Volcanos and other risks

April 21, 2010

Predictably, there are stories in the news about how the volcano in Iceland is disrupting world supply chains  (e.g., BBC story on Nissan and BMW).

One of the trends over the last 20 years is to source globally and to source from fewer suppliers. The reasoning is good. By sourcing globally you find the supplier with the lowest costs and best skills/technology. By sourcing from fewer suppliers you avoid variation in your products and you are better able to concentrate improvement efforts, both in the design of the component as well as in the manufacturing process. But the strategy also means that you become more vulnerable to idiosyncratic risk.

As risks go, this one is relatively mild (that a volcano erupts under a glacier sending ash into a big cloud that blankets Europe).  Assuming the volcano stop spitting ash into the sky soon, the consequences of this disruption are rather limited – about 7-10 days of lost production. This is a significant disruption, but nothing compared to the disruption that would occur if your supplier’s facility were buried in ash. Put another way, the risks to really worry about are ones that disrupt your supply chain for a much longer period of time – hurricanes, floods, earthquakes, lightning/fire, and domestic unrest all come to mind as potentially more severe in the sense that they can knock you out for months. Thus, I would be much more concerned about sourcing from a supplier in Indonesia (earthquake and volcano prone) than a supplier in Ireland.

The first step to managing these risks is to identify them – you can’t manage what you don’t know. (Or put another way, it is harder to manage “unknown unknowns” than “known unknowns”.)

The second step is to establish a monitoring system – when would you know if a supplier in your network has been disrupted? We all heard about the volcano in Iceland, but when would you hear if a key Tier 2 supplier had a fire in their facility? The sooner you know, the sooner you can get to step 3…

The third step is to have a contingency plan in place. Do you have potential second sources of supply? What assistance can you provide to shorten the length of the disruption?

To conclude, disruption risk is real, but it doesn’t mean that you shouldn’t source globally for a limited set of suppliers.

The promise of cloud computing

April 19, 2010

Cloud computing is all the rage! In case you are not sure what “cloud computing” is, it is having another company manage your data and applications. For example, instead of having a spreadsheet program on your desktop PC along with spreadsheet files, with cloud computing, the code and all of your data reside on a server somewhere, accessible via any device hooked up to the web.

It has been popular with small start up companies, but even the internet “giant” Netflix has joined the party – Amazon may be a competitor in the retail side of the business, but Amazon now manages Netflix’s data (see NY Times, 4/19/10).  And Amazon is by no means the only provider of these services – they are in good company with the likes of Microsoft and Google.

While lots of people talk about how cool cloud computing is, there is very little discussion regarding whether anybody will make a lot of money out of it. Maybe this is because while it could grow to be a huge revenue business, it is likely to remain a rather unprofitable one. Why? There are several reasons.

Let’s start with the comparison set. What other business sells “business process services”? … why of course, contract manufacturers (Foxconn, Flextronics, Jabil Circuits, etc.). Instead of making your latest mobile/cell/smart phone, have somebody else do it for you. There is a profitable transaction here because the contract manufacturer pools the demands across many clients. Consequently, the contract manufacturer experiences less volatility in demand than any one of the clients. It follows that the contract manufacturer can operate its assets with a higher utilization while providing at least as good service.  In other words, the contract manufacturer can do something that the clients cannot do as well, hence they can charge a “premium” for this service. However, the gross margin of contract manufacturers are generally in the 5-10% range – those are rather thin gross margins indeed. In other words, if cloud computing is like contract manufacturing, then there is reason to believe the cloud computing firms will have small margins as well.

Now consider the fundamentals of this business. Selling cloud computing means you are dealing with highly price sensitive customers (Netflix will negotiate more aggressively than Sally’s Organic Honey) and you are selling a product that arguably is a commodity – is one server better than another server? Put another way, it is unlikely, like the contract manufacturers, that the cloud computing providers will have any brand equity.  Finally, when you are competing against the likes of Amazon, Microsoft and Google, you shouldn’t expect the competition to be gentile.

