Comparative Advantage in the Age of Automation: The Impact of AI on Global Trade Dynamics

Artificial intelligence, or AI, is the ability of computers to generate output in a style and manner similar to humans.  Before user-friendly AI programs, such as ChatGPT, workers had to pull desired information from a variety of sources and compile it into the desired output.  This could take hours, or even days, of research.  AI allows the same output to be generated in minutes, with computers finding in seconds what might take workers hours.  So, what ramifications will AI have for the labor market?

The ability of AI to generate polished, human-like output will change the makeup of opportunity cost, which is the value of the next best alternative foregone.  In simple terms, it means what you give up when making a decision.  Based on the limits of the production possibilities frontier, when you are operating at capacity you must sacrifice some units of X to make an additional unit of Y, or vice versa.  AI will inevitably reduce opportunity costs related to communications work like designing and compiling reports and presentations.  It also helps further automate many office functions like data entry and bookkeeping, as well as legal research.

Effects of AI on Producers’ Comparative Advantage

By reducing the cost of office work, AI reduces the opportunity cost of focusing resources on physical production.  This will change the dynamics of comparative advantage, where nations specialize in making goods and services for which they have lower opportunity costs (give up less of another good or service).  Wealthy and highly developed nations trade with underdeveloped nations because wealthy nations tend to have high opportunity costs: producing more automobiles means giving up computers.  Underdeveloped nations, by contrast, may give up virtually nothing to produce more of their singular export, be it fish, agriculture, timber, or a mineral.  Even though the wealthy nation could produce more fish than its poor neighbor, it would have to give up other valuable products to do so.  Thus, the wealthy nation will buy fish from its poor neighbor so it can continue to produce its most valuable goods, such as smartphones.

AI may advantage developing nations that are well-known for physical labor in production (manufacturing) but are seen as less skilled when it comes white collar tasks like research-and-development.  No longer will corporations have to maintain large office staffs to complement factory production.  A small staff can use AI tools to generate the work of a formerly large office, especially by automating many data-related tasks.  AI can streamline bookkeeping and tax-related tasks, as well as answer basic questions about such processes.

Impacts on Labor Costs

Labor costs will likely decrease for some white collar workers and low-skill laborers due to AI automating many of their tasks, reducing demand for such workers.  Jobs that require physically complex tasks or socio-emotional skills will be more immune to AI automation.  However, decreased demand for some workers may depress wages overall due to an increased supply of available labor.  Some workers who become unemployed due to AI will compete for jobs that cannot be automated by AI, giving employers more power to avoid raising wages.  

Nations that will benefit are those focused on complex manufacturing, which cannot easily be automated.  In fact, corporations may take advantage of AI tools to shift white collar jobs from higher-income nations, such as Germany and the United States, to lower-income nations like China and India.  Lots of white collars jobs may be outsourced as AI makes it easier for foreign workers to generate content that is virtually indistinguishable from content made by a native speaker.  Multinational corporations will face high opportunity costs of not outsourcing white collar tasks, as they will be paying higher wages to workers at home when they could be using lower-cost labor abroad.

Therefore, nations known for heavy manufacturing may receive the most benefits of the AI revolution, as they will gain white collar jobs to complement their AI-insulated manufacturing jobs.  If a factory is in Shanghai and cannot easily be more automated by AI, while the white collar tasks are done in London, where wages are much higher than in Shanghai, why not move the white collar tasks to Shanghai as well?

Potential for Deindustrialization in Developing Economies

While AI will advantage some manufacturers, it will likely hurt others.  More complex manufacturing that involves reprogramming equipment to make different products will likely benefit, as AI can assist in this reprogramming.  Simple manufacturing may suffer as improved efficiency in complex manufacturing allows it to capture more of the market, especially if the increased efficiency leads to decreased prices.

An increase in manufacturing efficiency in developed countries may reduce opportunity costs of making simple goods, reducing the comparative advantage of developing countries.  For example, consumers may buy simple, lower-quality imports if the price is low enough…but they will switch back to buying higher-quality domestic products if there is no longer a substantial difference in price.  Developing nations are at risk of their low-cost goods being “absorbed” by increased efficiency in manufacturing in more developed nations.  This is especially true if trade restrictions like tariffs are imposed on the imports from developing nations. 

Policy Responses for Mitigating AI-Caused Unemployment

Some analysts feel that AI could cause significant changes in the labor market, especially in terms of a transfer of income from workers to employers.  The economic theory of stick prices and wages explains that workers’ wages will not likely increase even if the increase in efficiency from AI leads to higher company revenue.  Economic gains from AI will likely be concentrated among the owners of capital.  Wages may even fall if the efficiency gains from AI automation decrease demand for labor.  Even if there are no mass layoffs caused by AI, workers are likely to see a decline in their real income as wages are not increased by the same rate as inflation.  

If AI does lead to mass layoffs, there will be pressure on governments to support unemployed workers.  It is unlikely that governments would attempt to ban the use of AI by private companies, as this would violate free market principles and be seen as authoritarian.  Instead, governments will likely provide tax incentives or subsidies for private sector employers to retrain workers for non-automatable jobs.  This could involve a program similar to the PPP (Paycheck Protection Program) in the United States during the Covid pandemic of 2020-21: employers receive a de facto grant to retrain workers - or have them take training courses at a nearby institution - to take on new jobs within the company.

Companies that retain those retrained workers will not have to repay the loan, while those who lay off those workers will have to pay.  This incentivizes companies to find a use for workers whose original jobs were automated by AI.  In some cases, however, it is inevitable that no suitable new job will be found.  For companies with many employees, the “AI PPP” loan would not have to be paid if a certain percent of workers, say 80 percent, were retained over a certain amount of time, say four years, after the grant was taken.  If companies laid off too many workers too quickly, they would be forced to repay the loan. 

Historical Analogy and Controversy

In 1930, economist John Maynard Keynes famously predicted that his grandchildren would have a 15-hour work week thanks to automation.  The grandkids of the Greatest Generation, Generation X, are in their late forties to early sixties today - and are definitely not working just 15 hours per week!  Keynes’ turned out to be mistaken about the labor-saving aspect of automation.  Instead of using automation to not work as much, we simply do different tasks.  A similar phenomenon has occurred in the home, where the boom of convenience appliances in the 1950s did little to give us weeks of leisure.

Rather, expectations around work have changed to largely nullify the increased efficiency of technology.  The same may prove true of AI, making it the new email rather than a groundbreaking change in employment itself.  Even workers who do manage to do their original jobs in a fraction of the time using AI will have an incentive to take on other tasks, or at least appear busy, to preserve their jobs.  Although AI will certainly cause significant changes in at least some industries, it definitely will not be the “end of work” as some predict.