Findings · one level down
What the data says about Ohio
Cracked Buckeye holds one of the deepest piles of Ohio records anywhere: census counts back to 1810, every fatal crash since 1975, 3.9 million home records, a million building permits, every water system, county weather since 1895, and much more. We asked it questions any Ohioan might ask, and kept only the answers that held up when we tried to knock them down.
Facts wherever they land. Every number below was computed from records we already hold, then re-derived by a second, independent analysis. Where we were wrong, we say so on the page.
These are patterns across counties and towns, not facts about any one person who lives in one.
The short version
- 75 of 88 counties have more deaths than births, for two different reasons: age on the lakeshore and in the old industrial east, early death in the south.
- At the same income, southern Ohioans die younger, and lung disease plus overdose account for the whole gap.
- Franklin County is Ohio's nursery and port of entry; the biggest counties now grow only because of immigrants.
- Neighborhoods a few miles apart in Ohio's cities differ by up to 20 years of life.
- Ohio loses two-thirds fewer households to other states than in 2005 but only a fifth less income; Florida takes most of it.
- Ohio stopped building the starter home; since 2012 it rose faster than any other house, and a company owns about one in nine.
- A company-owned house gets less than half the roof, furnace and plumbing permits of the owner-occupied house next door.
- An Ohio roof costs about half again what it did in 2016 after inflation; the dearest homes get five times the permitted dollars.
- Permitted work comes back at the sale at about 90 cents on the dollar.
- The typical Ohio home did not beat inflation from 2000 to 2020; when you bought mattered far more than where.
- The new-business boom is one-person ventures: filings up 77% since 2019, those planning to hire up 7%; half the growth is driving.
- One thin-paycheck map: the counties that lean on benefits also pay the dearest power and the heaviest water bills.
- Gas bills halved since 1984, so a $54 fixed charge now carries most of a small home's gas bill.
- Ohio produces 20% more per job than in 2001 but pays 13% more; in the gas counties output tripled while paychecks stood still.
- Ohio's warmer winters show up in motorcycle deaths: winter rider deaths are up nearly 30% since the late 1970s.
- A six-inch month is when Ohio floods: 7% of months, 47% of flood-insurance claims; counties now get 56% more of them.
- Every bit of the fall in road deaths since the 1970s was people under 25.
- Where your water is drawn decides what is in it; wet seasons raise byproducts at lake and reservoir intakes, not rivers or wells.
Part
People: births, deaths, moving and health
Who is born, who dies young, who moves, and why the map looks the way it does.
Ohio's counties are shrinking for two different reasons, and in the south the problem is deaths, not births
75 of Ohio's 88 counties had more deaths than births in 2021-25. On the lakeshore and in the old industrial east that is mostly age: there are simply many older people. Along the southern tier it is early death. Lawrence County buries 17.0 people per 1,000 a year where its age mix predicts 12.4, about 267 extra deaths a year; Scioto 16.8 against 12.3, about 333. And the south is not short of babies: 12 of the 19 "dying young" counties have above-median birth rates.
Deaths per 1,000 residents a year beyond what each county's age mix predicts, 2021-25: the ten highest and the five lowest counties
Why it happens. Two mechanisms with two maps. Decades of slow growth left an old population in places like Ottawa, Erie, Lake and Belmont, so deaths outrun births even at ordinary death rates. In the southern tier people die before 75 at rates no age mix explains.
How we checked. If natural decrease were one story, the oldest counties and the counties that die youngest would be the same counties. They are not: among the 75 decrease counties, the share aged 65+ has no link with years of life lost (rank r -0.02), and the ten oldest and the ten that die youngest share one name, Jefferson. Age explains only about a third of county death rates; what is left lines up with years of life lost before 75 (rank r 0.83).
Your county. Most deaths beyond the age mix: Lawrence, Scioto, Pike, Fayette, Morgan, Crawford, Hardin, Gallia, Adams. Fewest: Delaware (-3.9 per 1,000), Geauga, Medina, Union, Morrow. More births than deaths: Holmes (+10.0 per 1,000 a year), Franklin (+4.6), Mercer, Union, Delaware.