So why are companies jumping into the cloud computing space? Well, for one, it will be big business on the revenue side. Contract manufacturers  need to find a set of clients that have similar manufacturing needs. This means that you can serve Nokia and Apple, but it is unlikely that you can also serve CitiBank, DuPont and Ryder Trucking with the same manufacturing equipment and labor. In other words, while the market for a particular kind of manufacturing is “limited”, the market for computing services must be orders of magnitude larger due to the fact that every business uses computers and data processing.  And, while the margins are not likely to be large, they will be positive because these companies are creating value.  Hence, a small percentage of a very large revenue number cannot be ignored. Nevertheless, the return on invested capital for this business will not create eye popping numbers like Google’s ad business.

What is wrong with Toyota

April 14, 2010

What is wrong with Toyota? Isn’t the Toyota Production System flawed, after everything the news has written in about the accidents resulting from unintended acceleration?

If one wants to make sense from the current crisis (or, even better, learn for the future), it is important to separate the hype from the facts. First, one should not confuse Toyota with the Toyota Production System. Toyota might have failed, but this says nothing about the Toyota Production System itself. If you see your doctor smoke, does this mean what he told you about lung cancer was not true?

Second, as operations experts, we have to look at the numbers. True, every life lost is tragic, and so the 50+ fatalities currently associated with unintended acceleration are not to be taken lightly. But, let’s keep in mind that:
(a) in the US 30,000 people die in car accidents every year (of which 5 to 10 per year were a result of unintended acceleration). This is less than 0.1%.
(b) the number of reported incidents has sky rocketed after the story was in the news, and many of the stories that now appear on television are highly questionable
(c) Experts estimate that every year, >50,000 people die in the US because of infections they acquired in the hospital (nosocomial infections). The main reason: doctors and nurses don’t wash their hands often enough.
Thus, it appears that if you are concerned about your life or the life of others, there might be better opportunities of making this country safer. And, as we explained in this blog before, hospitals are turning to the Toyota Production System as a way to improve patient safety.

Finally, it is helpful to contemplate the question “Why did Toyota not react to all of this?”. There are two reasons for this:
(a) Cars have become increasingly complex and the number of failure opportunities has grown dramatically, especially at the interface between electronics (modern gas pedals basically are like a computer mouse that translate mechanical movements into electronic signals)
(b) There are about 20 Million Toyotas on the road in the US. Everyone is correctly
accelerating some 10,000 times per year. We have about 2,000 defects (a conservative estimate).
So the probability of a defect is 2,000 : 10,000 * 20,000,000, which is a 1:100,000,000
Thus, one is more likely to win the lottery jack-pot than to observe a failure in a Toyota gas pedal, which makes it very hard for the Toyota employees to fix this type of problems.

So, is everything good? No. Toyota has been growing too quickly and their ratings in many of the global consumer satisfaction reports / defect reports have decreased (well before the current crisis).

For more insider details on the causes of the Toyota problems:


WalMart in India

April 12, 2010

Retailing, like politics, is said to be “local”.  WalMart clearly knows the U.S. market, but to expand it needs to learn other markets as well. For two years now it has been pushing into India (see NY Times 4/12/10) and WalMart isn’t in Kansas anymore.

First, India has a quant law that prevents foreign companies from selling directly to consumers – a potentially big problem for a retailer. So WalMart has a 50/50 joint venture with an Indian company to get around that problem.

Next, to WalMart’s credit (and retailing smarts), they are not trying to replicate their hyper-efficient big-box model in India. It simply won’t work. Instead, they are learning how to compete in a new market. In particular, transportation in India is slow and costly, so sourcing has to be done locally. In addition, there are no large suppliers, like P&G. Finally, it has to sell food because consumer durables cannot be the main product category (the Indian consumer has to allocate a large fraction of their budget to food). 

The one “habit” that WalMart is transporting to India is their propensity for proactive supplier management. They are not content to sit back and buy what is presented to them. They see inefficiencies in food production and distribution, so they are directly addressing those inefficiencies. They are giving farmers productivity advice and providing logistical support to ensure timely deliveries of quality produce. In short, they are using their scale to invest in their supply base. To make these investments profitable, it is important to make sure that their competitors cannot take advantage of their suppliers’ efficiency gains – the last thing WalMart wants is to improve a farmer’s yield only to have the farmer start selling his produce to another retailer. I suspect this doesn’t happen because (i) WalMart is willing to pay a good price, (ii) WalMart can buy up all of the farmer’s good produce and most importantly (iii) farmers develop a sense of loyalty to WalMart for helping them out. Economists have a real hard time with loyalty, but in the real world it is a powerful force. WalMart seems to understand this.