Why Do Managment

7/29/2019 Why Do Managment

1/22

Journal o Economic PerspectivesVolume 24, Number 1Winter 2010Pages 203224

EE conomists have long puzzled over the astounding dierences in productiv-conomists have long puzzled over the astounding dierences in productiv-ity between frms and countries. For example, looking at disaggregatedity between frms and countries. For example, looking at disaggregateddata on U.S. manuacturing industries, Syverson (2004a) ound that plantsdata on U.S. manuacturing industries, Syverson (2004a) ound that plantsat the 90at the 90thth percentile produced our times as much as the plant in the 10percentile produced our times as much as the plant in the 10thth percentilepercentileon a per-employee basis. Only hal o this dierence in labor productivity couldon a per-employee basis. Only hal o this dierence in labor productivity couldbe accounted or by dierential inputs, such as capital intensity. Syverson lookedbe accounted or by dierential inputs, such as capital intensity. Syverson lookedat industries defned at the our-digit level in the Standard Industrial Classifcationat industries defned at the our-digit level in the Standard Industrial Classifcation(SIC) system (now the North American Industry Classifcation System or NAICS) like(SIC) system (now the North American Industry Classifcation System or NAICS) likeBakeries and Tortilla Manuacturing or Plastics Product Manuacturing. Foster,Bakeries and Tortilla Manuacturing or Plastics Product Manuacturing. Foster,Haltiwanger, and Syverson (2008) show large dierences in total actor productivityHaltiwanger, and Syverson (2008) show large dierences in total actor productivityeven within very homogeneous goods industries such as boxes and block ice. Someeven within very homogeneous goods industries such as boxes and block ice. Someo these productivity dierences across frms and plants are temporary, but in largeo these productivity dierences across frms and plants are temporary, but in largepart they persist over time. At the country level, Hall and Jones (1999) and Jonespart they persist over time. At the country level, Hall and Jones (1999) and Jonesand Romer (2009) show how the stark dierences in productivity across countriesand Romer (2009) show how the stark dierences in productivity across countriesaccount or a substantial raction o the dierences in average per capita income.account or a substantial raction o the dierences in average per capita income.Both at the plant level and at the national level, dierences in productivity areBoth at the plant level and at the national level, dierences in productivity aretypically calculated as a residualthat is, productivity is inerred as the gap betweentypically calculated as a residualthat is, productivity is inerred as the gap betweenoutput and inputs that cannot be accounted or by conventionally measured inputs.output and inputs that cannot be accounted or by conventionally measured inputs.

Why Do Management Practices Dieracross Firms and Countries?

Nicholas Bloom is an Associate Professor of Economics, Stanford University, Stanford, Cali-Nicholas Bloom is an Associate Proessor o Economics, Stanord University, Stanord, Cali-fornia. John Van Reenen is a Professor of Economics, London School of Economics, London,ornia. John Van Reenen is a Proessor o Economics, London School o Economics, London,United Kingdom. Bloom is an International Research Fellow and Van Reenen is Director ofUnited Kingdom. Bloom is an International Research Fellow and Van Reenen is Director othe Center for Economic Performance, London School of Economics, London, United Kingdom.the Center or Economic Perormance, London School o Economics, London, United Kingdom.Both authors are also Faculty Research Fellows, National Bureau of Economic Research, Cam-Both authors are also Faculty Research Fellows, National Bureau o Economic Research, Cam-bridge, Massachusetts. Van Reenen is Research Fellow in the Centre for Economic Policy, London,bridge, Massachusetts. Van Reenen is Research Fellow in the Centre or Economic Policy, London,United Kingdom. Their e-mail addresses areUnited Kingdom. Their e-mail addresses [email protected]@stanord.edu [email protected]@lse.ac.uk..doi=10.1257/jep.24.1.203

Nicholas Bloom and John Van Reenen


2/22

204 Journal o Economic Perspectives

For this reason, Abramovitz (1956) labeled total actor productivity at the countryFor this reason, Abramovitz (1956) labeled total actor productivity at the countrylevel a measure o our ignorance. Productivity dierences at the frm level havelevel a measure o our ignorance. Productivity dierences at the frm level havelong been a measure o our ignorance, too. For example, one potential hypothesislong been a measure o our ignorance, too. For example, one potential hypothesishas been that persistent productivity dierentials are due to hard technologicalhas been that persistent productivity dierentials are due to hard technologicalinnovations as embodied in patents or adoption o new machinery. Although thereinnovations as embodied in patents or adoption o new machinery. Although therehas been substantial progress in improving our measures o technology, therehas been substantial progress in improving our measures o technology, thereremain substantial productivity dierences even ater controlling or such actors.remain substantial productivity dierences even ater controlling or such actors.In this paper, we present evidence on another possible explanation or per-In this paper, we present evidence on another possible explanation or per-sistent dierences in productivity at the frm and the national levelnamely, thatsistent dierences in productivity at the frm and the national levelnamely, thatsuch dierences largely reect variations in management practices. As two British-such dierences largely reect variations in management practices. As two British-born academics, we are accustomed to reports that blame Britains relatively lowborn academics, we are accustomed to reports that blame Britains relatively lowproductivity on bad management. Indeed, this view is so common in the Unitedproductivity on bad management. Indeed, this view is so common in the UnitedKingdom that it has generated a vibrant export industry o television shows onKingdom that it has generated a vibrant export industry o television shows onbad management, in wholesale (bad management, in wholesale (The Ofce), private services (), private services (Fawlty Towers), and the), and thepublic sector (public sector (Yes, Minister). Now that). Now thatThe Ofcehas been so successully importedhas been so successully importedinto the United States, this raises the question, is Michael Scott (the inamously badinto the United States, this raises the question, is Michael Scott (the inamously badAmerican manager in the show) representative o U.S. frms?American manager in the show) representative o U.S. frms?But while ascribing dierences in productivity to management practices has longBut while ascribing dierences in productivity to management practices has longbeen popular or television shows, business schools, and policymakers, it has been lessbeen popular or television shows, business schools, and policymakers, it has been lesspopular among economists or two broad reasons. First, much o the managementpopular among economists or two broad reasons. First, much o the managementliterature is based on case studies, rather than on systematic empirical data acrossliterature is based on case studies, rather than on systematic empirical data acrossfrms and countries. To tackle this problem, we have, over the last decade, undertakenfrms and countries. To tackle this problem, we have, over the last decade, undertakena large survey research program to measure management practices systematicallya large survey research program to measure management practices systematicallyacross frms, industries, and countries. We begin by describing our survey approach,across frms, industries, and countries. We begin by describing our survey approach,which ocuses on aspects o management like systematic perormance monitoring,which ocuses on aspects o management like systematic perormance monitoring,setting appropriate targets, and providing incentives or good perormance.setting appropriate targets, and providing incentives or good perormance.A second reason that economists have tended to shy away rom manage-A second reason that economists have tended to shy away rom manage-ment-based explanations or productivity dierences is a sense that changingment-based explanations or productivity dierences is a sense that changingmanagement seems a relatively straightorward process. To be sure, theremanagement seems a relatively straightorward process. To be sure, thereare always adjustment costs and agency costs, but i the potential gains romare always adjustment costs and agency costs, but i the potential gains romimproved management are indeed sizeable, it seems as i such barriers shouldimproved management are indeed sizeable, it seems as i such barriers shouldbe surmountable. In turn, this insight suggests that perhaps managementbe surmountable. In turn, this insight suggests that perhaps managementdierences are rooted in deeper inormational, social, legal, and technologi-dierences are rooted in deeper inormational, social, legal, and technologi-cal dierences. Thus, once we have explained how we measure managementcal dierences. Thus, once we have explained how we measure managementand identifed some basic patterns in our data, we turn to the question o whyand identifed some basic patterns in our data, we turn to the question o whymanagement practices vary so much across frms and nations. What we fnd ismanagement practices vary so much across frms and nations. What we fnd isa combination o imperectly competitive markets, amily ownership o frms,a combination o imperectly competitive markets, amily ownership o frms,regulations restricting management practices, and inormational barriers allowregulations restricting management practices, and inormational barriers allowbad management to persist.bad management to persist.As a oretaste o our argument, here are ten conclusions we will discuss in thisAs a oretaste o our argument, here are ten conclusions we will discuss in thispaper based on our management data:paper based on our management data:First, frms with better management practices tend to have better peror-First, frms with better management practices tend to have better peror-mance on a wide range o dimensions: they are larger, more productive, growmance on a wide range o dimensions: they are larger, more productive, growaster, and have higher survival rates.aster, and have higher survival rates.