What it does not show. The age model is fitted to Ohio's own counties, so "expected" already includes Ohio's high mortality. Prison counties (Noble) fool it: inmates are young adults who rarely die.
Sources: Census population estimates (Vintage 2025) births and deaths; ACS 2020-24 ages; County Health Rankings 2025.
At the same income, southern Ohioans die younger, and lung disease plus overdose account for all of the gap
Years of life lost before age 75 run from 19,460 per 100,000 residents in Vinton County and 16,720 in Scioto to 4,329 in Delaware. Even comparing counties with the same income, the farther south, the more years lost (moderate link). Hold the cancer registry's lung-cancer rate fixed and most of the north-south gap disappears; hold overdose deaths fixed too and none is left.
How strongly being farther south predicts years of life lost before 75 (rank partial correlation, 88 counties), as each explanation is held fixed
Correction. Our first pass headlined "Ohioans live about nine months longer for every 69 miles farther north". The slope is real, but by the rank test this page uses the link with life expectancy is only weak to moderate (0.36), and it is fully accounted for by lung disease and overdose. We report it on years of life lost instead.
Why it happens. Two causes with different weights. Overdose dominates years lost, because each overdose death takes 30 to 40 years of life. Smoking-era disease dominates the count of deaths. Even with every overdose death removed, 16 of the 19 dying-young counties would still bury more people than their age mix predicts.
How we checked. Lung-cancer incidence comes from the Ohio cancer registry, not from death certificates, so it is an independent record of smoking history (its statewide rate, 63.5 per 100,000, matches the published 63.6). It is a gradient, not a border: a flag for the 14 Ohio River counties adds nothing once latitude is known, and the "Appalachian" label adds nothing either.
Your county. Lung-cancer incidence (2018-22): highest Lawrence 98.8 per 100,000, Pike 98.3, Vinton 94.9, Ross 92.7, Adams 90.0; lowest Holmes 38.1, Putnam 43.1, Delaware 44.6, Geauga 45.5. Overdose deaths 2020-22: Scioto 126 per 100,000, Vinton 115, Meigs 90, Lawrence 82.
What it does not show. Lung incidence marks a smoking history, not lung cancer alone. Vinton has about 12,800 people, so its figures carry wide ranges.
Sources: County Health Rankings 2025 (years of potential life lost, overdose); State Cancer Profiles 2018-22; ACS income.
Franklin County is Ohio's nursery and its port of entry, and Ohio's biggest counties now grow only because of immigrants
From mid-2020 to mid-2025 Ohio gained 101,976 people: 44,281 more deaths than births, 27,732 more people moving out to other states than in, and 176,435 net arrivals from abroad. Franklin County had 13.6% of Ohio's births with 11.3% of its people, took a third of its immigrants, and lost 52,686 people to moves, 89% of them to other Ohio counties. Franklin, Hamilton, Butler and Montgomery grew only because of immigration. Of Ohio's 39 growing counties, 29 have more deaths than births.
Population change July 2020 to June 2025, by source: births minus deaths, moves to and from the rest of the U.S., and immigration
Why it happens. Columbus behaves like a port of entry: people arrive from abroad, from other states and from Ohio's small towns, form families, and the surplus moves one county out. The ring around it barely has more births than deaths (Fairfield +5 over five years, Licking -564, Pickaway -313); about half the jobs held by Delaware, Pickaway, Fairfield and Madison residents are in Franklin County. This fits a housing-driven spillover but does not prove it.
How we checked. Two sources that share no inputs agree on the moves: Census estimates and IRS tax-return address changes (-35,473 against -38,169 for Franklin over the same years; rank r 0.88 across all 88 counties).
Your county. Summit shrinks through deaths (-6,178) more than moves (-1,633); Lucas through departures (-9,172) more than deaths; Cuyahoga loses both ways and still took 22,407 immigrants. Counties growing only by people moving in while in natural decrease include Lorain, Clermont, Licking, Greene, Pickaway and Miami.
