- CPA’s economic modeling of a U.S.-U.K. free trade agreement that cut tariffs to zero between the two countries shows that such an agreement would increase unemployment by more than 2,000 jobs and reduce U.S. GDP by $142 million.
- This economic model includes for the first time a methodology for estimating job loss due to worker displacement, i.e. jobs lost in each country due to increased import penetration.
- The model finds that the U.K. would suffer job losses of 1,841 jobs but see growth in GDP of $202 million. The U.K. gains in GDP in part because it increases exports to the U.S., which is its largest market in many commodities.
- The U.S. job losses are primarily in the manufacturing sector, including motor vehicle and machinery production. The U.K. would gain jobs in manufacturing but lose jobs in financial services and other sectors where the U.S. would increase its share of the U.K. market.
- Both countries would see a small rise in their trade deficit, by $1.3 billion in the U.S. and $1.1 billion in the U.K.
Since the United Kingdom left the European Union, the U.K. government has sought a bilateral free trade agreement (FTA) with the United States. For Boris Johnson’s U.K government, a bilateral deal with the U.S. is part of its campaign to convince British voters that a go-it-alone Britain won’t be shut out of the world economy. Last year, the U.K.’s Department of International Trade (DIT) published a detailed analysis forecasting that such a trade agreement would add roughly $5 billion to the U.K. economy and $10 billion to the U.S. economy.
However, a new report from the CPA finds different results: our model of the effects of a U.S.-U.K. trade agreement based on a standard trade model finds benefits for both countries, but much smaller benefits than those found in the U.K. government studies. The benefits to gross domestic product (GDP) are reduced by more than 90%. Further, standard trade models assume trade shocks like a new FTA cannot cause unemployment. We have modified the standard model to allow for unemployment caused by imports and to estimate its impact on both economies. We call this a Worker Displacement trade model. This model finds that a U.S.-U.K. trade agreement would reduce U.S. GDP by $142 million and increase U.S. unemployment by 2,020 jobs. The model finds that the U.K. economy would grow by $202 million but suffer increased unemployment of 1,841 jobs. (Further details are available in our complete paper, here.)
The changes in GDP and job totals are very small in the context of the two multi-trillion-dollar economies. Nevertheless, they illustrate two important phenomena that some economists still deny: first, that trade agreements can and do cause job loss and reduced production. Secondly, they illustrate that GTAP, the most widely-used trade model in the world, can indeed be modified to estimate the levels of unemployment caused by trade. In view of this new CPA methodology, there is no reason why all modelers analyzing the impact of trade agreements should not incorporate unemployment into their models.
The U.S. and U.K. Economies
Both the U.S. and U.K. have been leaders of the move to reduce or eliminate tariffs since 1945. Both countries have some of the lowest tariffs in the world today, on each other’s goods and on other countries’ goods. The trade-weighted average of U.S. applied tariffs on U.K. imports is 2%, while the figure for U.K. tariffs on U.S. goods is also 2%. So any further reduction in tariffs would likely have a small effect on the two economies. However, trade agreements can have surprising effects. For example, before NAFTA, U.S. tariffs on Mexican-produced autos and auto parts were low, around 3%. Mexico then exported virtually no automobiles to the U.S. In practice, NAFTA turned out to be more of an investment agreement than a trade agreement: it was a signal for the world’s automakers to begin building state-of-the-art auto plants south of the border. Today, Mexico produces around 3.9 million cars a year and exports around 2.5 million of them to the U.S. Most of those autos represent production and jobs that would still be in the U.S. if there had been no NAFTA.
Like Mexico, the U.K. is also focused on the U.S. auto market. Before COVID, the U.K. produced around 1.4 million vehicles a year. The U.S. is the number one destination market for the U.K. auto industry. Americans buy around 18% of UK auto exports each year, far more than any other country. While U.K-built autos have lost market share in Europe in the decade up to 2019, the U.K. gained share in the U.S., due largely to the success of distinctively British brands Mini, Jaguar, and Land Rover/Range Rover.
