Five Arguments, One Worldview
Noah Smith's seventy-eighth roundup post covers five seemingly unrelated topics -- AI productivity, housing construction, global poverty, American wealth, and immigration economics -- but a single thread runs through all of them. In each case, Smith marshals evidence to argue that markets work, growth is real, and the pessimists are wrong. It is a potent distillation of what might be called roboliberalism: the conviction that technology, trade, and construction will, on balance, make most people better off.
The roundup format suits Smith well. Rather than belaboring a single thesis across thousands of words, he delivers quick punches backed by recent research and leaves readers to draw the connections.
The AI Productivity Puzzle
Smith opens with the hottest question in economics right now: whether artificial intelligence is already boosting aggregate productivity. Erik Brynjolfsson claims the answer is yes, pointing to revised 2025 payroll data showing fewer jobs alongside robust GDP growth. Smith quotes Brynjolfsson's striking claim:
This decoupling -- maintaining high output with significantly lower labour input -- is the hallmark of productivity growth...My own updated analysis suggests a US productivity increase of roughly 2.7 per cent for 2025. This is a near doubling from the sluggish 1.4 per cent annual average that characterised the past decade.
But Smith does not simply accept this. He immediately introduces Martha Gimbel of the Yale Budget Lab, who urges caution. The data is noisy. The GDP figures have not yet been revised. Immigration policy changes could explain the job shortfall without invoking AI at all.
Smith then pivots to micro-level evidence, citing Alex Imas's survey of studies showing individual productivity gains from AI that have not yet surfaced in aggregate statistics:
At the macro level, these gains have not yet convincingly shown up in aggregate productivity statistics. While some studies show a slow down in hiring for AI-exposed jobs -- which suggests that individual workers are either becoming more productive or tasks are being automated -- the extent and timing of these dynamics are currently being debated.
A European study by Aldasoro and colleagues finds something even more counterintuitive. AI adoption appears to increase employment at the firms that adopt it, not reduce it:
We find no evidence that AI reduces employment in the short run...AI augments worker output -- enabling employees to complete tasks faster and make better decisions -- without displacing labour.
Smith's conclusion is admirably restrained: nobody really knows yet. That intellectual honesty is welcome in a debate where both AI boosters and doomers tend to cherry-pick. Still, by leading with the optimistic Brynjolfsson data and closing with the pro-employment European findings, Smith tilts the framing toward the bullish case. A more balanced presentation might have given equal weight to the survey finding that over 80 percent of firms report no AI impact on either employment or productivity over the past three years -- a figure that cuts against both the techno-optimists and the job-loss alarmists.
Yuppie Fishtanks and the Housing Supply Evidence
The second entry is Smith at his most triumphant. Years ago he coined the term "Yuppie Fishtank" to describe how building high-end housing absorbs wealthy renters who would otherwise bid up prices in older, cheaper neighborhoods. New research from Honolulu proves the mechanism works exactly as predicted.
What the researchers found was that the new housing freed up older, cheaper apartments, which, in turn, became occupied by people leaving behind still-cheaper homes elsewhere in the city, and so on...The paper estimates the tower's 512 units created at least 557 vacancies across the city.
Bloomberg reporting from Austin and Denver confirms the pattern at city scale:
Rents got cheaper in several major cities this past year, thanks to an influx of luxury apartment buildings opening their doors and luring tenants to vacate their old homes...Rents for older units have fallen as much as 11%, and some are now on offer at rates as low as homes that are usually designated as "affordable."
The evidence here is genuinely strong, and Smith is right to celebrate it. The YIMBY position has accumulated an impressive body of research over the past decade. That said, housing filtering takes time and works best in cities with sustained building -- a condition that zoning politics can easily undermine. The Honolulu study traces a single tower's impact; whether these chains persist across economic cycles and tighter credit conditions remains an open question.
Global Poverty: An Optimist Catches a Pessimist Retreating
Smith takes visible pleasure in pointing out that David Oks, who argued in 2023 that global development since 1990 was a failure, has now implicitly conceded the opposite. Oks's new post focuses on rising poverty projections for Africa alone -- which means he is acknowledging that poverty reduction elsewhere was so successful it went to completion.