What it does not show. Immigration is the most revised Census number and the latest year is provisional: the direction is safe, the last digit is not. The Census does not say where immigrants come from by county.
Sources: Census population estimates, Vintage 2025 components; IRS county-to-county migration; LEHD commuting data.
Only 28 of 88 counties are at their population high, and the hill counties that emptied first have stopped emptying
28 counties set their population record in 2020, holding 42% of Ohioans. 17 had more people at some census before 1950 than today: Monroe County is half its 1900 size, Morgan peaked in 1860, and ten counties have fewer people than in 1900. Coal mines, not farm country, mark those early peaks: 9 of the 12 counties that peaked before 1950 have mapped underground mines under more than 1% of their land, against 10 of the other 76. Farm-heavy counties peaked late; most are still growing.
Ohio counties by the census year of their highest population
Show the numbers
| Group | Counties | Share of Ohioans | Below peak (median) | Moves in minus out, per 1,000 a year | Births minus deaths, per 1,000 a year |
|---|---|---|---|---|---|
| growth belt (peak 2000+) | 51 | 49.5% | 0% | +2.2 | -1.2 |
| hill and coal (peak before 1950) | 12 | 3.0% | 30% | +1.2 | -4.8 |
| industrial (peak 1950-99) | 25 | 47.4% | 9% | +0.1 | -2.6 |
Why it happens. The hill-and-coal counties lost most of their people before 1970. Today 9 of those 12 gain people from moves, though all 12 have more deaths than births. The counties still losing residents are the industrial counties that peaked in 1950-99, from Cuyahoga and Mahoning to Montgomery and Lucas, home to 47% of Ohioans.
How we checked. Census and IRS migration records, which share nothing, agree on the hill counties (9 of 12 and 8 of 12 gaining). The idea that one "long decline" runs from early peak to early death fails: once income is held fixed, a county's peak year has no link with years of life lost (-0.05). County population counts match Census state totals exactly for 1810, 1900 and 2020.
Your county. Deepest below their record: Monroe -50.5% (peak 1900), Morgan -37.6% (1860), Jefferson -34.2% (1960), Noble -33.2%, Paulding, Meigs, Harrison, Belmont, Cuyahoga -26.5% (1970). Counties 10% or more below their peak have a median home value of $151,600, against $244,750 in counties at a record.
What it does not show. Peaks are measured at census years. The mine map shows where abandoned underground mines were drawn, not how many people worked in them.
Sources: NHGIS county censuses 1810-1960, decennial censuses 1970-2020, Ohio DNR abandoned-mine maps, Census and IRS migration.
In 1900 farms covered 94% of Ohio; today 52%, and Ohio once had nearly two sheep for every person
Ohio had 276,719 farms at the 1900 census and 76,009 in 2022, down 72%. Land in farms fell from 93.9% of the state to 52.2%. Mid-sized farms of 50 to 175 acres are down 85%, while farms of 500 acres or more grew 5.6 times. In 1870 Ohio had 4.9 million sheep, 1.85 for every resident; today one for every 81. Oats, the feed of the horse era, went from 1.79 million acres to 13,512, while corn output quadrupled and soybeans, almost unknown in 1929, now cover more Ohio land than corn.
Show the numbers
| Then | Now | |
|---|---|---|
| Farms | 276,719 (1900) | 76,009 (2022) |
| Average farm | 88.5 acres | 179.6 acres |
| Sheep | 4,928,635 (1870) | 145,381 |
| Horses on farms | 910,224 (1910) | 91,418 (horses and ponies) |
| Oats | 1.79 million acres (1909) | 13,512 acres |
| Wheat | 3.21 million acres (1899) | 480,146 acres |
| Corn harvest | 152 million bushels (1899) | 609 million bushels |
| Soybeans | 96,944 acres (1929) | 4.83 million acres |
Why it happens. The tractor freed the land that grew horse feed and cut the labor each acre needed: Ohio had 21 horses and mules per tractor in 1925 and 10 in 1930. Output rose while farms, farmland and farm workers fell.