3/22

Nicholas Bloom and John Van Reenen 205

Second, management practices vary tremendously across frms and coun-Second, management practices vary tremendously across frms and coun-tries. Most o the dierence in the average management score o a country is duetries. Most o the dierence in the average management score o a country is dueto the size o the long tail o very badly managed frms. For example, relativelyto the size o the long tail o very badly managed frms. For example, relativelyew U.S. frms are very badly managed, while Brazil and India have many frms inew U.S. frms are very badly managed, while Brazil and India have many frms inthat category.that category.Third, countries and frms specialize in dierent styles o management. ForThird, countries and frms specialize in dierent styles o management. Forexample, American frms score much higher than Swedish frms in incentives butexample, American frms score much higher than Swedish frms in incentives butare worse than Swedish frms in monitoring.are worse than Swedish frms in monitoring.Fourth, strong product market competition appears to boost average manage-Fourth, strong product market competition appears to boost average manage-ment practices through a combination o eliminating the tail o badly managedment practices through a combination o eliminating the tail o badly managedfrms and pushing incumbents to improve their practices.frms and pushing incumbents to improve their practices.Fith, multinationals are generally well managed in every country. They alsoFith, multinationals are generally well managed in every country. They alsotransplant their management styles abroad. For example, U.S. multinationalstransplant their management styles abroad. For example, U.S. multinationalslocated in the United Kingdom are better at incentives and worse at monitoringlocated in the United Kingdom are better at incentives and worse at monitoringthan Swedish multinationals in the United Kingdom.than Swedish multinationals in the United Kingdom.Sixth, frms that export (but do not produce) overseas are better-managedSixth, frms that export (but do not produce) overseas are better-managedthan domestic non-exporters, but are worse-managed than multinationals.than domestic non-exporters, but are worse-managed than multinationals.Seventh, inherited amily-owned frms who appoint a amily member (espe-Seventh, inherited amily-owned frms who appoint a amily member (espe-cially the eldest son) as chie executive ofcer are very badly managed on average.cially the eldest son) as chie executive ofcer are very badly managed on average.Eighth, government-owned frms are typically managed extremely badly.Eighth, government-owned frms are typically managed extremely badly.Firms with publicly quoted share prices or owned by private-equity frms are typi-Firms with publicly quoted share prices or owned by private-equity frms are typi-cally well managed.cally well managed.Ninth, frms that more intensively use human capital, as measured by moreNinth, frms that more intensively use human capital, as measured by moreeducated workers, tend to have much better management practices.educated workers, tend to have much better management practices.Tenth, at the country level, a relatively light touch in labor market regulationTenth, at the country level, a relatively light touch in labor market regulationis associated with better use o incentives by management.is associated with better use o incentives by management.We hope and expect that these fndings will be tested and refned as researchWe hope and expect that these fndings will be tested and refned as researchcontinues in the area o quantitative analysis o management practices. At thecontinues in the area o quantitative analysis o management practices. At theend o the paper, we suggest some directions or uture research, and oer someend o the paper, we suggest some directions or uture research, and oer somebroader discussion on interpreting the quality o management both as shaped bybroader discussion on interpreting the quality o management both as shaped bynational actors and as reecting the decisions o frms.national actors and as reecting the decisions o frms.

How Can Management Practices Be Measured?How Can Management Practices Be Measured?To measure management practices, we have developed a new survey methodol-To measure management practices, we have developed a new survey methodol-ogy described in detail in Bloom and Van Reenen (2007). We use an interview-basedogy described in detail in Bloom and Van Reenen (2007). We use an interview-basedevaluation tool that defnes and scores rom 1 (worst practice) to 5 (best prac-evaluation tool that defnes and scores rom 1 (worst practice) to 5 (best prac-tice) 18 basic management practices. Table 1 lists the 18 management practicestice) 18 basic management practices. Table 1 lists the 18 management practicesand also gives some sense o how each is measured on a scale rom 1 to 5. In ourand also gives some sense o how each is measured on a scale rom 1 to 5. In ourview, a high score represents a best practice in the sense that a frm that has adoptedview, a high score represents a best practice in the sense that a frm that has adoptedthe practice will, on average, increase their productivity. The combination o manythe practice will, on average, increase their productivity. The combination o manyo these indicators reects good management as commonly understood, with ouro these indicators reects good management as commonly understood, with ourmain measure o management practices simply the average o these 18 scores.main measure o management practices simply the average o these 18 scores.


4/22


Table 1

The Management Practice Dimensions

Categories Score rom 15 based on:

1) Introduction o modernmanuacturing techniques

What aspects o manuacturing have been ormally introduced,including just-in-time delivery rom suppliers, autonomation,exible manpower, support systems, attitudes, and behavior?

2) Rationale or introduction omodern manuacturingtechniques

Were modern manuactur ing techniques adopted just becauseothers were using them, or are they linked to meeting businessobjectives like reducing costs and improving quality?

3) Process problemdocumentation

Are process improvements made only when problems arise, or arethey actively sought out or continuous improvement as part oa normal business process?

4) Perormance tracking Is tracking ad hoc and incomplete, or is perormance continuallytracked and communicated to all sta?

5) Perormance review Is perormance reviewed inrequently and only on asuccess/ailure scale, or is perormance reviewed continuallywith an expectation o continuous improvement?

6) Perormance dialogue In review/perormance conversations, to what extent is thepurpose, data, agenda, and ollow-up steps (like coaching)clear to all parties?

7) Consequence management To what extent does ailure to achieve agreed objectives carryconsequences, which can include retraining or reassignment toother jobs?

8) Target balance Are the goals exclusively fnancia l, or is there a balance o fnancialand nonfnancial targets?

9) Target interconnection Are goals based on accounting value, or are they based onshareholder value in a way that works through business units andultimately is connected to individual perormance expectations?

10) Target time horizon Does top management ocus mainly on the short term, or does itvisualize short- term targets as a staircase toward the mainocus on long-term goals?

11) Targets are stretching Are goals too easy to achieve, especially or some sacred cowsareas o the frm, or are goals demanding but attainable or allparts o the frm?

12) Perormance clarity Are perormance measures ill-defned, poorly understood, andprivate, or are they well-defned, clearly communicated, andmade public?

13) Managing human capital To what extent are senior managers evaluated and heldaccountable or attracting, retaining, and developing talentthroughout the organization?

14) Rewarding highperormance

To what extent are people in the frm rewarded equallyirrespective o perormance level, or are rewards related toperormance and eort?

15) Removing poor perormers Are poor perormers rarely removed, or are they retrained and/ormoved into dierent roles or out o the company as soon as theweakness is identifed?

16) Promoting high perormers Are people promoted mainly on the basis o tenure, or does thefrm actively identiy, develop, and promote its top perormers?

17) Attracting human capital Do competitors oer stronger reasons or talented people to jointheir companies, or does a frm provide a wide range o reasonsto encourage talented people to join?

18) Retaining human capital Does the frm do relatively little to retain top talent or do whateverit takes to retain top talent when they look likely to leave?

Note:The ull set o questions that are asked to score each dimension are included in Bloom and VanReenen (2006).


