• Overview what are management practices, how we measure
them, why they vary and what effect they have on performance
• Highlight how little is rigorously known – management is one of
the major holes in social science, and a great research area
Lecture 2: Measuring and explaining organizational practices
• As above overview what are organizational practices, how we
measure them, why they vary and what effect this has
• Again, large holes in rigorous large sample causal evidence
2
Other points
Meeting up: I am around until late Thursday so e-mail me if you
would like to talk for individually about work or topics
Lectures: I will post all the lectures on my website (teaching
page) on Thursday PM
http://www.stanford.edu/~nbloom/ (or Google Nick Bloom)
Breaks: I’ll take a 10 minute break at 10:30
Questions: Please feel free to ask any questions and/or make
comments. I’ve not prepared material assuming this
3
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Management practices across firms and countries
4. Explaining why management practices vary
5. The effect of management practices on performance
4
Large GDP & TFP differences across countries
Average US worker makes more in
2 weeks than a Tanzanian in 1 year
Source: Jones and Romer (2009). US=1
5
These TFP & GDP differences are often persistent
Source: Maddison (2008) Data is smoothed by decade
6
Productivity differences across firms are also large
• US plants at 90th percentile have 4x higher labor productivity
than plant at the 10th percentile (Syverson, 2004 REStat)
• Controlling for other inputs, TFP difference is about 2:1
• In China and India this gap is about 5:1 (Hsieh and Klenow,
2009 QJE)
• Not just mismeasured prices: in detailed industries with plant
level prices like white pan bread, block ice, concrete
productivity differences still 2:1 (Foster et al, 2008 AER)
7
TFP dispersion is particularly large in low
competition industries
Low competition
High competition
Source: Syverson (2004, JPE)
8
Productivity difference between firms in one obvious
motivation for looking at management
Persistent TFP differences a key part of many Macro, Trade, IO
and Labor models – but these are typically silent on the casues
Could this be in part because of differences in management –
even Adam Smith’s 1776 wealth of nations suggests this matters
Today will present a bunch of evidence showing that management
differences do seem to be a major factor driving TFP differences
Of course, might also be interesting in management for a range of
other reasons around growth, strategy and theory of the firm
9
Note, good productivity overview paper, Syversson
(2010, NBER WP)
Since completing your lecture outline noticed a very nice overview
of the productivity literature has been complied by Syversson
“What determines productivity”, NBER WP 15712 and forthcoming
in the Journal of Economic Literature
http://home.uchicago.edu/~syverson/productivitysurvey.pdf
10
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Management practices across firms and countries
4. Explaining why management practices vary
5. The effect of management practices on performance
11
Before discussing “management practices”, want to
point out that this is different from “managers”
There is also a large literature looking at CEOs (managers) – for
example Jack Welch, Bill Walsh or Alex Ferguson
Best known paper is Bertrand and Schoar (2003, QJE)
They build a panel dataset tracking managers across US firms
over time, and allow for firm and manager fixed effects
Focus on large US publicly traded firms – average of about 10,000
employees – so represents impact of strategy by the top manager
12
Summary of Bertrand and Schoar (2003)
Interesting results, and highly cited, finding:
1. Manager fixed effect exist, particularly for M&A, dividend policy,
debt ratios and cost-cutting
2. Managers have styles - more/less aggressive and
internal/external growth focus
3. Managers are also absolutely “better” or “worse” – performance
fixed effects exist, and linked to compensation and governance
4. These styles and fixed effects also correlated with manager
characteristics – particularly CEO age and having an MBA
13
Measuring management practices
Also a literature on management practices, which I will focus on in
these lectures, as these are more about firms than individuals
Historically been strongly case study based – e.g. Ford, GM,
Toyota, GE, Mayo Clinic, Citibank, Dabbawala etc.
Case-studies helpful for intuition and illustration, but potentially
misleading because very selected sample – e.g. Enron
More recently work has focused on trying to systematically
measure management practices in large samples of firms
• First generation, single country studies & direct questions
• Second generation, international studies & indirect questions
14
Challenges to measuring management practices
Despite sounding easy, “measuring management” is fraud with
difficulties, which has held back research.
1) How to quantify (as in put numbers on) management practices
2) How to get data from firms – surveys are tough to do
3) How to get the truth – will badly managed firms ‘fess-up’
4) Building a representative population – e.g. not just targeting
Compustat firms – especially important for cross-country work
15
First generation surveys: single-country focus with
direct survey techniques
Black and Lynch (2001, REStat) is a good example of a well
executed single country management survey
Surveyed about 3,000 establishments with the US Census bureau
1. Quantify: Asked a series of questions on employee recruitment,
work organization, meetings and modern production practices
2. Get data: Administered by the US Census Bureau
3. Truth: Told respondents their answers were confidential
4. Population: stratified from the Census establishment database
Found large variations in management, and strong correlation of
management practices and performance
16
Second wave surveys: cross countries and tries to
address response bias with indirect surveys
Cross country comparisons: identification of many factors driving
management typically require cross-country data
Problems with direct surveys: unfortunately people typically do not
tell the complete truth in open surveys:
•
•
•
Schwartz (1999, American Pschologist)
Opinion poll-evidence
Bertrand and Mullainathan (2001, AER P&P).
