Introduction to the symposium on reproducibility and replicability in economics: Part II
本文是经济学可重复性与可复制性研讨会第二部分的导言，总结了11项复制研究，涵盖移民创新、文化演变、贸易语言、阿片政策、投资者情绪、公共品合作、技术不平等、刑事司法、商品价格预测和气候经济等领域，强调复制研究对巩固经济学实证基础的作用。
The second part of the symposium on reproducibility and replicability in economics is a direct continuation of the first part, featuring further reproductions and replications of influential studies in the economics literature. This emphasis aims to rigorously scrutinize established findings within the economic literature and ultimately highlights the progress and the systemic challenges that persist in reproduction and replication efforts. The second part of the symposium includes 11 studies which span a diverse array of economic inquiries—including immigration and innovation, cultural evolution, trade and language, opioid policy, investor sentiment, public goods cooperation, technological inequality, criminal justice policy, commodity price forecasting, and historical climate-economy relationship—and at the same time, demonstrates how replication deepens our understanding of both micro- and macro-level phenomena while reinforcing the empirical foundations of economics as a self-correcting science. We provide a short summary of each article below. The first article, “Replication of ‘How Much Does Immigration Boost Innovation?’,” examines the impact of skilled immigration on innovation in the United States, measured by patenting activity. A concern in this line of work is the endogenous location choice of skilled immigrants, which is typically addressed using shift–share instruments Recent research has raised questions about the validity of such instruments—specifically, that the identifying variation comes from historical shares rather than from shifts in inflows—and has developed diagnostic tools to identify which groups drive the identification. Using reconstructed data and modern shift–share instrumental variable techniques, the author confirms that skilled immigrants significantly boost innovation. However, the analysis also shows that the identifying variation in the instrument is concentrated among a few origin countries, raising potential concerns about exogeneity. Despite this nuance, the positive relationship between skilled immigration and innovation remains robust across multiple econometric specifications. In the second article, “Understanding Cultural Persistence and Change: A Replication of Giuliano and Nunn (2021),” the authors revisit the link between ancestral climatic variability and individual-level cultural traits. The reproduction carefully examines the alignment between the paper's stated methods and its implementation, identifying several discrepancies in how samples were constructed and variables coded. After applying corrections to better reflect the original methodological descriptions, many of the main effects become weaker or statistically insignificant. The article titled “Trade and Ethnolinguistic Differences: A Replication and Extension” revisits the hypothesis that historical inter-ethnic trade opportunities influence linguistic similarities between neighboring ethnolinguistic groups. The authors refine the empirical strategy to better isolate inter-ethnic trade effects, distinguishing them from intra-ethnic interactions. Their replication confirms that trade plays a role in shaping linguistic convergence, though disentangling the precise mechanisms remains challenging. Next, the paper “A Closer Look at Doleac and Mukherjee (2022) and the Effects of Naloxone Access Laws on Opioid ER Admissions” critically re-evaluates the claim that expanding access to naloxone—a life-saving opioid overdose reversal drug—may unintentionally encourage riskier drug use. The author challenges the original study's conclusions by identifying flaws in their data sources, law timing, and statistical methods. Using a corrected triple-difference design, he finds no evidence that naloxone access increases emergency room admissions due to opioids. The paper “Re-examining Investor Sentiment and Stock Returns: A Replication and Extension of Baker & Wurgler (2006)” analyses how investor sentiment affects stock returns. The authors confirm the original findings using both the historical U.S. dataset (1963–2002) and a new dataset covering 2002–2023. They introduce a monthly sentiment measure and expand the analysis to the Chinese stock market. While the sentiment index continues to show predictive power in the U.S., its effects in China are inconsistent and often disappear when key market conditions—like valuation dispersion—are not met. Their findings highlight that the cross-sectional impact of sentiment is context-dependent and may shift or invert over time. The importance of dilemma-specific cooperation in public good environments where individuals' endowments are equal or unequal is explored in the paper titled “Reciprocity and the Tragedies of Maintaining and Providing the Commons: A Replication and an Extension to Income Inequality.” The authors replicate earlier findings using non-student samples and additionally consider scenarios where income inequality is introduced through heterogeneous endowments. Their findings confirm that cooperation is lower in maintenance compared to provision public good games, reinforcing previous conclusions in the literature when broader, more representative samples are used. The article “Mitigating Technology Gaps' Contribution to International Income Inequality” explores how cross-country differences in innovation efficiency contribute to global income disparities. The authors use broader patent metrics from PATSTAT, expand the country sample and time frame, and test alternative model calibrations. Their findings confirm that international technology gaps significantly influence