5/22

Why Do Management Practices Dier across Firms and Countries? 207

This evaluation tool was developed by an international consulting frm, and itThis evaluation tool was developed by an international consulting frm, and itcan be broadly interpreted as attempting to measure management practices in threecan be broadly interpreted as attempting to measure management practices in threebroad areas: 1)broad areas: 1) monitoringhow well do companies monitor what goes on insidehow well do companies monitor what goes on insidetheir frms and use this or continuous improvement; 2)their frms and use this or continuous improvement; 2) targetsdo companies setdo companies setthe right targets, track the right outcomes, and take appropriate action i the twothe right targets, track the right outcomes, and take appropriate action i the twoare inconsistent? 3)are inconsistent? 3) incentivesare companies promoting and rewarding employeesare companies promoting and rewarding employeesbased on perormance, and trying to hire and keep their best employees?based on perormance, and trying to hire and keep their best employees?11We hired MBA students to carry out the interviews because they generally hadWe hired MBA students to carry out the interviews because they generally hadsome business experience and training. The survey was targeted at plant manag-some business experience and training. The survey was targeted at plant manag-ers, who are senior enough to have an overview o management practices but not soers, who are senior enough to have an overview o management practices but not sosenior as to be detached rom day-to-day operations. We interviewed these manag-senior as to be detached rom day-to-day operations. We interviewed these manag-ers using what we call a double-blind technique.ers using what we call a double-blind technique.One part o this double-blind technique is that managers are not told they areOne part o this double-blind technique is that managers are not told they arebeing scored or shown the scoring grid. They are only told they are being inter-being scored or shown the scoring grid. They are only told they are being inter-viewed about management practices. To do this, we used open questions in theviewed about management practices. To do this, we used open questions in thesurvey. For example, on the frst monitoring dimension, we start by asking the opensurvey. For example, on the frst monitoring dimension, we start by asking the openquestion tell me how you monitor your production process, rather than closedquestion tell me how you monitor your production process, rather than closedquestions such as do you monitor your production daily [yes/no]. We continuequestions such as do you monitor your production daily [yes/no]. We continuewith open questions ocusing on actual practices and examples until the inter-with open questions ocusing on actual practices and examples until the inter-viewer can make an accurate assessment o the frms practices. For example, theviewer can make an accurate assessment o the frms practices. For example, thesecond question on that monitoring dimension is What kinds o measures wouldsecond question on that monitoring dimension is What kinds o measures wouldyou use to track perormance? and the third is I I walked round your actoryyou use to track perormance? and the third is I I walked round your actorywhat could I tell about how each person was perorming? The combined responseswhat could I tell about how each person was perorming? The combined responsesto this dimension are scored against a grid which goes rom 1 (out o 5), which isto this dimension are scored against a grid which goes rom 1 (out o 5), which isdefned as defned as Measures tracked do not indicate directly i overall business objectives are beingmet. Tracking is an ad-hoc process (certain processes arent tracked at all), up to 5 which), up to 5 whichis defned as is defned as Perormance is continuously tracked and communicated, both ormally andinormally, to all sta using a range o visual management tools. The ull list o dimen-. The ull list o dimen-sions and questions used to score these are given in Bloom and Van Reenen (2006).sions and questions used to score these are given in Bloom and Van Reenen (2006).The other side o our double-blind approach is that our interviewers are notThe other side o our double-blind approach is that our interviewers are nottold in advance anything about the frms perormance. They are only providedtold in advance anything about the frms perormance. They are only providedwith the company name, telephone number, and industry. We randomly samplewith the company name, telephone number, and industry. We randomly samplemedium-sized frms, employing between 100 to 5,000 workers. These frms aremedium-sized frms, employing between 100 to 5,000 workers. These frms arelarge enough that the type o systematic management practices chosen are likelylarge enough that the type o systematic management practices chosen are likelyto matter. However, these frms are generally small enough that they are not usu-to matter. However, these frms are generally small enough that they are not usu-ally reported in the business press, so the interviewers generally have not heard oally reported in the business press, so the interviewers generally have not heard othese frms beore and so should have no preconceptions. By contrast, interviewerthese frms beore and so should have no preconceptions. By contrast, interviewerpreconceptions might be more o a problem i the interviewers knew they were talk-preconceptions might be more o a problem i the interviewers knew they were talk-ing to an employee o well-known frms like General Electric, Boeing, or Honda.ing to an employee o well-known frms like General Electric, Boeing, or Honda.1 These practices are similar to those emphasized in earlier work on management practices, by, orexample, Ichinowski, Shaw, and Prennushi (1997) and Black and Lynch (2001). Bertrand and Schoar(2003) ocus on another important anglethe management style o chie executive ofcers and chie

fnancial ofcerswhich will capture dierences in management strategy (say over mergers andacquisitions) rather than management practices per se.


6/22


We used a variety o procedures to obtain a high response rate and to removeWe used a variety o procedures to obtain a high response rate and to removepotential sources o bias rom our estimates. First, we obtained governmentpotential sources o bias rom our estimates. First, we obtained governmentendorsements or the surveys in each country covered. Second, we positioned theendorsements or the surveys in each country covered. Second, we positioned thesurveys as a piece o work on lean manuacturing, never using the word sur-surveys as a piece o work on lean manuacturing, never using the word sur-vey or research, as telephone switchboards usually block surveys and marketvey or research, as telephone switchboards usually block surveys and marketresearch. Third, we never ask interviewees or fnancial data, instead obtainingresearch. Third, we never ask interviewees or fnancial data, instead obtainingsuch data rom independent sources or company accounts. Fourth, the interview-such data rom independent sources or company accounts. Fourth, the interview-ers were encouraged to be persistent, so they ran about two interviews a day lastingers were encouraged to be persistent, so they ran about two interviews a day lasting45 minutes each on average, with the rest o the time spent repeatedly contacting45 minutes each on average, with the rest o the time spent repeatedly contactingmanagers to schedule interviews. These steps helped to yield a 44 percent responsemanagers to schedule interviews. These steps helped to yield a 44 percent responserate which was uncorrelated with the (independently collected) perormance mea-rate which was uncorrelated with the (independently collected) perormance mea-sures or the frmthus, we were not disproportionately interviewing successul orsures or the frmthus, we were not disproportionately interviewing successul orailing frms.ailing frms.22 We also collected a series o noise controls on the interview pro-We also collected a series o noise controls on the interview pro-cess itsel (such as the time o day and the day o the week), characteristics o thecess itsel (such as the time o day and the day o the week), characteristics o theinterviewee and the identity o the interviewer. Including these in our regressioninterviewee and the identity o the interviewer. Including these in our regressionanalysis typically helps to improve the precision o our estimates by stripping outanalysis typically helps to improve the precision o our estimates by stripping outsome o the measurement error.some o the measurement error.33

International Patterns o Management and ProductivityInternational Patterns o Management and ProductivityIn discussing the patterns we fnd in our management data, it is importantIn discussing the patterns we fnd in our management data, it is importantto remember that our main data is essentially cross sectionalacross many frmsto remember that our main data is essentially cross sectionalacross many frmsand countries at roughly the same point in timeand so clearly establishing theand countries at roughly the same point in timeand so clearly establishing thecausal eect o how changes in management aect productivity is not possible.causal eect o how changes in management aect productivity is not possible.A wealth o feld experiments, surveyed in Lazear and Oyer (orthcoming), doesA wealth o feld experiments, surveyed in Lazear and Oyer (orthcoming), doesstrongly suggest the importance o incentive-based pay or increasing productivity.strongly suggest the importance o incentive-based pay or increasing productivity.Nevertheless, examining both the patterns o management across countries andNevertheless, examining both the patterns o management across countries andthe correlation between our measures o management and various measures othe correlation between our measures o management and various measures ofrm perormance is an important frst step in determining the extent to which ourfrm perormance is an important frst step in determining the extent to which ourmeasurements o management are economically meaningul.measurements o management are economically meaningul.

International Patterns o ManagementInternational Patterns o ManagementThe average country-level management practice score rom the almost 6,000The average country-level management practice score rom the almost 6,000frm interviews we have carried out since 2006 appear in Figure 1. These frms werefrm interviews we have carried out since 2006 appear in Figure 1. These frms were2As another step to validate our survey data, we resurveyed 5 percent o the sample using a secondinterviewer to independently survey a second plant manager in the same frm. Two independent man-agement interviews on dierent plants within the same frms should help to reveal how consistentlywe are measuring management practices. We ound that in the sample o 222 additional interviews,the correlation between our independently run frst and second interview scores was 0.51. Part o thisdierence across plants within the same frms is likely to be real internal variations in managementpractices, with the rest presumably reecting survey measurement error. However, the correlationacross the two interviews is highly signifcant (p-value o under 0.001), which suggests that while our

management score is clearly noisy, it is picking up signifcant management di erences across frms.3 See our paper Bloom and Van Reenen (2009) or a detailed description o the sur vey process.


7/22


randomly sampled rom the population o all public and private manuacturingrandomly sampled rom the population o all public and private manuacturingfrms with 100 to 5,000 employees. The median frm in every country is privatelyfrms with 100 to 5,000 employees. The median frm in every country is privatelyowned, employing around 350 workers, and operating across two production plants.owned, employing around 350 workers, and operating across two production plants.The United States has the highest management practice scores on average,The United States has the highest management practice scores on average,ollowed by Germany, Japan, Sweden, and Canada and then ollowed by a blockollowed by Germany, Japan, Sweden, and Canada and then ollowed by a blocko mid-European countriesFrance, Italy, Ireland, the United Kingdom, ando mid-European countriesFrance, Italy, Ireland, the United Kingdom, andPolandand Australia. At the bottom are countries in southern Europe like GreecePolandand Australia. At the bottom are countries in southern Europe like Greeceand Portugal, along with developing countries like Brazil, China, and India.and Portugal, along with developing countries like Brazil, China, and India.We can separate these overall management scores into three broad catego-We can separate these overall management scores into three broad catego-ries relating to monitoring, to targets, and to incentives, with country-level scoresries relating to monitoring, to targets, and to incentives, with country-level scoresshown in Table 2. For ease o comparison, average scores are given in the bottomshown in Table 2. For ease o comparison, average scores are given in the bottomrow o the table. U.S. management has by ar the largest advantage in incentivesrow o the table. U.S. management has by ar the largest advantage in incentives(with Canada and Germany ollowing), and the second-largest advantage in the(with Canada and Germany ollowing), and the second-largest advantage in thecategories o monitoring and target-setting (behind Sweden and Germany, respec-categories o monitoring and target-setting (behind Sweden and Germany, respec-tively). However, this data also describes how management styles dier acrosstively). However, this data also describes how management styles dier acrosscountries. In the United States, India, and China, managerial use o incentivescountries. In the United States, India, and China, managerial use o incentives

Figure 1

Management Scores across Countries

Source:Bloom, Genakos, Sadun, and Van Reenen (2009).Notes:Averages are taken across all frms within each country. There are 5,850 observations in total.Firms per country are in the right column.