Autos are not the only sector where the U.K. sees a potential for gains in the U.S. market. In sectors like machinery, the U.K. also has a reasonable opportunity to increase U.S. market penetration. Table 1 shows U.S. goods exports to the U.K. by major sector, and U.S. imports to the U.K. In 2019, we ran a small goods surplus with the U.K. of $5.8 billion. Our deficit in motor vehicles was substantial, at $8.5 billion, partially offset by a U.S. surplus in aerospace equipment.
A U.S.-U.K. trade agreement has attracted little attention in the U.S. Some of the U.S. discussion has focused on potential U.S. benefits in the food manufacturing and agriculture sectors, where U.S. costs are likely to be lower than U.K. costs due to the U.K.’s membership of the European Union’s price-fixing Common Agricultural Policy for most of the past half-century.
|Table 1. U.S.-U.K. Trade by Sector, 2019, Selected Sectors (All figures in Millions of U.S. Dollars)|
|Commodity||Exports to U.K.||Imports from U.K.||U.S. Surplus (Deficit)|
|Oil and Gas||$5,974||$1,511||$4,463|
|Food & Kindred Products||$675||$521||$154|
|Beverages & Tobacco Products||$438||$2,192||-$1,754|
|Textiles & Textile Mill Products||$211||$253||-$43|
|Apparel & Accessories||$333||$140||$193|
|Leather & Allied Products||$60||$82||-$23|
|Paper and Printed Matter||$997||$733||$264|
|Petroleum & Coal Products||$1,245||$2,445||-$1,199|
|Plastics & Rubber Products||$900||$761||$139|
|Nonmetallic Mineral Products||$292||$235||$58|
|Primary Metal Mfg||$12,328||$1,356||$10,973|
|Fabricated Metal Products, Nesoi||$1,725||$1,677||$48|
|Machinery, Except Electrical||$3,842||$6,667||-$2,824|
|Computer & Electronic Products||$5,241||$4,033||$1,208|
|Electrical Equipment, Appliances & Components||$2,223||$1,725||$498|
|Motor Vehicles, Bodies & Parts||$2,633||$11,121||-$8,487|
|Aerospace Products & Parts||$10,338||$5,604||$4,734|
|Railroad, Ships, & Other||$116||$291||-$175|
|Furnitures & Fixtures||$225||$297||-$72|
|Miscellaneous Manufactured Commodities||$2,891||$1,848||$1,044|
|Source: U.S. Census|
In manufacturing, the U.K.’s lower labor costs than the U.S. provide support for its aspiration to find larger markets in the U.S. Table 2 shows the labor cost differentials in several major markets. The table, using data gathered by the business association the Conference Board, shows that the compression of manufacturing wages that has taken place in the last two decades of “hyperglobalization” has affected some countries more than others. The “Direct Pay” column refers to the wage received by the worker, while hourly compensation costs include labor taxes and other social charges paid by the employer.
Today, Germany, Denmark, and Switzerland all have higher hourly labor costs and higher manufacturing wages than the U.S. Those countries often support their manufacturing industries with government grants or tax incentives. A case in point is Ireland, where direct manufacturing wages have now surpassed the U.S.
U.K. governments have been content to see manufacturing industries decline and as a result, the U.K. has some of the lowest manufacturing wages among advanced nations. These highly competitive wages (27% cheaper labor costs than the U.S.) provide some basis for the U.K.’s ambition to take a larger share in the U.S. market for autos and other manufactured goods. However, the U.K.’s manufacturing industry is far from dynamic. The U.K. suffers from many of the same ailments as the U.S. economy, including persistent trade deficits, growing import penetration, and a culture biased against manufacturing production. Nevertheless, U.S. multinationals have shown interest in investing in the U.K as a production base, and a trade deal could increase that interest. Further, the U.K. government continues to subsidize important industries including auto production, typically to maintain employment levels. Both Nissan and Stellantis (the renamed Fiat-Chrysler) have recently announced new investment into electric vehicle production at U.K. locations, with investment support from Boris Johnson’s government. Since the 2009 bailouts of General Motors and Chrysler, the U.S. has not provided financial support to its auto industry.
