August 05, 2019 Feature

Economists Must Be Careful in Their Use of IMPLAN to Analyze Public Interest Issues in Section 337 Cases

Robert Rogowsky and Jeffrey Klenk

©2019. Published in Landslide, Vol. 11, No. 6, July/August 2019, by the American Bar Association. Reproduced with permission. All rights reserved. This information or any portion thereof may not be copied or disseminated in any form or by any means or stored in an electronic database or retrieval system without the express written consent of the American Bar Association or the copyright holder.

Some recent cases before the U.S. International Trade Commission (USITC) have made use of an economic model known as IMPLAN to estimate the possible effects of an exclusion order prohibiting the importation and subsequent sale of a product found to infringe a valid and enforceable U.S. patent. These cases arise under Section 337 of the Tariff Act of 1930, which empowers the USITC to issue such an exclusion order.1 Before issuing an exclusion order, the USITC must weigh the importance of protecting intellectual property rights against any potential adverse effects of that exclusion order.2 As part of this process, economists are often retained to analyze how the domestic economy might respond if an infringing product were removed from the marketplace or if domestic production of the infringing product was curtailed. Although IMPLAN may have valid uses for certain types of policy analysis, its use in assessing the potential effects of an exclusion order can yield misleading results.

Explanation of the IMPLAN Model

IMPLAN is one of a set of models known as regional input-output models.3 Simply stated, input-output models are premised on the fact that certain economic factors are interrelated. Using data from a variety of government sources, including the Bureau of Economic Analysis, the Bureau of Labor Statistics, and the Census Bureau, IMPLAN estimates the interrelatedness of different sectors of the U.S. economy (i.e., the extent to which different industries buy from and sell to one another) as well as the interrelatedness of different regions of the United States (i.e., the extent to which different geographic areas of the United States buy from and sell to one another).4

Results from input-output models, such as IMPLAN, are in the form of “multipliers.”5 As explained by IMPLAN, its multipliers estimate, all else equal, “how for a given change in a particular industry a resultant change will occur in the overall economy.”6 IMPLAN’s multipliers thus return the estimated effects on the broader U.S. economy of a particular change in a specific industry, broken down into three components. These three components are the direct effect, which estimates the potential impact on the specific industry in question; the indirect effect, which estimates the potential impact on related industries, such as through changes in supply chains; and the induced effect, which estimates the potential impact on consumer spending as a result of possible changes in household incomes.7

Using IMPLAN to Consider the Potential Effects of Exclusion Orders

In patent infringement cases brought before the USITC, the only remedy available to a complainant if a product is found to infringe is the exclusion of that product from importation (as opposed to a damages award based on either a reasonable royalty or lost profits as might be available in district court). Respondents will often claim, however, that the proposed exclusion order will result in the prohibition of the domestic production of products incorporating the infringing product, to the extent that noninfringing alternatives cannot be found. Such a decrease in production is argued to have deleterious effects rippling across the broader economy far beyond the original industry of the product at issue. These deleterious effects are then claimed by respondents to outweigh any potential benefits accruing to the complainant of keeping the infringing product off the market.

To estimate the supposed broader effects on the U.S. economy of a proposed exclusion order, economic experts retained by respondents will often conduct an analysis using IMPLAN. As one input into their IMPLAN analysis, respondents’ experts generally assume that proposed exclusion orders will curtail production within a particular industry by some dollar amount. Based on historical relationships among other industries related to the industry of interest, IMPLAN returns multipliers putatively showing that for every dollar lost in the industry of interest, the broader economy would lose some multiple greater than that amount.

In one recent USITC investigation, the respondents’ expert used IMPLAN to estimate the effects that “production declines,” putatively stemming from an exclusion order in the automotive sector, might have on the broader U.S. economy.8 In a separate USITC investigation, the respondents’ expert used IMPLAN to “assess the economic effects of an exclusion order” in the computer industry.9 In particular, this expert took estimates of the supposed decline in supply stemming from an exclusion order and then used IMPLAN to “trace out [the] ripple effects” that might result.10

Given the methodology underlying IMPLAN, though, it is essentially the case that an analysis based on a decline in output in a particular industry will always show that for every dollar lost in that industry, there will be a greater loss in the broader economy. Such a showing, rather than suggesting a harmful effect resulting from the exclusion order, instead reflects a set of restrictive assumptions that very well may not hold in actuality.

