This issue summarises four important recent studies that use instrumental variables and natural experiments to address high-stakes empirical questions in labour economics, trade, and health policy.

1. Borusyak et al.(2022) — Quasi-experimental Shift-Share Research Designs

Citation: Borusyak, K., Hull, P., and Jaravel, X. (2022). Quasi-experimental shift-share research designs. Review of Economic Studies, 89(1), 181–213.

Research Question: When is the shift-share (Bartik) instrument valid, and what conditions must hold for consistent estimation?

Identification Strategy: The paper provides a new theoretical framework. The shift-share instrument \(Z_i = \sum_k s_{ik} g_k\) combines industry shares \(s_{ik}\) (local industry composition) and industry shocks \(g_k\) (national employment growth by industry). Prior literature (e.g., Goldsmith-Pinkham et al.(2020)) derived validity conditions based on the exogeneity of shares. Borusyak, Hull, and Jaravel instead derive validity from the exogeneity of shocks \(g_k\), showing that this leads to a different estimator and different standard errors (clustered at the industry level rather than the regional level).

Key Result: The "many shocks" interpretation: the shift-share estimator is consistent when there are many industries with bounded concentration, and the industry-level shocks are orthogonal to industry-level confounders (e.g., industry demand shocks unrelated to the policy or treatment of interest). Standard errors must account for correlation across regions sharing the same industries.

Takeaway: Applied researchers using Bartik instruments must decide whether validity rests on the exogeneity of shares or shocks, as the appropriate specification, standard errors, and overidentification tests differ substantially between the two interpretations. The BHJ framework, implemented in the ssaggregate Stata/R package, is now the standard reference.

2. Autor et al.(2013) — The China Syndrome: Local Labor Market Effects of Import Competition

Citation: Autor, D. H., Dorn, D., and Hanson, G. H. (2013). The China syndrome: Local labor market effects of import competition in the United States. American Economic Review, 103(6), 2121–2168.

Research Question: What is the causal effect of Chinese import competition on US local labour markets, including wages, employment, and public transfer receipt?

Identification Strategy: The key challenge is that US regions with higher exposure to Chinese imports may differ from low-exposure regions in ways that also affect employment. The authors use an IV strategy: Chinese import growth in other high-income countries (Australia, Denmark, Finland, Germany, Japan, New Zealand, Spain, Switzerland) instruments for Chinese import growth in the US, by industry. The logic: if China's global comparative advantage drives export growth to all developed countries, growth in non-US destinations reflects Chinese supply-side factors rather than US demand shocks.

Key Result: Regions more exposed to Chinese imports experience larger declines in manufacturing employment, higher unemployment, and increased government transfer payments. A $1,000-per-worker increase in import exposure reduces manufacturing employment by 0.6 workers per 1,000. Effects persist for over a decade and are not offset by employment growth in other sectors. The paper is credited with resurrecting empirical interest in trade and local labour markets.

Takeaway: Trade liberalisation with China created large, geographically concentrated, and persistent costs for US manufacturing workers — costs that had been invisible in aggregate employment and wage data. The shift-share IV design has since been used across dozens of contexts in international trade and labour economics.

3. Lee et al.(2022) — Valid \(t\)-ratio Inference for IV

Citation: Lee, D. S., McCrary, J., Moreira, M. J., and Porter, J. (2022). Valid \(t\)-ratio inference for IV. American Economic Review, 112(10), 3260–3290.

Research Question: The standard Wald/2SLS \(t\)-ratio is invalid when instruments are even slightly weak. How weak must instruments be before bias becomes problematic, and can we construct a simple, valid \(t\)-ratio-based test for IV regressions?

Identification Strategy: Methodological. The paper derives the exact (finite-sample) distribution of the 2SLS \(t\)-ratio and shows that it has correct size only when instruments are very strong. Even for first-stage \(F\)-statistics of 100 (far above the conventional rule of thumb of 10), the \(t\)-ratio can over-reject substantially when using two-sided tests at conventional significance levels. The authors propose a simple correction: inflate the standard error by a factor that depends on the first-stage \(F\), producing a valid \(t\)-ratio at any level of instrument strength.

Key Result: The corrected \(t\)-ratio, implemented as a one-line adjustment to standard 2SLS output, has correct coverage for all values of the first-stage \(F \geq 0\). Applied to a sample of published IV papers, the correction reverses statistical significance in a substantial fraction of cases where \(F < 100\).

Takeaway: Researchers should routinely apply the Lee et al.\ correction to IV estimates, especially when the first-stage \(F\) is below 100. The conventional 2SLS standard error is anti-conservative under weak instruments; the corrected standard error is substantially larger but correctly sized.

4. Nakamura and Steinsson(2018) — Identification in Macroeconomics

Citation: Nakamura, E. and Steinsson, J. (2018). Identification in macroeconomics. Journal of Economic Perspectives, 32(3), 59–86.

Research Question: Can causal methods developed in microeconomics — natural experiments, IV, DiD — be applied to macroeconomic questions? What are the unique challenges of identification in macroeconomics?

Identification Strategy: The paper is a methodological overview, but it uses the authors' own research as illustrations. Their estimate of the fiscal multiplier uses cross-sectional variation in military spending across US states as a natural experiment: states receive disproportionate military contracts during wartime (because of existing military infrastructure, congressional representation, and industrial capacity), creating quasi-random variation in government spending. The instrument is state-level military spending growth, purged of endogenous local demand effects by using the national defence spending cycle.

Key Result: The "open economy relative multiplier" — the effect of a dollar of government spending on local GDP — is approximately 1.5 in the short run. Whether this implies a similar aggregate multiplier depends on the degree of monetary offset and cross-regional spillovers. The paper reviews how similar natural experiment logic has been applied to monetary policy, international trade, and financial shocks.

Takeaway: The identification revolution that transformed microeconomics is beginning to reshape macroeconomics, but macro identification faces additional challenges: SUTVA violations (spillovers across regions), endogenous monetary policy, and the difficulty of randomising or exploiting exogenous shocks at an aggregate level. The Nakamura–Steinsson programme of "sufficient statistic" estimation provides a bridge between micro identification and macro theory.