Unlike Boris Johnson’s government, the Biden administration has shown little interest in a U.S.-U.K. trade agreement. On June 17th, when the U.S. and U.K. government reached an agreement to suspend tariffs related to a dispute over large aircraft subsidies, U.S. Trade Representative Katherine Tai pointedly did not mention a U.S.-U.K. trade agreement. Instead she spoke about the U.S. and U.K. working together to address “common challenges from China and other non-market economies.” Tai has often invoked the concept of “worker-centric” trade policies, to distinguish the Biden administration’s trade policies from the past 30 years of trade policies which played a key role in the loss of some five million manufacturing jobs. We have yet to see what “worker-centric” trade policies mean in practice.
To manage the effects of trade agreements, one first must be able to measure their effects objectively. Our aim in this modeling exercise is to make trade modeling more objective, realistic, and relevant.
|Table 2. International Manufacturing Labor Costs Compared, Total and Direct Pay, 2016 Data|
|Country||Hourly Compensation Costs in Mfg inc. Social Costs||Index (US=100)||Direct Pay||Index (US=100)|
|United States||$ 39.03||100||$ 29.66||100|
|United Kingdom||$ 28.41||73||$ 24.29||82|
|Denmark||$ 45.32||116||$ 40.06||135|
|France||$ 37.72||97||$ 26.25||89|
|Germany||$ 43.18||111||$ 33.72||114|
|Ireland||$ 36.23||93||$ 30.22||102|
|Italy||$ 32.49||83||$ 23.07||78|
|Slovakia||$ 11.57||30||$ 8.41||28|
|Switzerland||$ 60.36||155||$ 49.25||166|
|Mexico||$ 3.91||10||$ 2.74||9|
|Japan||$ 26.46||68||$ 21.67||73|
|South Korea||$ 22.98||59||$ 18.59||63|
|Taiwan||$ 9.82||25||$ 8.35||28|
|Source: U.S. Conference Board|
Modeling the Effects of Trade Agreements
In 2020 the U.K. government’s Department of International Trade (DIT) published a 184-page economic studyestimating the benefits of a U.S.-U.K. free trade agreement. The DIT study is based on a computer-generated equilibrium model of the world economy. The report does not specify what model is used, but the database used comes from GTAP (Global Trade Analysis Project), the trade model developed at Purdue University and widely used for such exercises. The model used by the U.K. government economists may be a customized version of GTAP.
We have written previously of the notoriously bad record of trade models at forecasting actual results from free trade agreements. In testimony to the U.S. International Trade Commission last October, we identified several ways in which the GTAP trade model focuses on the positive benefits of freer trade and disallows negative impacts from influencing the model’s forecast. “The most important negative consequence of increased imports caused by trade action, namely increased unemployment in the affected industries, is not allowed to happen in the standard GTAP trade model,” we wrote.
Our work built on earlier work by Professor Tim Kehoe of the University of Minnesota. In a 2003 paper, Professor Kehoe looked at the results of economic models of the NAFTA trade agreement which took effect in 1994. “These models drastically underestimated the impact of NAFTA on North American trade. Furthermore, the models failed to capture much of the relative impacts on different sectors,” Kehoe wrote. In a 2018 paper, Professor Kehoe looked again at NAFTA model results compared to the actual performance of the U.S., Mexican, and Canadian economies. This time his conclusion was even more severe. “The average correlation between model prediction and post-reform data is 0.0—that is to say, the GTAP model had essentially zero predictive accuracy.”