Bloom and Van Reenen (2007, QJE) is a good example of a
second wave of management survey, which I’ll cover in detail
17
The Bloom and Van Reenen (2007) approach
1) Quantifying: use scoring grid from a consulting firm
•Scores 18 monitoring, targets and incentives practices
•≈45 minute phone interview of manufacturing plant managers
2) Truth: use “Double-blind”
•Interviewers do not know the company’s performance
•Managers are not informed (in advance) they are scored
•All interviews run from a single location with rotation by country
3) Getting data: a variety of tricks
•Introduced as “Lean-manufacturing” interview, no financials
•Official Endorsement: Bundesbank, PBC, CII & RBI, etc.
•Run by 75 MBAs types (loud, assertive & business experience)
4) Population: sample randomly medium and large firms (100-5000
employees) from population databases across countries
18
Monitoring - i.e. “How is performance tracked?”
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
19
Survey Video
20
Getting representative cross country samples
•
So far interviewed about 7,000 firms across about 20 countries
• Obtained 45% coverage rate from sampling frame (with
response rates uncorrelated with performance measures)
•
Currently being extended to Charities, Hospitals, Law Firms,
Retail firms, PPPs, Schools and Tax Collection Agencies
• So basic concept easily transported across industries
21
Internal survey validation – useful exercise
suggesting double-blind methodology may work
2
3
Firm-level
correlation
of 0.627
1
2nd interview
4
5
Re-interviewed 222 firms with different interviewers & managers
Firm average scores (over 18 question)
1
2
3
management_2
1st interview
4
5
22
External survey validation – another useful exercise
suggesting double-blind methodology may work
Performance country c
measure
c
c
c
yi  MNGi  l li
management
(average z-scores)
c
 k ki
c
 mhi
c
  ' xi
c
 ui
ln(capital)
other controls
ln(labor)
ln(materials)
•
Use most recent cross-section of data (typically 2006)
•
Note – not a causal estimation, only an association
23
External validation: better performance is correlated
with better management
Dependent Productivity Profits 5yr Sales Share Price
variable
(% increase) (ROCE) growth
(Tobin Q)
Estimation
Firm sample
Management
Firms
Exit
OLS
OLS
OLS
OLS
Probit
All
All
All
Quoted
All
28.7***
2.018***
0.047***
0.250***
-0.262**
3469
1994
1883
374
3161
Includes controls for country, with results robust to controls for industry,
year, firm-size, firm-age, skills etc.
Significance levels: *** 1%, ** 5%, * 10%.
Sample of all firms where accounting data is available
24
External validation – robustness across countries
(the “ooh la la” question)
Performance results robust in all main regions:
•
Anglo-Saxon (US, UK, Ireland and Canada)
•
Northern Europe (France, Germany, Sweden & Poland)
•
Southern Europe (Portugal, Greece and Italy)
•
East Asia (China and Japan)
•
South America (Brazil)
25
0
10
20
30
Share of firms exporting
40
50
Consistent with Helpman, Melitz and Yeaple (2004,
AER) well managed firms also export more
1
1.5
2
2.5
3
3.5
4
4.5
Management score (rounded to nearest 0.5)
26
Well managed firms also more energy efficient
• Many US firms not operating using energy efficient
technology (De Cannio, 1993 EP ), due in large part to a mix of
management problems (Howarth, Haddad and Paton, 200 EP)
• We find similar results in our management data (below)
Energy use, log( KWH/$ sales)
1 point higher
management score
associated with about
20% less energy use
Management
27
Source: Bloom, Genakos, Martin and Sadun, (2010, EJ). Analysis uses Census of production data for UK firms
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Management practices across firms and countries
4. Explaining why management practices vary
5. The effect of management practices on performance
28
US management practices score highest on
average, with developing countries lowest
US
Germany
Sweden
Japan
Canada
France
Italy
Great Britain
Australia
Northern Ireland
Poland
Republic of Ireland
Portugal
Brazil
India
China
Greece
Also have data from Chile,
Mexico and New Zealand, but
not yet publicly released
2.6
2.8
3.2
3
mean of management
Average Country Management Score
3.4
29
Variation even greater across firms than across countries
Brazil
Canada
China
France
Germany
Great Britain
Greece
India
Ireland
Italy
Japan
Poland
Portugal
Sweden
US
0
.5
1
0
.5
1
0
.5
1
0
.5
1
Australia
1
2
3
4
5
1
2
3
4
5
1
2
3
4
5
Firm-Levelmanagement
Management Scores
1
2
3
30
4
5
Relative management practices also vary by country
Sweden
France
Australia
Italy
Portugal
Germany
Japan
Greece
Canada
Great Britain
Brazil
Northern Ireland
US
Republic of Ireland
China
Poland
India
Relatively better at
‘operations’
management
(monitoring, continuous
improvement, Lean etc)
Relatively better at
‘people’
management
(hiring, firing, pay,
promotions etc)
-.4
-.2
0
of peo_ops
(hiring,mean
firing,
pay &
.2
.4
People management
promotions) –
operations (monitoring, continuous improvement and31Lean)
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Management practices across firms and countries
4. Explaining why management practices vary
5. The effect of management practices on performance
32
So why does management vary across countries
and firms?
