Overview

The United States of America traditionally embraced the free flow of financial capital along with the free flow of people or the free flow of goods. In the 1920s and 1930s, after policymakers raised tariffs and imposed quotas on immigrants while maintaining free international capital flows (except gold), growth in total factor productivity, or TFP – also called multi-factor productivity, or MFP — hit all-time highs. Will TFP (output per all inputs) also soar if the second Trump administration continues to restrict the inflow of people and goods but not capital?

The goal of this note is to gain insight into the macroeconomic effects of current policies if U.S. borders, for just the second time in the nation’s history, remain largely shut to trade and people, while international capital continues to flow freely. This is important because the three main facets of international economic policy — barriers to the flow of goods, people and capital — are rarely considered together despite considerable empirical evidence that they are intricately intertwined (Taylor and Wilson 2011).

Many economists rightly remain skeptical of claims that tariffs can spur productivity-enhancing innovation because companies may respond to import taxes with merely defensive innovations made solely in response to trade barriers, or with rent-seeking innovations, such as smuggling techniques (Albrecht 2025). This note focuses on TFP, rather than patents or other looser metrics of innovation, because it best captures innovations that increase productivity.

This note should not be construed as endorsing any set of policies as it does not assess the overall impact of immigration, trade or capital barriers on real per capita output. Tariffs, for example, almost always reduce domestic and world output (Pugel 2020) and they certainly hurt labor productivity (output per hour worked) at times, including from 1870 to 1909 (Klein and Meissner 2024). During that period, however, immigrants came to the United States legally in large numbers, dampening incentives for employers to invest in productivity enhancing innovation. Because both people and goods inflows faced high barriers in the 1920s and 1930s, incentives to invest in productivity enhancements waxed.

Table 1 summarizes the combinations of barriers to the international flow of trade, people, and assets implemented by U.S. policymakers in various periods since 1790, along with their putative effects.

Establishing the causation of the putative effects in Table 1 must remain outside the scope of this note. Correlation may be coincidental, not causal, even if the putative effects are theoretically justified. Also, reverse causation may be a factor. By protecting incumbents from competition, for example, occupational licensing and scope of practice regulations increase political support for freer immigration.

Immigration Barriers

Matching the policy categories in Table 1 to precise years is difficult, mostly due to the see-saw tariff rates of the earlier periods, whether measured as the average tariff rate on all imports or the average tariff rate on only dutiable goods. The frequent ups and downs were not due to changing legislation but rather from changing import prices (Irwin 2017, 5-7). Although some tariffs were imposed on an ad valorem basis, i.e., as a fixed percentage of the import price, legislation set most as specific duties, i.e., as a fixed dollar amount. As the price of an import with a specific duty increased (decreased), the tariff rate automatically decreased (increased). A $5 specific duty on a bolt of a certain cloth priced at $10 was 50 percent but if the cloth’s import price increased to $20 per bolt, for example, the $5 fixed dollar tariff became a 25 percent tariff.

As Figure 1 shows, periods of tariffs generally rising (1790-1830; 1921-1933; 2018-present), generally falling (1831-1861; 1903-1920; 1934-1979), generally staying high and range bound (1862-1902), and generally low and range bound (1980-2018), can be discerned. Tariff history can be divided in other ways. Irwin (2017, 8), for example, sees three periods based on the three “R” goals of tariff policy, revenue from the Founding until the Civil War, then restriction for protection of domestic producers until the Great Depression, then reciprocity until the near present.

It remains unclear whether the key variable for TFP is the tariff level or its trend, as indicative of expectations of future tariffs. Regardless, key events mark each transition: the Nullification Crisis of 1832-33; the U.S. Civil War; the Great Mess (my term for the two world wars and the Depression); presidential elections.

The difference between average tariffs on all imports and on dutiable goods arises because various tariff laws and free trade agreements allow(ed) some goods to enter free, typically because the goods were intermediate inputs or because no domestic producers existed to protect (Irwin 2017, 5-6). Note that the measures track each other closely but that the size of the spread varied, especially between about 1875 and 1975, due to changes in the number and value of free goods relative to dutiable imports.