2.6 2.8 3 3.2 3.4

United States

Germany

Sweden

Japan

Canada

France

Italy

Great Britain

Australia

Northern IrelandPoland

Republic o Ireland

Portugal

Brazil

India

China

Greece

Management scores, rom 1 (worst practice) to 5 (best practice)

# frms

695

336

270

122

344

312

188

762

382

92231

102

140

524

171

620

559


8/22


(relative to the average country) are substantially greater than their use o monitor-(relative to the average country) are substantially greater than their use o monitor-ing and targets (relative to the average). However, in Japan, Sweden, and Germany,ing and targets (relative to the average). However, in Japan, Sweden, and Germany,managerial use o monitoring and targets (relative to the average) ar exceeds theirmanagerial use o monitoring and targets (relative to the average) ar exceeds theiruse o incentives (relative to the average). There could be many reasons or this pat-use o incentives (relative to the average). There could be many reasons or this pat-tern o specialization across countries. One actor we will examine below is that thetern o specialization across countries. One actor we will examine below is that thelighter labor market regulations in the United States make it easier to remove poorlighter labor market regulations in the United States make it easier to remove poorperormers and to reward high perormers.perormers and to reward high perormers.What does the distribution o management practices look like within coun-What does the distribution o management practices look like within coun-tries? We can plot a frm-level histogram o management practices by country, astries? We can plot a frm-level histogram o management practices by country, asshown in Figure 2. The frst histogram shows this data or the United States, whereshown in Figure 2. The frst histogram shows this data or the United States, wherethe bars show the actual data and the dark line is a smoothed (kernel) ft o thethe bars show the actual data and the dark line is a smoothed (kernel) ft o theU.S. data. Other advanced economies in western Europe, Japan, Canada, and Aus-U.S. data. Other advanced economies in western Europe, Japan, Canada, and Aus-tralia have some resemblance to the U.S. distribution, except they have a somewhattralia have some resemblance to the U.S. distribution, except they have a somewhatthicker tail o badly managed frms. To illustrate this we show the histograms othicker tail o badly managed frms. To illustrate this we show the histograms oFrance and Southern Europe (Portugal and Greece). These diagrams also showFrance and Southern Europe (Portugal and Greece). These diagrams also showthe smoothed value or the U.S. economy, so that management in these countriesthe smoothed value or the U.S. economy, so that management in these countriescan be readily compared to the United States. Histograms or Brazil and Indiacan be readily compared to the United States. Histograms or Brazil and Indiashow that although they both have numerous well-managed frms, their averageshow that although they both have numerous well-managed frms, their average

Table 2

Management Practice Scores by Country

Country

Overall

management

Monitoring

management

Targets

management

Incentives

management

# o frms

in the

sample

Australia 2.99 3.27 2.96 2.76 382Brazil 2.69 2.81 2.68 2.60 559Canada 3.13 3.35 3.02 3.02 344China 2.64 2.72 2.53 2.66 524France 3.00 3.28 2.98 2.78 312Germany 3.18 3.40 3.24 2.95 336Great Britain 2.98 3.16 2.93 2.88 762Greece 2.65 2.90 2.56 2.50 171India 2.65 2.62 2.66 2.67 620Italy 2.99 2.98 2.80 2.73 194Japan 3.15 3.20 3.25 2.90 188Northern Ireland 2.91 3.01 2.84 2.86 92Poland 2.88 2.88 2.93 2.85 231Portugal 2.79 3.07 2.72 2.61 140Republic o Ireland 2.84 2.95 2.76 2.81 102Sweden 3.18 3.54 3.22 2.86 270United States 3.33 3.44 3.23 3.30 695Average 2.94 3.09 2.91 2.84 344

Notes:Overall management is the average score in across all 18 questions. Monitoring managementis the average score across questions 1 to 6 in Table 1. Targets management is the average score

across questions 8 to 12. Incentives management is the average score across questions 7 and 13 to 18.The lowest and highest country-level scores in each column are highlighted in bold.


9/22


frms have much lower management scores than the United States. Finally, Chinafrms have much lower management scores than the United States. Finally, Chinahas a more compressed distribution, possibly because Chinese frms are muchhas a more compressed distribution, possibly because Chinese frms are muchyounger and so have less variation in terms o vintages o management practices.younger and so have less variation in terms o vintages o management practices.In one sense this cross-country ranking is not surprising, since it approximatesIn one sense this cross-country ranking is not surprising, since it approximatesthe cross-country productivity ranking. Although we cannot oer a rigorous argu-the cross-country productivity ranking. Although we cannot oer a rigorous argu-ment here about the magnitude o any causal eect, it certainly appears plausiblement here about the magnitude o any causal eect, it certainly appears plausiblethat management practices should be viewed as part o the determinants o nationalthat management practices should be viewed as part o the determinants o nationalproductivity. A regression o GDP per capita on management practices across theproductivity. A regression o GDP per capita on management practices across thesample o 17 countries yields ansample o 17 countries yields an R-squared o 0.81. Since some o this is simply-squared o 0.81. Since some o this is simplya contrast between more- and less-developed countries, ocusing the regressiona contrast between more- and less-developed countries, ocusing the regressionon the eleven OECD nations with good manuacturing productivity data (Inklaaron the eleven OECD nations with good manuacturing productivity data (Inklaarand Timmer, 2008) yields anand Timmer, 2008) yields an R-squared o 0.66. Either way, management practices-squared o 0.66. Either way, management practicesappear to be potentially quantitatively important.appear to be potentially quantitatively important.Countries can improve average management practices and thereore aggre-Countries can improve average management practices and thereore aggre-gate productivity in two distinct ways. The frst is by promoting actors that increasegate productivity in two distinct ways. The frst is by promoting actors that increaseaverage management quality inaverage management quality in each frm(say through better business education)(say through better business education)and thereore raise productivity within the average frm. The next sub-sectionand thereore raise productivity within the average frm. The next sub-sectionrelates to this mechanism.relates to this mechanism.

Figure 2

Management Practice Scores across Firms

Source:Bloom, Genakos, Sadun, and Van Reenen (2009).Notes:Bars are the histogram o the actual density. The line is the smoothed (kernel) o the U.S.

density or comparison. Southern Europe combines Greece and Portugal.

Fra

ctionofrms

Firm-level average management scores, rom 1 (worst practice) to 5 (best practice)

