While interrupted from time to time by the business cycle – exceptionally so at the moment – automotive demand has expanded steadily since the end of WWII in the US, Japan and the EU. As a consequence the normal challenge for auto companies has been expansion. That process was not always orderly, as the different divisions of the larger firms fought for product for their dealers. Such product overlap is particularly severe at Toyota and GM, with the most sales channels. For dealers there is modest downside; additional models mean additional inventory. But each dealer hopes to pick up some incremental sales. Unfortunately that includes sales cannibalized from other sales channels – Pontiac sales robbing from those of Chevrolet – and so the benefits to the manufacturer are smaller. Overall however the bias (as in most franchising systems) is to have too many sales points and too many products from the standpoint of the system as a whole. As disorderly as the expansion path may appear, with firms responding to "hits" by their rivals with product that may mesh poorly with the overall vehicle lineup, reversing that process is much more difficult. Doing it in an orderly manner is well-nigh impossible.

inancial implications aside, trimming capacity presents formidable strategic challenges. First, the menu of vehicle programs must be developed with a 6-8 year time horizon. The allocation of engineering resources is a complicated dance, and has to also mesh with assembly capacity. A firm can work on only 1-2 vehicles in a size class at a time, and delaying a program until later then means some future program must be moved forward or otherwise shifted. So the dance has to position the players with a vision of where the firm wants to be 8 years down the road. That is hard enough when most vehicles programs are merely (?!) developing a replacement to existing product. Downsizing will almost inevitably mean that a firm is set to develop the wrong vehicles in the wrong order.

econd, if products are located rationally in consumer space, canceling one model creates a hole in the overall product lineup. That means that there are greater benefits from moving adjacent vehicles to the head of the development queue. So one less model increases the pressure to re-engineer two or more adjacent vehicles. Doing that of course costs money rather than saves money, so a hole must remain. Such a hole affects dealerships; if a particular sales channel is starved for product, then it will be more difficult for the dealers to maintain the presence in the market in overall staffing and advertising needed to support vehicles that ought to sell well. Downsizing, in other words, can amplify an initial decline in sales. And that's without taking into consideration any negative publicity from media coverage of the process.

hird, developing vehicles entails teamwork; downsizing means breaking up teams. Trying to "cherry pick" the good engineers is hard, and hurts morale – and of course a team without spirit is not much of a team, as is a team that has never played together before, even if (especially if?) it is composed of all-stars. Voluntary buyouts may force a company to "buy back" workers if too many in a give area quit. And in either case a firm will find it difficult to continue recruiting young engineers and other functions where learning the trade takes time.

ourth, uncertainty looms large. CAFE (fuel efficiency) mandates skew incentives for where to allocate resources, with an impact that varies from firm to firm depending on their mix of domestic and imported vehicles. "Green" incentives that accrue to vehicles that may in fact not be very efficient, the lack of an energy policy that creates uncertainty in the path of energy prices (and which in the US makes diesel more expensive rather than cheaper than gasoline), and the presence of state and local economic development incentives that make a new plant for a company with an expanding market share cheaper than an old plant for a firm with a declining share all make life more complicated.

hen there are legacy costs. Incumbents in the US already had large number of retirees, due to the cumulate impact of increases in longevity and increases in productivity, that left them with an unfavorable ratio of retirees to workers. Downsizing makes things worse. New entrants have no such problems. They have virtually no retirees, their healthcare costs are further lowered by the relatively young age of their workforce, and they have "modern" benefit plans rife with deductibles and co-pays, things that were of little or no concern when such benefits were negotiated by the incumbents in the 1950s.

inally, what scale is needed for survival? Electric vehicles, hybrid diesels, hybrid gasoline vehicles, natural gas vehicles, fuel cell vehicles – it is unclear which of the next generation of propulsion plants will prove fruitful. All require significant expenditures now that will not generate revenue in the near term. A large firm can have multiple platform teams, very small cars in Korea, small cars in Germany, midsized-cars and light trucks in the US, real-wheel drive projects in Australia, and so on. There are potentially large benefits from that, though to date the track record of "world" cars is poor, in part because regulatory barriers and variations in manufacturing infrastructure make it expensive and time-consuming to adapt a European car for the US market.

f course there is excess capacity in the market alongside too many models. The ease of entry means that will not change in the medium term – remember, there are 14 producers in NAFTA, not to mention importers. Dealers face their own problems, particularly for those in urban areas, as the internet undermines the value of a large, expensive physical footprint. Downsizing individual firms affects suppliers in an uneven manner, but few suppliers depend on a single customer. Viewed from the other end, a Toyota or a Honda is reliant on the health of suppliers for whom one or another Detroit firm is a major customer.

n short, downsizing can unravel along multiple dimensions. And probably will.