Due to what former U.S. Deputy Treasury Secretary (1993-94) Roger Altman termed “creeping protectionism” (Altman 2009, 2), average tariff levels, however measured, became an increasingly less reliable measure of trade barriers as various non-tariff barriers (NTBs, sometimes now NTMs for “non-tariff measurers”) proliferated over the last half century or so. Every year since 1984, the United States, rivaled occasionally by India, has implemented the most measures to counter dumping, trade subsidies, and import surges under the WTO system (https://www.worldbank.org/en/data/interactive/2021/03/02/temporary-trade-barriers-database). The U.S. remains the world’s heaviest user of NTBs (Thompson 2025, 3, 14). Accounting for tariffs, NTBs, and other trade barriers, including emerging barriers to digital trade, the Trade Barrier Index (TBI) ranks the United States 61^st lowest in trade openness in the world in 2025, just ahead of Poland and just behind Guyana. If Trump’s present trade policies persist, and other countries’ policies stay the same, the US will drop to 113^th place out of 122 in the TBI in 2026 (Hannan 2025).

Other researchers using other metrics also find that US trade barriers are already higher than a casual look at average tariff levels would suggest. Before Trump’s 2025 tariff increases, some industries, like textiles and footwear, enjoyed double digit tariff protection. Moreover, NTMs that cover 80 or more percent of all imports protect textiles, chemicals, and food products (Leibovici and Dunn 2023).

Trade Barriers

The net flow of people into the United States as a percentage of the nation’s population shows just two major policy legal regimes, largely free legal entry until the Great War, followed by largely restricted legal entry since.

During the open period, three great immigration waves can be discerned, along with cycles that match the relative position of the U.S. economy vis-à-vis that of Europe. During the period of restriction, net migration was sometimes negative, indicating that more people left the country each year than entered it on a permanent basis. In fact, net migration turned slightly negative from 1931 until 1933, was zero in 1934 and 1935, and was positive but de minimis the rest of the decade and throughout World War II.

Those figures likely represent a lower bound as it has long been suspected that official statistics undercount immigrants. Even critics of the government’s numbers, however, note that its series tracks the accounts in numerous narrative sources about immigration flows (see, e.g., Chickering 1848, 23-25).

Capital Barriers

Barriers to international capital flows include taxes, quotas, bans, and other capital controls on the international exchange of assets. Such barriers are so infamously difficult to reduce to a single number that one research team cheekily cited “introspection” on a graph purporting to measure international capital mobility since the mid-nineteenth century. According to their stylized view, capital mobility increased (barriers decreased) globally after 1860, plummeted during the Great War, rebounded slightly before 1929, then plummeted to new lows by the end of World War II, increased slightly and slowly during the Bretton Woods era, then increased rapidly after 1980, attaining new heights by 2000 (Obstfeld and Taylor 2004, 28).

Generally, we know that in the US barriers to international capital flows were low to nil during the gold standard era, a.k.a. the first wave of financial globalization, and the post-Bretton Woods float period, a.k.a. the second wave of financial globalization (Ghosh and Qureshi 2016, 7-8).

U.S. barriers to international asset exchange were relatively high only during the end phase of the Bretton Woods system of fixed exchange rates. For much of that period, nations other than the U.S. imposed capital controls that American policymakers typically refused to help them to enforce (Ghosh and Qureshi 2016, 17). In fact, Quinn estimates that aside from the gold ban, the US effectively had zero capital controls from 1950 until through 1965 and from 1974 to the present (Quinn 2023; Ghosh and Qureshi 2016, 21). In the intervening period, it restricted international asset flows with the Interest Equalization Tax (IET), which imposed a variable rate surcharge on debt or equity instruments issued by various foreign issuers purchased by U.S. entities (Neely 1999, 24; Ghosh and Qureshi 2016, 18-19).

