The Political Economy of Firm Networks: CEO Ideology and Global Trade

Elisabeth Kempf, Mancy Luo and Margarita Tsoutsoura. Presented at NBER SI International Economics and Geopolitics, July 10, 2025 (discussant: David Yang); no recording exists. Video used here: a nine-minute Swedish House of Finance interview with Kempf (October 2025) — an interview, not a seminar, so there is no discussant or Q&A to report. Paper: the August 10, 2026 draft, now retitled simply “CEO Ideology and Global Trade” (tsoutsoura.com); the SI 2025 version carried the longer title.

Private sanctions, administered by mood

Governments restrict trade with tariffs, sanctions and export controls, and a large literature measures what that does. Kempf, Luo and Tsoutsoura are after a trade barrier that no government imposes and no one votes on: the political opinions of the person who signs off on where the firm buys its inputs. Their earlier work documented that American executive teams have sorted themselves into partisan camps — Democrats working with Democrats, Republicans with Republicans — and the natural follow-up is whether a partisan CEO runs a partisan supply chain. The answer is yes, in a specific and slightly embarrassing way: not toward adversaries, where one could at least tell a national-security story, but toward ordinary democratic trading partners, depending on who won their last election.

How you would identify that

The problem with “ideologically close countries trade more” is that ideological closeness travels with geography, language, culture and trade policy. The trick is to hold the country fixed and let its government change. The authors combine transaction-level U.S. import data from Panjiva with CEOs’ party registrations from voter files, score foreign governments’ ideology with the Manifesto Project, and look at foreign national elections that move the government left or right. Italy in 2013 goes from center-right to center-left: for a Republican-led U.S. firm that is an increase in ideological distance from Rome, for a Democratic-led firm a decrease. Compare the two kinds of firms importing the same product from the same country around the same election — in the tightest version, from the same foreign supplier — and whatever is happening to Italian demand, Italian prices, Italian trade policy or the supplier’s own fortunes washes out. What is left is the CEO.

The firm whose CEO just became more distant is 4.1 percent less likely to keep importing from that country, relative to the firm whose CEO became closer, and the gap grows to 9.7 percent of the mean two years after the election. The effect is concentrated among CEOs with strong political views and among shorter trade relationships. (Kempf offers two readings of the second fact: short relationships have low switching costs, so they are the cheap ones to drop; or long relationships come with enough first-hand information that a “political gut reaction” to an election has less to work with.) Firms do not import less in total. They reallocate to other origin countries — which are, on average, ideologically closer to the CEO and more expensive, with higher unit values. That is the paper’s quiet damning detail: the firm pays more to source from places its CEO finds congenial. The interview adds the cost side from an earlier draft — foreign partners dropped this way lose about nine percent of export revenue, and among U.S. firms with heavy import exposure, cost margins rise about one percent when distance increases — which Kempf summarizes as the behavior being “not in their economic interests.”

The mechanism, which is not what you’d guess

The obvious story is that CEOs are making a political statement, a boycott. Kempf says explicitly she does not think so, and the evidence points the other way: the executives think they are making an economic decision. When a CEO becomes more distant from a foreign government, the firm’s earnings calls talk about that country more negatively and with more references to uncertainty (using the Hassan–Schreger–Schwedeler–Tahoun country-sentiment measures). A sentence-level pass with a large language model sharpens this: the extra negativity and uncertainty show up in how executives describe the foreign macroeconomy, not in anything firm- or industry-specific. A government that shares your worldview looks, from the C-suite, like a country in safe hands; one that does not looks like a place where bad states of the world are more likely. Ideology does not override the economic analysis. It is upstream of it, and that is worse, because it means financial literacy is no protection — a point Kempf makes about her whole research agenda in the first minute of the interview.

Why it matters beyond the anecdote

Two reasons, both flagged in the paper and the interview. First, scope: the Manifesto Project countries are overwhelmingly democracies, not adversaries, and they account for about two-thirds of U.S. foreign trade in the sample period, so this is a friction operating across the friendly part of the trading system, where no policy rationale exists. Second, networks: firms that experience rising distance trade with fewer foreign partners, and their sourcing becomes more concentrated, so partisan executives are narrowing supply chains in the name of risk management while making them more exposed to any single origin. Kempf’s other worry is about the receiving end — large U.S. firms can impose real damage on foreign suppliers, and possibly on local economies and future elections, on the basis of an election result the supplier had nothing to do with.

(Because the only recording is an interview, I cannot report David Yang’s discussion from the SI session or the NBER room’s questions. The obvious ones are whether the effect survives election-by-industry or election-by-state shocks — the paper says the more demanding specifications strengthen it — and whether reallocation toward pricier origins is a persistent loss or a temporary overreaction that unwinds as information accumulates.)