America's Manufacturing Advantage

A recent issue of Manufacturing Engineering contained a NewsDesk article which concluded that migration of (our) manufacturing to China was not going to abate any time soon. Chinese advantages quoted in the article are all real.

If you are being courted to export your manufacturing to Asian-owned

I'm a consultant who designs factories and business systems for Asian investors. In my experience, the Chinese honor their exclusivity agreements to the letter. But that doesn't keep those same investors from taking advantage of what they have learned to start a new, independent company down the road with similar equipment. If automation can be bought, and it can, China is no longer simply the home of low-cost labor.

The Manufacturing Engineering article addressed only Chinese advantages, not drawbacks. There are a series of disadvantages to Chinese manufacture. But if your business has caught the eye of an Asian investor here are three alternatives you should consider.

The first is to take them up on their offer. Reputable Chinese firms deliver what they promise when they say they will. But you'd better have an iron-clad defense of your market. Your supplier fully intends to capture your business.

The second alternative involves the first. But also do an unemotional analysis of what a determined low-cost competitor will do to your business. Believe the numbers. Plan a business-exit strategy.

The third alternative is to convert your manufacturing into a strategic advantage. It must have value to your customers that cannot be replicated in China.

Because of logistics limitations, as well as the fact Chinese producers have no interest in anything other than traditional manufacturing, mass production rules. If you could build to order for next-day delivery at, say, double your present efficiency, would that be of value to your customers?

Decades ago, as a result of a crisis not of Toyota's making, most staff and production supervision at that company had to be reassigned. The tyranny of data and simple methods for problem solving were substituted for a traditional management structure. Results were beyond anyone's wildest dreams. They achieved first pass defect levels of under 200 ppm for assemblies. This near-perfection allowed the legendary 150-200 inventory turns that gave rise to the JIT movement. There was another advantage created. They achieved heijunka (highly predictable interchangeable hours). Had computers been available with the power of our present personal computers, Toyota could have accomplished alternative three.

Unfortunately, by the time American researchers discovered what Toyota was doing (and named it Lean), Toyota was already evolving away from Lean-as-a-production-management-system. Even though first-pass rejections have risen by perhaps ten times, Toyota continues to best their competitors.

In the most recent Industry Week/Manufacturing Performance Institute census of 530 manufacturers, median first pass yield was 97% (30,000 ppm nonconforming). A 2004 survey of representatives of 53 companies that had major manufacturing improvement initiatives in effect for more than six months revealed none had first pass yields exceeding 99.0%. Shingo and Baldrige award winners were included.

A process yield of 99% doesn't seem much different than the 99.98% reached by Toyota in the seventies. Indeed, if the 1% were uniform across products and runs, then allowances in leadtime and inventories would be easy and inexpensive. Instead, imperfect manufacturing processes often result in "20% of something" situations rather than "1% of everything". Necessary allowances for such uncertainty keep leadtimes and inventories high.

The great news about alternative three comes in four parts:

* Even the best American companies have plenty of room to improve.

* It's fast. The bulk of Toyota's results came in months, not years. The few companies who have implemented the original Lean have been able to replicate Toyota's performance.

* There is virtually no capital investment involved.

* It's easy to know if it is working. If you haven't seen a 50% throughput improvement within six months you probably aren't getting it right. Within two years you should be able to eliminate routine quality-data collection.