I will discuss five factors that seem important
• Competition
• Family firms
•
•
•
Multinationals
Labor market regulations
Education
But, before that, I that want to raise one informational constraint
for why every firm does not adopt best practices
33
Wanted to find out if firms were aware of their
practices being good/bad?
We asked:
“Excluding yourself, how well managed would you say your
firm is on a scale of 1 to 10, where 1 is worst practice, 5 is
average and 10 is best practice”
We also asked them to give themselves scores on operations
and people management separately
.3
.4
To the extent they are honest, most managers seem
to think they are well above average
“Average”
“Best
Practice”
0
.1
.2
“Worst
Practice”
0
2
4
6
8
Their self-score: 1 (worst practice), 5 (average) to 10 (best practice)
10
The Brazilians and Greeks overscored the most, the
US and French the least
Brazil
Greece
India
Portugal
China
Republic of Ireland
Northern Ireland
Australia
Canada
Italy
Great Britain
Poland
Germany
Japan
Sweden
France
US
0
.5
1
1.5
of gap
Self score (normalized to 1 to 5mean
scale)
– Management score
2
0
Correlation
0.032*
-6
-4
-2
Labor Productivity
labp
These self-scores also appear not only too high on
average, but also uncorrelated
with actual practices
Lowess smoother
0
2
4
6
8
Their self-score: 1 (worst practice), 5 (average) to 10 (best practice)
bandwidth = .8
10
Self scored management
* In comparison the management score has a 0.295 correlation with labor productivity
So seems many firms are unaware of their poor
management, consistent with a range of evidence
that management practices are a type of technology
A number of studies, including several I will discuss later on this
lecture, provide evidence that management is a technology
Innovations include the American System of Manufacturing,
Scientific Management, Mass Production, M-form firm, Quality
Movement and Lean
So one reason for bad management is like any other technology
there is a diffusion curve, with many firms below the curve
Even so, many well informed firms are badly managed, hence
why I will discuss a range of other factors
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
• Competition
• Family firms
• Multinationals
• Labor market regulations
• Education
4. The effect of management practices on performance
Tough competition appears strongly linked to better
management practices
Competition proxies
Import penetration
(SIC-3 industry, 1995-99)
Dependent variable:
Management
0.066**
(0.033)
“1-Rents” measure1
(SIC-3 except firm itself, 1995-99)
1.964**
(0.721)
# of competitors
(Firm level, 2004 and 2006)
0.158***
(0.023)
Observations
2499
2980
3589
Full controls2,3
Yes
Yes
Yes
= 1- (operating profit – capital costs)/sales
2 Includes 108 SIC-3 industry, country, firm-size, public and interview noise
(analyst, time, date, and manager characteristic) controls
3 S.E.s in ( ) below, robust to heteroskedasticity, clustered by country-industry
1 1-Rents
40
We also have some management panel data, and
find similar results
Competition proxies
Dependent variable: Change in
Management 2006-2004
Change in Import penetration
0.013**
(0.005)
1.006**
Change in “1-Rents” measure1
(0.415)
0.120**
Change in Number of rivals
(0.052)
Observations
421
404
432
= 1- (operating profit – capital costs)/sales
S.E.s in ( ) below, robust to heteroskedasticity, clustered by country-industry
1 1-Rents
UK, US, France and Germany only
41
Competition also appears linked to selection
An additional point on the management score is associated with
an increase of employment
US 715 more workers
UK 546 more workers
India 263 more workers
Competitive forces of reallocation weak in India compared to US
42
So appears to be a mix of ways competition can
improve management
“Incentives” (e.g. “Boot up the ass effect”) – competition forces
badly managed firms to improve performance
“Selection” – competition selects out badly managed firms
“Learning” – competition provides more firms in and industry,
increasing experimentation and learning.
43
Studies on TFP find similar results of competition
on performance
Syversson (2004, JPE) looks at the concrete industry and finds
that more competitive markets had higher average levels of TFP
and less dispersion.
Pavcnik (2002, REStud) and Olley-Pakes (1996, Econometrica)
also at changes in competition from trade-reforms and
deregulations respectively, finding this weeds out low TFP firms
Schmitz (2005, JPE) shows great lakes iron-producers responded
heavily to import competition
Nickell (1996, JPE) shows changes in competition lead to faster
TFP growth within a panel of firms
44
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
• Competition
• Family firms
• Multinationals
• Labor market regulations
• Education
4. The effect of management practices on performance
45
Management practices vary strongly with
ownership, even after controlling for industry,
country, skills, size etc.