The period book-ended by the two world wars and dominated by the Great Depression is somewhat difficult to parse de facto but de jure flows into and out of the US were open, except for gold beginning in 1933 (Mitchener and Wandschneider 2015, 191). Internationally, capital controls decreased after the Great War (Ghosh and Qureshi 2016, 11), only to reappear with a vengeance in the 1930s as the world’s largest nations, including the United States, one-by-one gave up the retail gold standard to gain more monetary policy discretion (Neely 1999, 13, 17; Ghosh and Qureshi 2016, 13). What one team of researchers termed “a dizzying decade of destabilizing hot money flows to which countries responded with a variety of exchange controls on outflows” resulted (Ghosh and Qureshi 2016, 4). Although U.S. policymakers banned Americans from owning, importing, or exporting gold, they otherwise did not impose capital controls despite potentially destabilizing long- and short-term capital inflows (Ghosh and Qureshi 2016, 14-15).

As Figure 3 shows, US policymakers during the Great Mess also tried to maintain fixed exchange rates with Britain. Prior to 1933, adherence to the gold standard and free capital flows more generally produced stable exchange rates at gold parity. Periods of disruption coincide with Britain being off gold. After 1933, neither country adhered to the gold standard, so stable exchange rate conditions indicate active management by both countries through stabilization funds (League of Nations 1944, 146-50). The dollar-sterling exchange rate strayed from the traditional gold parity of $4.86 per pound in 1933 but stayed range-bound, with help from the 1936 Tripartite Agreement between the US, Britain, and France, until World War II broke out in 1939. During the war, Treasury worked with British monetary authorities to manage the fixed exchange rate of $4.03 per pound agreed to in the 1936 agreement with wartime foreign exchange controls. After the war, the Bretton Woods fixed exchange rate regime took effect and allowed pound devaluation in 1949.

Summary of the Quantitative Evidence

In the 1920s and 1930s, U.S. tariff rates, however measured, increased, immigration flows slowed and even briefly reversed, and capital, except for gold after 1933, continued to flow into and out of the country with few barriers.

Today, U.S. tariff rates and other trade barriers are increasing dramatically and immigration has slowed considerably, and could possibly reverse. Legal barriers to international capital flows are almost nil, as increasingly are the transaction costs of such flows.

In short, in terms of international economic policy, the 1920s-30s can be compared to the policies of the Trump administration.

TFP Growth in the 1920s and 30s

This note focuses on the 1920s-30s because it was the only other time in U.S. history when policymakers simultaneously restricted immigration and international trade while allowing most capital assets to flow freely across its borders. Careful examination of the period may therefore provide some guidance about future TFP growth if present policies persist.

Economist Alexander J. Field argues that TFP growth rates hit historical highs in the 1920s and especially the 1930s (2011, 1), a.k.a. the Roaring Twenties and the Great Depression. The latter may seem intuitively improbable because TFP measures the efficiency by which an economy turns inputs like labor and capital into output (Zymek 2024). As Figure 4 shows, TFP levels in the postwar period generally rose year-over-year during expansions and decreased during recessions, but the level fell four times without a recession, and it often rose during recessions.

Figure 4

To see how TFP can move independently of output, it can be^[2] represented as:

Y = A * K^.3 * L^.7

Or in plain English, output (Y) equals TFP (A) times physical capital (K) weighted by its input share (traditionally thought to be .3) times labor input (L) weighted by its input share (traditionally .7). (The equation can also be rendered as percentage changes in all the variables over some period, like a quarter or year.)

Algebraically, a higher (lower) A, K, or L will increase (decrease) output, with other variables constant. That said, all three right hand variables can move simultaneously at different rates or even signs. In other words, Y could decrease, as during the Great Depression and some postwar recessions, due to decreases in K and L while A increases.

Drawing on numerous previous studies, Field shows that A, i.e., TFP, indeed increased in the 1920s and 30s at over 2 percent annually, significantly higher than any decade before, or since (27, 43). The only rival occurred during the long postwar Golden Era (1948-1973), when TFP growth averaged 1.88 percent per year by Field’s calculation.