0

.5

1

0

.5

1

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

United States Brazil China

France India Southern Europe


10/22


The second is throughThe second is through improved reallocation across frms. This actor turns out. This actor turns outto be empirically important in explaining cross-country dierences in aggregateto be empirically important in explaining cross-country dierences in aggregateproductivity: that is, high-productivity countries like the United States appear toproductivity: that is, high-productivity countries like the United States appear tobe better at getting efcient frms to grow larger, while low-productivity countriesbe better at getting efcient frms to grow larger, while low-productivity countrieslike China and India are not (Hsieh and Klenow, orthcoming). The implication islike China and India are not (Hsieh and Klenow, orthcoming). The implication isthat actors like product market competition should generate a stronger relation-that actors like product market competition should generate a stronger relation-ship between management quality on the one hand and frm size and growth onship between management quality on the one hand and frm size and growth onthe other, and thereore lead to higher aggregate productivity. We discuss this laterthe other, and thereore lead to higher aggregate productivity. We discuss this laterwhen we turn to the determination o management practices.when we turn to the determination o management practices.Associations between Management Quality and Firm PerormanceAssociations between Management Quality and Firm PerormanceWe examined the correlation between our measure o management practicesWe examined the correlation between our measure o management practicesand frm perormance in terms o productivity, proftability, growth rates, survivaland frm perormance in terms o productivity, proftability, growth rates, survivalrates, and market value. For these measures o frm perormance, we used companyrates, and market value. For these measures o frm perormance, we used companyaccounts data that were available or 3,380 o the frms. We ound that, or ouraccounts data that were available or 3,380 o the frms. We ound that, or oursample o manuacturing frms, higher management scores are robustly associatedsample o manuacturing frms, higher management scores are robustly associatedwith better perormance.with better perormance.44Table 3 reports the results o some ordinary least squares regressions. OurTable 3 reports the results o some ordinary least squares regressions. Ourdependent variables are dierent measures o frm perormance, including salesdependent variables are dierent measures o frm perormance, including salesper employee, proftability, Tobinsper employee, proftability, Tobins q(the ratio o a frms stock market value to its(the ratio o a frms stock market value to itscapital stock), the growth o sales, and survival. Our key explanatory variable is thecapital stock), the growth o sales, and survival. Our key explanatory variable is themeasure o the companys management quality. In some o the regressions, we alsomeasure o the companys management quality. In some o the regressions, we alsoadjust or capital per employee, and the share o the workorce with a college degree.adjust or capital per employee, and the share o the workorce with a college degree.We also employ other control variables including country and industry dummy vari-We also employ other control variables including country and industry dummy vari-ables, frm-level control variables or hours worked and frm age, and a set o noiseables, frm-level control variables or hours worked and frm age, and a set o noisecontrols that (as discussed earlier) include dummy variables or our interviewerscontrols that (as discussed earlier) include dummy variables or our interviewersas well as or the job tenure o the manager, the day o the week the interview wasas well as or the job tenure o the manager, the day o the week the interview wasconducted, the time o day the interview was conducted, the length o the interview,conducted, the time o day the interview was conducted, the length o the interview,and a judgment rom the interviewer on the reliability o the inormation collected.and a judgment rom the interviewer on the reliability o the inormation collected.In column 1, the dependent variable is the logarithm o sales per employee, aIn column 1, the dependent variable is the logarithm o sales per employee, avery basic measure o frm productivity. Our management score is an average acrossvery basic measure o frm productivity. Our management score is an average acrossall 18 questions. The coefcient suggests that frms with one point higher averageall 18 questions. The coefcient suggests that frms with one point higher averagemanagement score have about 45 log points (about 57 percent) higher labor produc-management score have about 45 log points (about 57 percent) higher labor produc-tivity. So a one-standard deviation change in management (o 0.664) is associatedtivity. So a one-standard deviation change in management (o 0.664) is associatedwith about a 38 percent increase in sales holding employment constant. Columnwith about a 38 percent increase in sales holding employment constant. Column2 controls or country and industry to reect dierent accounting standards and2 controls or country and industry to reect dierent accounting standards andprices across countries and industries. The management coefcient drops inprices across countries and industries. The management coefcient drops inmagnitude to 0.208, but remains highly signifcant. Column 3 adds controls ormagnitude to 0.208, but remains highly signifcant. Column 3 adds controls orcapital per employee, the percentage o the workorce with a college degree, andcapital per employee, the percentage o the workorce with a college degree, and4Our sample contained 90 percent private frms and 10 percent publicly listed frms. In most countriesaround the world, both public and private frms publish basic accounts. In the United States, Canada,and India, however, private frms do not publish (sufciently detailed) accounts, so while we still sur-

veyed these frms, no accounting perormance data is available or them. Hence, these perormanceregressions use data or all frms except privately held ones in the United States, Canada, and India.


11/22


our controls or survey noise. These additions slightly reduce the coefcient onour controls or survey noise. These additions slightly reduce the coefcient onthe management variable to around 0.172, because better-managed frms tend tothe management variable to around 0.172, because better-managed frms tend tohave more fxed capital and human capital. These correlations are not driven byhave more fxed capital and human capital. These correlations are not driven bythe Anglo-Saxon countries, as one might expect i the measures were culturallythe Anglo-Saxon countries, as one might expect i the measures were culturallybiased. The relationship between productivity and management is strong across allbiased. The relationship between productivity and management is strong across allregions in the data.regions in the data.In column 4 o Table 3, we look at proftability as measured by return on capitalIn column 4 o Table 3, we look at proftability as measured by return on capitalemployed (defned as profts over equity plus debt capital) and fnd that this isemployed (defned as profts over equity plus debt capital) and fnd that this is

Table 3

Estimates o Firm Perormance Equations

Dependent variable

(1) (2) (3) (4) (5) (6) (7)

ln(Sales/

Employee)

ln(Sales/

Employee)

ln(Sales/

Employee)

Proftability

(ROCE)

Ln

(Tobins Q)

Sales

growth

Survival

Management 0.450***(0.028)

0.208***(0.021)

0.172***(0.024)

1.804***(0.668)

0.150**(0.062)

0.044***(0.014)

0.55*a

(0.30)a

ln(Capital/Employee) 0.106***(0.014)

% College degree 0.076***(0.014)

Country & industrydummies

No Yes Yes Yes Yes Yes Yes

General controls No No Yes Yes Yes Yes YesNoise controls No No Yes Yes Yes Yes YesFirms 3,380 3,380 3,380 2,369 524 2,298 3,627Observations 29,390 29,390 29,390 20,141 3,505 19,568 3,627

Source:Bloom, Genakos, Sadun, and Van Reenen (2009).Notes:All columns estimated by ordinary least squares with standard errors are in parentheses undercoefcient estimates clustered by frm, except or column (7), which is estimated by probit (we reportmarginal eects at the sample mean). The sample is o all frms with available accounts data atsome point between 2000 and 2008. The management score has a mean o 2.973 and a standard

deviation o 0.664. Country and industry dummies includes a ull set o 17 country and 162 SIC3-digit dummies. General controls comprise frm-level controls or ln(average hours worked) andln(frm age). Noise controls are 78 interviewer dummies, the seniority and tenure o the managerwho responded, the day o the week the interv iew was conducted, the time o the day the interviewwas conducted, the duration o the interviews, and an indicator o the reliability o the inormation ascoded by the interviewer. All regressions include a ull set o time dummies. Management is the frm-level management score. % College degree is the share o employees with a college degree (collectedrom the survey). Proftability is ROCE which is Return on Capital Employed and Sales growthis the 5-year growth o sales. Survival is equal to zero i a frm exited due to bankruptcy/liquidationby spring 2009 and one otherwise.a This is the marginal eect and standard error multiplied by 100. The sample mean o nonsurvivalis 2.64% so the marginal eect o 0.53 implies one management point is associated with 20.1%(0.53/2.64) lower exit rate.

