Foxconn Subsidy Will Lose Money for Wisconsin, Says Study Group
The $3 billion deal between the Wisconsin Economic Development Corporation (WEDC) and Foxconn for subsidies for an LCD manufacturing plant was agreed by then-Governor Scott Walker, a Republican, and has had many critics from the left of the political spectrum, including his successor, the current Governor Tony Evers, in the 2018 election campaign. This week the deal was also criticized by a study group on the conservative side of the spectrum, which indicated that the deal will result in Wisconsin’s economy being smaller by tens of billions of dollars over the next 15 years.
Foxconn Groundbreaking, June 2018
“The Economics of a Targeted Economic Development Subsidy” (hereafter, “ETEDS”) was released last week by the Mercatus Center of George Mason University in Arlington, Virginia. Mercatus describes itself as “the world’s premier university source for market-oriented ideas—bridging the gap between academic ideas and real-world problems, and says its “economic toolkit” draws from the works of Nobel Laureates Friedrich A. Hayek, Elinor Ostrom, Douglass North, and George Mason University’s own James Buchanan and Vernon Smith. Mercatus is the Latin word for “market”, and the center addresses public policy issues from a market-oriented perspective.
ETEDS is written as a general criticism of targeted subsidies by states, but takes the Foxconn Wisconsin deal as an example of why Mercatus believes that such subsidies are counterproductive, leading to worse economic results overall. This is in direct conflict with the claims of the deal’s proponents, including ex-Governor Walker and President Trump. When the deal was made, a study from the University of Wisconsin (UW) claimed that the Foxconn Gen 10.5 plant would add about $39 billion to the output (GDP) of the state of Wisconsin over 15 years.
ETEDS refutes the claim that subsidies boost economic growth with a wide-ranging and comprehensive argument drawn from standard economic theory. The entire report reaches 70 pages and includes generous use of the classic supply/demand curves you learned about in Economics 101 (though I would characterize this study as more like a 300-level discourse), but I will try to summarize the argument here:
- Subsidy proponents fail to recognize and account for the fact that subsidies rarely determine location decisions. The UW assumes with 100% certainty that the Foxconn plant would not locate in Wisconsin except for the subsidy. ETEDS states that location decisions are generally based on main economic factors such as talent availability, labor costs, or similar factors, and subsidies simply add incentive to decisions that companies would already make on their own. ETEDS quotes a review of academic research concluding that subsidies “probably tip somewhere between 2% and 25% of incented firms toward making a decision favoring the location providing the incentives.”
- Because of #1, the proper estimate of the benefits of a subsidy program should be the value of the economic gain ($39 billion in the UW study) times the probability that the subsidy would effect the outcome (2% to 25%); this greatly reduces the benefits.
- The UW study and similar studies apply a multiplier to the benefits of the subsidy, but fail to apply the same multiplier to the costs of the subsidy. The $3.6 billion of subsidies could be used for other government purposes or returned to taxpayers, where it would generate a multiple of economic activity. Note that this argument is a generalized version of an earlier study commissioned by Governor Evers’ office, which calculated the effect of the cost of the subsidy on a reduction in K-12 education (or an increase in case the subsidy did not happen), and the resulting economic impact of a less-educated labor force.
- Subsidy proponents (such as the UW study) describe ‘clustering’ as a likely benefit, which the Wisconn Valley name suggests will happen, but ETEDS refers to several academic studies that conclude that there is no evidence that government subsidies can encourage clustering, and some evidence that subsidies “may discourage the sort of beneficial clustering that would occur naturally.”
- Instead of providing a subsidy of $3.6 billion to Foxconn, Wisconsin could choose to reduce its tax burden by a similar amount. A reduction in the sales tax, for example, would increase economic activity by making goods more affordable and reducing the “deadweight loss” that such taxes impose on an economy. ETEDS calculates the deadweight loss impact of the $3.6 billion Foxconn subsidy as between $5.7 billion and $34.3 billion, with an “average” impact of $20 billion.
- Considering both items #2 and #6 above, the proper calculation of the net benefits of the subsidy should be: [(economic growth from the new business) * (probability that subsidy will affect decision)] – (deadweight loss of taxation). In their baseline calculation this means: $39.3 billion * 25% - $20 billion = -$10.2 billion impact from the subsidy. ETEDS gives a probable range of -$369 million to -$19.2 billion as the likely impact of the subsidy.
- ETEDS applies the same logic and math to the smaller case of a Gen 6 fab, which they cite industry experts (which turns out to be DSCC) as requiring only ¼ the investment. ETEDS reaches the same conclusion for Gen 6, with all the numbers divided by 4.
- ETEDS also brings in a supply-side deadweight loss description as a hidden cost of a subsidy, with the following description: “Firms produce up to the point at which another unit of production would not be worth the cost. Subsidies, however, encourage firms to produce beyond this point.”
