Free Trade

Published on 22 October 2024

The mainstream discourse around trade is dispiriting. Economists who should be elevating the debate instead mindlessly regurgitate the Ricardian model given to undergrads. The neoprotectionists seem unable to articulate an argument against this model, even though their central intuition – that outsourcing all good jobs is bad, is a genuinely defensible belief. So I thought I’d have a go.

Note I’m not contributing anything novel to trade theory here at all; nothing here is really all that contentious, it’s just mysteriously absent from the doctrinaire screeds coming from economists.

In the basic textbook model, nations have fixed amounts of capital, labour, land, etc. The question is how best to use those factors of production. When countries allow free trade and focus production on goods in which they have comparative advantage, all nations end up being better off. The basic intuition is that nations are freeing up resources from producing goods that they’re ‘bad’ at producing so they can allocate them to producing goods which they’re ‘good’ at producing, and when everyone does this there’s a net gain that ends up being shared.

The most common model used by trade departments to assess free trade agreements and so forth are static CGE models which more or less are just sophisticated versions of above. There are a great variety of academic trade models which don’t have the same constraints, but they’re not what I’m criticising.

Let’s take labour supply for example. There’s ample evidence that the China-shock led to permanent reductions in participation rates in newly deindustrialised areas. Labour was not ‘freed up’ to produce other things, many became welfare dependants or criminals. Indeed to receive welfare many had to commit fraud and pretend they were disabled, and there seems to be some weird mechanism where if you induce a lot of people to pretend to be disabled, eventually they really do become disabled. The result is shown in surveys for things like ‘experienced pain in the last week’, and undoubtedly fed into the opioid epidemic in the US. Note that to fund these welfare dependents you end up levying higher taxes on industries and individuals which are productive, which might be more onerous on said industries than the tariffs etc. ever were.

I’ll repeat myself one last time on this point, the above claim that free trade can reduce labour supply is widely accepted by economists, it’s just not in the models that are used for decision making and economists similarly ignore it when making free trade arguments in the public arena. This motte and bailey technique is very common in economics.

Next let’s consider human capital. Economists claim that deindustrialisation, if anything, improved human capital. To address this we need a digression on what ‘human capital’ means.

Mainstream economists, following on from Mincer, equate years of schooling with human capital. The aforementioned CGE models also typically have two categories of labour, skilled and unskilled, where skilled means having a higher education. Anyone remotely familiar with modern degree-mills knows how absurd this is. Empirically years of schooling doesn’t even robustly predict GDP growth, despite human capital supposedly being a core-feature of the MRW model (essentially the Solow growth model with human capital added in).

The upshot of all this, is that while Western technical expertise was being transferred to the Chinese in special economic zones, Western economists instead insisted that increasing human capital meant churning out more baristas with communication degrees. This growth strategy, adopted most enthusiastically by the UK who greatly increased higher-education attainment rates, has catastrophically failed to result in any sort of labour productivity surge whatsoever. Somehow blame for this is typically directed to British politicians, even though they were just following the advice of mainstream economists.

In reality, a lot of human capital is acquired on the job or through apprenticeships. This is why people in their first few years in a career will see very rapid wage growth (among other causes). Many manufacturing jobs were semi-skilled but employed people without a HE degree. When those jobs disappeared the workers, if they did remain in the labour pool, were forced to take less-skilled jobs in hospitality and retail. This decline in skill-level was reflected in wages. However some workers, and their progeny, will ultimately respond to the disappearance of well-paid semi-skilled jobs by enrolling in higher education, hence the claim that empirically deindustrialisation improved human capital. I’m extremely skeptical though that the deindustrialised heartlands have more human capital than they did 50 years ago, regardless of HE completion rates. The lost human capital, e.g. how to operate a machine lathe, was simply was never measured by economists, so when it disappeared they didn’t notice.

In the textbook models, human capital is treated as exogenous, a function of education policy or some such. In reality, people respond to incentives when choosing how much to invest in skills. For example, if you visit schools in the bay area you’ll find the children are good at coding. While some of that is due to heredity, it’s also the case that if they become a decent coder they’ll be extremely well-remunerated given they live near employers who are desperately in demand for those skills. And the converse is equally true, if you rip all the good jobs out of an area, the children will invest less in themselves because they can see it is futile. This is one reason why it’s so difficult to improve educational outcomes in disadvantaged areas.

So occupational structure can causally change human capital accumulation, a fact acknowledged in passing in the literature but again ignored in the models used to make trade decisions, and equally ignored by economists who claim free trade can only result in net-gains.

Next we turn to physical capital. As is oft-remarked, Ricardo explicitly said his model only holds if capital isn’t mobile. Unlike the previous issues, some dynamic models really do model future investment decisions as a product of trade policy. But again, these are not the standard models used by trade bureaus, nor do economists mention it much in the media. Standard economic theory predicts that capital should flow from capital-rich to capital-poor countries, and the resultant capital shallowing should make those rich countries poorer for it. As with ‘human capital’, to understand why economists deny this, you have to look into the utter confusion they’ve got themselves into due to definitions.