***, **, and * indicate signifcance at the 1, 5, and 10 percent levels, respectively.


12/22


about 1.8 percentage points higher or every one point increase in the managementabout 1.8 percentage points higher or every one point increase in the managementscore. In Column 5 we look at Tobinsscore. In Column 5 we look at Tobins qor the subsample o publicly quoted frmsor the subsample o publicly quoted frms(where Tobins(where Tobins qis calculated as the stock market value o the frm divided by theis calculated as the stock market value o the frm divided by thebook value o the frm) and fnd a one-point increase in management is associatedbook value o the frm) and fnd a one-point increase in management is associatedwith a 15 log point increase in Tobinswith a 15 log point increase in Tobins q. Column 6 uses the fve-year sales growth. Column 6 uses the fve-year sales growthrate as the outcome. Here, a unit improvement in the management practice score israte as the outcome. Here, a unit improvement in the management practice score isassociated with 4.4 percent higher annual sales growth. Finally, Column 7 looks atassociated with 4.4 percent higher annual sales growth. Finally, Column 7 looks atthe post-survey survival rates and shows that better-managed frms are more likelythe post-survey survival rates and shows that better-managed frms are more likelyto survive while worse-managed frms are more likely to go bankrupt.to survive while worse-managed frms are more likely to go bankrupt.Another key measure o perormance is frm size; in equilibrium, better-Another key measure o perormance is frm size; in equilibrium, better-managed frms should be larger (Lucas, 1978). This is partly because the marketmanaged frms should be larger (Lucas, 1978). This is partly because the marketwill allocate these frms a greater share o sales, but also because larger frms havewill allocate these frms a greater share o sales, but also because larger frms havethe resources and incentives to employ better management. When we plotted aver-the resources and incentives to employ better management. When we plotted aver-age management score against the number o employees in a frm (as a measure oage management score against the number o employees in a frm (as a measure ofrm size) we ound that frms with 100200 employees had average managementfrm size) we ound that frms with 100200 employees had average managementscores o about 2.7. The management score then rose steadily with frm size, so thatscores o about 2.7. The management score then rose steadily with frm size, so thatfrms with 20005000 employeesthe largest frms in our samplehad averagefrms with 20005000 employeesthe largest frms in our samplehad averagemanagement scores o about 3.2.management scores o about 3.2.The international data revealed some patterns o specialization by country inThe international data revealed some patterns o specialization by country inmanagement style, in terms o whether management in certain countries places amanagement style, in terms o whether management in certain countries places ahigher relative weight on monitoring and target-setting or on incentives. Althoughhigher relative weight on monitoring and target-setting or on incentives. Althougha frm that is good at one dimension o management tends to be good at all (that is,a frm that is good at one dimension o management tends to be good at all (that is,the answers to the individual questions tend to be positively correlated), a patternthe answers to the individual questions tend to be positively correlated), a patterno specialization in dierent styles o management is also observable at the frmo specialization in dierent styles o management is also observable at the frmlevel. Firms operating in industries like pharmaceuticals that are relatively human-level. Firms operating in industries like pharmaceuticals that are relatively human-capital intensive tend to have better incentive management practices than frmscapital intensive tend to have better incentive management practices than frmsoperating in industries like textiles and apparel that have more unskilled workersoperating in industries like textiles and apparel that have more unskilled workers(Bloom and Van Reenen, 2007).(Bloom and Van Reenen, 2007).The association o management with frm perormance is also clear in otherThe association o management with frm perormance is also clear in othersectors outside manuacturing. In Bloom, Propper, Seiler, and Van Reenen (2009),sectors outside manuacturing. In Bloom, Propper, Seiler, and Van Reenen (2009),we interviewed 181 managers and physicians in the orthopedic and cardiologywe interviewed 181 managers and physicians in the orthopedic and cardiologydepartments o U.K. hospitals. We ound that management scores were signif-departments o U.K. hospitals. We ound that management scores were signif-cantly associated with better perormance as indicated by improved survival ratescantly associated with better perormance as indicated by improved survival ratesrom emergency heart attack admissions and other kinds o general surgery as wellrom emergency heart attack admissions and other kinds o general surgery as wellas shorter waiting lists.as shorter waiting lists.MightMightbettermanagement also be associated with worse outcomes or workersmanagement also be associated with worse outcomes or workersand or the environment? In an earlier 2004 survey wave, we also collected inor-and or the environment? In an earlier 2004 survey wave, we also collected inor-mation on aspects o work-lie balance such as child-care acilities, job exibility,mation on aspects o work-lie balance such as child-care acilities, job exibility,and sel-assessed employee satisaction. Well-managed frms actually tended toand sel-assessed employee satisaction. Well-managed frms actually tended tohave better acilities or workers along these dimensions (Bloom, Kretschmer, andhave better acilities or workers along these dimensions (Bloom, Kretschmer, andVan Reenen, 2009). We also ound that energy efcency is strongly associated withVan Reenen, 2009). We also ound that energy efcency is strongly associated withbetter frm-level management, probably because good management practices (likebetter frm-level management, probably because good management practices (likelean manuacturing) tend to economize on energy use (Bloom, Genakos, Martin,lean manuacturing) tend to economize on energy use (Bloom, Genakos, Martin,and Sadun, orthcoming).and Sadun, orthcoming).


13/22


What Causes Dierences in Management Practices?What Causes Dierences in Management Practices?Management practices vary substantially across countries and across frms,Management practices vary substantially across countries and across frms,which raises a difcult question. I improved management oers proftability gains,which raises a difcult question. I improved management oers proftability gains,why would frms not adopt better management practices? To address this, we ocuswhy would frms not adopt better management practices? To address this, we ocuson product market competition, labor regulation, multinational status, ownership,on product market competition, labor regulation, multinational status, ownership,and education. O course, some o these reasons may be better suited to explain-and education. O course, some o these reasons may be better suited to explain-ing dierences within countries or across industries, while other reasons may being dierences within countries or across industries, while other reasons may bebetter-suited to explaining dierence between countries.better-suited to explaining dierence between countries.

Product Market CompetitionProduct Market CompetitionWhen product market competition is not very intense, some low-productivityWhen product market competition is not very intense, some low-productiv ityfrms will be able to survive. This insight is consistent with our earlier argumentfrms will be able to survive. This insight is consistent with our earlier argumentthat the United States, which generally has very competitive product markets bythat the United States, which generally has very competitive product markets byinternational standards, does not have as much o a tail o badly-managed frmsinternational standards, does not have as much o a tail o badly-managed frmsas some other countries. Syverson (2004b) showed that in a very homogeneousas some other countries. Syverson (2004b) showed that in a very homogeneousindustry in the United States (ready-mix concrete), more competitive geographicindustry in the United States (ready-mix concrete), more competitive geographicmarkets had a smaller tail o less-productive plants.markets had a smaller tail o less-productive plants.In our surveys, we asked managers how many competitors they aced, andIn our surveys, we asked managers how many competitors they aced, andound the average management score was signifcantly higher when frms reportedound the average management score was signifcantly higher when frms reportedacing more competitors. Using other measures o competition not reported byacing more competitors. Using other measures o competition not reported bymanagers, like the import penetration rates (measured by imports as a share omanagers, like the import penetration rates (measured by imports as a share odomestic production) or Lerner indices o competition, yields a similar generaldomestic production) or Lerner indices o competition, yields a similar generalresult that management quality tends to increase with competitive intensity.result that management quality tends to increase with competitive intensity.55In general, we interpret this fnding as showing that competitive prod-In general, we interpret this fnding as showing that competitive prod-uct markets are associated with better management practices. This result coulduct markets are associated with better management practices. This result couldarise through a variety o channels. For example, one route or competition toarise through a variety o channels. For example, one route or competition toimprove management practices may be through selection, with badly run frmsimprove management practices may be through selection, with badly run frmsexiting more speedily in competitive markets. A second route may be throughexiting more speedily in competitive markets. A second route may be throughincentives to improve practices, which could be sharper when competition raisesincentives to improve practices, which could be sharper when competition raisesthe stakes either because efciency improvements have a larger impact on shit-the stakes either because efciency improvements have a larger impact on shit-ing market share or because managers are more earul o losing their jobs. Ining market share or because managers are more earul o losing their jobs. Inongoing work described in Bloom, Genakos, Sadun, and Van Reenen (2009), weongoing work described in Bloom, Genakos, Sadun, and Van Reenen (2009), wefnd evidence that both eects are at play. On the selection story, the relationshipfnd evidence that both eects are at play. On the selection story, the relationshipbetween management and frm growth appeared to be particularly strong in thebetween management and frm growth appeared to be particularly strong in theUnited States, where competition is more intense. Further, using our short panelUnited States, where competition is more intense. Further, using our short paneldata or our countries, we ound that increases in competition were associateddata or our countries, we ound that increases in competition were associatedwith increases in management quality or surviving frms. In any case, the bottomwith increases in management quality or surviving frms. In any case, the bottom

5 The Lerner index is calculated as one minus the average proftssales ratio o all other frms in thecountry industry cell over the last fve years. High values suggest low long-run profts, which are sug-gestive o tough competition. When we use this and the import measure data, we add country and

industry dummies to control or things like country size and dierent reporting requirements. SeeBloom and Van Reenen (2007) or details.


14/22


line is that encouraging tougher competition should improve average managementline is that encouraging tougher competition should improve average managementpractices.practices.Labor Market RegulationLabor Market RegulationLabor market regulations that constrain the ability o managers to hire, fre,Labor market regulations that constrain the ability o managers to hire, fre,pay, and promote employees could reduce the quality o management practices.pay, and promote employees could reduce the quality o management practices.Figure 3 plots each countrys average management scores on incentives manage-Figure 3 plots each countrys average management scores on incentives manage-ment (questions 7 and 13 to 18 on hiring, fring, pay, and promotions) against anment (questions 7 and 13 to 18 on hiring, fring, pay, and promotions) against anemployment rigidity index rom the World Bank, which ocuses on the difcul-employment rigidity index rom the World Bank, which ocuses on the difcul-ties that frms ace in hiring workers, fring workers, and changing their hoursties that frms ace in hiring workers, fring workers, and changing their hoursand pay. Tougher labor market regulation is signifcantly negatively correlated withand pay. Tougher labor market regulation is signifcantly negatively correlated withthe management scores on incentives. In contrast, more restrictive labor marketthe management scores on incentives. In contrast, more restrictive labor marketregulations are not signifcantly correlated with management practices in otherregulations are not signifcantly correlated with management practices in otherdimensions like monitoring or targets.dimensions like monitoring or targets.Obviously a number o other actors also vary across countries, so the pat-Obviously a number o other actors also vary across countries, so the pat-tern shown in Figure 3 does not conclusively demonstrate labor market regulationstern shown in Figure 3 does not conclusively demonstrate labor market regulations

Figure 3

Labor Market Regulation and Incentives Management

Source:Bloom, Genakos, Sadun, and Van Reenen (2009).Note:World Bank index rom the Doing Business database, http://www.doingbusiness.org/ExploreTopics/EmployingWorkers/.