How we checked. Oats and horses still go together today: across 47 counties they rank together (rank r 0.65), led by Holmes, Wayne and Geauga, and the same counties kept the most farms (Holmes 66% of its 1900 farm count). The 2022 count matches USDA's published 76,009 exactly.
Your county. Farms left in 2022 as a share of 1900: Holmes 66%, Wayne 53%, Geauga 45%; Cuyahoga 2%, Hamilton 8%, Franklin 10%.
What it does not show. The census definition of a farm changed several times, and no farm series between 1950 and 1992 is held.
Sources: NHGIS agricultural censuses 1850-1950; USDA Census of Agriculture 1992-2022.
Ohio's baby belts are Amish country and the western farm counties, and teen births fell by more than half
A woman in Holmes County is 3.3 times as likely to have a baby in a year as one in Athens County (96.0 births per 1,000 women against 28.8; Ohio 56.8). After Holmes come Mercer, Shelby and Putnam, western farm counties. Teen births fell from 40.1 per 1,000 girls in 2007 to 17.6 in 2020, in all 88 counties; between 2010 and 2019 the drop in births to teenagers was larger than the entire fall in Ohio's births.
Births per 1,000 women aged 15-44, 2021-25: highest and lowest counties (Ohio 56.8)
Why it happens. Fertility follows the farm economy: agriculture's share of a county's economy goes with higher birth rates even at the same population size. Holmes adds the Amish signature (45% of residents are Amish adherents). The teen decline came fastest where more adults had college degrees and higher incomes; where teen births stayed highest is also where people die youngest (rank r 0.75).
How we checked. An independent earlier estimate (Ohio Department of Health infant counts per woman, 2017-19) ranks counties the same way (rank r 0.83). Raw teen-birth counts agree with the federal model estimates (0.89). Statewide, our rate (56.78) matches the Ohio Department of Health's published 56.88.
Your county. Highest teen birth rates 2018-20: Pike 41.7, Jackson 38.1, Adams 36.7, Marion 34.9; lowest Delaware 4.1, Geauga 5.0, Medina 6.1. Franklin produces 13.6% of Ohio's babies.
What it does not show. College towns are not low only because of students: Athens stays last even per woman aged 25-44. Noble County is left off the chart because its denominator includes a large prison.
Sources: Census births 2021-25; ACS women 15-44; NCHS model-based teen birth rates; County Health Rankings.
Holmes County is one Amish story, and Ohio looks the same without it
Holmes County (44,000 people, 45.5% Amish adherents) had Ohio's fourth-largest surplus of births over deaths in 2021-25 (+2,222), behind only Franklin, Hamilton and Delaware, counties many times its size. It has Ohio's highest birth rate, youngest median age (32.6), lowest lung-cancer rate (38.1 per 100,000 against Ohio's 63.6) and the most farms kept since 1900. It also sends people away: its leavers settle in Wayne, Tuscarawas and Coshocton.
Why it happens. Among the 14 counties at least 3% Amish, the Amish share lines up with farms kept, low teen births and fewer early deaths, so those read as one Amish effect. On fertility it does not scale: in Geauga (10% Amish) and Wayne (8%) the other nine-tenths of the county is ordinary Ohio.
How we checked. Remove Holmes and every statewide finding on this page moves by 0.03 or less. Two data faults turned up: one Amish population table assigns whole multi-county settlements to a single county, and the County Health Rankings smoking model ranks Holmes first in Ohio for smoking while the cancer registry ranks it last for lung cancer. The model guesses smoking from schooling and income, which the Amish keep low by choice.
Your county. The 14 Amish counties: Holmes, Geauga, Wayne, Coshocton, Knox, Ashtabula, Tuscarawas, Ashland, Monroe, Carroll, Hardin, Morrow, Jackson, Noble.
What it does not show. Nothing here measures Amish individuals; every figure is a county with an Amish share. Holmes's income and poverty rate are ordinary for Ohio, so its health figures are not an income story.
Sources: 2020 U.S. Religion Census; Census estimates; State Cancer Profiles; USDA.