Distribution of firm management scores by ownership. Overlaid dashed line is approximate density for
dispersed shareholders, the most common US ownership type
Family, external CEO
Family, family CEO
Founder
Government
Managers
Other
Private Equity
Private Individuals
0
.5
1
0
.5
1
0
.5
1
Dispersed Shareholders
1
2
3
4
5
1
2
3
4
5
1
Average Management Score
2
3
4
46
5
Ownership differences are another factor behind
cross-country variations in management practices
Sweden
Japan
US
France
Poland
Canada
Australia
Germany
China
Great Britain
Republic of Ireland
Northern Ireland
Italy
Brazil
Portugal
Greece
India
share family CEO (2nd+ generation)
share founder CEO (1st generation)
share government owned
0
.2
.4
.6
.8
share of ownership (for types associated with low management scores)
mean of family
47founder
mean of
Results again consistent with other studies using
other performance metrics
One nice study is by Perez-Gonzalez (2006, AER) looking at the
impact of a family CEO
• Finds stock prices fall the day a firm’s founder announces
they are passing the CEO position down to one of their kids
• Drops particularly for hand-downs to kids who went to nonselective schools
Another clever study by Bennedsen, Nielson, Perez-Gonalez,
and Wolfenzon (2007, QJE) on family firms and performance
• Use gender of the first born to instrument for family control
• Find family CEOs reduce profitability and growth rates
48
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
• Competition
• Family firms
• Multinationals
• Labor market regulations
• Education
4. The effect of management practices on performance
49
Multinationals appear to always be well managed,
consistent with selection and most trade models
Foreign multinationals
Domestic firms
US
Japan
Sweden
Germany
Canada
Australia
Italy
Great Britain
France
Poland
Northern Ireland
Republic of Ireland
India
China
Portugal
Brazil
Greece
2.4
2.6
2.8
3
3.2
Average Management Score
3.4
50
Multinational presence also linked to cross-country
differences in average management practices
Foreign multinationals
Domestic multinationals
India
Brazil
China
Greece
Japan
Poland
Italy
Northern Ireland
Republic of Ireland
Portugal
Canada
US
Great Britain
Australia
Germany
France
Sweden
0
.2
.4
.6
share of multinationals
.8
51
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
• Competition
• Family firms
• Multinationals
• Labor market regulations
• Education
4. The effect of management practices on performance
52
3.4
3.2
US
3
Canada
Germany
2.8
GreatJapan
Britain
Northern Ireland
Australia
Poland
Sweden
Republic of Ireland
France
Italy
2.6
India
China
Portugal
Brazil
Greece
2.4
Average incentives management
(hiring, firing, pay and promotions)
peop_mean
Labor market regulations appear linked to worse
management, particularly incentives management
0
20
40
WB_RigidityEmployment
World Bank Employment Rigidity Index
60
53
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
• Competition
• Family firms
• Multinationals
• Labor market regulations
• Education
4. The effect of management practices on performance
54
80
Finally, education is highly correlated with better
management, but no idea on causation
60
40
20
0
Percent with a degree
Non-managers
Managers
1
1.5
2
2.5
3
3.5
4
4.5
Management score (rounded to nearest 0.5)
mean of degree_nm
mean of degree_m
55
MY FAVOURITE QUOTES:
The traditional British Chat-Up
[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….”
MY FAVOURITE QUOTES:
The traditional Indian Chat-Up
Production Manager: “Are you a Brahmin?’
Interviewer “Yes, why do you ask?”
Production manager “And are you married?”
Interviewer “No?”
Production manager “Excellent, excellent, my son is looking
for a bride and I think you could be perfect. I must contact
your parents to discuss this”
MY FAVOURITE QUOTES:
The difficulties of defining ownership in Europe
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” ?”
Americans on geography
Interviewer: “How many production sites do you have abroad?
Manager in Indiana, US: “Well…we have one in Texas…”
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…]
References for the management data
The management data I showed you comes from two sources:
● Bloom, Nicholas, and John Van Reenen (2010) “Why do
management practices differ across firms and countries”, Journal
of Economic Perspectives, 24(1).
http://www.stanford.edu/~nbloom/JEP.pdf
● Bloom, Nicholas, Sadun, Raffaella and John Van Reenen
(2010), “Recent advances in the empirics of organizational
economics”, forthcoming Annual Review of Economics
http://www.stanford.edu/~nbloom/AR.pdf
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
4. The effect of management practices on performance
61
Estimating effect of management on performance
Is there really “bad” management, or are management
variations just response to different environments?
• For example, in Brazil does corruption make is better not to
monitor performance?
Management discipline is big on “contingent” management
(Woodward, 1958), while the Chicago school would claim bad
managed firms would be driven out of the market.
Three types of approaches to investigating this:
• Repeated arms-length surveys
• Longitudinal ground-based studies
• Experiments on management
62
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
4. The effect of management practices on performance
• Repeated arms-length surveys
• Longitudinal ground-based studies
• Experiments on management
63
Black and Lynch (2004) and Cappelli and Neumark
(2001) as good examples of panel survey-data
The survey data in Black and Lynch (2001, REStat) was actually
collected in two waves (1994 and 1997)
This was matched into performance panel data to create a
management and performance panel
Black & Lynch (2004, EJ) and Cappelli & Neumark (2001, ILRR)
run panel regressions of performance on management finding:
• Significant effects in the cross-sectional regressions
• Nothing when fixed effects are included
Appears to be because the Census management data is noisy
and management changes slowly (e.g. too little signal:noise)
64
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
4. The effect of management practices on performance
• Repeated arms-length surveys
• Longitudinal ground-based studies
• Experiments on management
65
Longitudinal ground based surveys
Classic paper is Ichniowski, Shaw and Prennushi (1997, AER)
(and more recently Bartel, Ichniowski and Shaw (2007, QJE))
Ichniowski et al. (1997) collected detailed monthly performance
and management data on 36 steel lines owned by 17 firms.