Australia

Brazil

Canada

China

France

Great Britain

Germany

Greece

India

Republic o Ireland

Italy

JapanNorthern Ireland Poland

Portugal

Sweden

US

2.4

2.6

2.8

3

3.2

3.4

0 20 40 60

World Bank Employment Rigidity Index

Incen

tivesmanagementscore


15/22


constrain some orms o management practicesbut it is certainly supportive oconstrain some orms o management practicesbut it is certainly supportive othis eect.this eect.Ownership and Meritocratic Selection o the Chie Executive OfcerOwnership and Meritocratic Selection o the Chie Executive OfcerThe frms in our sample can be divided up by ultimate ownership: includingThe frms in our sample can be divided up by ultimate ownership: includingdispersed shareholders, amily ownership with an external chie executive ofcer,dispersed shareholders, amily ownership with an external chie executive ofcer,amily ownership with a amily chie executive ofcer; owned by the ounder, theamily ownership with a amily chie executive ofcer; owned by the ounder, thegovernment or the managers o the frm; and owned by private equity or privategovernment or the managers o the frm; and owned by private equity or privateindividuals. Figure 4 plots a frm-level histogram by ultimate ownership category.individuals. Figure 4 plots a frm-level histogram by ultimate ownership category.The bars display the distribution o management practices within ownershipThe bars display the distribution o management practices within ownershipgroup. The dark line is the kernel density or dispersed shareholderswhich is thegroup. The dark line is the kernel density or dispersed shareholderswhich is themost common ownership category in the United Statesor comparison. The di-most common ownership category in the United Statesor comparison. The di-erences shown across the categories are not primarily explained by dierences inerences shown across the categories are not primarily explained by dierences incountries or in type o industry.countries or in type o industry.One interesting group are the amily frms, defned in our research as frmsOne interesting group are the amily frms, defned in our research as frmsowned by the descendants o the ounder (so sons, daughters, grandsons, etc).owned by the descendants o the ounder (so sons, daughters, grandsons, etc).Those that are amily owned and also amily managed (Family, amily CEO)Those that are amily owned and also amily managed (Family, amily CEO)have a large tail o badly managed frms, while the amily owned but externallyhave a large tail o badly managed frms, while the amily owned but externally

Figure 4

Ownership and Management Scores

Source:Bloom, Genakos, Sadun, and Van Reenen (2009).Notes: Graphs show the distribution o frm management scores or frms with dierent types o

management. The overlaid line is the kernel density or dispersed shareholders, the most commonU.S. ownership type.

0

.5

1

0

.5

1

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

5+ shareholders

Family, external CEO

Family, amily CEO

Founder

Government

Managers

Private equity

Private individuals

Fr

actionofrms

Firm-level average management scores, rom 1 (worst practice) to 5 (best practice)


16/22


managed (Family, external CEO) look very similar to dispersed shareholders.managed (Family, external CEO) look very similar to dispersed shareholders.The reason appears to be that many amily-owned frms adopt a rule o primo-The reason appears to be that many amily-owned frms adopt a rule o primo-geniture, so the eldest son becomes the chie executive ofcer, regardless o talentgeniture, so the eldest son becomes the chie executive ofcer, regardless o talentconsiderations. Many governments around the world also provide strong tax subsi-considerations. Many governments around the world also provide strong tax subsi-dies or amily frms; or example, the United Kingdom has many more amily-rundies or amily frms; or example, the United Kingdom has many more amily-runand -owned frms than the United States and Germany, which is likely to be relatedand -owned frms than the United States and Germany, which is likely to be relatedto the estate tax exemption or inherited business assets in the United Kingdom.to the estate tax exemption or inherited business assets in the United Kingdom.Since amily frms typically have less debt, product market competition maySince amily frms typically have less debt, product market competition maynot be as eective in driving them out o business i they are badly managed. With-not be as eective in driving them out o business i they are badly managed. With-out debt, frms only have to cover operating costs (like salaries and wages) but notout debt, frms only have to cover operating costs (like salaries and wages) but notcapital costs like the rent on property or equipment because these were typicallycapital costs like the rent on property or equipment because these were typicallybought outright many years ago. Hence, amily frms can continue to generate posi-bought outright many years ago. Hence, amily frms can continue to generate posi-tive cash-ow while generating economic losses, because their amily owners aretive cash-ow while generating economic losses, because their amily owners aresubsidizing them through cheap capital.subsidizing them through cheap capital.Firms owned by private equity appear well managed, in particular when com-Firms owned by private equity appear well managed, in particular when com-pared to amily and government-owned frms (Bloom, Sadun, and Van Reenen,pared to amily and government-owned frms (Bloom, Sadun, and Van Reenen,2009b). Thus, the pattern in recent years o private equity frms purchasing frms2009b). Thus, the pattern in recent years o private equity frms purchasing frmsin Europe and Asia that were previously under amily or government managementin Europe and Asia that were previously under amily or government managementmakes some economic sense.makes some economic sense.A perhaps surprising result is that Founder frmswhere the current chieA perhaps surprising result is that Founder frmswhere the current chieexecutive ofcer ounded the frmare also badly managed. We are still tryingexecutive ofcer ounded the frmare also badly managed. We are still tryingto understand this phenomenon, but one potential explanation is that the entre-to understand this phenomenon, but one potential explanation is that the entre-prenuerial skills required o a start up, like creativity and risk taking, are not theprenuerial skills required o a start up, like creativity and risk taking, are not theprimary skills required when a frm grows large enough to enter our sample oprimary skills required when a frm grows large enough to enter our sample ofrms with at least 100 employees. A mature frm needs to move beyond inormalfrms with at least 100 employees. A mature frm needs to move beyond inormalrules, and these may be implemented more eectively by a proessional manager.rules, and these may be implemented more eectively by a proessional manager.We have also examined how the distribution across these ownership categoriesWe have also examined how the distribution across these ownership categoriesvaries across countries, since ownership can account or up to 40 percent o cross-varies across countries, since ownership can account or up to 40 percent o cross-country dierences in management practices.country dierences in management practices.66 In particular, we ocused on the threeIn particular, we ocused on the threeownership categories associated with the lowest management scores in our sample:ownership categories associated with the lowest management scores in our sample:amily frms with a amily chie executive ofcer, ounder frms, and government-amily frms with a amily chie executive ofcer, ounder frms, and government-owned frms. In developed economies like Germany, Japan, Sweden, and the Unitedowned frms. In developed economies like Germany, Japan, Sweden, and the UnitedStates, these categories as a group make up about 20 to 30 percent o the sampledStates, these categories as a group make up about 20 to 30 percent o the sampledfrms. By contrast, around three-quarters o our Indian frms are owned either by thefrms. By contrast, around three-quarters o our Indian frms are owned either by thefrms ounder or one o his descendants. In Italy, Brazil, Portugal, and Greece, thefrms ounder or one o his descendants. In Italy, Brazil, Portugal, and Greece, theshare o frms in our sample that all into these three categories is roughly 60 percent.share o frms in our sample that all into these three categories is roughly 60 percent.One likely explanation or this dierence is that the underdevelopment oOne likely explanation or this dierence is that the underdevelopment ofnancial markets and poor rule o law in many developing countries makes thefnancial markets and poor rule o law in many developing countries makes theseparation o ownership and control extremely difcult. For example, amilies mayseparation o ownership and control extremely difcult. For example, amilies may

6 For example, including a ull set o dummies or dierent ownership types reduces the R-squared o

country dummies in frm-level management regressions by 40 percent, which suggests about 40 per-cent o the cross-country var iation in management is associated with dierences in ownership.


17/22


be reluctant to hire outside managers because the law is not strong enough tobe reluctant to hire outside managers because the law is not strong enough toprotect them rom thet (La Porta, Lopez-de-Silanes, Shleier, and Vishny, 1997).protect them rom thet (La Porta, Lopez-de-Silanes, Shleier, and Vishny, 1997).Multinationals and ExportersMultinationals and ExportersFigure 5 plots the management scores by country or domestic frms (thoseFigure 5 plots the management scores by country or domestic frms (thosewith no production acilities abroad) and oreign multinationals. Two resultswith no production acilit ies abroad) and oreign multinationals. Two resultsstand out. First, oreign multinationals are better managed than domestic frms,stand out. First, oreign multinationals are better managed than domestic frms,presumably reecting the selection eect that better managed frms are morepresumably reecting the selection eect that better managed frms are morelikely to become multinationals. Second, oreign multinationals seem able tolikely to become multinationals. Second, oreign multinationals seem able topartially transport their better practices abroad despite oten difcult localpartially transport their better practices abroad despite oten difcult localcircumstances (Burstein and Monge-Naranjo (2009) oer a model consistentcircumstances (Burstein and Monge-Naranjo (2009) oer a model consistentwith these fndings). We also fnd that multinationals transplant other eatureswith these fndings). We also fnd that multinationals transplant other eatures

Figure 5

Multinationals Are Well Managed in All Countries

Source:Bloom, Genakos, Sadun, and Van Reenen (2009).

