Annals of technology diffusion

Evidence of $100 bills lying on (Indian) sidewalks:

we ran a management field experiment on large Indian textile firms. We provided free consulting on modern management practices to a randomly chosen set of treatment plants and compared their performance to the control plants.

We find that adopting these management practices had three main effects. First, it raised average productivity by 11% through improved quality and efficiency and reduced inventory. Second, it increased decentralization of decision making, as better information flow enabled owners to delegate more decisions to middle managers. Third, it increased the use of computers, necessitated by the data collection and analysis involved in modern management.

Since these practices were profitable this raises the question of why firms had not adopted these before. Our results suggest that informational barriers were a primary factor in explaining this lack of adoption. Modern management is a technology that diffuses slowly between firms, with many Indian firms initially unaware of its existence or impact. Since competition was limited by constraints on firm entry and growth, badly managed firms were not rapidly driven from the market.

A new paper by Bloom, Mahajan, McKenzie and Roberts.

Never forget that management–basically the efficient organization of production–is possibly the great technological achievement of our age. And has yet to fully diffuse.

It’s these $100 bills that make me a development optimist.

17 Responses

  1. The most interesting part of that study lies in its findings on control and delegation. It paints a fascinating picture of these family-controlled firms – the male family members work incredibly hard (on average, 68 hours/week) but don’t trust their subordinates. That’s partly because they lack the data to monitor them. But it’s also because India’s dysfunctional legal system makes it impractical to enforce honest conduct. Trust doesn’t extend beyond the family.

    The practical consequence is that firms can’t own more plants than they have male family members, and individual plants have trouble expanding beyond the scope of a single individual’s oversight. The most efficient firm in the study has only one plant, because its owner has neither sons nor brothers.

    So they key intervention here wasn’t management practices per se, but rather, the institution of systems of data collection that allow the family owners to monitor and hold accountable their employees in real-time – and that, in turn, allow them to risk greater delegation of authority, of the sort necessary to implement the recommended changes.

    Almost all of the other factors cited pale in comparison. Even if most firms labor under other burdens the study identified – lack of information, incorrect information, procrastination, and financial constraints – the fundamental barrier remains span of control. All it takes is a few firms, capable of unlimited expansion, to adopt best practices and drive out their competitors. So the same constraints, applied in an environment in which any single firm can expand without limitation, cease to be relevant. And removing all the other constraints – but maintaining the limits on size imposed by male kinship networks – still results in an inefficient industry, in which a separate (and presumably, periodic) intervention is required for every cluster of family-held plants.

    Daniel McCallum had this figured out 155 years ago, when his annual report for the Erie Railroad explained in painstaking detail the link between good data and accountability – and how this, in turn, enabled much more complete delegation of authority. It’s how America solved the problem of scale at the dawn of our railroad age. And it’s discouraging to see developing nations still facing the same hurdles all these years later.

    You can hire all the McKinsey Consultants you want, and send them into the world’s factories one by one. Or you can fix the legal systems of the developing world, and allow the best-managed firms to expand beyond the control of a single family. One solution costs a quarter-million dollars for each separate intervention, with uncertain results. The other is costlier, in the short term, and vastly more complicated – but the good that will flow from it will be far, far more broad in its effects.

  2. For those interested in a succinct writeup of some of the underlying data from Bloom on management practices around the world–including the finding that the best Indian companies are as well managed as any others but that there is a vast stock of very poorly managed companies who do not face enough competitive pressure to be forced to improve–he wrote about the findings on the HBR blog last week:

    http://blogs.hbr.org/cs/2011/06/why_american_management_rules.html

  3. Only 11%? That’s not a $100 bill, that’s a dime and a penny. India’s per-capita income is what, 10% of ours?

    Not the worst thing in the world, but I’m honestly astonished that the gains were so small, given how ridiculous the levels of waste generally are around untrained managers.