Smith frames this as a significant intellectual retreat. The concern about Africa is real: poverty rates there are not declining, and rapid population growth could push the global headcount of extremely poor people upward. But Smith points to possible upside surprises from solar power, AI, and falling fertility rates. His parting shot captures the mood of the entire roundup: "Our goal should be to keep the pessimists embarrassed."
America's Wealth Is Not a Statistical Illusion
The fourth section tackles the persistent disbelief among foreigners that America is as rich as the GDP numbers suggest. Smith cites a remarkable Globe and Mail investigation in which Canadian journalists traveled to Alabama -- supposedly one of America's poorest states -- and found a booming economy.
Alabama is also home to five million people...and its economy is booming. The state's unemployment rate is now just 2.7 per cent, versus 6.5 per cent in Canada, and its major employers include Airbus SE and giant defence contractor Northrop Grumman Corp. The state has also morphed into an auto manufacturing powerhouse.
Smith acknowledges Alabama's higher poverty rate and lower life expectancy compared to Canada, though he notes the Globe and Mail glosses over crime statistics. His core point stands: per capita GDP in Canada is lower than in Alabama, and the gap in median incomes is roughly 18 percent and growing.
The section is effective but arguably too quick to dismiss the non-GDP factors. Health outcomes, incarceration rates, and inequality-adjusted measures tell a more complicated story about material well-being. GDP per capita is a powerful metric, but it is not the only one that matters to the people living in these places.
The Borjas Pattern
Smith closes with a detailed prosecution of George Borjas, the Harvard economist whose immigration research consistently finds negative effects that other researchers cannot replicate. Drawing on his own Bloomberg columns from the mid-2010s, Smith recounts how Borjas's Mariel Boatlift study used a sample of only 17 to 25 workers and excluded groups most likely to be affected by immigration:
Borjas ignores young workers and non-Cuban Hispanics -- two groups of workers who should have been among the most affected by competition from the Mariel immigrants. When these workers are added back in, the negative impact that Borjas finds disappears.
A new study by Jiaxin He and Adam Ozimek finds similar problems in Borjas's recent work on H-1B visa holders. The most damaging error is a temporal mismatch: comparing H-1B wages from 2020-2023 with native-born wages from 2023 only, without adjusting for 15 percent inflation in the intervening years. When corrected, the wage gap shrinks by roughly half.
Nowhere did the paper mention controlling for inflation or wage growth. Given 15.1 percent inflation and an 18.7 percent wage increase for software occupations alone from 2020 to 2023, comparing wages of H-1B workers from 2020 to 2023 to native-born wages from 2023 only produces negatively biased results.
Smith's case against Borjas is damning in its accumulation. One questionable methodological choice could be an honest mistake. A pattern of choices that all push in the same anti-immigration direction, across multiple papers and multiple decades, is harder to explain away.
Bottom Line
This roundup is Noah Smith at his most characteristic: data-forward, relentlessly optimistic about markets and growth, and impatient with what he sees as fashionable pessimism. The five topics are well-chosen to reinforce each other. AI might boost productivity. Building housing lowers rents. Development works. America is rich. Immigration helps. The common denominator is that letting markets operate -- with appropriate nudges -- tends to produce good outcomes.
The evidence Smith assembles is genuinely strong in most cases, particularly on housing supply and the Borjas critique. Where the post is weakest is in its tendency to present ambiguous data through an optimistic lens. The AI productivity section, for instance, honestly acknowledges uncertainty but still leaves readers with a bullish impression. The America-is-rich section could have engaged more seriously with non-GDP measures of well-being.
But these are quibbles with a roundup that does exactly what it sets out to do: survey the latest evidence on five major economic questions and find that, more often than not, the liberal growth-oriented view holds up. Readers looking for a clear-eyed counterweight to economic doom-mongering will find this post useful. Those looking for a more nuanced treatment of the losers in each of these stories will need to look elsewhere.