Everyone Wants a Fab

In the 1960s and 1970s, every country wanted an auto industry. In a world obsessed with the automobile, being in the car business was central to national pride. In today’s economy, the chip fab seems to have taken the place of the auto industry. Nations and U.S. states offer dramatic incentives to capture these massive manufacturing facilities. A modern chip fab—those producing 300 mm wafers—cost more than $3 billion to build and incorporates the latest manufacturing technology.

Enticing a chip fab has long been a goal of the Pataki administration. The Luther Forest Technology Park in Saratoga County received substantial state and federal money to support site assembly, permitting and infrastructure. The Luther Forest brochure promises “up to 10,000 good paying jobs for our young people” for the estimated 2 million square feet of manufacturing space. So far, so good.

But site assembly, pre-permitting and infrastructure weren’t enough to persuade Advanced Micro Devices (AMD) to award the prize to New York. To close the deal, New York promised AMD a cool billion dollars now, with more to come in future savings.

Is the “juice is worth the squeeze?” Particularly the squeeze on taxpayers? Or did the thrill of the moment overrule common sense?

These are good jobs, 1,200 of them according to the Governor’s press release. But what happened to the 10,000 jobs that were going to be created at Luther Forest? At this rate, this massive and expensive site—planned to hold two million square feet—will house only 2,000 workers at build-out. 1.2 million square feet (sixty percent of the site) will have been paved over by a single plant.

AMD’s workers are likely to earn just over $100 million per year (based on averages reported by Minnesota IMPLAN group for the semiconductor industry). And as AMD buys goods and services and its employees spend their hard-earned pay, more jobs will result Doing a rough calculation, this will probably total another 900-1,000 jobs earning an additional $40+ million (the “spillover” jobs are largely from consumer spending and don’t pay as well as the AMD jobs). Theodore Geisel, chairman of Albany’s Center for Economic Growth is quoted by The Business Review (Albany) as saying that “economists estimate that there will be a 3:1 ratio for the stimulation of additional jobs in the region.” Given the current configuration of industry in New York State, that’s wildly optimistic. Maybe firms will grow up around the AMD fab and bring supplier jobs to New York. And maybe not.

I can’t argue that the construction phase will make a dent in the regional economy for a few years. A chip fab is an expensive and complex facility. We’re told to expect a project costing in the neighborhood of $3.2 billion. But much of this cost is for highly specialized equipment—$2.6 billion, according to the press release. Most of this specialized equipment will be sourced outside the state. So we have $600 million in industrial construction. Let’s assume 10% of the equipment is purchased in New York for another $260 million in equipment manufacturing. If correct, then total wages for the construction period (including direct and spillover earnings, both wages and proprietor’s earnings) might total $800 million. But this depends a lot on whether the construction crews are local or not. The more specialized the project, the more likely it is that national firms will be engaged and their crews recruited from other locales.

If we count all the money paid to all the people who are likely to benefit from the chip fab, then the billion dollar investment makes sense. We break even very soon—maybe in year three.

But that’s pretty screwy math. All of New York’s taxpayers throw money into the pot but only the lucky few who either work at the plant or supplier firms—plus AMD’s shareholders—benefit? This is a transfer payment, not a cool-headed business investment.

There is a benefit for taxpayers. While I wonder if AMD will ever pay much tax, its employees will. Figuring roughly, I’m guessing that we can expect maybe $7 million per year in personal income tax, post construction. Sales tax might come in at another $4 million annually, split between the state & localities. And there are other taxes, of course (this IS New York, after all). If we think of the taxpayer as the “owner” of the business (the State of New York), we’re spending a billion dollars or more in the early years and getting a 25-year stream of tax revenue worth a lump sum of perhaps $225 million today (including construction period taxes).

That assumes that the plant will be around for 25 years. The semiconductor business is notoriously competitive and cyclical. We only need to think back to the early part of this decade to find our last IT recession. AMD closed two fabs in 2002 and fired 15% of its workforce—and announced even more cuts for 2003. Nearly every other major player in the semiconductor business cut jobs in 2002, 4,000 at Intel alone. Four years later, Intel is back to firing workers. The world’s leading microprocessor manufacturer announced in September that it was cutting 10,500 jobs.

Chip fabs aren’t forever: The fabs making 300 mm wafers are making the older, 200 mm plants obsolete. While many 200 mm plants are being rebuilt to accommodate 300 mm technology, companies making these investments expect a whopping public subsidy. New York gave IBM $475 million for its Fishkill plant overhaul. Many older fabs will simply be closed, particularly as an expected 35 new fabs come online in 2007, increasing total capacity by 17% (according to Strategic Marketing Associates). Will AMD be back looking for more support a decade after this plant goes online? Or will competitive pressure from global competitors (particularly from fast-growing Asian “foundries”) leave New York with a 1.2 million SF paint-ball emporium?

The entire budget for programs administered by the Empire State Development Corporation and the New York Office of Science, Technology and Academic Research (NYSTAR) was $130 million in 2005-06. What share of that goes for small business development? For small loans to struggling new businesses scattered about the state? For retraining for displaced workers? What could a billion dollars buy in expanding these other programs?

The AMD deal does not fit into a sustainable or scalable economic development policy. Let’s focus on improving the business climate for all companies, not “buying” jobs from AMD or any other firm. If this deal made economic sense, we could save the trouble of dealing with private industry and just put everyone on the public payroll.