10 conclusions of an excellent analysis of firm management: Firms with “better” management practices tend to have better performance on a wide range of dimensions: they are larger, more productive, grow faster, and have higher survival rates.
You are right observational studies cannot establish causality. That is why they have a randomized trial where some textile factories are treated by a management consultant and others are not, based on the criteria laid out in this paper. They find huge differences in any metric of firm performance that you can think of. Do you honestly think that they do not know how to compute standard errors? The name of the firm is anonymous because that was the condition under which they could do this research.
Your strong tone of self-righteousness is completely uncalled for. The JEP is a summary journal, not a technical journal for economists. If you want more technical detail, read any of their other papers.
The paper isn't really clear about what they have done.If I understand right:1) They made the performance metric based on theoretical assumptions before they looked at the data.They weighting of the different question is determined before they look at the data."This evaluation tool was developed by an international consulting firm"Why isn't the name of the firm disclosed? It's something we should know when we want to know whether the metric is biased.Consulting firm have an incentive to create metrics that tell companies that the consulting firm improves their management and aren't neutral arbiters of the academic quest for truth.2) After admitting "clearly establishing thecausal effect of how changes in management affect productivity is not possible." they make two arguments for their metric:2.1) Per country measures correlate with GDP. If good GDP means good management we could look directly at the GDP and don't really need their metric if we want to compare countries based on GDP.2.2) When they try to predict values like profit they get significant results. Even through they have 3,627 firms in their survey they don't manage to get a result that's significant (p<0.05).
There metric is better than guessing at random but I don&#039t see that it&#039s powerful at predicting success.Even a bad performance metric can be significant when it&#039s better than random and you have enough data points.
They should have given confidence intervals for the country ratings of overall management performance.Why should I trust someone who doesn&#039t know the size of his own error?
They could have trained a model for their metric based on their questions and ~1500 firms and used the other ~1500 firms to evaluate the country differences and differences between different types of ownership.
I don't they asked subjective questions like "do you motivate your employees?" but instead objective questions like "what percentage of your employees's salary is tied to performance?"
There is a very interesting new subfield in economics that studies the determinant of micro-level firm productivity. See this survey http://home.uchicago.edu/~s...for an overview. The concerns of the posters above are a little silly, do you guys think that academic microeconomists at top universities are not careful about their data work? If there was obvious bias in the sample, their work would get shredded at seminars.
"Over 850 British firms (i.e., Great Britain and Northern Ireland – why no Scotland or Wales?) but less than 700 American firms. America is MUCH bigger. That screams biased sample"
RE, the author of this study, Nick Bloom is British. He did this research back when he was at LSE using mostly British grad students in a call centre in central London. There is no conspiracy, but there is a bias in his sampling.
Here is an interesting question: are managers in certain cultures more likely to participate in lengthy phone interviews with academics? I heard that a relatively high %age of people in the US refuse to complete the census.
Over 850 British firms (i.e., Great Britain and Northern Ireland - why no Scotland or Wales?) but less than 700 American firms. America is MUCH bigger. That screams biased sample.
Having worked in "best practices" research for some time, it is as "black swan" blind as research can get. The number of firms that I've seen profiled for following "best practices" that no longer exist (or are now state-owned) is laughable.
Maybe this research tells us that the U.S., Germany, and Sweden have better managed firms than Brazil, India, China, and Greece, but that's it.
#2 suggests that letting bad firms go out of business is an important factor here. (I assume there are other explanations, but this is the one that seems to me the obvious first guess.) The 1978 bankruptcy law reform had an explicit goal of keeping more bankrupt firms intact, pushing firms toward chapter 11 rather than liquidation, and giving incumbent management more control in chapter 11 at the expense of the creditors (who are, at that point, the primary residual claimants). I wonder whether this has lead to a deterioration in management quality in the US in the last 31 years.
One has to wonder how important the median firm is on a macro level. I'd be more interested in a comparison of firms at the ninetieth or ninety-fifth percentile - intuitively, I would expect the best managed companies to have a disproportionate effect on growth.
You know, it is interesting, that the scale is 1 to 5 and the US 3.3 is made to look good because the scale is cut off on the right side because it doesn't extend to 5.
Agree with this point in the article based on experience in dealing with management at various firms:
"Seventh, inherited family-owned firms who appoint a family member (especially the eldest son) as chief executive officer are very badly managed on average."
Fixed; thanks.
Dead link. "Why Do Management Practices Differ across Firms and Countries?", Bloom & Van Reenen 2010 https://pubs.aeaweb.org/doi... http://houdekpetr.cz/!data/...