Neighborhoods a few miles apart in Ohio's cities differ by up to 20 years of life
In Franklin County the shortest-lived neighborhood with a reliable estimate expects 64.2 years of life and the longest-lived 84.2, a 20-year gap between neighborhoods 6.5 miles apart. In Montgomery County 13.3 years separate neighborhoods 2.9 miles apart. Most of Ohio's life-expectancy gap is inside counties: 87% of the variation between neighborhoods lies within a county, not between counties.
Shortest- and longest-lived neighborhoods (census tracts with reliable estimates) in eight large counties, life expectancy at birth, deaths 2010-15
Why it happens. Neighborhood income does most of the sorting: each doubling of a tract's income goes with 4.3 more years of life, in every big county. The richest fifth of neighborhoods outlives the poorest fifth by 7.1 years on average. Once income and poverty are held fixed, a neighborhood's racial makeup adds nothing further at this scale.
How we checked. Neighborhood life tables (NCHS) and tax-record income data (Opportunity Insights) share no inputs. Only tracts with a standard error under 1.5 years are quoted, because the raw extremes are mostly noise. The weighted average of all tracts (77.42) is within 0.2 years of the published Ohio figure (77.6).
Your county. Largest gaps between neighborhoods: Franklin 20.0 years, Lucas 14.1, Stark 14.0, Butler 13.4, Montgomery 13.3. Among cities of 10,000 or more (2010-15): Avon 83.3 years, Mason 83.2, Solon 83.2; Dayton 72.0, Portsmouth 72.6, Maple Heights 72.9.
What it does not show. These are deaths from 2010-2015, before the fentanyl and pandemic years; no newer neighborhood data is held. Patterns across neighborhoods are not facts about any one person.
Sources: NCHS/NAPHSIS USALEEP tract life expectancy 2010-15; Opportunity Insights tract data.
Ohio's overdose epidemic moved: from the Cincinnati-Dayton corridor to the Ohio River counties and the Mahoning Valley
In the late 2010s the Cincinnati-Dayton corridor ran at 1.5 times Ohio's overdose death rate; by the 2025 release it was at the state average. The Ohio River counties and the Mahoning Valley moved to the front: Scioto County now has the highest rate (126 per 100,000 in 2020-22). Heroin deaths fell from 1,472 in 2015 to 56 in 2025 as fentanyl took over, and all overdose deaths fell from a peak of 5,405 (2021) to 3,144 in 2024.
Overdose death rate relative to Ohio (1.0 = the state rate), by region, County Health Rankings releases 2014-2025
Show the numbers
| Year (12 months to December) | Synthetic opioids excl. methadone (fentanyl class) | Heroin | Natural and semi-synthetic opioids (prescription pills) | Cocaine | Psychostimulants incl. methamphetamine | All drug overdose deaths |
|---|---|---|---|---|---|---|
| 2015 | 1244 | 1472 | 693 | 705 | 112 | 3336 |
| 2016 | 2305 | 1469 | 786 | 1137 | 242 | 4326 |
| 2017 | 3550 | 1016 | 857 | 1576 | 551 | 5138 |
| 2018 | 2818 | 712 | 493 | 1112 | 576 | 4004 |
| 2019 | 3150 | 536 | 405 | 1235 | 838 | 4279 |
| 2020 | 4126 | 293 | 440 | 1251 | 1069 | 5215 |
| 2021 | 4209 | 157 | 412 | 1537 | 1361 | 5405 |
| 2022 | 3997 | 86 | 371 | 1649 | 1374 | 5144 |
| 2023 | 3600 | 79 | 329 | 1777 | 1306 | 4745 |
| 2024 | 2084 | 48 | 247 | 1164 | 850 | 3144 |
| 2025 | 1344 | 56 | 208 | 945 | 631 | 2452 |
Why it happens. The timing fits fentanyl reaching the Cincinnati-Dayton-Springfield corridor first and the Ohio River counties later, but county records do not say which drug, so that is an inference from timing. The fall since 2021 was steepest in Appalachian Ohio and the corridor, slowest around Cleveland, Toledo and the Mahoning Valley.
