Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

51
Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning
  • date post

    23-Jan-2016
  • Category

    Documents

  • view

    224
  • download

    0

Transcript of Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Page 1: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

LABOR TOPICS

Nick Bloom

Learning

Page 2: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Technologies – like pineapples - are not used by everyone. Question is why?

Suri (2011, forthcoming Econometrica)

Page 3: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

A few classic learning papers

A learning related paper I know well…

Page 4: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Conley and Udry (2008) is based around a learning story, with some key points

• Learning appears to happen slowly over time – pineapple does not immediately spread to every farmer in every village

• Information spreads best through friends and close contacts, suggesting people do not trust all information equally

• Spread also depends on success of trusted contacts, suggesting process of discovery – not everything known at t=0

Page 5: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

The original classic – Griliches (1957) – also focused on learning and discovery

Page 6: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

• Hybrid seen corn is a way of developing appropriate corn for different growing conditions – breeding is done for each area

• A single impactful technology that spread slowly across the US

• So Griliches splits adoption delays into– The “acceptance” problem (the lag in uptake by farmers) which is learning within markets

– The “availability” problem (breeding appropriateseed corn by market) which is discovery acrossmarkets, driven by profits

The original classic – Griliches (1957) shows gradual learning about hybrid seed corn

Page 7: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Duflo, Kremer and Robinson (2010) suggest other non-learning stories• Experiment on fertilizer use in Kenya where returns to fertilizer is about 50% to 100% per year – so a highly profitable investment

• Despite this farmers do not take up fertilizer, and this is despite being a well known effective technology (i.e. not learning issues)

• They has a model around hyperbolic discounting, and show in experiments with pre-commitment get large (profitable) uptake

– Discount at harvest (rather than planting) time increases adoption by 17%, equivalent to at 50% subsidy

• Interestingly, these are not persistent – it appears to be a commitment issue rather than a learning story

Page 8: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Suri (2010) suggests a heterogeneity interpretation instead• Looks at hybrid maize adoption in Kenya over 1996-2004

• Stable rates of adoption and 30% of households switch (upside of using panel data, which Besley and Case 1993 also push)

• Find heterogeneity in costs and returns explains apparent adoption paradox, in particular three groups of households:

– Small group very high returns, but blocked by distance to seed/fertilizer distributors– Larger group of adopters with high returns– Larger group of switchers that have about zero returns

Page 9: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 9

A few classic learning papers

A learning related paper I know well…

Page 10: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Does management matter: evidence from India

Nick Bloom (Stanford)Benn Eifert (Berkeley)Aprajit Mahajan (Stanford)David McKenzie (World Bank)John Roberts (Stanford)

NBER WP16658

Page 11: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

2.6 2.8 3 3.2 3.4mean of management

USJapan

GermanySwedenCanada

ItalyFrance

Great BritainAustralia

New ZealandPolandIreland

PortugalChile

MexicoGreece

BrazilChina

ArgentinaIndia

11

Management appears to be better in rich countries

Average country management score, manufacturing firms 100 to 5000 employees(monitoring, targets and incentives management scored on a 1 to 5 scale)

Source: Bloom, Sadun and Van Reenen (2010, Annual Review)

Page 12: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 12

0.2

.4

.6

.8

De

nsity

1 2 3 4 5management

0.2

.4

.6

.8

De

nsity

1 2 3 4 5management

US, manufacturing, mean=3.33 (N=695)

India, manufacturing, mean=2.69 (N=620)

De

nsi

tyD

en

sity

Firm level management score, manufacturing firms 100 to 5000 employeesSource: Bloom and Van Reenen (2010, JEP)

Developing countries have more badly managed firms

Page 13: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

But do we care - does management matter?

• Long debate between business practitioners versus academics

• Evidence to date primarily case-studies and surveys. In fact Syverson’s (2010) productivity survey stated on management

“Perhaps no potential driver of productivity differences has seen a higher ratio of speculation to actual empirical study than management”

• So in India we ran a management field experiment

13

Page 14: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Investigate in large Indian firms

14

Took large firms (≈ 300 employees) outside Mumbai making cotton fabric. Randomized treatment plants get 5 months management consulting, controls plants get 1 month consulting.

Collect weekly data on all plants from 2008 to 2010

1) Management ‘improves’

2) Productivity and profits up by about 10% to 20%

3) Decentralization of decision making within firms

4) Increased computerization

Page 15: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Exhibit 1: Plants are large compounds, often containing several buildings.