If Philippon (2022) is correct, TFP is linear, i.e., the growth rate does not compound but rather increases additively over long periods at a highly stable rate calculated from the base year. If he is correct, recent declines in TFP stem solely from the mistaken assumption that TFP growth should be exponential. “We should not have expected growth rates to be constant,” he explains (32). Like Field, however, Philippon’s model shows that 1933, the pit of the Depression, constituted a break point after which TFP increased its growth trend slope (26-27).

In other words, whether TFP growth is linear or exponential, the 1930s still loom large and thus require explanation. Field notes that selective employee retention cannot explain the increase in labor productivity during the Great Depression, which he defines as 1929-1941, because most workers returned to the labor force in the late 1930s and early 1940s, presumably with no significant net gain in their human capital. That is because an additional year of education increases human capital no more on average than an additional year of job experience does and of course not all unemployed people went back to school for the duration of their unemployment. Workers culled by disability or death were indeed replaced by younger workers with higher average education but they were too few and too similar in education to earlier generations to account for much of the increased labor productivity. Similarly, depressed economic conditions culled the least efficient companies, like those still using steam power, but other inefficient companies eventually replaced them (Field 2011, 36-40).

What, then, caused sustained growth in TFP? Philippon credits implementation of a general-purpose technology (GPT) with TFP break points, which he claims also occurred in the eighteenth and nineteenth centuries, coinciding with the first and second industrial revolutions identified by economic historians (31-32). Field points to specific improvements in R&D, transportation, and telecommunications, starting with the manufacturing revolution of the 1920s. TFP in the industrial sector grew at a whopping 5.12 percent (continuously compounded) that decade, largely because the proliferation of electric motors allowed shop floors to be reconfigured in more efficient ways than possible when machines were powered by belts linked to waterwheels or steam engines (44-46).

Electrification of shop floors was largely complete by the 1930s, prompting Fields to credit industrial TFP increases in that decade in large part to increased investments in research and development. AT&T, GE, GM, IBM, RCA, Alcoa, Dupont, and Kodak indeed contributed disproportionately to TFP advances (47). After the initial shock from the Depression, they returned to profitability well before the buildup to World War II began. R&D employment increased from 6,274 in 1927 to 10,918 in 1933 to 27,777 in 1940. Employment of researchers grew fastest in chemical, electrical machinery, petroleum, and primary metals R&D and unsurprisingly that was where many important improvements occurred (55-56).

A new chemical process, for example, made it possible to extract more sugar from beets without any additional capital or labor inputs, while “topping” made it possible to recycle excess heat from high pressure boilers with a minimal increase in capital (48). Other new chemicals helped railroad ties to last longer and automobile paint to dry considerably more quickly (49). New materials, from plastics to tungsten carbide blades, also increased productivity, as did new more precise and efficient instrumentation (49-50).

According to Field, the cheap production and distribution of electricity, gasoline, and petrochemicals, along with improvements in aviation and trucking, made possible the wartime and postwar expansion (i.e., increases in K and L) of agriculture and industry (40-41).

TFP Growth in the Context of International Economic Policy

Field’s story seems complete and compelling until one asks why those productivity boosts came in the 1920s and 1930s instead of the 1910s or 1940s. Why, in other words, did companies invest in electrical motors, shopfloor reorganizations, and chemical and materials research in the 1920s and 1930s, not before or after?

Technological advance is often assumed to be exogeneous, coming randomly whenever somebody solves a technical problem.^[3] Technological advance, however, can be considered endogenous, a function of its expected return relative to alternative projects posing similar risks (heterogeneity of returns). Here is where trade, immigration, and capital flows can enter the analysis.

Unrestricted immigration may reduce real wages, or it may simply create producer expectations that real wages can be capped, or at least managed. That expectation, apart from the level of real wages per se, influences producers’ research and capital expenditures. Immigration restrictions, by contrast, unmoor real wage expectations, making investments in labor-saving capital, or research into new labor-saving capital, more attractive, all else equal.

