50 Years of Eviction Risk, 1976–2026
Every published eviction dataset in the country starts around the year 2000, because that is when court filing records become reliable enough to count. This report goes back to 1976. It can, because it measures something different: not how many evictions courts processed, but how favorable each state's legal machinery was to the landlord side in each year, reconstructed from the statutes in force. Session laws are dated, numbered, and archived, so the notice periods and procedural rules of 1976 are as verifiable today as those of 2026. The reconstruction rests on 490 dated legislative instruments, each carrying its session-law citation in the downloadable dataset.
Four findings from the series
1. Vermont moved further than any state in the country
Vermont entered the period at 1.89, among the most landlord-favorable legal environments in the nation, and now scores 5.17, a rise of 3.28 points on a 10-point scale. No state moved further. The District of Columbia rose almost as much (5.90 to 9.10) but started from the strictest baseline in the country, which makes Vermont the cleaner story: a rural New England state that rebuilt its landlord-tenant law from one era into another. Maryland, Rhode Island, and Colorado follow, each up roughly 2.8 points.
2. New Jersey was already the strictest state in 1976, and it barely moved
New Jersey opened the series at 6.12, far above every other state, driven by its 1974 Anti-Eviction Act, and closed at 6.36: a 50-year change of just 0.24. The lesson for operators is that the hardest legal environments are not always recent creations. Some states set their posture two generations ago and have simply maintained it.
3. The gap between states keeps widening
In 1976 the spread between the most and least landlord-favorable states was 4.61 points. Today it is 5.74. State choice now dominates almost every operating variable a landlord controls: the difference between owning in Tennessee (3.71 today) and owning in the District of Columbia (9.10) exceeds the practical difference between a model tenant and a difficult one in most markets.
4. Legislative attention is wildly uneven
Massachusetts has enacted 20 material changes to its eviction process since the milestone record begins, and New York 17. Several states show three or fewer in fifty years. Where the statute book is quiet, a landlord's real risk drivers are local court backlog and tenant finances, not pending legislation; where it is loud, underwriting must price the next session, not just the current code.
Every state, 1976 to 2026
Population-weighted score per state at three anchor years. Click a column to sort; click a state for its full profile.
| State | 1976 | 2000 | 2026 | 50-yr change | Cited milestones |
|---|---|---|---|---|---|
| Vermont | 1.89 | 4.65 | 5.17 | +3.28 | 9 |
| District of Columbia | 5.90 | 7.90 | 9.10 | +3.20 | 12 |
| Maryland | 2.71 | 4.14 | 5.52 | +2.81 | 8 |
| Rhode Island | 2.82 | 5.02 | 5.63 | +2.81 | 8 |
| Colorado | 2.24 | 3.46 | 5.02 | +2.78 | 11 |
| Oregon | 3.57 | 4.44 | 6.34 | +2.77 | 9 |
| Washington | 3.53 | 4.45 | 6.11 | +2.58 | 9 |
| Pennsylvania | 2.51 | 4.26 | 4.91 | +2.40 | 9 |
| North Carolina | 1.56 | 3.75 | 3.94 | +2.38 | 8 |
| New York | 5.90 | 7.04 | 8.28 | +2.38 | 17 |
| California | 5.25 | 5.96 | 7.58 | +2.33 | 13 |
| Alabama | 1.59 | 3.15 | 3.65 | +2.06 | 10 |
| Mississippi | 1.90 | 3.73 | 3.93 | +2.03 | 7 |
| Nevada | 2.52 | 3.61 | 4.52 | +2.00 | 8 |
| South Carolina | 1.79 | 3.59 | 3.77 | +1.98 | 9 |
| Idaho | 1.56 | 3.12 | 3.52 | +1.96 | 7 |
| Arkansas | 1.58 | 3.40 | 3.53 | +1.95 | 10 |
| Illinois | 3.20 | 4.36 | 5.11 | +1.91 | 13 |
| North Dakota | 1.51 | 3.04 | 3.36 | +1.85 | 7 |
| Texas | 1.81 | 3.46 | 3.63 | +1.82 | 12 |
| Virginia | 2.43 | 3.54 | 4.25 | +1.82 | 10 |
| Georgia | 2.35 | 3.64 | 4.13 | +1.78 | 11 |
| Nebraska | 1.85 | 3.06 | 3.56 | +1.71 | 9 |
| Wyoming | 1.78 | 3.03 | 3.41 | +1.63 | 8 |
| Florida | 2.31 | 3.65 | 3.93 | +1.62 | 10 |
| Louisiana | 2.58 | 3.90 | 4.19 | +1.61 | 9 |
| Utah | 1.88 | 3.04 | 3.47 | +1.59 | 8 |
| South Dakota | 1.87 | 3.03 | 3.40 | +1.53 | 8 |
| Montana | 2.18 | 3.32 | 3.66 | +1.48 | 9 |