- The study also cites additional negative implications of subsidies that are difficult to quantify, including the anti-competitive effects of subsidizing one company in a competitive market and the efforts expended in getting the subsidy.
- Finally, ETEDS describes some of the political problems of subsidy programs, where benefits are concentrated to a single entity and costs are diffuse across the entire state. This creates an environment of political discrimination which impedes efficient economic decision-making.
The author of the ETEDS study was quoted in the Milwaukee Journal Sentinel as saying that criticism of targeted economic incentives “brings together interesting ideological…bedfellows. If you’re a free market person there’s reason to oppose favoritism on the grounds that it messes with the market process, and if you’re a progressive there’s reason to oppose it because it favors the wealthy and well-connected at the expense of the poor and the unknown.”
Besides the ETEDS study, I have been asked about the Foxconn project and its likely trajectory, and my answers may appear in Wisconsin-based media this week, so I thought it important to share my thoughts with our readers. Foxconn has stated publicly that they will be producing TFT LCD panels in their Wisconsin facility before the end of 2020. We think that won’t happen with new equipment; our logic is as follows:
- DSCC have seen no evidence of Foxconn ordering equipment for the Wisconsin fab.
- Based on the extensive work DSCC does on equipment orders, if Foxconn was making orders we are "extremely confident" we would be able to track them.
- Because they have not yet made the orders, Foxconn will not be capable of making TFT LCD backplanes and LCDs until 2021 at the earliest.
- It is possible that Foxconn could move equipment from existing fabs in Asia, and install that equipment to make TFT LCD by the end of 2020.
- If Foxconn was moving equipment from Asia, DSCC may hear about it, but we are less confident that lack of such rumors is conclusive proof.
- If Foxconn is installing module assembly equipment, they could be producing LCD modules before the end of 2020.
I recently reviewed a report from Foxconn’s fire-protection engineering consultant, submitted earlier this year to the state of Wisconsin as part of Foxconn’s request to omit sprinklers in certain sections of their 1 million square foot manufacturing facility. The consultant describes the various sections of the planned facility in language that clearly identifies TFT LCD production. This suggests that Foxconn plans eventually to make complete TFT LCDs at the Wisconsin site; we just think it won’t happen in 2020.
- The facility is organized by key operational areas which include the Array, Cell, Assembly, Support, and Office areas. The manufacturing process starts in the Array area, where glass substrates are cleaned and coated with semiconductor material and circuitry. This involves thin film transistor (TFT) and color filter (CF) deposition, photoresist coating, developing and etching, and photoresist removal operations conducted on glass substrates. Glass substrates move from each process in a linear fashion, where substrates at each step are stored in stockers, are processed, and moved to the next step in the process.
- The Cell process follows Array and consists of the TFT glass and CF glass assembly. When assembled, liquid crystal is vacuum injected into the space between the TFT and CF substrates to form a completed mother substrate that is then cut into smaller panels for further processing in Assembly. Similar to the Array area, substrates are stored in Cell stockers as they move through the process.
- The Assembly process follows Cell. Here completed panels from Cell are fully assembled into working flat panel displays. Support areas located to the south of the Cell and Array areas contain material storage, chemical and gas distribution systems, utilities, etc., that support the Cell and Array operations. Office space is located to the north of the Cell area.
In other Foxconn Wisconsin developments:
- Foxconn announced an additional $31 million in construction awards for work on its Wisconn Valley Science and Technology Park. The new awards, to four different Wisconsin-based contractors, involved fire protection and various functions for the Office and Assembly areas of the site, including mechanical, electrical and security sections. According to Foxconn, this brings the total value of construction contracts for the Wisconsin project to approximately $350 million. With respect to the building description from Foxconn’s fire-protection engineering consultant above, It’s noteworthy that these contracts cover only the Office and Assembly areas, and not the Array or Cell areas.
- Foxconn also stated that they are nearing completion of roof installation for the nearly 1 million square foot advanced manufacturing facility at the Wisconn Valley site.
- Foxconn founder Terry Gou visited Wisconsin to celebrate progress in the company’s Wisconn Valley project, reviewing the project site in Mount Pleasant and attending the company’s holiday party in Milwaukee. “Wisconsin is Foxconn’s North American home,” said Founder Terry Gou. “We have achieved a great deal over the past year, and it is due in great part to the hard work of the hundreds of Foxconn Wisconsin employees and the many partners who are supporting the construction of the new facilities. Next year is projected to be even more significant as we bring facilities at the Wisconn Valley Science and Technology Park on stream.”
- My contacts in Wisconsin have indicated that Foxconn has not yet made any effort to renegotiate or amend the contract with the state. The office of Governor Tony Evers has clearly communicated that under the current conditions of the project Foxconn is not in compliance with the contract, and will not be eligible for any of the subsidies in the contract, but that the state is willing to engage in a discussion to update the contract.