April 15, 2020
A few years ago, I went to a talk given by an acquaintance on the topic of Lean Manufacturing. I may be mistaken since it's been a while, and maybe the exact sub-category of management philosophy wasn't "Lean," but it was in that family. In general terms, the talk was about all of the ways in which it is beneficial to cut anything not absolutely necessary to production. The message of the talk bothered me at the time, and it still does.
Nobody can argue with the idea that cutting waste is good, and there's no question that there can be all kinds of ancillary benefits to running Lean. For instance, if the river of processes and materials that feed into production is too wide, then there is a cost to that. The cost of continuously carrying an investment in these materials is obvious, but there are less obvious costs too. All of that stuff needs storage, tracking it takes time and effort, and if it's held in or near the assembly facility, then there can be a subtle psychological cost since the shop floor can feel chaotic and disorganized, no matter how well organized it is in the abstract. On the surface, and properly done, Lean sounds like a no-brainer.
After the talk, the presenter and I talked about my concerns. I recall using the analogy of traffic on a busy highway. If everyone is driving at full speed bumper-to-bumper, then one small glitch can lead to a massive pileup. People die, dozens of cars are totaled and the highway is closed for an entire day. My friend's response was essentially that, with good planning, the likelihood of a serious problem is negligible.
But human beings are terrible at judging risks, and we are especially bad at high-stakes risks. In the financial world, the standard model of stock prices assumes a log-normal distribution of returns. Everyone who works in finance knows that this model is wrong (or they should). The log-normal assumption underestimates the risks and does not take into account sudden shocks or the ways in which herd behavior plays out, but it is a very good model under ordinary circumstances. There is some debate about this, but a more correct model is the Levy-Pareto distribution. The problem is that it's much harder to work with (the math is messier) and it's difficult to know exactly what the parameters should be. We simply don't know what the correct distribution is. Stock pricing is a simple system compared to the economy as a whole, and even people whose job it is to think dispassionately about these risks and probabilities don't know the right answer. Paraphrasing Donald Rumsfeld, "there are known unknowns and unknown unknowns."
These unknowns are what bother me about the Lean approach, and we're seeing some of the shortcomings play out right now as the economy is jolted by the coronavirus. Donald Trump said that the US "wasn't built to be shut down," and he's right. His being right about this may be a monkey-typewriter-Shakespeare kind of situation, but he is right, although my guess is that he was complaining primarily about the fact that there's no magic button he can push to shut everything down. Where he is right is that any attempt to shut down portions of the economy leads to unforeseeable outcomes.
The economy is too complicated for us to rely on the assumption that what worked yesterday or last year will work today, and that is precisely what a Lean approach does assume. I am reminded of a kayaking trip through a large marshy area near Juneau, Alaska, just as the tide was reaching its low point. The marshes extend for miles, and depending on the exact level of the water relative to all of the various drainages, the current might be with me or against me. One minute I was being pushed along nicely, and the next minute I was fighting the current and losing; and a few minutes later it was with me again. It was an odd and frustrating experience, and it was only later that I realized what must have been happening. The interactions of the various inputs to the system were hidden and too complex for me to understand and predict.
Normally, when currents shift in the economy, it happens slowly – small changes happen over days or months, and larger shifts happen over years – and we have time to respond in a way that minimizes shocks and dislocations. If the change is sudden, then some currents are blocked and they cut a new and potentially destructive path; some currents dry up leaving those upstream high-and-dry; and some currents swirl chaotically without giving us time to understand them and turn them in a useful direction before their energy is wasted. A Lean approach to supply chains only magnifies this problem.
A couple of examples are well-known. The famous N95 masks are made in only a few places, and the supply chain was, in hindsight, too Lean. Every day in the news, we hear of other examples in the area of medical supplies and equipment. Meat packing is another example that's been in the news recently. The bulk of US meat production passes through a small number of facilities which act as bottlenecks. If too many people who work in those places get sick, then the entire chain from farm to table suffers dislocation. This is not to say that there is currently a serious risk of dangerous shortages, but there will be economic costs.