Dynamic Responses of Economic Agents

Embedded in the inner workings of IMPLAN, as well as in similar input-output models, is the implicit assumption that relationships among various economic agents are fixed over time and are not themselves affected by the change being modeled. IMPLAN assumes, for example, that key variables, such as wage levels, prices, input costs, property values, labor supply, and productivity, remain constant. The fact that economic agents respond to incentives is a well-accepted principle, though, at least among economists.

Consider the example of the five-cent per bag tax levied on plastic bags in Washington, D.C., and other municipalities around the country. Originally expected to generate around $4 million dollars in revenue annually, the bag tax in D.C. has in fact annually generated about half that amount.11 The reason is that in response to a nominal fee, customers dramatically curtailed their usage of plastic bags in lieu of reusable bags. Although the actual outcome still had a desirable effect—fewer plastic bags littering the waterways around D.C.—the expected level of revenue generation fundamentally failed to account for how individuals changed their shopping habits in response to a change in policy.

Just as customers dramatically responded to a five-cent tax, one would expect other economic agents similarly to respond to changes in their environment. By relying solely on historical data, and without any predictive ability to model how economic agents might respond to change, IMPLAN can only provide estimates of the potential effects of a change in policy based on what economic relationships looked like in the past. IMPLAN does not account for how those relationships might themselves change in response to the change being analyzed. Continuing with the D.C. bag tax as an example, an analyst using IMPLAN to model potential implications of that tax likely would have generated overstated results because IMPLAN is a static model that does not account for possible responses to the policy change (e.g., the reduction in bag usage).

Unless one has a reason to believe that a change in a policy (or the implementation of an exclusion order) is unlikely to cause any change in behavior of the economic actors affected by the policy, a reliance on IMPLAN may be misplaced. This inability to dynamically capture the implications of changes within an industry is one of the primary factors limiting the reliability of IMPLAN in assessing the possible effects of an exclusion order.

Limitations of the IMPLAN Model in Analyzing Exclusion Orders

Another way of framing the inherent failure of IMPLAN (or similar input-output models) to account for the dynamic responses of economic agents is through the consideration of opportunity costs. An opportunity cost represents the next best alternative that must be given up in pursuit of a particular decision. In a world of limited resources, every decision necessarily implies a trade-off. Were a certain economic activity to be curtailed, such as might occur through the imposition of an exclusion order, the resources devoted to the pursuit of that activity would most likely be diverted to their next best alternative. As an example, employees producing an infringing product subject to an exclusion order would presumably seek employment elsewhere if that product could no longer be produced. Alternatively, the factory where those workers are employed might switch production over to a noninfringing product.

Thus, rather than sitting idle, as IMPLAN assumes would happen, the financial capital, physical infrastructure, and human labor used in an industry potentially affected by an exclusion order would find some other productive use. Although economic theory indicates that this diversion of economic resources to other productive uses would occur, IMPLAN fails to account for the value that those redeployed resources would generate. An analysis of an exclusion order that relies solely on IMPLAN to show a potentially detrimental effect on the economy likely would be incomplete without an accompanying analysis to show how the affected economic resources might be redeployed.

There are a number of other reasons as well for why an analysis making use of IMPLAN may overstate the detrimental effects potentially stemming from an exclusion order. In many industries, for example, there is no reason to suppose that a change in output necessarily results in a proportional change in employment, as IMPLAN inherently assumes.12 As observed by one academic article considering the merits of input-output models, “if an industry can increase its output by extending mainly the number of hours that existing employees work,” then the estimated results “will overstate the actual increase in local employment.”13 The opposite is also true: if an industry can reduce output by tweaking its production processes rather than by laying off workers, then there is no reason to believe that the reduced output figures fed into IMPLAN will return reliable estimates based on an assumption of reduced employment.