More photos and some basic video footage on http://worldmanagementsurvey.org/

Page 16: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Exhibit 2a: Plants operate continuously making cotton fabric from yarn

Fabric warping

Page 17: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 Fabric weaving

Exhibit 2b: Plants operate continuously making cotton fabric from yarn

Page 18: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011Quality checking

Exhibit 2c: Plants operate continuously making cotton fabric from yarn

Page 19: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Exhibit 3: Many parts of these Indian plants were dirty and unsafe

Garbage outside the plant Garbage inside a plant

Chemicals without any coveringFlammable garbage in a plant

Page 20: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Exhibit 4: The plant floors were often disorganized and aisles blocked

Page 21: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Exhibit 5: There was almost no routine maintenance – instead machines were only repaired when they broke down

Page 22: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Exhibit 6a: Inventory was not well controlled – firms had months of excess yarn, typically stored in an ad hoc way all over the factory

Page 23: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Exhibit 6b: Inventory was not well controlled – firms had months of excess yarn, typically stored in an ad hoc way all over the factory

Page 24: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 24

Management practices before and after treatment

Performance of the plants before and after treatment

Why were these practices not introduced before?

Decentralization and IT

Page 25: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Intervention aimed to improve core textile management practices in 6 areas – e.g.

25

Page 26: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Treatment plants

Control plants

Sha

re o

f key

text

ile m

anag

emen

t pra

ctic

es a

dopt

ed

Excluded plants(not treatment or control)

Adoption of these 38 management practices did rise, and particularly in the treatment plants

.2.3

.4.5

.6

-10 -8 -6 -4 -2 0 2 4 6 8 10 12Months after the diagnostic phase

Treated

Control

Page 27: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Management practices before and after treatment

Performance of the plants before and after treatment

• Quality

• Inventory

• Output

Why were these practices not introduced before?

Decentralization and IT

27

Page 28: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Poor quality meant 19% of manpower went on repairs

Workers spread cloth over lighted plates to spot defectsLarge room full of repair workers (the day shift)

Defects lead to about 5% of cloth being scrappedDefects are repaired by hand or cut out from cloth

Page 29: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Previously mending was recorded only to cross-check against customers’ complaints

29

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

Page 30: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Now mending is recorded daily in a standard format, for analysing by loom, shift, & weaver

3030

Page 31: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 31

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

Page 32: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

02

04

06

08

01

00

12

01

40

-20 -10 0 10 20 30 40weeks since diagnostic phase

2.5th percentile

Figure 3: Quality defects index for the treatment and control plants

Control plants

Treatment plants

Weeks after the start of the diagnostic

Qua

lity

defe

cts

inde

x (h

ighe

r sc

ore

=lo

we

r qu

ality

)

Start of Diagnostic

Start of Implementation

Average (+ symbol)

97.5th percentile

Average (♦ symbol)

97.5th percentile

End of Implementation

2.5th percentile

Page 33: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Differences are not driven by one firm

QDI fell in every treatment firm by at least 10%.

02

46

8

-1 -.5 0 .5 1 -1 -.5 0 .5 1

De

nsi

ty

Before/after difference in log(QDI)

TreatmentControl

Page 34: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Can also run weekly performance regressions

Instrument “Management” with log(1+weeks of consulting)

Calculate standard errors using clustered bootstrap, and also using small-sample permutation and t-asymptotic tests

34

OUTCOMEi,t = αi + βt + θMANAGEMENTi,t+νi,t

Page 35: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Quality (a Quality Defects Index)

Note: standard errors bootstrap clustered by firm. Instrument in second column in log(1+weeks treatment). ITT is intention to treat and regresses log(QDI) on a 0/1 indicator for treatment. IV instruments management with log (1+weeks of consulting)

OLS IV ITTManagementi,t -0.561 -2.028***

Interventioni,t -0.386**

Small sample robustness

Ibragimov-Mueller (95% CI) (-4.46,-0.53) (-5.03,-0.98) (-0.69,-0.38) (90%CI) (-4.09,-0.90) (-4.65,-1.36) (-0.66,-0.41)Permutation Test I (p-value) 0.02IV Permutation Tests (95% CI) (-6.05,0.35) (90% CI) (-6.00,-0.03)Time FEs 113 113 113Plant FEs 20 20 20Observations 1732 1732 1732

Quality (log QDI)

Page 36: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 36

Management practices before and after treatment

Performance of the plants before and after treatment

• Quality

• Inventory

• Output

Why were these practices not introduced before?