As a publicly available, open-source model, GTAP can be modified and customized by any users. The GTAP founders reinforce this point with their slogan: “If you don’t like it, help fix it.” Therefore the standard GTAP model should be viewed as a starting point for serious analysis and customization should be used to capture real-world effects.
The U.K. DIT model found that in the case where a U.S.-U.K. trade agreement reduced all bilateral tariffs to zero (and also reduced non-tariff barriers), U.K. GDP would rise by £3.4 billion ($4.8 billion) or 0.16%.
Using the GTAP model, the Coalition for a Prosperous America ran its own simulation of the effects of a U.S.-U.K. trade agreement. We then modified the model to track the effect on different sectors and include the impact of unemployment. We found much smaller benefits than the DIT model found, including only $303 million in benefits for the U.K. economy, 94% smaller than the DIT estimate, and only $98 million for the U.S. economy, 99% smaller. We suspect this huge disparity is due to the DIT’s inclusion of abolition of non-tariff barriers in their model. We did not alter non-tariff barriers as changing such barriers (for example, health and safety regulations that keep out foreign imports), involves some guesswork as to how that will affect import volumes.
In our Worker Displacement model, we made a significant modification to the standard GTAP model. The purpose of a free trade agreement for any individual country is to enable trading partners to provide goods to that country more cheaply than it can do itself. The standard economic assumption is that the resources thus freed up in the importing country will move into other more productive uses. It’s clear that the disruption caused by increased imports will lead to unemployment for a time. The level and duration of this unemployment has been quantified in many economic studies. We used these estimates to develop what we call the Worker Displacement version of GTAP.
The U.S. Bureau of Labor Statistics has been surveying U.S. workers about their experiences following mass layoff events for more than 20 years. The results of these surveys are very consistent, finding that between 25% and 40% of workers are not working, following layoff events that took place between one and three years before the survey. For example, the 2020 survey found that 29.1% of displaced workers were still non-employed in January 2020. (We use the term non-employed because it combines two BLS categories, the unemployed, and those who have dropped out of the labor force.) For the manufacturing sector, 36.4% of displaced manufacturing workers were non-employed in January 2020.
For the Worker Displacement version of our trade model, for all sectors that saw a reduction in output due to the FTA, we inserted job loss figures of 36% for the manufacturing sectors and 29% for non-manufacturing sectors.
We obtain similar estimates for the U.K. from two studies of the British economy. One study found that displaced workers lost 24% of their income over a period of five years after the displacement event. Another British study of the 2015 closure of a British steel plant in northeastern England found that for 18% of the survey respondents, it took up to two years to find a new job and 24% of the former steelworkers who described themselves as employed were either self-employed or employed part-time. We use the 24% figure for worker displacement in the sectors of the U.K. economy that suffer output declines.
When we insert job losses into the model, we obtain very different results for the impact of the trade agreement. The reduction in employment in sectors that lose output reduces sectoral output and national GDP. Other sectors still see a growth in output and employment. Note that we do not include secondary employment effects: for example, when General Motors shut its Lordstown facility in 2020, it led to employment declines in many local service businesses that supported Lordstown workers. This sort of effect is not in our modified model.
For our Worker Displacement model, we took all the sectors where output declined and assumed that the workforce declined by the same percentage (just as the standard model would do). We then applied our worker displacement parameters to those workers and instead of assuming that all workers found new jobs in other sectors, we made them unemployed in the model, reducing sectoral output. In the U.S. and the U.K., 15 out of 24 sectors saw a decline in output. . For example, in the U.S. output in manufacturing sectors including transportation equipment (motor vehicles), computers, and furniture declined. In the U.K., the transportation equipment, chemical manufacturing, and petrochemical sectors grew relatively strongly, while food production, furniture manufacturing, and financial services declined.