In many industries involving an assembly line, output can be reduced, without a proportionate decrease in employment, by increasing the takt time of the line. Simply put, takt time represents the length of time that it takes a completed product to roll off a production line.14 For instance, if an automotive plant generally produces a completed car every 60 seconds, it could reduce production by 10 percent by slowing its takt time down to 66 seconds. Alternatively, an assembly line running beyond full capacity, such as by utilizing overtime or weekend work, could also slow down production by scaling back such overtime efforts. While the slowing down of a production line, or the curtailing of overtime or weekend work, might result in reduced wages, that is a very different economic phenomenon than the assumption of layoffs implicit in the IMPLAN model.

However, even if it were reasonable to assume that a reduction in output brought about by an exclusion order would result in layoffs, there is no reliable reason to assume that those laid-off workers would be unable to obtain alternative employment.15 In the IMPLAN model, once a worker is laid off, that individual’s earnings are simply deducted from the local economy and are assumed to no longer be spent in support of local businesses. Such a modeling approach fails to account for the likely possibility that an unemployed worker finds alternative employment. By way of example, in January 2018 there were 7.6 million job openings while there were 5.6 million individuals separated from their jobs and 5.8 million individuals hired at new jobs.16 In an economy in which millions of individuals switch jobs in any given month and in which over seven million jobs are unfilled, assuming as IMPLAN does that workers who may be adversely affected by an exclusion order cannot find alternative employment seems unreasonable.

A related issue is IMPLAN’s assumption that a decrease in production necessarily affects domestic employment. As observed by Robert Lawrence, a former member of the Council of Economic Advisors and currently a professor at Harvard, “in reality the raw materials, transportation, and other intermediate inputs used in manufacturing need not be produced within the same country.”17 Professor Lawrence goes on to conclude: “Thus the multiplier does not necessarily refer to domestic jobs.”18

Based on its use of industry-wide (as opposed to firm-specific) data, an input-output model such as IMPLAN reflects the average extent of an industry’s impact on domestic production and employment. Often, though, a proposed exclusion order is directed at a particular firm within an industry. Whether or not the impact of that particular firm on domestic production and employment conforms to its industry’s average is an assumption that cannot be made without firm-specific evidence. Even within a firm, the effect of a proposed exclusion order on domestic production and employment may be ambiguous. The 2015 Toyota Camry, for example, was adjudged to be the “most made-in-the-U.S. car you can buy” according to an assessment by; yet, the 2015 Toyota Prius was solely assembled abroad and from completely foreign-manufactured parts.19 As a result, the potential effects on the domestic economy of an exclusion order affecting the Toyota Camry versus the Toyota Prius could be very different, but an analysis conducted using IMPLAN would return the same results.

A consideration of opportunity costs is important on the consumption side as well. If an exclusion order curtailed the purchase of a particular product, consumers would likely divert their savings from not purchasing that product to a substitute product or some other consumer good. Without an explicit effort to model those diverted expenditures, though, IMPLAN, on its own, would not do so. In other words, were a modeler only to assume a loss in consumer expenditures, IMPLAN, due to its static nature, would not account for how those expenditures otherwise might be redeployed in the economy.

The need to explicitly account for both alternative production and consumption patterns that might result from an exclusion order highlights yet another potential drawback of IMPLAN: that it is only as reliable as the data being fed into it. As noted by one economics professor specializing in the analysis of economic development: “The results of any economic impact model will be only as accurate and realistic as the assumptions and data used to produce them.”20 This professor further notes that “standard input-output models have a significant limitation to the extent that they do not reflect how an economy adjusts over time to changes in macroeconomic conditions, regional industrial structure, public policies, and technological advances.”21


For the USITC to accurately balance the right of a complainant to obtain relief from infringement versus the detrimental effects of that relief potentially stemming from an exclusion order, it is often necessary to assess the effects on the domestic economy of that exclusion order. Input-output models such as IMPLAN provide a static estimate of how interrelated a particular industry is with other industries. While that estimate can be helpful, alone it is insufficient to answer the question of the extent to which a proposed exclusion order would harm the economy as a whole. Instead, a complete analysis of that question requires an assessment of how the affected economic resources might be redeployed. In addition, nuances within IMPLAN may render its use in any particular circumstance inappropriate without further investigation of its applicability.