| Minnesota | 3.10 | 3.96 | 4.53 | +1.43 | 9 |
| Indiana | 2.28 | 3.16 | 3.69 | +1.41 | 8 |
| Maine | 3.21 | 4.23 | 4.62 | +1.41 | 8 |
| West Virginia | 2.18 | 3.51 | 3.57 | +1.39 | 8 |
| New Hampshire | 2.78 | 3.85 | 4.15 | +1.37 | 7 |
| Oklahoma | 2.19 | 3.43 | 3.52 | +1.33 | 8 |
| Missouri | 2.37 | 3.45 | 3.67 | +1.30 | 7 |
| Kansas | 2.21 | 3.15 | 3.50 | +1.29 | 9 |
| Connecticut | 4.47 | 5.11 | 5.72 | +1.25 | 11 |
| Iowa | 2.38 | 3.39 | 3.60 | +1.22 | 9 |
| Massachusetts | 5.19 | 6.15 | 6.41 | +1.22 | 20 |
| Arizona | 2.35 | 3.18 | 3.55 | +1.20 | 10 |
| Ohio | 2.81 | 3.48 | 3.98 | +1.17 | 8 |
| Delaware | 3.06 | 3.64 | 4.16 | +1.10 | 10 |
| New Mexico | 3.17 | 3.84 | 4.14 | +0.97 | 8 |
| Hawaii | 4.95 | 5.48 | 5.88 | +0.93 | 9 |
| Kentucky | 2.77 | 3.60 | 3.67 | +0.90 | 13 |
| Michigan | 3.38 | 3.90 | 4.16 | +0.78 | 9 |
| Wisconsin | 3.08 | 3.71 | 3.86 | +0.78 | 9 |
| Tennessee | 3.02 | 3.64 | 3.71 | +0.69 | 11 |
| Alaska | 3.24 | 3.48 | 3.76 | +0.52 | 8 |
| New Jersey | 6.12 | 5.85 | 6.36 | +0.24 | 13 |
How the series is built
Each state-year value is the population-weighted average of the model's city scores for that state in that year, on the same 1.0–9.9 scale used across this site. Values from 2005 forward incorporate observed court and survey data. Earlier values are a reconstruction from state session laws and statute history: a rigorous reading of what each state's law required in each year, anchored by 490 dated legislative instruments whose citations ship in the milestones file. This is a reconstruction of the legal environment, not an estimate of court behavior. An estimate could be wrong by an unknowable amount; a reconstruction can be checked against the session law it cites, and the observed and reconstructed segments are labeled in every download.
Contested cases, jury demands, appeals, and court congestion push real-world outcomes beyond what any statute prescribes, in every era. That is exactly why the statutory baseline is the right instrument for comparing states across 50 years: it isolates what legislatures did from what dockets did. The full method is documented on the methodology page.
Cite this report
NextGen Properties Research Team (Kerstner, C., ed.). 50 Years of Eviction Risk by State, 1976–2026. EvictionRiskMap.com, 2026. https://evictionriskmap.com/50-years-of-eviction-risk/
The underlying data is free under CC BY 4.0: the state-year series and the cited milestones. A frozen copy of this year's release lives at /open-data/2026/ and is never revised, so citations to it stay reproducible.
Questions researchers ask
How can you have data from before 2000 when Eviction Lab starts there?
Eviction Lab counts court filings, and filing records before 2000 are too incomplete for national statistics. We measure the legal environment instead. A 1978 notice-period requirement is preserved verbatim in that state's session laws, so the legal posture can be reconstructed for any year and verified against the enacting statute. The two datasets answer different questions and complement each other where they overlap.
Are the pre-2005 numbers estimates?
They are reconstructions from primary legal sources, scored through the same model as the modern data. The milestones file lists the enacting instrument for every material change; any reader can pull the cited session law and check the direction and timing of each move. Treat reconstructed segments as legal-environment readings, not filing counts, and keep the segment label in any analysis.
Why does the score move in years with no cited milestone?
The model also carries economic and demographic inputs (ACS series and their historical antecedents), so scores drift between statutory anchor points. The milestones mark the discrete legal breaks; the series between them reflects the slower-moving conditions the model tracks.
Who produced this, and can I use it commercially?
The research team at NextGen Properties, a multifamily operator managing 750+ units across six states, founded in 2000 and led by Chris Kerstner. The data is CC BY 4.0: use it commercially, adapt it, republish it, with attribution and a link.