The results of our Worker Displacement model are shown in Table 3. The net effect is that the U.K. economy still grew, but growth fell to just $202 million or one tenth of one percent of the U.K. economy. The U.S. result flipped from GDP growth of $98 million to a fall in GDP of $142 million. The job loss figures, when all sectors are added up, come to 2,020 jobs lost in the U.S. and 1,841 jobs lost in the U.K. The vast majority (83%) of the jobs lost in the U.S. are in manufacturing sectors, while in the U.K, the manufacturing sectors see a net gain of 220 jobs.
|Table 3. Model Results Compared|
|UK DIT||CPA-Baseline||CPA-Worker Displacement|
|GDP change (%)||0.05%||0.0006%||-0.0008%|
|GDP change ($M)||$ 10,500||$ 98||$ (142)|
|Trade Balance change ($M)||NA||$ (1,415)||$ (1,294)|
|Total Employment change (no. workers)||NA||NA||-2020|
|Manufacturing Employment change (no. workers)||NA||NA||-1670|
|GDP change (%)||0.16%||0.01%||0.01%|
|GDP change ($M)||$ 4,760||$ 303||$ 202|
|Trade Balance change ($M)||NA||$ (1,146)||$ (1,095)|
|Total Employment change (no. workers)||0.1%||NA||-1841|
|Manufacturing Employment change (no. workers)||NA||NA||220|
|Source: U.K. Department of International Trade; CPA|
These are all quite small figures in comparison with U.K. GDP which is around $2.8 trillion and U.S. GDP at $21 trillion, but the point is to show two things: first, that real-world unemployment effects can be introduced into the standard economic model of trade; and secondly that once you do so, it is very possible that a trade shock involving trade liberalization can lead to declines in national output and employment for a period of several years.
The automotive sector is the best illustration of the differing interpretations of how trade affects jobs. The standard trade model works by assuming lower tariffs reduce prices of motor vehicles and parts in both countries, leading to greater consumption at home and greater exports. In other words, both countries take a greater share of the global market due to their lower prices. Experience, however, teaches us something different. When the U.S. signed a trade deal with Mexico (NAFTA), and again with South Korea (KORUS), the U.S. did see auto and auto parts prices fall (relative to all consumer prices), but instead of that boosting U.S. production, production migrated to the so-called partner country. The U.S. lost global market share and thousands of workers lost their jobs. U.S. brands might have gained global market share, but U.S. factories and workers did not.
In a sense, the CPA model and the DIT model are not directly contradictory. Our model represents a forecast of the medium-term impact of a trade agreement, the effects on both economies after a period of some two to five years. The DIT model represents a forecast of the long-term effect, assuming all workers made unemployed by any trade shocks find re-employment at the same wage. Economists recognize the huge assumptions hidden within their models, and acknowledge them. The DIT explains these assumptions, but like many such studies, buries them so as not to inconvenience the politicians who will wish to cite the report as evidence of the benefits of trade agreements. Here are the frank comments of the DIT economists, buried on page 69 of their report:
“The CGE model estimates long-run impacts…the model does not provide estimates of the magnitude of any potential short-run impacts, such as the impact on unemployment associated with workers moving jobs within or across sectors or within and across geographical nations and regions of the UK economy…the CGE model assumes both the supply of labour and overall rates of employment and unemployment in the economy are fixed in the long-run…This is appropriate as over the long term…FTAs would not be expected to influence the underlying drivers of the long-run employment and unemployment rates.”
Our view is that trade agreements and other trade shocks can and do impact domestic unemployment levels locally, regionally and nationally, and these effects can last for years or decades. One has only to visit Detroit, Michigan, or Oldham, Lancashire, to see the effects of trade and import competition in destroying jobs and making cities poor and deprived. The short-term and medium-term impacts of bad policy choices are much more important than the long-term effects, because the short- and medium-term impacts change the long-term outcome. The long-term outcome is of course very important, but a long-term forecast made in 2020 and based on rosy, fallacious assumptions of what will happen in 2035 will inevitably turn out to be completely wrong.
