Nationally and in fast-building metros, yes — but a straight cross-county comparison over 2013-2023 shows the opposite, because builders chase demand

Summary

The published research says yes, decisively, when the comparison is made carefully: metros and ZIP codes that added the most housing supply relative to their own baseline saw meaningfully slower rent growth, and the clearest recent case studies (Austin, Minneapolis) show outright rent declines following supply surges. The National Multifamily Housing Council found that between 2015 and 2024, a 10% increase in a metro's housing supply was associated with about 5% less rent growth over that period, and Pew's 2025 ZIP-code study found a similar dose-response pattern concentrated in older, lower-cost units.

But when we ran the equivalent test ourselves in AskAmerica's own government data — 574 U.S. counties with 100,000+ residents, cumulative Census Building Permits per capita for 2014-2023 against the change in ACS median gross rent from 2013 to 2023 — we got the OPPOSITE raw answer: counties that built the most housing per capita saw rents rise the MOST (quintile means of 36% up to 55%, r=0.45, p<0.001). The reason is a textbook confound, not a contradiction of the literature: fast-building counties are overwhelmingly fast-GROWING counties (Austin's Travis County, Phoenix's Maricopa County, and similar Sun Belt metros), and population growth alone predicts most of both the building and the rent growth. Once we hold population growth fixed, the partial correlation between building intensity and rent growth collapses to essentially zero (r=0.02, p=0.62), and building MORE than population growth alone would require shows no robust negative effect on rent growth either (coefficient positive but not significant, and leave-one-out testing shows the significance is not stable across states).

Net answer: the causal claim from the literature — build more, relative to your own demand, and rent growth slows — is real and well-supported by within-metro, high-frequency data. A simple decade-long, cross-county comparison of who-built-most cannot detect it, because raw building volume is dominated by demand growth, not a proxy for it. This is exactly the endogeneity the more careful published studies control for and our simple version does not.

What the published literature finds

• NMHC / RealPage analysis (2015-2024, metro grain): markets with higher supply growth recorded lower rates of rent growth, especially for lower-tier (1- and 2-Star) apartments; a 10% increase in metro housing supply correlated with about 5% less rent growth and roughly $470/year in savings per renter.
• Pew Charitable Trusts (2025, ~1,600 ZIP codes, 2017-2024): a 10% increase in a ZIP code's housing supply was associated with rents growing about 1.4% less than in ZIP codes with no new supply; the effect was largest for older, Class C apartments that lower-income renters occupy, and 11 large metros that added 10%+ housing 2017-2023 saw outright declines in older-apartment rents 2023-2024.
• Austin case study: Austin issued the most multifamily permits per capita nationally (roughly 957 apartments per 100,000 residents, 2021-2023) and median asking rents fell about 16% from December 2021 to January 2026; large (50+ unit) buildings saw a 7% rent decline in 2023-2024 alone, the steepest of any large U.S. metro.
• Minneapolis case study (Minneapolis Fed): following the 2019 Minneapolis 2040 zoning reform and a roughly 12% increase in housing supply 2017-2022, the metro recorded an 11% year-over-year rent decrease; permitting has since fallen sharply (down ~92% from the 2019 peak by 2024), a reminder that the relationship runs on the supply pipeline, not on a permanent structural condition.
• Zillow Research (national, 2024-2026): a nationwide apartment-construction boom pushed vacancy higher and slowed multifamily rent growth to roughly 1.9% year-over-year by early 2026, down from ~16% at the 2022 peak — with a caution that multifamily permits are down about 31% from their 2022 peak, so the supply-driven relief may not persist.

The common thread across all of these: the strongest, cleanest evidence comes from WITHIN a metro or ZIP-code panel (comparing a place to its own trend, or to peer places after controlling for demand shocks), or from recent high-frequency data (Zillow Observed Rent Index, monthly) capturing the 2023-2025 supply-driven correction — not from a simple decade-apart, place-vs-place level comparison.

What we computed directly: a decade-apart county cross-section

Method: using housing.building_permits (Census Building Permits Survey, county grain) summed 2014-2023, and census.acs_housing_tenure (ACS 5-year median gross rent) at the 2013 and 2023 vintages, we built a panel of 574 U.S. counties with 2013 population of 100,000+ (census.acs_population). We computed permits per 1,000 2013 residents as the supply measure and percent change in median gross rent 2013-to-2023 as the outcome.

Raw relationship: a quantile-binning dose-response test (5 bins) shows a monotonically INCREASING relationship — mean rent growth rises from 36.1% in the lowest-building quintile to 54.9% in the highest-building quintile (trend p=0.0008). An OLS regression controlling for 2013 rent level (mean reversion) and log county population still finds a positive, highly significant coefficient on permits per capita (+0.13 percentage points of rent growth per additional permit per 1,000 residents, p<0.001).

Controlling for the obvious confound — population growth: once we add 2013-2023 population growth (the direct proxy for demand growth) to the regression, the coefficient on permits per capita drops by roughly 75% (from 0.134 to 0.033) and loses significance (p=0.08). A partial correlation test confirms this directly: the raw correlation between building intensity and rent growth is r=0.45 (p<0.001), but the correlation net of population growth is r=0.02 (p=0.62) — population growth explains essentially all of the raw relationship.

Testing 'excess' supply directly: we also constructed a measure of supply relative to demand — permits per capita in excess of what population growth alone implied — and regressed rent growth on it, controlling for 2013 rent level, county size, and population growth. The coefficient is positive (+0.33), not statistically significant at conventional levels (p=0.08), and a leave-one-out sensitivity test across the 49 states represented shows the significance is NOT robust: dropping California or Texas flips it across the p=0.05 threshold in either direction. In other words, this county cross-section provides no reliable evidence, in either direction, that building beyond what population growth required moved rent growth over this decade.

Why the two results differ, and which to trust for what

This is not a contradiction so much as a difference in what's being measured. The published city/ZIP-level studies isolate the causal supply effect by comparing similar places (within a metro, or before/after a specific zoning change or permit surge) and by using recent monthly data that captures the sharp 2023-2025 correction directly. Our county cross-section instead compares different places to each other over a full decade using smoothed 5-year ACS estimates (the 2023 vintage averages 2019-2023, which mutes the sharpest recent declines in places like Austin) and does not isolate supply from demand — a fast-growing county builds more AND has more rent pressure from new residents, for reasons unrelated to its housing policy.

Read together: the literature's answer — the more careful, demand-controlled comparison — is that yes, building more housing (relative to a place's own demand trajectory) predictably slows rent growth, with real, sizeable, and recently very visible effects in the most aggressive builders (Austin, Minneapolis). Our own county-level check confirms why that framing matters: a naive 'who built the most, did their rents rise less' comparison across places gives the wrong sign, because it is dominated by the demand growth that drives both variables together. A reader should trust the demand-controlled, within-place literature findings over a raw cross-place ranking.