Summarize Paper logo SummarizePaper.com

← Economics and finance this week · 26 October 2026 · 5 min read

Economics and finance this week: When better rules change the game

From AI-assisted sales to automated umpiring, this week’s preprints explore how technology, incentives and measurement shape economic outcomes—often in unexpected ways.

How to check the findings and keep reading? Illustrated guide

Better leads, fewer sales #

Giving workers more promising customers sounds like a straightforward productivity improvement. At a Chinese online health-insurance brokerage, however, deploying AI voice agents to handle low-intent leads was followed by worse customer acquisition among human workers precisely where their opportunities improved most. In more than half a million outreach records, the probability that workers’ first three calls reached a high-intent lead rose by about 23 percentage points, while customer acquisition fell by about 5 percentage points. Both effects faded over the workday.

The researchers’ explanation is that better prospects made workers less willing to persist with the customer already on the line. Workers reallocated early leads more often and shortened those calls by about 13%. The preprint uses a regression-discontinuity-in-time design around the rollout, rather than a randomized experiment. Its practical message is nonetheless important: reallocating tasks between people and AI can change workers’ expectations and effort, not merely improve the raw material they receive.

Cheaper AI does not guarantee cheaper prices #

A simulation of two competing sellers asks whether lower costs of running AI models translate into lower prices—and whether the answer depends on how quickly sellers see rivals’ prices or can revise their own. In the setting with identical firms, the freshest information and the most frequent price changes, the two sellers passed on 55.6% and 44.0% of an imposed cost reduction. Across all 16 experimental conditions, consumer surplus under the learned pricing policies was below the corresponding static Nash benchmark, a conventional model of non-cooperative competition.

The limits of the result matter as much as the headline. The estimated interaction between information delays and repricing frequency was close to zero, with uncertainty intervals spanning zero; related comparisons were also inconclusive. These are simulated sellers using tabular reinforcement learning, not observations of actual AI providers. The study shows that incomplete pass-through can arise in this environment, not that real firms are colluding.

An automated umpire changes who benefits #

South Korea’s top baseball league replaced human ball-strike judgments with an automated system in 2024. Using more than a million pitches from 2021 to 2026, researchers distinguish the removal of human discretion from the introduction of a differently shaped strike zone. Automation eliminated count-dependent calls, the benefit catchers gained by presenting pitches favorably, and a small advantage enjoyed by star batters. Those advantages had been worth only a few runs per player-season.

The new zone also redistributed strikes toward tall batters and across matchups. Yet relatively few pitches were exposed to these changes, and estimated effects on plate-appearance outcomes were close to zero. Pitchers adjusted quickly by throwing higher; the researchers found no evidence that batters adjusted. Beyond baseball, the preprint illustrates why replacing judgment with a rule is not simply a matter of removing bias: the rule’s design, who encounters it and how people adapt all determine its consequences.

Procurement rules leave fingerprints in the numbers #

Colombian public buyers can use faster, less competitive procedures below legal contract-value thresholds. A study of procurement records finds sharp concentrations of estimated values just below those cutoffs, including roughly 49,600 excess processes below the small-purchase threshold. Crucially, the spikes follow each entity’s legal threshold as it changes with the minimum wage, rather than staying attached to convenient round numbers. Territorial governments led by elected officials bunch more than appointed national agencies, even after adjustments for several differences in their purchasing.

Near the tender threshold, processes just below the cutoff attract about half as many bidders as those just above, secure smaller discounts and are more often enlarged after award. Timing rules also prompt adaptation: direct contracting falls by 95% during the four-month pre-presidential-election ban but rises by 209% in the preceding two weeks. Related evidence from Chile strengthens the case that buyers respond to the rules themselves. The policy dilemma is clear: thresholds that simplify small purchases also create incentives to keep contracts on the less competitive side of the line.

A simple ratio for a complicated inequality problem #

Could average wealth divided by mean salary offer a useful shorthand for inequality? Earlier agent-based simulations suggested that a higher wealth-to-salary ratio went together with more unequal wealth and income distributions. A new preprint tests that relationship using World Inequality Database and OECD data, finding positive associations across several empirical settings, with the strongest relationships in mature and advanced economies.

The attraction is simplicity: the ratio compares accumulated assets with earnings from work, making a broad economic relationship easier to track. But it is an indicator, not a demonstrated cause of inequality or a replacement for distributional measures. Differences in definitions and measurement across countries and periods constrain comparisons—a particularly important caution for a statistic built to summarize so much.

Solar potential is about timing, not just sunshine #

A global simulation of rooftop solar paired with electric vehicles used as batteries finds that these systems can supply up to 84% of annual building and mobility electricity demand in the settings studied. Across 5,000 grid cells, the key is not simply how much electricity panels generate, but how well generation matches climate-shaped demand and how EV storage can move electricity between hours. Self-sufficiency is highest where daytime demand aligns with solar output and seasonal variation is limited.

That distinction changes the map of opportunity. Screening locations using solar capacity factor alone undervalues 21.9% of the world’s 1,000 most populous cities, according to the study. Many low-latitude cities emerge as promising locations. These are modeled energy potentials, not proof of commercial viability, but they suggest that investment screening should examine the whole supply-demand-storage system rather than sunshine in isolation.

These six preprints, drawn from a week in which 143 papers were announced in the field, share a concern with what simple headline measures miss. Better sales leads need not mean more sales; cheaper computation need not deliver equivalent price cuts; more sunshine need not identify the best solar opportunity. Rules and technologies operate through timing, exposure and human responses. Understanding those connections is often more useful than assuming that an improvement in one part of a system will improve the whole.

Ask all the papers of this letter Save the papers in a collection RSS

Was this letter useful to you?

Written by AI from the 143 papers of the week in this field. Most are preprints, not yet peer reviewed: the text says what their authors report.

Get one letter by email every week with a free account, all of them with Premium. Create a free account