    1. It’s a good point – the direct short-run (within one year) impacts are not huge. That’s about 4 or 5 years productivity growth. I think in the long-run they could be much larger as I think well managed firms could be yielding an extra 10% a year for probably decades in India given how low the base is. That was certainly the experience of Toyota in Japan which went for being low productivity in the 1950s to best in the world by the 1990s through continuous improvements.

  4. Charles Kenny discusses “process technologies,” which include management practices, in “Getting Better.” He says that process technologies tend to be “sticky,” i.e. they do not flow easily across borders and cultures. He argues that the centrality of process technologies gor growth, or rather their stickiness, was an important factor in the divergence of development outcomes between the West and the rest. That’s what I remember, anyway (I’m on the train so I can’t factcheck my memory). Maybe you will find it to be a useful reference..?

    1. Thanks very much and we’ll follow that up and sounds very intuitive – hard produce technologies are easier to codify (think patents) while process technologies are much harder. Completely agree and I think it’s a big reason as you suggest for slow spread of modern management.

  5. I saw N. Bloom present this, and he actually said–if I remember right–that this intervention didn’t clearly pass a cost-benefit analysis (they used expensive consultants). Maybe however a simpler one would? Would be interesting to discuss that aspect further.

    1. Yes – very much. Basicall in the first year they would about break even and only make profits from the second year onwards. Now if you assume these changes last that’s a long-run profit, but discount rates in India can be high and these changes may not last (we are hoping to revisit this summer to find out).

      So it’s not obvious it they be profitable for any individual firm to hire in Accenture, although it probably would make sense to get in the local consultants that could do 50% of the work for 10% of the cost. Although as noted above the main reason they don’t if they think they are well managed to have no need for consultants…

  6. But how do you manage to control for the effect where you change something and efficiency improves just because it’s new. Didn’t Frederick Winslow Taylor find that? I think with a group of women workers, they changed to a new working method and efficiency improved, then to another and efficiency improved again. Then back to the start and again it improved.

    1. Yes great point – the classic reference were the Hawthorn studies were basically the finding was just having guys with clipboards taking data led to improvements in performance. So simply running an experiment can potentially induce improvements even if the intervention would normally not be helpful.

      This is definitely something we worried about a lot. We don’t this this is a major because because these improvements are very long-lasting (in the Hawthorne experiments they were short lived), and occur in the areas impacted by the management changes despite collecting data across the whole plant. It is also the case our firms copied these changes to other plants (so they also believed they were helpful) and our control firms were also monitored (so both had guys collecting data crawling around).

      But again totally agree that designing these experiments is tricky and these measurement issues are critical to get right. I’m actuall a big fan of Taylor who originally pushed scientific management.

  7. This is one of the reasons explaining the very poor management in many public hospitals…….

    1. Neither doctors nor lawyers are trained in management; they are trained for their professions. Unfortunately, the assumption in many workplaces involving such professionals is that if someone is good at his profession, he will be good at managing other people to do that job as well. It’s not necessarily the case.
      Also, I think Atul Gawande is right that medicine in particular must move toward a more systemic understanding of patient care in order to reduce both errors and over-expenditure.

      1. Yes – that’s a great point and I think motivates doing studies on management.

        One school of thought is that practices are all contingent – that is what works for the Americans won’t work for the Japanese and visa-versa. Think drunken Karoke with your boss….. The other school of thought is there are some universal best practices – one example is the Toyota Production system, which came from Japan but is now globally copied in manufacturing, retail and most recently healthcare.

        In terms of why they were bad the main reason seems to be informational. They simply were either not aware of the modern practices, or if they had heard of them were skeptical. If you read popular books like “The machine that changed the world” about the US car industry slowly waking up to Japanes management, or “Moneyball” about Baseball learning to use player statistics you can see how if even in the US if good practices take decades to spread this could easily occur in India.

  9. This is fascinating, Chris.

    I’d be interested to know what creates the barriers to diffusion. One possibility is that more of the key ‘technology’ needed to run a business is protected by intellectual property (e.g. software licences).