This is why British economist John Maynard Keynes broke with tradition in 1930 to support tariffs for the U.K. economy. The point of Keynes’s dictum that in the long run we are all dead was not to say that the long run does not matter, but instead to point out that short term effects can render the long-term equilibrium forecasts of traditional economics irrelevant. Keynes explained this clearly in his General Theory, where he expressed his disdain for trade economist David Ricardo, and pointed out the danger for economists of relying on equilibrium models that bear little relation to reality:
“The completeness of the Ricardian victory is something of a curiosity and a mystery…although the doctrine itself has remained unquestioned by orthodox economists up to a late date, its signal failure for purposes of scientific prediction has greatly impaired, in the course of time, the prestige of its practitioners. For professional economists…were apparently unmoved by the lack of correspondence between the results of their theory and the facts of observation—a discrepancy which the ordinary man has not failed to observe, with the result of his growing unwillingness to accord to economists that measure of respect which he gives to other groups of scientists.”
Eighty years before the U.S. elected Donald Trump president and U.K. voters voted to leave the European Union, Keynes saw that denying what people can see with their own eyes would discredit economists.
The U.K. government’s trade model accurately reflects the thinking of the British government in one important respect: that with lower costs, the U.K. ought to be able to take a larger share of important U.S. manufactured goods markets, thereby increasing its global share via the economies of scale that would follow. One can doubt whether this effect would actually occur, because the U.K., like the U.S., is not culturally or politically equipped to exploit export opportunities, at least not in manufacturing. The U.K., like the U.S., is an economy with a track record of decades of trade deficit, manufacturing decline, a disgruntled middle class, and an elite professional class enamored of a self-harming free trade doctrine. The free trade doctrine may not be the fundamental cause of economic distress. But like an incurable infection, it makes every other economic ailment worse, and unless arrested, will ultimately destroy the patient.
By modifying the standard trade model to show how trade agreements can and do generate win-lose outcomes, we hope to bring the discussion of trade and economic outcomes onto a more realistic level.
 Statement by Ambassador Katherine Tai, June 17, 2021. https://ustr.gov/about-us/policy-offices/press-office/press-releases/2021/june/statement-ambassador-katherine-tai-us-uk-cooperative-framework-large-civil-aircrafts.
 U.K. Department of International Trade, UK-US Free Trade Agreement, 2020. (henceforth, “DIT”)
 Ferry, Jeffrey, CPA Testimony on Economic Modeling of Trade Agreements, October 2020, pg. 3.
 Kehoe, Timothy J., An Evaluation of the Performance of Applied General Equilibrium Models of the Impact of NAFTA, April 2003, Federal Reserve Bank of Minneapolis Research Staff Report 320. http://users.econ.umn.edu/~tkehoe/papers/NAFTAevaluation.pdf
 Kehoe, Timothy J., Rossbach, Jack, and Pujohas, Pau S., Improving the Analysis of Trade Policy, January 2018, Federal Reserve Bank of Minneapolis, https://www.minneapolisfed.org/article/2018/improving-the-analysis-of-trade-policy
 U.S. Bureau of Labor Statistics, Displaced Workers Summary, Worker Displacement 2017-2019, Aug. 27, 2020, https://www.bls.gov/news.release/disp.nr0.htm
 Hijzen, Alexander, Upward, Richard, and Wright, Peter W., The Income Losses of Displaced Workers, Journal of Human Resources, 2008.
 Coats, David, A Just Transition, Managing the Challenges of Technology, Trade, Climate Change and COVID-19, Chapter 3. Cold Blue Steel and Sweet Fire: The Closure of the Redcar Steel Works, Alex Ferry Foundation, 2015.
 DIT, pg. 69.
 Eichengreen, Barry, Keynes and Protection, Journal of Economic History, June 1984.
 Keynes, John Maynard, The General Theory of Employment, Interest, and Money, Harcourt, 1964, pg. 32-33.