Decentralization and IT

Page 37: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 37

Organizing and racking inventory enables firms to slowly reduce their capital stock

Page 38: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

60

80

10

01

20

-20 -10 0 10 20 30 40 50weeks since diagnostic phase

2.5th percentile

Figure 4: Yarn inventory for the treatment and control plants

Control plants

Treatment plants

Weeks after the start of the intervention

Ya

rn in

ven

tory

(n

orm

aliz

ed

to

10

0 p

rior

to d

iag

no

stic

)

Start of Diagnostic

Start of Implementation

Average (+ symbol)

97.5th percentile

Average (♦ symbol)

2.5th percentile

97.5th percentile

End of Implementation

Page 39: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 39

Many treated firms have also introduced basic initiatives (called “5S”) to organize the plant floor

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.

Page 40: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 40

Spare parts were also organized, reducing downtime (parts can be found quickly) and waste

Nuts & bolts sorted as per specifications

Tool

storage organized

Parts like gears,

bushes, sorted as per specifications

Page 41: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 41

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

daily into a computer)

Page 42: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Daily performance boards have also been put up, with incentive pay for employees based on this

42

Page 43: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

70

80

90

10

01

10

12

01

30

-20 -10 0 10 20 30 40 50weeks since diagnostic phase

2.5th percentile

Figure 5: Output for the treatment and control plants

Control plants

Treatment plants

Weeks after the start of the intervention

Start of Diagnostic

Start of Implementation

Average (+ symbol)

97.5th percentile

Average (♦ symbol)

2.5th percentile

97.5th percentile

End of Implementation

Ou

tpu

t (n

orm

aliz

ed

to

10

0 p

rior

to d

iag

no

stic

)

Page 44: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 44

Management practices before and after treatment

Performance of the plants before and after treatment

Decentralization and IT

Why were these practices not introduced before?

Page 45: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Better management improved information flow enabling owners to trust managers more

45

The India firms hierarchical: owners take all major decisions

Reason is owners fear theft by managers:- punishment is limited (Indian courts are ineffective)- risk of getting caught is limited (little information to monitor)

Better management, increases information, so better monitoring

So owners delegate more: visit factories less, take less decisions

Page 46: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Better management led to decentralization in firms

Decentralization index is the principal component factor of 7 measures of decentralization around weaver hiring, manager hiring, spares purchases, maintenance planning, weaver bonuses, investment, and departmental co-ordination.

Page 47: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Better management also increased computerization

(pre-experiment mean=10)

Computerization index is the principal component factor of 10 measures around computerization, which are the use of an ERP system, the number of computers in the plant, the number of computers less than 2 years old, the number of employees using computers for at least 10 minutes per day, and the cumulative number of hours of computer use per week, an internet connection at the plant, if the plant-manager uses e-mail, if the directors use of e-mail, and the intensity of computerization in production.

Page 48: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 48

Management practices before and after treatment

Performance of the plants before and after treatment

Decentralization and IT

Why were these practices not introduced before?

Page 49: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

Why does competition not fix bad management?

49

Bankruptcy is not (currently) a threat: at weaver wage rates of $5 a day these firms are profitable

Reallocation appears limited: Owners take all decisions as they worry about managers stealing. But owners time is constrained – they already work 72.4 hours average a week – limiting growth.

Entry is limited: Capital intensive ($13m assets average per firm), and no guarantee new entrants are any better

Page 50: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011 50

So why did these firms not improve themselves – limited information/learning

Collected panel data on reasons for non implementation, and main (initial) reason was a lack of information• Firms either never heard of these practices (no information)• Or, did not believe they were relevant (wrong information)

Later constraints after informational barriers overcome primarily around limited CEO time and CEO ability

Page 51: Nick Bloom, Labor Topics 247, 2011 LABOR TOPICS Nick Bloom Learning.

Nick Bloom, Labor Topics 247, 2011

.2.3

.4.5

.6

-10 -8 -6 -4 -2 0 2 4 6 8 10 12Months after the diagnostic phase

Treatment plants (on-site) Control plants

(on-site)

Sha

re o

f key

text

ile m

anag

emen

t pra

ctic

es a

dopt

ed

Excluded plants in treatment firms

Adoption of these management practices was spread by firms to non-experimental plants: learning