A more personal example is my local power company. We had a moderately bad snow storm here in Maine, and we were without power for five days (it came back this morning). Due to restrictions in how the repair crews are able to work (one person in each truck, etc.), and the difficulty of bringing in crews from outside the state, it was more time-consuming to make the repairs. The power company's repair infrastructure was too Lean for the circumstances. Given how often we suffer power outages, and the fact that so many of my neighbors choose to have backup generators, I think it's always too Lean, but that's a discussion for another day.
If the Lean philosophy leads to these kinds of problems, then what's the solution? It's a tragedy of the commons situation. A Lean approach does make sense almost all of the time, and for almost every business, under the normal environment, but when things go wrong, everyone suffers.
It's not reasonable to attempt a solution by asking 3M (say), out of the goodness of their heart, to build their N95 mask factories so that these factories normally run at 70% capacity. Their competition would eat their lunch. The government can't mandate this kind of short-term inefficiency for 3M and all of its competitors either, since other countries would then eat our nation's collective lunch. And stockpiles aren't the complete answer. The risk of dislocation in certain areas is so dire that sometimes a stockpile makes sense, like the Strategic Petroleum Reserve or the various stockpiles held by FEMA, but we can't have a federal stockpile for every contingency, foreseen and unforeseen.
A modern market-driven economy is Lean by design. We've chosen to walk on this knife-edge because the benefits are so obvious and so large. But maybe the knife-edge can be made more like an elevated path with some room to maneuver.
Mandated paid leave (vacation plus sick time) for everyone is an easy place to start. According to Pew Research, something like 24% of all US workers get no paid leave at all (33.6 million people), and the less a worker is paid, the less likely he or she is to get paid leave. Pandemic or not, that's ridiculous. Nobody should ever feel pressure to work while they're sick, if for no other reason than that they'll make other people sick. Showing up for work sick should be socially unacceptable under any circumstances. Even a simple cold lowers productivity, so it's not clear that this would cost the broader economy anything at all, even under ordinary (non-pandemic) circumstances.
The use of contractors in the gig economy is another place to look. It's too easy for employers to eliminate these employees and this exacerbates swings in the broader economy. Aside: Yes, I am aware of the semantics here; the companies involved would prefer that we think of their employees as "contractors" and the companies as the "client" of these contractors. That's dishonest.
A limited amount of protectionism is appropriate. No country should be in the position of relying on another country for crucial items like medical supplies or food. If an item is of true strategic importance, then taking protectionist action would not be seen as offensive or aggressive by other countries. Smart countries should encourage others to be more self-sufficient when it comes to basic well-being. Once a certain aspect of a country's economy (like medical equipment) is somewhat isolated from global competition, it would be possible to require that they overbuild their capacity.
In the same vein, a stricter anti-trust regime should be put in place. Basic materials and products should not be subject to supply bottlenecks, as they tend to be when production becomes concentrated in fewer hands. If too much of a given item is coming from a single country or only a few facilities, then that's a problem. Production and distribution should be dispersed, with supply chains that are shorter and more transparent.
All of these ideas can be summed up by saying that the economy should be less Lean and less focused on short-term efficiency, with more stored potential energy and less active kinetic energy. That seems like a hard pill to swallow, but it's entirely unclear whether the usual measures of efficiency mean anything when applied to the economy as a whole. According to the book Bullshit Jobs: A Theory, by David Graeber, an anthropologist at the London School of Economics, close to half of all jobs are pointless or even pernicious. He defines a "bullshit job" as one which even the person who holds the job considers to be socially useless. Even if Graeber's theory is way off base, and only 20% of all jobs are bullshit, that is still a huge amount of waste. The economy is already extremely inefficient; it's just inefficient in the wrong way.