1. See Understanding Investigations of Intellectual Property Infringement and Other Unfair Practices in Import Trade (Section 337), U.S. Int’l Trade Commission, (last visited June 12, 2019).

2. 19 U.S.C. § 1337(d)(1). In particular, the USITC is required to consider the impact of an exclusion order on (1) public health and welfare, (2) competitive conditions in the U.S. economy, (3) domestic production of like or directly competitive articles, and (4) U.S. consumers.

3. Regional in this context means different regions of the United States, not different regions of the world.

4. United States Economic Data, IMPLAN, (last visited June 12, 2019).

5. See, e.g., Rebecca Bess & Zoë O. Ambargis, U.S. Bureau of Econ. Analysis, Input-Output Models for Impact Analysis: Suggestions for Practitioners Using RIMS II Multipliers 2 (Mar. 2011),

6. Phil Cheney, General Information about Multipliers, IMPLAN (Nov. 2, 2017),

7. See, e.g., David Mulkey & Alan W. Hodges, Using IMPLAN to Assess Local Economic Impacts 5 (UF/IFAS Food & Res. Econ. Dep’t, Document No. FE168, 2000),

8. Transcript of Open Session at 601–02, Certain Thermoplastic-Encapsulated Electric Motors, Components Thereof, and Products and Vehicles Containing Same II, Inv. No. 337-TA-1073 (USITC July 24, 2018). The authors were retained by the complainant in this matter.

9. Transcript of Open Session at 493, Certain Microprocessors, Components Thereof, and Products Containing Same, Inv. No. 337-TA-781 (USITC Aug. 22, 2012).

10. Id. at 494–95.

11. Officials Rejoice over Low 5-Cent Bag Fee Revenue, WTOP (Oct. 2, 2012),

12. An assumption of proportional change also fails to account for how economies of scale might develop within an industry.

13. Bess & Ambargis, supra note 5, at 8.

14. See, e.g., Takt Time, iSixSigma, (last visited June 12, 2019).

15. This assumes that any local and federal legal requirements for conducting layoffs or terminations are met. Some states, such as South Carolina, require a notice period before certain plant closings and “mass” layoffs. See WARN Act, S.C. Dep’t Emp. & Workforce, (last visited June 12, 2019).

16. Bureau of Labor Statistics, USDL-19-09404, Job Openings and Labor Turnover—January 2019 (2019),

17. Robert Z. Lawrence, Does Manufacturing Have the Largest Employment “Multiplier” for the Domestic Economy?, Peterson Inst. Int’l Econ. (Mar. 22, 2017),

18. Id.

19. Nathan Bomey, Made in the USA? Not These Cars, USA Today (Sept. 16, 2016),; see also Kelsey Mays, The 2015 American-Made Index, (June 29, 2015),; Kelsey Mays, The 2017 American-Made Index, (June 26, 2017), It is also worth noting that after an apparent change in methodology between 2015 and 2017, the Toyota Camry no longer was one of the top-10 most made-in-America vehicles.

20. Jonathan Q. Morgan, Analyzing the Benefits and Costs of Economic Development Projects, 79 UNC Community & Econ. Dev. Bull., no. 7, Apr. 2010, at 5, Also noted by Morgan is the need for users of IMPLAN to have the requisite expertise to use IMPLAN correctly.

21. Id. at 6.


Robert Rogowsky is a special advisor to Berkeley Research Group, where his practice focuses on Section 337 cases and the analysis of issues related to the public interest and domestic industry. He was previously the chief economist at the USITC.

Jeffrey Klenk is a director at Berkeley Research Group, where his practice focuses on issues related to antitrust and intellectual property damages.


The views and opinions expressed here are those of the authors and do not necessarily reflect the opinions, positions, or policies of Berkeley Research Group LLC or its other employees and affiliates.