About The Authors

Thursday, April 2, 2009

Zero (Economic) Profits

Previously I analyzed GM's decline and the rise of new firms in the US market. It was simply (but ironically) sound short-run strategy for all involved. Some time soon I extend that, arguing that downsizing in the auto industry presents almost insuperable challenges.

Here I argue that overcapacity and low profitability are intrinsic to the industry. Even if GM shuts down completely, removing a big block of production capacity, the overall impact will be fleeting. When the economy recovers, none of the volume OEMs will roll in profits, even Toyota. At core new entry is simply too easy. For years the new entrants – Detroit South – benefited from a pricing umbrella, as the Detroit Three priced to cover fixed costs. Henceforth, the volume players will have lost that last vestige of pricing power.

Caveats. First, I am less sure that the luxury producers will fall into the same trap; my instinct is they will. Second, this logic is specific to OEMs, not to suppliers or to distribution and finance. Third and finally, I avoid the jargon of a professional economist, but for any fellow members of the secret society of PhDs, this is a standard application of the model of monopolistic competition.

Bear with me while I explain.

Back in the bad old days of the 1960s, developing a new car was a long and costly undertaking, paling in comparison only to that needed to develop a new engine. As traced in Takahiro Fujimoto's Harvard Business School PhD thesis, and summarized in The Machine That Changed the World and elsewhere, firms in Detroit employed a sequential process that began with stylist sketches, then sculpted that into a clay model, and after concept approval moved to hand-crafting prototype parts. Once those checked out firms built tooling, which then was carted to an assembly plant for prototype production. Not everything would go right the first time around, and so some things would get thrown back into the pipeline, delaying the entire process. All in all it took 4 years and frequently more from initial approval to "Job One," ate up capital, and led to lots of incomplete fixes.

Meanwhile the market may have moved on. GM (and Toyota) were thus both explicit about it: if a model sold, tweak it but no more. Besides, with large market shares they had very little upside potential, and if the resulting car was blah, it should still sell OK. (If it was edgy, it could flop, which for a volume vehicle would be a true disaster.) Given all this, a car needed to be kept in production for 4-6 years to pay back up front costs, though perhaps it might be "freshened" midway with minor changes to the body and trim.

Japanese firms had grown up in a different environment: after the initial chaos of multiple firms entering and exiting in the 1950s, some 11 firms remained. Rivalry was thus more intense than in the US (or in Europe), and capital was scarce. A leisurely design process would be a route to failure, between the launch of new models by rivals and the sheer expense of the process. In addition, the same capital constraints (reinforced by the structure of the union movement in Japan) meant a greater role for suppliers in the process. Partly in response, the OEMs used an overlapping rather than a strictly sequential process. For example, you could begin work on structural components as soon as the attachment points of the sheet metal skin were fixed. Forcing cooperation across functions improved intrinsic design quality. But the real bottom line was speed.

During the 1990s such project management techniques diffused throughout the industry. Meanwhile the IT revolution was moving apace as was a revolution in materials science. For the former, CATIA, computer-aided design (CAD) software developed by the French aerospace firm Dassault Systemes, helped set the standard. CAD systems not only saved labor; they saved time while improving accuracy. Designs could be recycled, stretched, modified at the stroke of a light pen, and then sent electronically to suppliers or to a machine tool that could work on a prototype while the engineers slept. Eventually even the prototype was eliminated: structural testing and even mock assemblycould be done in a purely virtual world. Again, the real bottom line was speed.

Speed not only meant it was possible to better match market trends. Speed lowered costs. The "platform" concept pushed things further. Much of the engineering hours went to the brake-suspension-steering assemblies, and the design of the crush zones front and back, and other systems that were to some extent dependent on the size and weight parameters of a design. Once a new model was done, those core systems could be modified to create a whole family of vehicles in that size class, with some assurance that handling and safety performance and cost would be acceptable. That translated into more savings in development time and project cost.

So … in the 1950s there were around only 60 models on the road. There are now 60 new models a year. In extremis, a new vehicle can be designed in under 2 years; routinely, in under 3. Of course international trade complements these changes; it now makes sense to develop vehicles with modest shares in several markets, and a wider global palette of vehicles means a firm may be able to paint in a hole in a market that they would have ignored in the past.

At the project level the logic is inexorable. There are also lots of players. In order of US sales they are (subject at any time to Chapter 11!) the Big Three [GM, Toyota and Ford], the Middle Four [Chrysler, Honda, Nissan and Hyundai] and the ] and the Little Seven [VW, Mazda, BMW, Mercedes, Subaru, Suzuki and Mitsubishi]. And these 14 exclude firms that are solely importers! The chance that a vehicle will prove a hit is slim; the pressure to redesign and attack new segments immense. Almost everyone is optimistic at launch; almost everyone is disappointed. So on average there is everywhere excess capacity; it is better to be able to build more than to be caught short, and you don't want to drop margins. For years sales per model have trended down. And the irony is that while the improved economics of the design process has encouraged more vehicles, the aggregate effect is that overall costs haven't fallen in a commensurate manner; indeed, they may well have risen. And so another bottom line appears: red ink, or at least meager margins.

There is in my mind no way to reverse this process. No firm is large enough to take into account its impact on the overall market from model proliferation. So all continue to throw more models at the consumer. Of course the process built up a head of steam during the Greenspan-Bush Bubble; to date we've not seen many project cancellations. But to repeat, no longer is any single firm large enough in the market to mute the logic of this process.

Again, a bottom line, the bottom line: on a structural basis, an OEM is now and will remain a business that will barely manage a positive return on assets in good times and will bleed badly in poor times. It is a fundamentally bad business to be in and it will not get better, only less bad.