While the sample size is small, by looking at one very detailed
industry – steel finishing – can control for host of other factors
Also collected a many other control variables from frequent plant
visits - so it’s a mix of case-study and econometrics approaches
66
Summary Ichniowski, Shaw and Prennushi (1997),
slide (1/2)
Key findings:
1) Strong cross-sectional and panel correlation between
adoption of modern management practices and productivity
• Appears robust to controls for other factors like
rotation of individual managers to external threats
2) Clustering of management practices in firms, and empirical
results strongest for clusters rather than individual practices
• Both consistent with complementarity across practices
67
Summary Ichniowski, Shaw and Prennushi (1997),
slide (2/2)
Main issues:
1) Performance and management relationship not fully identified
– practices and performance could change in response to an
external shock. No good instrument, although plausible story
2) Clustering of management practices, and empirical results
being strongest for clusters, does not prove complementarity
• Clustering could equally reflect differences across firms
• Question level measurement error gives the results that
clusters are more significant than individual practices
Despite this hugely cited and shows that impact that a good
empirical management paper can have
68
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
4. The effect of management practices on performance
• Repeated arms-length surveys
• Longitudinal ground-based studies
• Experiments on management
69
Very recently, academics have started running
experiments on changing management practices
Running management experiments is expensive, so this is to
date this has been limited to:
•
Developing countries, typically on micro-enterprises (i.e. 5 to
10 person firms), or
•
Single firms (i.e. fruit-picking firms) in developed countries
70
Evidence from micro-enterprises in developed
countries (1/2)
A few projects are in progress (nothing published) - Karlan and
Valdivia (2010, R&R REStat) in Peru; Bruhn, Karlan and Schoar
in Mexico; Karlan and Udry in Ghana; McKenzie and Woodruff
in Sri Lanka
These provide a limited amount (≈50 hours) of basic trainings to
small firms – e.g. accounting, marketing, pricing, strategy etc.
This training is provided randomly and performance measured
before and after the intervention
71
Evidence from micro-enterprises in developed
countries (2/2)
Data so far extremely preliminary – my evidence on them is
mainly from discussing this directly with the authors.
Some studies find evidence of impact of management training
on performance, others do not (so far)
Maybe management does not matter? But I’ll present in detail
another study I’ve been involved in showing large effects
Or maybe these firms are very small, so management matters
less in small firms?
Or maybe to have an impact need extensive intervention – a few
hours of training not sufficient to have much effect?
72
Evidence from the single firms in developing
countries (1/2)
The team of Bandiera, Barankay and Rasul have produced an
impressive set of papers (2005, QJE; 2007, QJE; 2009,
Econometrica; and 2010, REStud)
Run experiments on incentives for workers and managers, team
selection and task division on a fruit picking farm
Typically introduce managerial changes part-way through
season to look at change in performance, plus use last seasons
output as seasonality controls
73
Evidence from the single firms in developing
countries (2/2)
Bandiera, Barakalay and Rasul find large effects of varying
management practices on fruit-picking performance:
• Worker incentive pay increases their performance,
especially absolute (rather than relative) incentives
• Peer monitoring effects are strong between workers when
absolute group incentives exist
• Manager incentive pay improves team selection (less
favoritism) and the effort they put into monitoring workers
74
Finally, a management experiment on large firms
The only experiment I know on panels of large firms is Bloom,
Eifert, Mahajan, McKenzie and Roberts (2010).
Randomize management practices delivered by Accenture to 20
plants in large (300 person) textile firms in Mumbai, India
Control firms get one month of diagnostic (≈ 200 hours of help)
and then monthly monitoring (≈ 20 hours a month).
Treatment firms get one month of diagnostic, four months of
intervention (≈ 800 hours of help) and then monthly monitoring.
Collect weekly data for all plants from 2008 to 2010
Before discussing the details show some pictures for context
75
Exhibit 1: Plants are large compounds, often containing several buildings.
Plant entrance with gates and a guard post
Plant surrounded by grounds
Front entrance to the main building
Plant buildings with gates and guard post
Exhibit 2: The plants operate 24 hours a day for 7 days a week
producing fabric from yarn, with 4 main stages of production
(1) Winding the yarn thread onto the warp beam
(2) Drawing the warp beam ready for weaving
(3) Weaving the fabric on the weaving loom
(4) Quality checking and repair
The production technology has not changed much over time
Krill
Warp
beam
The warping looms at Lowell Mills in 1854, Massachusetts
Exhibit 3: Many parts of these Indian plants were dirty and unsafe
Garbage outside the plant
Garbage inside a plant
Flammable garbage in a plant
Chemicals without any covering
Exhibit 4: The plant floors were disorganized
Instrument
not
removed
after use,
blocking
hallway.
