There has been a lot written recently about the state and future of making stuff in the US. Just in the past month, Strategy & Business has touted new technology and software as a way forward for American manufacturing (America’s Real Manufacturing Advantage, Jan 20) while Steven Rattner, the administration’s former Car Czar, took the New York Times pooh-poohing talk of a revitalized manufacturing base (The Myth of Industrial Rebound, Jan 25). As Rattner sees it, whatever boost there has been in manufacturing has been inconsequential and not beneficial for workers.
But we need to get real about the so-called renaissance, which has in reality been a trickle of jobs, often dependent on huge public subsidies. Most important, in order to compete with China and other low-wage countries, these new jobs offer less in health care, pension and benefits than industrial workers historically received. …
This disturbing trend is particularly pronounced in the automobile industry. When Volkswagen opened a plant in Chattanooga, Tenn., in 2011, the company was hailed for bringing around 2,000 fresh auto jobs to America. Little attention was paid to the fact that the beginning wage for assembly line workers was $14.50 per hour, about half of what traditional, unionized workers employed by General Motors or Ford received.
With benefits added in, those workers cost Volkswagen $27 per hour. Consider, though, that in Germany, the average autoworker earns $67 per hour. In effect, even factoring in future pay increases for the Chattanooga employees, Volkswagen has moved production from a high-wage country (Germany) to a low-wage country (the United States).
This all gets us to a different New York Times on Harley Davidson that describes changes the firm made to save itself (Building a Harley Faster, Jan 28). What is interesting is that Harley’s approach differs fundamentally from either of the approaches above. They are relying on humans over widespread automation and working with their existing, unionized workforce as opposed to hightailing it to a location with cheaper labor.
The setting here is Harley’s plan in York, PA.
Harley tore down the existing plant and built a new one. Unlike most factories I’ve seen lately, the new plant in York has people everywhere. There are no robots on the main assembly line (they have various peripheral jobs); instead, hundreds of workers, operating in teams of five or six, manually build each motorcycle. This seemed like an expensive way of doing business, but Magee said that experienced, skilled workers, unlike robots, can constantly adjust to new information. The York plant makes four basic styles of motorcycle, but each has an array of customizable options. There are around 1,200 different configurations, and a new bike starts its way through the production line every 80 seconds. Virtually each one is unique, and workers have no idea what’s coming 80 seconds later. Surprisingly, robots can’t adjust on the fly like that.
Human beings can also solve thorny problems that lead to major inefficiencies, like that plastic piece that took an extra 1.2 seconds to install. Dettinger and a small team quickly came up with a fix — a tiny plastic latch needed to be set at a different angle — and saved Harley millions. (On the day I visited, he solved two other problems.) In fact, his entire job is to continuously monitor his small section of the production line and search for better ways to make motorcycles. There are 150 problem-solvers like him in the factory.
There are a couple of points to be made here. First, as the article notes, Harley is in some ways hemmed in by its branding. It cannot easily pick up and start making bikes in, say, Brazil and project the same All American blue-collar image. A brazen, union-busting fight would similarly be problematic. Harley has to work with the workforce it had.
But that doesn’t explain everything. The argument about variety and complexity does matter. They produce a wide range of products and humans have some advantages over robots for doing this work. It seems that this is really a question of volume. For example, Ford might offer a similar level of variety for its pick ups if one counts different stereos together with cab configurations and engines. But Ford builds so many pick ups annually (more than half a million a year) that it is much easier to justify more automation than one sees at Harley.
A final point. One of the comments on the article comes from our colleague Gerard Cachon who notes that some of the numbers in the article don’t quite add up.
The comment that “1.2 seconds per bike added up to 2,200 lost bikes annually.” is a misleading exaggeration. First, assuming this one task takes 1.2 seconds longer to complete, this does not necessarily mean that the interval of time between the production of bikes increases by 1.2 seconds. That would occur if all other workers on the line take 1.2 seconds longer to complete their task, which is unlikely. Second, even if we did assume that this translated into 2,200 fewer bikes of regular production, “Millions could be lost in revenue”, is clearly wrong. Only if they company were stupid would they let actual production fall by 2,200 bikes, thereby losing millions in revenue. Instead, they would run about 5 to 6 overtime shifts to makeup production. The additional labor cost of those shifts would be far less than millions.
Continuous improvement and attention to small details is absolutely critical for manufacturing success, but it is important to keep in proper perspective the consequences of 1.2 seconds at a single assembly operation.