Glenn Reynolds, aka Instapundit, wrote a great op-ed piece in the Washington Examiner recently, on the impact the Japanese earthquake and tsunami are having on supply chains around the globe.

Quoting the Professor in full:

Japan’s earthquake/tsunami has occupied the news and also spurred a lot of thought. Among other things, it has underscored the fragile and interconnected nature of modern society, and caused some to question the wisdom of “just-in-time” manufacturing approaches in today’s unsettled world.Instead, it is suggested, we might want to focus on “just in case” approaches designed to be more resilient under stress.

Japan’s earthquake was in some ways a triumph of preparedness: Thanks to strict building codes, not a single building in Tokyo collapsed. But the earthquake, and the tsunami it produced, have had impacts that go well beyond the immediate.

In particular, the damage is exposing the extent to which modern supply-chain management has produced a system that is so lean it lacks the reserve capacity needed to cope with disasters.

In manufacturing, plants have been idled around the world because Japanese factories — or often, a single Japanese factory — serve as the sole source for a vital component. With the factories sidelined by damage or power outages, the components are unavailable, and production has to stop.

Ford Motor Company idled a plant in Belgium for five days over parts shortages; Toyota warned plants in the United States to be prepared to close for the same reason. A U.S. plant making car seats had to close because of a shortage of premium vinyl made only in Japan. Ford has suspended orders for some models in red and black because the paints come from a single factory in Japan, now closed. Tales like these abound.

Even the New York subway system is affected by the parts shortage: As National Public Radio’s “Marketplace” reported: “Steel from the north of Japan can’t get to Suzuki. Suzuki can’t make the parts for Hitachi. And Hitachi can’t send the parts to New York. The global supply chain breaks down with the removal of just one link.”

As Edward Tenner noted in the Atlantic: “The tsunami has exposed a weakness in global logistics long recognized in principle but disregarded in practice. Lean manufacturing plus heavy reliance on a single plant equals vulnerability to disruption.”

With managers under pressure to keep costs down, there has been a tendency to cut special deals with single suppliers, and to keep stocks of parts as low as possible. So long as everything goes smoothly, this saves money: Single suppliers give you the best price, and low inventories keep you from tying up working capital.

The problem is that we seem to be in a period where things aren’t going as smoothly as they did for a while. And when things don’t go smoothly, the lean approach means that it doesn’t take much to bring things to a halt.

I mentioned this to a friend who’s got a custom-car business, and he said his experience with disruptions in getting supplies from vendors has caused him to move from a “just-in-time” system to what he calls a “just in case” system, where stockpiles are bigger and alternative suppliers are identified in advance. I think we’ll be seeing a lot more of this, post-Japan.

But the problem goes well beyond cars and subways. Lots of more important systems are similarly vulnerable. My wife takes a heart-rhythm drug called Tikosyn; if she misses a dose, she could die.

Walgreen’s doesn’t want to keep it in stock, so they order a bottle by air-freight when her prescription is about to expire. Normally, that’s fine — but if something happened to interrupt shipping, she’d be in trouble.

She keeps a backup supply, but what would Walgreen’s do for others in a similar predicament? A few days of shipping problems and many pharmacies would be out of important drugs.

Likewise, grocery stores now keep only a small supply of food on hand, depending on regular deliveries for restocking. When those deliveries are interrupted, shelves start to empty pretty fast. (And government emergency food stockpiles are nothing like they were in the Cold War era).

Power plants used to keep a 60-day supply of coal in stock. Now they typically keep only 30 days’ worth. That saves utilities money but it means that there’s less margin if deliveries get interrupted. In the past, severe blizzards have left some utilities dangerously close to running out. Most cities have only a few days’ worth of gasoline.

We’ve come off a period of several decades in which weather was better than average, and in which other forms of societal disruption were fairly minor. The 21st century looks likely to be less placid.

As we make all sorts of plans, at the governmental, the business, or the personal level, it will pay to think more about the likelihood that things won’t go smoothly, and about ways we can prepare now to deal with the inevitable problems ahead.

A new subdiscipline called “resilience engineering” looks at how systems can be made more resistant to failure, and better able to recover when they do fail. That kind of thinking, it seems to me, is relevant to all of us, not just engineers.

The ride seems to be getting bumpier. In all sorts of areas, we need more of a cushion.

This advice is similar to advice I’ve been giving for years when it comes to a supply chain in China. Simply put, “your order will be late; plan accordingly”. When I was supplying parts to a large American motorcycle company, I knew that I had to buffer uncertainty with inventory. They often couldn’t give reliable forecasts (or I should say, their forecasts didn’t always filter through the other supply chain members reliably), so I had to keep weeks or months worth of product on the floor to compensate. And that was just the customer – the suppliers were a whole other ration of uncertainty. This is the dirty little secret of the JIT revolution – the inventory levels are often still there, just held by someone other than the OEM.

If you actually both to read past the first chapter in any book on JIT, you will see that stability is a prerequisite for a JIT system – stability in demand and in supply. The system can’t work in chaos. Most Chinese supply chains are far too chaotic to be part of a JIT system (for Apple this is not the case, but chances are if you are reading this blog, the rule applies).

I could go into the math of how to calculate an inventory level based on your target fulfillment goal, but suffice it to say, you are wise to keep some stock on hand. Much cheaper than airfreight.

As for single sourcing, some products don’t have the volume to have multiple sources, but whenever we have a client with substantial order quantities, I always recommend having at least two sources for each component. I also encourage clients to develop domestic sources to support 20% of the normal production, and make sure they have capacity to handle increased orders if there are problems with the Chinese suppliers.

So, when you come to China, do NOT expect to run the supply chain the way the new text books say. Go old school and put some fat in the system. Just in case.