Economists can now fix incomplete network data without guessing what's missing

Social scientists often study networks (who knows whom, who trades with whom) but can't observe everyone. A new method fills in the missing connections using math instead of assumptions. This means researchers can now analyze partial networks and get the same reliable answers they would from complete data.

For decades, studying incomplete networks meant either collecting expensive full data or accepting that your answers were probably wrong. This method solves that trade-off by using observed connections and covariates to impute missing links without assuming the network has special mathematical properties (like being low-rank, a common shortcut that doesn't fit reality). The result: economists studying microfinance, trade networks, social connections, or any domain where you can't observe everyone can now work with cheaper, partial data and trust their estimates. That matters because it lowers the cost of network research across fields—development economics, organizational behavior, epidemiology—where collecting complete information is either impossible or prohibitively expensive.