Models can mislead us on the impact of global trade

One of the most important lessons from the 2008 global financial crisis was that financial models don’t always work. The idea that you could throw thousands of variables into an algorithmic black box, shake them up with the millions of positions taken daily by banks, and extrapolate from it all a simple and easy-to-understand take on potential balance sheet losses or gains is now seen as naive. Risk can ricochet around in ways that we can’t begin to model mathematically, and market events often create their own momentum.

Likewise, since the Covid-19 pandemic and the war in Ukraine, there has been a rethinking of overly simplistic notions of shareholder “value”. We have left behind the era in which corporate leaders are expected only to raise share prices and lower consumer ones.

We now understand that stakeholders — from workers to communities to the public sector — must also be served. We see that “negative externalities” such as environmental degradation or low labour standards carry their own costs. That has forced a much deeper conversation about the true price of “cheap” goods and services.

But these sorts of realisations have not yet trickled down (no pun intended) to much of our thinking about global trade. Any questioning of free trade is still considered tantamount to a defence of protectionism. Tariffs are always bad.

Yet we seldom stop to consider the assumptions of the models that we have relied on for decades to help us arrive at these supposed truths. This is despite the fact that the events of the past 20 years have increasingly thrown into question our basic preconceptions about how countries do, or don’t, trade.

On that score we must consider everything from the rise of state-run capitalism and mercantilist China, through the successful use of industrial policy by the East Asian “tiger” countries, to the fact that most of the trade agreements signed over the past 30 years were less about removing cross-border restrictions and more about negotiating standards for workers, the environment, intellectual property and so on.

In such negotiations, multinational businesses have had a huge advantage relative to individual nation states and the workers within them. As the Indian politician Rahul Gandhi put it recently, the west “created” modern China as the factory of the world, since US and European multinationals favoured its “coercive” production model over those of other nations. Capital thrived by outsourcing production globally, while workers in places with hollowed-out job markets or polluted environments did not. 

These asymmetries are now prompting greater scrutiny of the models that policymakers have traditionally used to build support for free trade deals. Consider, for example, the general equilibrium models that economists use to analyse the impact of trade reform. They contain Panglossian assumptions about “full employment” and “costless switching”, according to which a laid-off auto worker in Detroit, say, can simply walk across the street and find a new job for the same pay.

Such models also fail to account for the tendency of capital to look for the places with the lowest production costs, or the broader economic and social effects of hollowing out communities. Likewise, they don’t tally the upside of growth from production rather than consumption, or the effects that stable jobs and capital stocks have on communities in the longer term. The upshot is that the models tend to underplay the costs of free trade and overplay, at least according to some analysts, the costs of tariffs.

In 2021, a group of Democratic senators complained to the US International Trade Commission about the assumptions and omissions in a report on the “small but positive effect on the US economy” of various trade agreements since 1984. Several years before, in 2018, a report from the Federal Reserve Bank of Minneapolis found that standard trade modelling failed to capture the real world effects of four recent bilateral trade liberalisations. In fact, the Minneapolis Fed researchers found that the model in question (the Global Trade Analysis Project model, or GTAP, for short) had “essentially zero predictive accuracy”.

Academics and trade groups are now experimenting with tweaking conventional trade assumptions. The Coalition for a Prosperous America, a bipartisan trade group representing US domestic producers and workers, recently modelled what would happen if the US placed 35 per cent tariffs on all manufactured goods from countries without free trade agreements with the US (including China), and 15 per cent tariffs on all non-manufactured goods. It also assumed “tariff productivity elasticities”, meaning that growth could happen through production rather than just cheap prices, and “factor supply elasticities”, meaning that the level of jobs and capital stock could increase.

The result was that gross domestic product went up by $1.7tn, 7.3mn new jobs were created and real household incomes rose by 17.6 per cent. Of course, this model didn’t account for the geopolitical fallout of such an action — just as conventional models didn’t account for the populism stoked by global trade paradigms running ahead of national politics. The point is simply that the assumptions we make matter when we think about trade.

rana.foroohar@ft.com

Letter in response to this column:

Economies suffer under newfangled tariff models / From Erica York, Senior Economist and Research Manager, Tax Foundation, Washington, DC, US