If they have calculated their standard errors why don't they publish those in their paper?
You are right observational studies cannot establish causality. That is why they have a randomized trial where some textile factories are treated by a management consultant and others are not, based on the criteria laid out in this paper. They find huge differences in any metric of firm performance that you can think of. Do you honestly think that they do not know how to compute standard errors? The name of the firm is anonymous because that was the condition under which they could do this research.
Your strong tone of self-righteousness is completely uncalled for. The JEP is a summary journal, not a technical journal for economists. If you want more technical detail, read any of their other papers.
The paper isn't really clear about what they have done.If I understand right:1) They made the performance metric based on theoretical assumptions before they looked at the data.They weighting of the different question is determined before they look at the data."This evaluation tool was developed by an international consulting firm"Why isn't the name of the firm disclosed? It's something we should know when we want to know whether the metric is biased.Consulting firm have an incentive to create metrics that tell companies that the consulting firm improves their management and aren't neutral arbiters of the academic quest for truth.2) After admitting "clearly establishing thecausal effect of how changes in management affect productivity is not possible." they make two arguments for their metric:2.1) Per country measures correlate with GDP. If good GDP means good management we could look directly at the GDP and don't really need their metric if we want to compare countries based on GDP.2.2) When they try to predict values like profit they get significant results. Even through they have 3,627 firms in their survey they don't manage to get a result that's significant (p<0.05).
There metric is better than guessing at random but I don&#039t see that it&#039s powerful at predicting success.Even a bad performance metric can be significant when it&#039s better than random and you have enough data points.
They should have given confidence intervals for the country ratings of overall management performance.Why should I trust someone who doesn&#039t know the size of his own error?
They could have trained a model for their metric based on their questions and ~1500 firms and used the other ~1500 firms to evaluate the country differences and differences between different types of ownership.
I don't they asked subjective questions like "do you motivate your employees?" but instead objective questions like "what percentage of your employees's salary is tied to performance?"
There is a very interesting new subfield in economics that studies the determinant of micro-level firm productivity. See this survey http://home.uchicago.edu/~s...for an overview. The concerns of the posters above are a little silly, do you guys think that academic microeconomists at top universities are not careful about their data work? If there was obvious bias in the sample, their work would get shredded at seminars.
Great Britain = England + Scotland + Wales.
"Over 850 British firms (i.e., Great Britain and Northern Ireland – why no Scotland or Wales?) but less than 700 American firms. America is MUCH bigger. That screams biased sample"
RE, the author of this study, Nick Bloom is British. He did this research back when he was at LSE using mostly British grad students in a call centre in central London. There is no conspiracy, but there is a bias in his sampling.
Here is an interesting question: are managers in certain cultures more likely to participate in lengthy phone interviews with academics? I heard that a relatively high %age of people in the US refuse to complete the census.
This is a bit weird.
Over 850 British firms (i.e., Great Britain and Northern Ireland - why no Scotland or Wales?) but less than 700 American firms. America is MUCH bigger. That screams biased sample.
Having worked in "best practices" research for some time, it is as "black swan" blind as research can get. The number of firms that I've seen profiled for following "best practices" that no longer exist (or are now state-owned) is laughable.
Maybe this research tells us that the U.S., Germany, and Sweden have better managed firms than Brazil, India, China, and Greece, but that's it.
Christian above, nailed it: this is just silly.
#2 suggests that letting bad firms go out of business is an important factor here. (I assume there are other explanations, but this is the one that seems to me the obvious first guess.) The 1978 bankruptcy law reform had an explicit goal of keeping more bankrupt firms intact, pushing firms toward chapter 11 rather than liquidation, and giving incumbent management more control in chapter 11 at the expense of the creditors (who are, at that point, the primary residual claimants). I wonder whether this has lead to a deterioration in management quality in the US in the last 31 years.
One has to wonder how important the median firm is on a macro level. I'd be more interested in a comparison of firms at the ninetieth or ninety-fifth percentile - intuitively, I would expect the best managed companies to have a disproportionate effect on growth.
Actually the scale exaggerates the differences by cutting off both sides of the scales. The scale is still centered so not too misleading.
You know, it is interesting, that the scale is 1 to 5 and the US 3.3 is made to look good because the scale is cut off on the right side because it doesn't extend to 5.
Much room for improvement.
david and Christian, see the paper - their metric consistently predicts many positive qualities like profits, sales growth, etc.
Agree with this point in the article based on experience in dealing with management at various firms:
"Seventh, inherited family-owned firms who appoint a family member (especially the eldest son) as chief executive officer are very badly managed on average."