Gravity models excel at explaining international trade and investment flows, though the reasons for their success are somewhat mysterious. Using a comprehensive transactions-level dataset of cross-border transactions in high-value commercial real estate, we show that a significant portion of the explanatory power of a gravity model for transaction flows in this market can be attributed to preferential matching between counterparties (buyers and sellers) of the same nationality, regardless of the location country of the investment. These elevated rates of same-nationality counterparty matching are quantitatively large and robust, and increase in poorly-governed markets. We build an equilibrium search model of the market embedding this new matching friction, and structurally estimate it. We find large counterfactual gains from eliminating the friction in our model on both liquidity and prices in the market. We conclude that frictions between counterparties hailing from different countries, including trust, impede cross-border investment and can help explain the persistent success of gravity models. Joint work with Cristian Badarinza and Chihiro Shimizu.