Competing for CapitalThe Diffusion of Bilateral Investment Treaties, 1960–2000

Imagine trying to convince a stranger to trust you with their life savings. You could make promises, or you could sign a contract in front of a judge who can actually make you pay if you break it. Bilateral investment treaties, or BITs, are governments doing the second thing. Starting in the late 1950s, countries began pairing up to set the ground rules for how foreign investors could operate on each other's soil and what would happen if those rules were broken. The punchline wasn't the promises themselves; it was who enforced them. These treaties lifted investor protections out of the murky world of domestic law and customary practice and strapped them to international arbitration, giving private investors a direct right to sue and the power to collect damages from a host state. What sits inside a typical BIT? The essentials are remarkably standard. They promise national treatment and most-favored-nation treatment. They protect contracts and let profits leave in hard currency. They limit performance requirements. And, crucially, they send disputes to international arbitration where an investor can bring a case even if their home government stays silent. That mix does two things at once. It clarifies what a host state owes and raises the ex post cost of reneging because an international tribunal can put a price on a breach, and reputations travel fast. Now, zoom out. This wasn't a niche experiment; it went global. As Elkins, Guzman, and Simmons recount, more than one hundred seventy-eight countries have signed at least one BIT, and by the 1990s, the world was adding well over one hundred new treaties a year. The United States jumped in the mid-1980s, long after European states had built the template. Over the same period, foreign direct investment exploded as a share of world output, rising from about one point two percent in 1970 to roughly eight point nine percent by 2000. Treaties and capital were moving together, and by the end of the century, BITs had become the dominant way governments tried to make themselves legible and safe to global investors. Here's the twist. This wave didn't spread because everyone embraced the same investment theology. It spread because countries were competing. In their account, BITs are tools in a tournament. They make a host's promises more believable, which can raise expected returns for investors. They also threaten costly arbitration if a government backslides, which deters opportunism. And they give officials a way to say, to a very skeptical market, "We're serious—on terms you recognize." In a world where investors can choose among dozens of plausible sites for a factory, that's not a small thing. If that's really what's going on, you should see a few patterns. Rivals should copy rivals, not just friends copy friends. Manufacturing-heavy economies, where there are many substitute locations, should scramble harder than extractive economies, where the oil is where the oil is. BITs should be signed more when the global pool of capital is swelling. And, maybe, countries with shaky domestic institutions should be more eager to borrow credibility from abroad. The last one is the most intuitive; it's also the most slippery. To sort this out, Elkins, Guzman, and Simmons build an event-history design that looks like a stopwatch on every potential treaty pair in every year. The unit is a dyad-year: one potential host and one potential home each year from 1958, just before the first BIT, to January 1, 2000. They assign the roles by relative income, calling the richer partner the home and the poorer the host. And they deliberately drop dyads where both are rich—defined as above six thousand dollars in 1987 dollars—to focus on pairs where investor protection is more likely to be at issue. That choice removes about one hundred twenty-five treaties from the estimation sample and keeps the lens on the most relevant contest. The core statistical engine is a Cox proportional hazards model. Think of it as asking, in any given dyad-year, how much various factors raise or lower the chance that a BIT appears. A hazard ratio above one means higher odds of signing; below one means lower odds. The clever part is how they capture competition. They build "spatial lags" that measure what a host's rivals did last year, weighted by how directly those rivals compete. One set of weights is based on selling into the same export markets. Another is based on exporting similar products. A third leans on similarities in workforce and infrastructure—educational attainment and physical capacity that matter to investors. They also create cultural networks—shared language, colonizer, and religion—to test whether treaties diffuse through emulation instead of rivalry. And they add two explicit channels: learning, measured from whether BITs seemed to boost a country's recent foreign investment inflows, and coercion, captured by whether the host was on International Monetary Fund, or IMF, credit that year. Before we dive into numbers, one more piece of scaffolding. They control extensively for the things you'd expect might also move BITs: host and home size and income, growth, illiteracy, openness, law and order, democracy; dyad-level ties like trade, common language, colonial history, and alliances; even a Cold War indicator and a global count of treaties by year. And to dial back the worry that competition today is reacting to treaties already signed yesterday, they lag the rivals' activity by one year. So what happens when you press "go"? The rivals' moves matter. Across three different competition networks, when a host's competitors sign more BITs, that host becomes more likely to sign one the next year. The effect is consistently positive and statistically sharp. Among the three ways of measuring rivalry, the signal is strongest when rivals are defined by exporting similar products, which is just where you'd expect manufacturing competition to bite. That's the signature of competitive diffusion: not simply that treaties are spreading, but that they're spreading most where hosts are substitutable. Now put the macro lens back on. When the global pool of foreign investment gets bigger, BITs become more likely. In three versions of the model, the hazard ratios on average world foreign direct investment are one point thirty-two, one point fifty-three, and one point forty-six. That's social science for "as the tide rises, more boats launch." It's also a nice reality check