A free flow of goods across borders may reduce the profit expectations of domestic firms facing competition from abroad, reducing their willingness and constraining their ability to finance capital expenditures or R&D. High trade barriers, by contrast, increase profit expectations for some companies, making investments in existing labor-saving capital, or research into new labor-saving capital, more easily financed and hence more attractive, especially if companies expect trade barriers to decrease in the medium to long term. In one well-documented case, Continental entrepreneurs indeed capitalized on Britain’s blockade of Napoleonic Europe to increase their rate of productivity-enhancing innovation, cognizant that the blockade would end when peace returned (Juhasz 2018).

A free flow of capital across borders may make it easier for domestic firms to finance projects but by broadening the menu of projects that domestic firms can undertake to include the acquisition of foreign businesses or securities, it may reduce the likelihood that domestic capital expenditures or research will be the best option. Capital flow restrictions, by contrast, force domestic firms to consider mostly, or even solely, domestic projects. Global depression and geopolitical fracturing can also limit capital flows de facto, especially outflows from relatively stable countries. That was the case in the 1930s and also today.

In the 1920s and 1930s, U.S. firms had three reasons to invest in research and development: fear of increased real wages; expectation of higher profits; a dearth of attractive foreign projects.

As Figure 5 shows, real wages increased early in the 1930s, then declined slightly and flattened, then showed robust growth. Despite high rates of unemployment, employers largely expected those increases because they knew about sticky wages and government policies, including immigration barriers and pro-union and minimum wage legislation, that increased the bargaining power of labor (Wright 2024, 76-78, 105, 128, 153-54). Unionized American workers never had more bargaining power than in the late 1930s (Pencavel 2023, 46).

As Figure 6 shows, stock prices generally rose in the 1920s until 1929, then fell through 1932, then rebounded until the 1937 recession. Those movements track what we know about corporate profits in the era (Epstein 1933, 3; Fabricant 1934, 1). Interestingly, the largest corporations remained profitable throughout the Depression, and unsurprisingly their share prices held up better than the indices (Fabricant 1934, 6).

International capital flows decreased significantly in the 1930s for a variety of reasons (Reinhart, Reinhart, and Trebesch 2017, 16-18). Capital flowed into the U.S. on net in the 1930s because, despite the Depression, it appeared relatively more stable than Europe, the only real alternative due to Japanese and Soviet capital controls (Mason 1992; Gregory and Sailors 2003). Moreover, exchange rate risks increased while hedging costs remained high. Finally, American banks and investors were extremely wary of international investment in the 1930s due to earlier losses on South American and European loans and securities (James 2016).

Implications for Today

If US policy continues to restrict immigration and trade and the geopolitical environment remains fractured, US companies may come to expect real wage and/or earnings increases large enough to induce them to invest in R&D and the proliferation of existing technologies, like AI and robotics. That, along with favorable regulatory and tax changes (Cardella 2025), could spur another big jump in TFP.

Recall, though, that TFP increases may prove insufficient to spur economic growth if labor and/or capital inputs decrease, for whatever reasons, including the very policies driving TFP growth (Gordon 2017). Unpopular, quixotic, rapidly-changing, unpredictable economic policies may induce firms to reduce production, as they certainly did during the 1930s (Baker, Bloom, and Davis 2016). As Figure 7 shows, U.S. trade policy uncertainty in 2025 is an order of magnitude higher than its pre-Trump peak in November 1993, the climax of the NAFTA ratification debate, and multiples higher than during the first Trump administration.

Moreover, the free international flow of capital could dampen domestic business investment to the extent that foreign projects offer competitive expected risk-adjusted returns. Given the high tariff wall, however, outbound FDI seems unlikely to prove appealing and continued geopolitical turmoil in much of the world may also dampen Americans’ demand for foreign assets, even portfolio investments.

In sum, TFP growth might be poised for an upside breakout like that experienced in the 1930s because America’s current international economic policies and the global situation again both provide incentives to increase research and development, especially in AI and robotics. That does not mean, however, that real per capita output will increase because labor and capital inputs might decline. It also does not mean that the benefits of TFP growth will outweigh the deadweight losses that restrictions on trade and immigration impose on the economy.