Capital, in the context of classical economics, meant durable goods used in the production of other goods and services, i.e. a tractor or a piece of machinery. Capital, in the context of finance and importantly the system of national accounts, simply means any economic resource or asset. Economists have consistently denied that free trade + free mobility of capital results in capital outflows to poor countries on the basis of balance of payments data. But this is to conflate the two definitions.

The Lucas paradox1, that contra standard theory capital flowed from poor countries to rich countries, can largely be explained away by digging into the data and noting that FDI and portfolio equity investments actually do flow from rich countries to poor countries (in the absence of capital controls), resulting in capital shallowing in rich countries where the marginal product of labour is lower, just as we would expect from economic theory. The net capital flows from say China to the US wasn’t the Chinese building factories in the US, but China buying debt which often is a form of ‘deferred consumption’, i.e. bankrolling American healthcare entitlements through the purchase of American public debt.

Note even without China systematically rigging its currency and sovereign debt purchases, we’d likely expect to see something like this. The one-child policy plus a bad pension system forces a high-savings rate which naturally will seek other markets for better returns. In addition, countries like the US have better developed capital markets (e.g. you’re less likely to be scammed) than developing nations so naturally portfolio investments are biased towards it.

One can quibble over the various explanations for the Lucas paradox (it’s still a matter of contention in the literature), the main point is that neoclassical theory actually predicts free trade + free movement of capital will impoverish western countries through capital shallowing and that the empirical evidence against this is mixed at best. The textbook model supports the protectionists.

To give a more concrete example, which is less debateable, I can confidently assert that:

  1. Fixed capital formation has fallen over the last few decades in the UK
  2. This is partly due to the outsourcing of capital-hungry industry.
  3. Ceteris paribus this results in lower national output than otherwise would be the case.
  4. The static models used to assess free trade agreements simply assume this can’t happen.

Also observe that manufacturing doesn’t just consume capital, the automation in that sector means that labour productivity gains year-on-year tend to be higher than other industries. Mainstream economists essentially offshored the industry with the best productivity gains and then were surprised that productivity gains across the economy fell… the productivity gains from switching to the sectors with supposed comparative advantage (e.g. finance) was always going to be a one-off windfall.

Economists will occasionally pivot and say that their theoretical confusion doesn’t matter, we can see that trade helps countries in the real world. Firstly, cross-sectional national comparisons are a bad idea, there’s too many variables, too few data-points. Typically when a nation embarks upon trade liberalisation it’s always associated with other changes which cannot be adequately captured in the models. Unfortunately academia abhors a vacuum and where respectable econometricians fear to tread, grifters jump in, typically reaffirming the prejudices of the profession which means their shonky data work escapes serious scrutiny (has anyone ever changed their minds after reading AJR?)

But even if we suppose trade liberalisation has been associated with growth, does anyone think those were Ricardian gains? Like Korea was actually bad at producing rice in 1960, and their subsequent growth was in realising that they were good at building smart-phones instead? There’s an alternative narrative, that these countries used trade-policy to create jobs which were skills and capital intensive. These jobs induced both human capital accumulation and physical capital accumulation, which is to say there are demand-side effects. Trade imbalances also helped – just as Marxist countries ran surpluses and then used the foreign currency to buy machinery, the Asian tigers did this within the capitalist context in a more hands-off manner. FDI also enabled skill transfers, and because manufacturing is a sector which naturally has high labour productivity gains they benefitted from that as well.

All of this has been stated many times before. I’m just trying to clarify why mainstream economists are so opposed to this alternative narrative. On the one-hand, they’re ardent supply-siders. But on top of that, they exogenise the factors of production – quantity of labour is a function of immigration policy, fertility policy and health policy, quality of labour is a function of immigration policy and education policy, and quantity of capital is a function of financial regulation and tax policy. Having established that none of those variables are allowed to change through trade, they’re left with the simple exercise of allocating scarce resources. But poor countries do not get rich by getting better at allocating their near non-existent resources, they get rich by accumulating productive capacity. I would venture that China has for the last few decades pursued a policy of wasteful growth, while nations like the UK have pursued a policy of efficient decline.

On a final note, there are third-order effects that probably can’t be modelled, but should be considered. Obviously a strong manufacturing base comes in handy during a world war for instance. But perhaps even more important, the quality of institutions seems to partly reflect occupational structure. Countries dependent on primary industries (there are exceptions like Australia) tend towards high level of inequality and corruption as comprador elites can capture the resource rents without too much difficulty. Conversely manufacturing tends to support a large middle-class which better supports democratic institutions and technological innovation. A historian who simply asserts that manufacturing countries tend to do well in the long-run might be making a better argument than any the doctrinaire free-traders have ever proffered.

  1. Trade theory is riddled with paradoxes, which has done nothing to diminish the confidence of economists expounding on the topic. ↩