Dirty and
poorly
maintained
machines
Old warp
beam, chairs
and a desk
obstructing the
plant floor
Tools left on
the floor
after use
Exhibit 5: The inventory rooms had months of excess yarn, often without
any formal storage system or protection from damp or crushing
Yarn without
labeling, order or
damp protection
Different types
and colors of
yarn lying mixed
Yarn piled up so high and
deep that access to back
sacks is almost impossible
A crushed yarn cone, which
is unusable as it leads to
irregular yarn tension
Exhibit 6: The spare parts stores were also disorganized and dirty
Spares without any labeling or order
No protection to prevent damage and rust
Spares without any labeling or order
Shelves overfilled and disorganized
Exhibit 7: The path for materials flow was often obstructed
Unfinished rough path along which several 0.6 ton
warp beams were taken on wheeled trolleys every day
to the elevator, which led down to the looms.
This steep slope, rough surface and sharp angle
meant workers often lost control of the trolleys. They
crashed into the iron beam or wall, breaking the
trolleys. So now each beam is carried by 6 men.
A broken trolley (the wheel snapped off)
At another plant both warp beam elevators had
broken down due to poor maintenance. As a result
teams of 7 men carried several warps beams down
the stairs every day. At 0.6 tons each this was slow
and dangerous - two serious accidents occurred in
our time at the plant.
Exhibit 8: Routine maintenance was usually not carried out, with repairs
only undertaken when breakdowns arose, leading to frequent stoppages.
Broken machine parts being repaired
Parts being cleaned and replaced on jammed loom
Workers investigating a broken loom
Loom parts being disassembled for diagnosis
1.5
These firms appear typical of large manufacturers in
India, China and Brazil
01
.5
1
Experimental Firms, mean=2.60
1
3
management
5
.8 .2
0
.4
.6
.8
Indian Textiles, mean=2.60
1
2
3
management
4
5
.8
0
.2
.4
.6
Indian Manufacturing, mean=2.69
1
2
3
management
4
5
.2
.4
.6
Brazil and China Manufacturing, mean=2.67
0
85
1
2
3
management
4
Management scores (using Bloom and Van Reenen (2007) methodology)
5
Sample of firms we worked with
86
Intervention aimed to improve 38 core textile
management practices in 6 areas (1/2)
Targeted
practices in 6
areas:
operations,
quality,
inventory,
loom planning,
HR and sales
& orders
87
Intervention aimed to improve 38 core textile
management practices in 6 areas (2/2)
Targeted
practices in 6
areas:
operations,
quality,
inventory,
loom planning,
HR and sales
& orders
88
Share of the 38 management practices adopted
.2
.3
.4
.6
.5
Adoption of these 38 management practices did
rise, and particularly in the treatment plants
Wave 1 treatment plants:
Diagnostic September
2008, implementation
began October 2008
Wave 2 treatment plants:
Diagnostic April 2009,
implementation began
May 2008
Control plants:
Diagnostic July 2009
Non-experiment plants:
No intervention
2008.25
April 2008
2008.5
July 2008
2008.75
October 2008
2009
ym
January 2009
April 2009
2009.25
2009.5
July 2009
2009.75
October 2009
89 practices over
Notes: Non-experiment plants are other plants in the treatment firms not involved in the experiment. They improved
this period because the firm internally copied these over themselves. All initial differences not statistically significant (Table 2)
Adoption of these 38 management practices led to
clear improvements in 3 areas of these India firms
• Quality
• Inventory
• Operational efficiency
90
Exhibit 10: Quality was so poor that 19% of manpower was spent on
repairing defects at the end of the production process
Large room full of repair workers (the day shift)
Workers spread cloth over lighted plates to spot defects
91
Defects are repaired by hand or cut out from cloth
Non-fixable defects lead to discounts of up to 75%
Previously mending was recorded only to crosscheck against customers’ claims for rebates
Defects log with
defects not
recorded in an
standardized
format. These
defects were
recorded solely
as a record in
case of
customer
complaints. The
data was not
aggregated or
analyzed
92
Now mending is recorded daily in a standard format,
so it can analyzed by loom, shift, design & weaver
93
93
The quality data is now collated and analyzed as
part of the new daily production meetings
Plant managers now meet
regularly with heads of
quality, inventory, weaving,
maintenance, warping etc.
to analyze data
94
100
120
140
quality)
score=lower
(higher
index
Quality defects
40
60
80
Figure 3: Quality defects index for the treatment and control plants
Start of Diagnostic
Start of Implementation
Spline, 95% upper CI
Control plants
Data (+ symbol)
Cubic Spline
Spline, 95% lower CI
Treatment plants
Spline, 95% upper CI
Cubic Spline
Data (♦ symbol)
Spline, 95% lower CI
-10
-5
0
5
10
15
20
timing
Weeks after the
start of the intervention
Notes: Displays the average quality defects index, which is a weighted index of quality defects, so a higher score means lower quality. This is
plotted for the 14 treatment plants (♦ symbols) and the 6 control plants (+ symbols). Values normalized so both series have an average of 100
prior to the start of the intervention. “Data” is plotted using a 5 week moving average. To obtain series (rather than point-wise) confidence
intervals we used a cubic-spline with one knot at the start of the implementation period. The spline estimate is labeled (“Cubic Spine”), the 95%
confidence upper and lower intervals labeled (“Spline, 95% upper CI”) and (“Spline, 95% lower CI”) from plant-wise block boostrap. Timing
based on weeks after the intervention (positive values) or before the intervention (negative values). For wave 1 treatment plants this is relative to
95Diwali and Ede
September 1st 2008, for Wave 2 treatment and control firms April 7th 2009. The control group’s rise in weeks 10+ are due to the pre
production increase, which usually leads to a deterioration in quality due to increased speeds of production.