2.4 2.6 2.8 3 3.2 3.4

United States

Japan

Sweden

Germany

Canada

Australia

Italy

Great Britain

FrancePoland

Northern Ireland

Republic o Ireland

India

China

Portugal

Brazil

Greece

Foreign multinationals

Domestic frms

Average management scores


18/22


o their organizational orm overseas such as the average degree o decentraliza-o their organizational orm overseas such as the average degree o decentraliza-tion (Bloom, Sadun, and Van Reenen, 2009a). We also distinguished by exporttion (Bloom, Sadun, and Van Reenen, 2009a). We also distinguished by exportstatus. Consistent with the predictions o papers such as Helpman, Melitz, andstatus. Consistent with the predictions o papers such as Helpman, Melitz, andYeaple (2004), there is a pecking order: average management scores were lowestYeaple (2004), there is a pecking order: average management scores were lowestor non-exporters (2.6), next lowest or nonmultinational exporters (2.8), andor non-exporters (2.6), next lowest or nonmultinational exporters (2.8), andhighest or multinationals (3.2).highest or multinationals (3.2).Human CapitalHuman CapitalEducation is strongly correlated with high management scores, whether oneEducation is strongly correlated with high management scores, whether onelooks at the education level o managers or o workers. We cannot iner a causallooks at the education level o managers or o workers. We cannot iner a causalrelationship rom this association, o course, but it is plausible that managers withrelationship rom this association, o course, but it is plausible that managers withan MBA or college education are more likely to be aware o the benefts o mod-an MBA or college education are more likely to be aware o the benefts o mod-ern management practices like lean manuacturing. More surprisingly perhaps, isern management practices like lean manuacturing. More surprisingly perhaps, isthat worker-level education is also positively associated with management scores,that worker-level education is also positively associated with management scores,suggesting that implementing many o these practices may be easier when thesuggesting that implementing many o these practices may be easier when theworkorce is more knowledgeable. Many o the best practices in Table 1 depend onworkorce is more knowledgeable. Many o the best practices in Table 1 depend onsignifcant initiative rom workers, such as the Japanese-inspired lean manuactur-signifcant initiative rom workers, such as the Japanese-inspired lean manuactur-ing techniques and higher-powered incentives.ing techniques and higher-powered incentives.Our belie is that more basic business educationor example around capi-Our belie is that more basic business educationor example around capi-tal budgeting, data analysis, and standard human resources practicescould helptal budgeting, data analysis, and standard human resources practicescould helpimprove management in many nations, especially in developing nations.improve management in many nations, especially in developing nations.InormationInormationThe slow evolution o management practices across the United States,The slow evolution o management practices across the United States,Europe, and JapanTaylors scientifc management, Fords mass production,Europe, and JapanTaylors scientifc management, Fords mass production,Sloans M-orm corporation, Demings quality movement, and Toyotas leanSloans M-orm corporation, Demings quality movement, and Toyotas leanproductionsuggest that management practices do have a resemblence toproductionsuggest that management practices do have a resemblence toprocess technologies that diuse slowly over time. Slow technological diusionprocess technologies that diuse slowly over time. Slow technological diusioncan have many causes (as surveyed by Hall, 2003), but one well-studied actorcan have many causes (as surveyed by Hall, 2003), but one well-studied actorisis inormation. New management practices are oten complex and hard to intro-. New management practices are oten complex and hard to intro-duce without the assistance o employees or consultants with prior experienceduce without the assistance o employees or consultants with prior experienceo these innovations. Firms learn rom the experiences (good and bad) o oth-o these innovations. Firms learn rom the experiences (good and bad) o oth-ers in experimenting with dierent practices, so not all will adopt immediatelyers in experimenting with dierent practices, so not all will adopt immediately(or example, Acemoglu, Aghion, Lelarge, Van Reenen, and Zilibotti, 2007). An(or example, Acemoglu, Aghion, Lelarge, Van Reenen, and Zilibotti, 2007). Anexample is the two-decade struggle o U.S. automotive frms to replicate the Japa-example is the two-decade struggle o U.S. automotive frms to replicate the Japa-nese lean manuacturing system.nese lean manuacturing system.In our survey, we directly ask managers the question In our survey, we directly ask managers the question Excluding yoursel, how wellmanaged would you say your frm is on a scale o 1 to 10, where 1 is worst practice, 10 is best

practice and 5 is average. By this measure, frms have an average score o 7.1well. By this measure, frms have an average score o 7.1wellabove what should be the averageand this score is uncorrelated with either theabove what should be the averageand this score is uncorrelated with either themanagement score we give them or their frms own perormance on the dimen-management score we give them or their frms own perormance on the dimen-sions in Table 3 such as productivity and proftability. Hence, this fnding suggestssions in Table 3 such as productivity and proftability. Hence, this fnding suggeststhat managers are not well inormed about how good their own management prac-that managers are not well inormed about how good their own management prac-tices are and which areas need improvement.tices are and which areas need improvement.


19/22


A project in which we have been involved in India attempts to test this romA project in which we have been involved in India attempts to test this romanother angle by taking a sample o textile frms and providing a randomly selectedanother angle by taking a sample o textile frms and providing a randomly selectedsub-group with ree management consulting and comparing this to a control groupsub-group with ree management consulting and comparing this to a control groupwithout assistance (Bloom, Eiert, Mahajan, McKenzie, and Roberts, 2009). We fndwithout assistance (Bloom, Eiert, Mahajan, McKenzie, and Roberts, 2009). We fndchanges in management practices lead to signifcant improvement in perormance,changes in management practices lead to signifcant improvement in perormance,and the reason frms most requently suggested or not introducing these practicesand the reason frms most requently suggested or not introducing these practicesearlier was simply lack o awareness o these. Again, this suggests that improvedearlier was simply lack o awareness o these. Again, this suggests that improvedmanagement practices are a type o process innovation that takes time to diusemanagement practices are a type o process innovation that takes time to diuseacross frms.across frms.

Contingent ManagementContingent ManagementThus ar we have been assuming that management resembles a technologyThus ar we have been assuming that management resembles a technologyand there can be technical progress in management, just as there is or machines.and there can be technical progress in management, just as there is or machines.An alternative perspective is that all management practices are contingent onAn alternative perspective is that all management practices are contingent onthe frms environment (or example, Woodward, 1958). In this view, every frm isthe frms environment (or example, Woodward, 1958). In this view, every frm isadopting its own best practices given the circumstance in which it fnds itsel.adopting its own best practices given the circumstance in which it fnds itsel.There is certainly some element o contingency in management choices inThere is certainly some element o contingency in management choices inat least three respects. First, dierent frms (and indeed countries) specialize inat least three respects. First, dierent frms (and indeed countries) specialize indierent aspects o the managerial practices. For example, Japan ocuses moredierent aspects o the managerial practices. For example, Japan ocuses moreon monitoring than incentives/people management. Firms in highly skilledon monitoring than incentives/people management. Firms in highly skilledindustries and/or with lighter labor market regulations ocus relatively more onindustries and/or with lighter labor market regulations ocus relatively more onincentives/people management than on monitoring. Second, many aspects o stra-incentives/people management than on monitoring. Second, many aspects o stra-tegic management, such as pricing or takeover decisions, will be very contingent ontegic management, such as pricing or takeover decisions, will be very contingent onspecifc circumstances with no typical good or bad practicewhich is why ourspecifc circumstances with no typical good or bad practicewhich is why oursurvey looks only at a subset o the more process-oriented management practicessurvey looks only at a subset o the more process-oriented management practiceswhere it appears there is a more universal set o good practices. Third, the man-where it appears there is a more universal set o good practices. Third, the man-agement practices we use have not been equally benefcial throughout history. Foragement practices we use have not been equally benefcial throughout history. Forexample, rigorously and systematically using data to deal with problems and makeexample, rigorously and systematically using data to deal with problems and makedecisions has been acilitated in more recent times by the dramatic all in the realdecisions has been acilitated in more recent times by the dramatic all in the realcost o inormation technology.cost o inormation technology.But with these elements o contingency readily acknowledged, our work sug-But with these elements o contingency readily acknowledged, our work sug-gests that contingency is not the whole story. As Table 2 showed, better-managedgests that contingency is not the whole story. As Table 2 showed, better-managedfrms within the same country and industry are

Documents

Why Do Managment