on the tournament logic: governments sprint hardest for treaties when there's more capital to catch. Sector matters, too. Places that lean on extractive exports, such as fuels, ores, and metals, are less likely to sign. The hazard ratios for extractive dependence sit around zero point seventy-two to zero point seventy-three, which means a material dampening of signing propensity. That squares with the intuition that when your comparative advantage is a hole in the ground, you don't need a treaty to persuade investors to come. In manufacturing, marginal credibility can matter a lot more. In the paper's comparisons, the difference in signing rates between more extractive and more manufacturing-oriented states can amount to about ten percentage points. That's not noise; it's policy-relevant. What about the domestic credibility story? Here the evidence is mixed. Some signals of orderliness help. Where perceptions of law and order are better, the hazard nudges up—around one point thirty-eight to one point thirty-nine when that measure is in the model. Democracy doesn't do much once you control for other features. And a common-law heritage shows positive associations in some versions, though it's not the main act. The broader point is that weak domestic institutions do not reliably predict treaty signing; governments with stronger rule-of-law signals also sign, perhaps because the legalistic framing of BITs complements institutions they already value. Dyadic ties cut both ways. A shared language between host and home makes a BIT more likely, with hazard ratios in the one point fifty-four to one point fifty-seven range. Investors and officials speaking the same tongue may just find it easier to do a deal. By contrast, a shared colonial heritage lowers the hazard, around zero point forty to zero point forty-one. That's a striking asymmetry: historical intimacy isn't acting like trust here; it looks more like wariness or an incentive to avoid locking in old patterns under new legal clothes. Two non-competitive channels also leave marks. Coercion matters. When a host is drawing on IMF credits, the likelihood of signing jumps, with hazard ratios in the one point thirty-nine to one point forty-four neighborhood. You can read that as bargaining power: when you need money, you accept conditions, and BIT-like reforms pair naturally with IMF programs. Learning matters, too. When a country's recent experience suggests that more BITs were followed by more foreign investment, the hazard rises—roughly one point eighty-three to two point thirteen across models. That's a behavioral twist on the same story: governments update from their neighbors' payoffs. Emulation through cultural networks, on the other hand, is weak. Sharing a religion, a colonizer, or a language with other signers doesn't, on its own, move much. There's a second, more descriptive piece of evidence pointing to competition: the fingerprints of who drives the action. When the authors map BIT signings over time, they find that host governments sign in bursts—programmatic flurries—while home countries show a more even hum. The contrast shows up in the shape of the distributions. The average kurtosis, a measure of how peaky a distribution is, is nine point eleven for hosts and four point forty-eight for homes. The standard deviation over time is lower for hosts, about seven point zero eight compared to nine point thirty-nine for homes. Translation: hosts bunch their BITs into campaigns; homes keep a steadier pace. That looks like hosts deciding, consciously, to run a play to compete for capital. And when you read the treaties themselves, another pattern pops. Core provisions—mandatory international arbitration, a private right of action for investors, monetary compensation for violations, national treatment, and most-favored-nation treatment—are almost uniform, even across different home countries. That uniformity tells you something about bargaining power. Home countries show up with a model and market power; hosts accept the terms to stay in the race. In a sense, hosts are price-takers in governance content even as they act like price-makers in timing. Stepping back, the big picture is a trade: sovereignty for credibility. By signing, a host delegates adjudicative authority and narrows its policy room—especially in crises—because violating a treaty can trigger not just arbitration awards but diplomatic and reputational costs. As Elkins, Guzman, and Simmons see it, the gamble is that these constraints buy enough capital, on good enough terms, to make the sacrifice worthwhile. BITs don't create investment out of thin air; they change its risk profile. What you shouldn't hear in these results is a verdict that treaties are a development engine. The models show credible commitments and competitive diffusion. They don't show that BITs raise growth. The authors are explicit about that distinction. BITs raise ex post costs for host governments and provide orderly dispute resolution for investors. Whether that translates into sustained, broad-based development is a separate question, and the evidence here stays agnostic. If you're a policymaker, two sobering lessons emerge. First, the tournament is real. When rivals sign, your odds of signing go up, and when the global pool of investment swells, the pressure to move intensifies. Second, the terms you'll sign are likely to look just like everyone else's. That uniformity makes it easier for investors to compare jurisdictions. It also means your leverage is mostly in whether and when you sign, not in rewriting the rules. Where does that leave us? With a cleaner understanding of the politics of credible commitment. BITs spread fastest where the payoff from signaling is biggest—manufacturing-heavy, capital-seeking economies facing many close substitutes. They're reinforced by moments of financial need and by learning from neighbors' apparent gains. And they're written in a language investors already trust, often more the language of home countries than of hosts. The next hard task is causal. If treaties are primarily a way to win a share of a growing pie, which countries actually gain, and which are just keeping up? Disentangling selection—who signs when—from payoff—who gets more and better investment because they signed—needs designs that go beyond diffusion. Natural experiments around sudden home-country template shifts or arbitral shocks could help. Until then, this study gives us the map of the race and the rules of the game. It doesn't pick the winners.