Management impact on quality, regressions
96
Adoption of these 38 management practices led to
clear improvements in 3 areas of these India firms
• Quality
• Inventory
• Operational efficiency
97
Organizing and racking inventory enables firms to
reduce capital stock and reduces waste
Stock is organized,
labeled, and entered
into an Electronic
Resource Planning
(ERP) system which
has details of the type,
age and location.
Bagging and racking
yarn reduces waste
from rotting (keeps the
yarn dry) and crushing
Computerized
inventory systems help
to reduce stock levels.
98
Sales are also informed about excess yarn stock so
they can incorporate this in new designs.
Shade cards now
produced for all
surplus yarn. These
are sent to the
design team to use
in future designs
99
And yarn for products ranges no longer made by
the firm (e.g. suiting fabric) was sold
This firms
used to make
suiting and
shirting yarn,
but stopped
making
suiting yarn 2
years ago
100
Management impact on inventory, regressions
101
Adoption of these 38 management practices led to
clear improvements in 3 areas of these India firms
• Quality
• Inventory
• Operational efficiency
102
The treated firms have also started to introduce
basic initiatives (called “5S”) to organize the plant
Worker involved in 5S initiative on
the shop floor, marking out the area
around the model machine
Snag tagging to identify the abnormalities
on & around the machines, such as
redundant materials, broken equipment, or
accident areas. The operator and the
maintenance team is responsible for
removing these abnormalities.
This is all part of the routine maintenance
103
Spare parts were also organized, reducing downtime
(parts can be found quickly), capital stock and waste
Nuts & bolts
sorted as per
specifications
Tool
storage
organized
Parts like
gears,
bushes,
sorted as per
specifications
104
Production data is now collected in a standardized
format, for discussion in the daily meetings
Before
(not standardized, on loose
pieces of paper)
After
(standardized, so easy to enter
105
daily into a computer)
Daily performance boards have also been put up,
with incentive pay for employees based on this
106
Management impact on efficiency, regressions
107
Estimated impacts on productivity and
profitability are large and rising
Estimate the intervention has increase profits by about
$250,00 per firm and productivity by 9% so far from:
- reduced repair manpower costs
- reduced wasted materials (from less defects)
- lower inventory
- higher efficiency levels
Full impacts of better management should be much larger:
- short-run impacts only
- narrow set of management practices (almost no HR)
108
So why did these firms have bad management?
Asked the consultants to investigate the non-adoption of each
practice in each firm every other month
They did this by discussion with the owners, observation of
factory, and from their experiences of changing practices
To collect this information systematically we developed a
“management non-adoption” flow-chart
109
“Non adoption flow chart” used to collect data
Legend
Is the reason for the non adoption
of the practice internal to the firm?
No
External factors (legal, climate etc)
Hypothesis
Yes
Conclusion
Was the firm previously aware
that the practice existed?
Lack of information
No
Limited incentives and/or
authority for employees
Can the firm adopt the practice with
existing staff & equipment?
Could the firm hire new
employees or consultants
to adopt the practice?
Lack of local skills
Did the CEO believe introducing the
practice would be profitable?
Would this adoption be
profitable
Not profit maximizing
Could the CEO get his employees to
introduce the practice?
Low ability of the owner
and/or procrastination
Do you think the CEO was correct
about the cost-benefit tradeoff?
Does the firm have enough internal
financing or access to credit?
Did the firm
realize this
would be
profitable?
Other reasons
Credit constraints
Incorrect information
110
Non adoption data in treatment firms
1) Lack of, or incorrect, information was the major reason. Firms
not aware of, or incorrectly evaluated, modern practices.
2) Took time to change incorrect information, as initial trust in
consultants was limited, but grew as their advice worked.
3) Blockages later arose with the owners ability or procrastination
4) Ongoing issues with managerial incentives – currently trying to
introduce management incentive systems to address this
111
Why does competition not fix badly managed firms?
Bankruptcy is not (currently) a threat: a weaver wage rates of $5
a day, so these firms can be profitable with bad management.
Reallocation appears limited: Owners take all decisions as they
worry about managers stealing. But owners time is constrained –
they already work 66.2 hours average a week – limiting growth.
As an illustration firm size is more linked to number of male
family members (corr=0.689) - who are trusted to be given
managerial positions - than management scores (corr=0.223)
Entry appears limited: Capital intensive ($13m assets average
per firm), and no guarantee new entrants are any better.
112
Lecture 1: Overview
1. Motivation: productivity across firms and countries
2. Measuring management practices
3. Explaining why management practices vary
4. The effect of management practices on performance
5. Conclusions
113
Summary
• Management practices do vary widely across firms and
countries, much like productivity
• Factors associated with good management are competition,
professional ownership (not families or Government),
international exposure, light regulations & education
• There is “good” and “bad” management, in that monitoring,
targets and incentives certain practices appear to causally
improve performance
• Lastly, change appears slow to tough with many badly run firms.
