Economic research and policy comparisons

I start with primary sources and official data whenever the question permits it. An economic table should lead back to the agency workbook, statistical release, law, or historical text from which its figures and definitions came. Secondary sources can supply context or point to evidence, but they should not become an invisible authority standing between a reader and the record.

Every changing number keeps its data vintage visible. A page published in 2026 can still rely on 2023 or 2024 data because official series arrive and are revised on different schedules. Naming the year is more honest than calling a value current. It also makes a future correction auditable: I can identify which release changed, rebuild the comparison, and explain why the result moved.

I distinguish descriptive comparison from causal inference. A difference between groups of states does not by itself show that an election result, tax system, or party caused the difference. Population, industry mix, housing costs, migration, federal facilities, and many other factors move together. The chosen measure matters too. Total output, output per resident, purchasing power, household income, and consumption answer different questions, so changing the measure can legitimately change the answer.

Sources and caveats belong beside the claim rather than in a decorative bibliography. When a reader finds a broken link, a calculation error, or a better primary source, I correct the page and update its verification date. AI agents help me gather, compare, code, and test the material. I review what they produce, choose what gets published, and remain accountable for every claim under my name.