Weekly links July 23: a causal inference checklist, lessons from innovation RCTs, long-term impacts are hard to measure, and more...

· Scott Cunningham’s 12-step checklist for a (non-experimental) causal inference project – useful suggestions for understanding where the treatment variation is coming from, and for building into an AI pipeline. · IGL blog on lessons from 29 RCTs they have funded on innovation, start-ups, entrepreneurship and firms. These include “People can be nudged into acting as innovators,... “Rational quitting” can be a positive outcome....Matching the type of support to the needs of the individual or business can be crucial...” and crucially for doing these experiments “Entrepreneurs don’t always know what they need....” And “Benign-sounding interventions can have unintended negative effects.” So that “If there’s one key lesson to learn from across the portfolio of studies, it’s that we can’t predict the impacts of an intervention from theory alone. Rigorous testing of the impacts is crucial.” · Also on the IGL blog, some lessons/challenges from longer-term follow-ups of several business support programs they evaluated. “From huge diversity of participants, to a huge range of outcomes: after their involvement in the support programmes, businesses went on to widely diverging fates, from shutting down entirely to becoming superstar high-growth companies. This variation was much higher than expected from the interim data used for the original power calculations, highlighting how outcomes can diverge strongly over longer timescales (our primary outcomes were measured over the 4-5 years after programme participation). Why does this matter? First, it highlights the importance of measuring outcomes over longer timescales to capture the true potential, and pitfalls, of interventions. Second, there are powerful methodological implications. The more variable and unpredictable are outcomes, the lower the statistical power and higher the risk that we cannot identify a true difference between participants and comparison SMEs, or between experimental condition” · The World Bank Research Observer (WBRO) seeks to publish policy relevant surveys of development issues, aimed at a broad audience including non-specialists. Papers for consideration at the Fall 2026 meeting of the WBRO Editorial Board should be submitted online to https://academic.oup.com/wbro no later than Wednesday, September 30, 2026. · Conference call for papers: the new thinking in industrial policy conference at Columbia university on November 6, submissions due September 14. · Anthropic’s Economic Futures Research Fund has launched with $200 million to support ambitious interventions to prepare society for the economic impact of AI. “we want to fund the most ambitious proposals possible. We aim to fund large-scale RCTs or ambitious, creative pilots or program evaluations that expand our shared understanding of what shows promise and in which contexts, fill gaps where evidence is thin, and inspire new solutions...We plan to primarily fund projects in the $5-30 million range, though we’re flexible upward for well-scoped, high-potential-impact projects. Based on what we learned from the Economic Futures program, and the ambition and scale we’re seeking in proposals, we won’t directly fund anything below $1 million from this fund.” – see also an interesting essay by Peter McCrory, the Head of Economics at Anthropic on why AI hasn’t increased U.S. unemployment so far.