Suggests good management is a type of technology, with
informational, incentive and behavorial barriers to adoption
114
My five outstanding research questions
The key questions which I think remain substantially unanswered
1. What fraction of the differences in TFP across firms and
countries can management causally explain?
2. What are the key factors causing difference in management?
3. Why do management practices take so long to change?
4. Are different management practices complementary, or are
their impacts more or less additive?
5. What broad types of management practices are universally
good and what types of contingent on firm’s environment
115
Backup slides
116
Summary reading list, plus papers cited in slides
9:00-12:00 Tuesday February 9th:
Readings by category: (* are core readings)
• Bertrand, Marianne, and Antoinette Schoar. 2003. “Managing with Style: The Effect of Managers on Firm Policies.” Quarterly
Journal of Economics, 118 (4): 1169-1208.
• * Ichniowski, Casey, Kathryn Shaw and Giovanna Prenushi. (1997), “The Effects of Human Resource Management: A Study
of Steel Finishing Lines”, American Economic Review, LXXXVII (3), 291-313.
• Black, Sandra, and Lisa Lynch. 2001. “How to Compete: The Impact of Workplace Practices and Information Technology on
Productivity.” Review of Economics and Statistics, 88(3): 434-45.
• * Bloom, Nicholas, and John Van Reenen (2007) “Measuring and Explaining Management Practices across Firms and
Countries”, Quarterly Journal of Economics, 122(4), 1341-1408.
• Burstein, Ariel, and Alexander Monge-Naranjo. 2009. “Foreign Know-how, Firm Control, and the Income of Developing
Countries.” Quarterly Journal of Economics, 124(1): 149-195.
• Bloom, Nicholas, and John Van Reenen (2010) “Why do management practices differ across firms and countries”, Journal of
Economic Perspectives, 24(1).
• Syverson Chad. 2004b. “Market Structure and productivity: A Concrete Example.” Journal of Political Economy, 112(6): 11811222.
• * (read the empirical part only) Foster Lucia, John Haltiwanger, and Chad Syverson. 2009. “Reallocation, Firm Turnover and
Efficiency: Selection on Productivity or Profitability.” American Economic Review, 98(1): 394-425.
• Hall, Robert, and Charles Jones. 1999. “Why Do Some Countries Produce So Much More Output Per Worker Than Others?”
Quarterly Journal of Economics, 114(1): 83-116.
• Hsieh, Chiang-Tai, and Pete Klenow. 2009. “Misallocation and Manufacturing TFP in China and India.” Quarterly Journal of
Economics,
• * Bloom, Nicholas, Eifert, Benn, Mahajan, Aprajit, McKenzie, David and John Robert. (2010), “Management matters:
evidence from Indian firms”, Stanford mimeo
• Bloom, Nicholas, Mahajan, Aprajit, McKenzie, David and John Robert. (2010), “Why do firms in developing countries have
low productivity”, forthcoming American Economic Review papers and proceedings
117
Summary reading list, plus papers cited in slides
9:00-12:00 Thursday February 11th: (* are core readings)
• * Rajan, Raghuram, and Julie Wulf (2006) “The Flattening Firm: Evidence from Panel Data on the Changing Nature of
Corporate Hierarchies”, Review of Economics and Statistics, 88(4), 759-773.
• Acemoglu Daron, Philippe Aghion, Claire Lelarge, John Van Reenen, and Fabrizio Zilibotti (2007) “Technology, Information
and the Decentralization of the Firm”, Quarterly Journal of Economics, 122(4), 1759–1799.
• * Bloom, Nicholas, Sadun, Raffaella and John Van Reenen. (2009). “The organization of firms across countries”, 2009,
NBER WP15129.
• Bloom, Nicholas, Sadun, Raffaella and John Van Reenen. (2010). “Does product market competition lead firms to
decentralize?”, forthcoming American Economic Review papers and proceedings
Organizational structures, IT and performance:
• * Bresnahan, Brynjolfsson and Hitt (2002, QJE) Bresnahan, Timothy, Erik Brynjolfsson, and Lorin Hitt (2002) “Information
Technology, Workplace Organization and the Demand for Skilled Labor: Firm-level Evidence”, Quarterly Journal of
Economics, 117(1), 339-376.
• Baker, George, and Thomas Hubbard (2003) “Make Versus Buy in Trucking: Asset Ownership, Job Design and Information”,
American Economic Review, 93(3), 551-572.
• Bloom, Nicholas, Raffaella Sadun, and John Van Reenen (2007) “Americans do IT Better: American Multinationals and the
Productivity Miracle”, NBER WP 13085, and revise and resubmit for the American Economic Review.
118
Lazear (2000, AER) study on Safelite glass
Another classic, which studies the introduction of one type of
management practice – piece-rate pay – on performance
The setting is Safelite Glass, who replace car windscreens, who
rolled out a switch from flat to piece-rate across regions.
Examines performance-data for 19 months before and after the
switch from hourly rates to piece-rate and finds:
• Increase in productivity of 44%
• Combination of selection and effort effects
119