wage and income inequality, but the strength of this relationship depends on how innovation is measured and which countries are included. The article “A Reappraisal of Real-time Forecasts of the Real Price of Oil” revisits a set of real-time forecasts for the real price of oil, replicating them using an updated dataset. The authors find that only futures-based forecasts consistently outperform the end-of-month no-change forecast, suggesting that previous conclusions no longer holds when their forecasts are compared to the end-of-month no-change forecasts. They also discuss refinements in the literature and the implications for other work using similar methods. Next, the paper “De-Prosecution and Death: A Comment on Hogan (2022)” critiques a study published in Criminology and Public Policy that claimed de-prosecution policies implemented by the Philadelphia District Attorney's Office led to an increase in homicides—about 75 additional homicides per year, equivalent to roughly 31% of the mean homicide count from the synthetic control. The authors first successfully reproduce Hogan's main results by independently reconstructing the synthetic control model based on his published description. They then assess a range of methodological choices, including the short pre-intervention period, the composition of the donor pool, the use of homicide counts rather than rates, limited covariate balance, and the use of alternative estimation methods such as the augmented synthetic control. They find that these alternative specifications reduce the estimated effects and, in some cases, render them statistically insignificant. The paper cautions against using the original study to inform criminal justice policy. The article “Gender Differences in Cooperation in Congress: Replicating Gagliarducci and Paserman (2022)” re-examines evidence on whether gender affects cooperative behavior among U.S. House Representatives. The replication corrects errors in the original data, adjusts the clustering of standard errors, and extends the analysis to 2011–2020. The results largely confirm the original conclusion that cooperation across party lines is primarily driven by ideological proximity rather than gender. However, the extension reveals a notable shift: in recent years, women legislators from both parties have tended to recruit more co-sponsors overall than their male colleagues, suggesting that female representation may now be associated with higher levels of legislative cooperation. Finally, the article “The Economic Effects of Long-Term Climate Change: Evidence from the Little Ice Age: Replication” finds a strong positive relationship between higher temperatures and economic growth during the Little Ice Age. The authors highlight important limitations of the original data and, consequently, of the main conclusions. In particular, cities with fewer than 1000 inhabitants—about half of all cities in the dataset—are coded as having a population of 500, even though their true populations lie somewhere between 1 and 999. While the original finding of a positive and statistically significant relationship between temperature and city size remains robust to a number of alternative treatments of these small-city observations, the replication uncovers a key heterogeneity: the positive relationship does not hold for larger cities. When the analysis excludes small cities (those below 1000 inhabitants), the relationship between temperature and city size becomes negative and significant. A couple of important lessons emerge from the two replication special issues. First, missing or incomplete data and documentation remain a major hurdle to reproducibility and replicability. Even when data and code archives accompany a paper, authors do not always provide the raw data or the code used to construct the processed datasets. The good news is that recognition of the importance of full transparency is growing, and more journals are implementing replication policies, appointing data editors, and investing resources to make replication feasible from the raw-data stage onward. We hope this trend continues—spreading to field journals as well—and that other outlets follow suit by encouraging replication studies, particularly of articles published in their own pages. Second, several replications revealed sensitivities in model specification or sample construction. These are not intended as “gotchas.” On the contrary, while they expose the limitations of a study, they often highlight the robustness of its core conclusions when tested under alternative approaches. In this sense, replications play a role analogous to post-marketing (Phase IV) studies in pharmaceuticals: once a result has been “released” into the literature, independent reanalysis helps monitor its robustness, generalizability, and potential side effects on inference. In this way, replications play an essential role in the production of economic knowledge. While refereeing improves research through dialog with authors and multiple robustness checks, independent replication extends that process by allowing other scholars to reengage with the data and code, uncover heterogeneity, and assess which modeling or data-cleaning choices matter most. Together, peer review and open replication form complementary stages of scientific progress. We close with a thank you to the authors, referees, and the journal's editor who made these replication issues possible. Their efforts demonstrate that replication has begun to take root as part of mainstream economic inquiry. Sustaining this momentum will require ongoing commitment—from journals, authors, and replicators alike—to ensure that transparency and reanalysis remain central to the discipline's scientific progress. Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.