How we checked. CDC's separate provisional counts for 2020-23 name the same top counties. The county counts sum to 98% of CDC's Ohio total for 2021. Change in overdose deaths explains more of each county's lost life expectancy since 2015-17 than COVID does.
Your county. Highest overdose rates, CDC 2023: Scioto 114 per 100,000, Vinton 80, Mahoning 69, Lawrence 69, Marion 69, Pike 67, Ross 65, Trumbull 65. Life expectancy fell in 87 of 88 counties between the 2019 and 2025 releases; most in Vinton (-7.6 years).
What it does not show. Regions are this project's own grouping. The 2025 figures are the least complete. Counts under 10 are suppressed in small counties.
Sources: County Health Rankings 2014-2025; CDC Vital Statistics Rapid Release provisional overdose counts.
Ohio loses two-thirds fewer households to other states than in 2005, but only a fifth less income, and Florida takes most of it
The net number of households Ohio loses to other states fell from 23,397 (2004-05) to 7,828 (2022-23), down 67%. The income those households take fell only 19%. Florida takes 28% of the households but 45% of the income, and 71% in the COVID years. Ohioans moving to the Naples area report $369,000 of income per tax return on average; people moving from there to Ohio, $164,000. The pattern reverses for big cities: Ohioans moving to Chicago or Manhattan earn less than the people coming the other way.
Average income per tax return (2024 dollars, thousands) of households moving between Ohio and each county, 2004-2023
Why it happens. What remains of the loss to other states is increasingly high-income households heading to a few coastal counties. That fits retirement and second-home moves, but the tax file has no ages, so it cannot show it: people 65 and over are only about 9% of Ohio's interstate leavers by headcount.
How we checked. The premium is present in every year from 2004 to 2023, and county sums reproduce the IRS state-to-state file to within 1.1% every year. The Census survey matches the overall income gap but cannot break it down by destination.
Your county. Most Florida-bound households each year come from Franklin, Cuyahoga, Hamilton, Summit and Montgomery; leavers from Hamilton County out-earn those who stay by 1.9 times.
What it does not show. Averages of a skewed distribution: a few very high incomes move the Naples figure. Income includes the year's home and investment sales, so it overstates ordinary income.
Source: IRS Statistics of Income county-to-county migration, 2004-05 to 2022-23.
Part
Homes: building, buying and fixing
What Ohio builds, what its houses have earned, and who keeps them up.
Ohio stopped building the starter home; since 2012 it has gained more than any other house, and it is the house a company most often owns
Of single-family houses built in the 1950s, 68% are under 1,400 square feet; of those built in the 2020s, 8%. The median new house went from 1,200 to 2,200 square feet with the same three bedrooms. So the small house is a fixed, aging stock: 85% of today's were built before 1980. Its price fell harder in the bust and rose faster after: a small old home's median sale price went from $95,300 in 2012 to $198,200 in 2024-25 (2024 dollars), +108%, against +66% for every other house. Sales under $150,000 fell from 38% of all sales to 19%. And about one in nine small old homes is owned by a company, against one in twenty other homes; inside a county, a company owns one in five of the cheapest tenth of homes.
Median sale price, 2024 dollars, 27 counties with deeds every year
Show the numbers
| Built | Share of new single-family houses under 1,400 sq ft |
|---|---|
| 1950-1954 | 69% |
| 1955-1959 | 68% |
| 1960-1964 | 58% |
| 1965-1969 | 43% |
| 1970-1974 | 44% |
| 1975-1979 | 36% |
| 1980-1984 | 39% |
| 1985-1989 | 26% |
| 1990-1994 | 19% |
| 1995-1999 | 18% |
| 2000-2004 | 13% |
| 2005-2009 | 10% |
| 2010-2014 | 10% |
| 2015-2019 | 7% |
| 2020-2024 | 8% |
Why it happens. A fixed supply meeting two kinds of demand at once, first-time buyers and rental investors. No builder puts up a 1,100-square-foot detached house today; condos fill only part of the gap (19% of new homes are small counting condos, against 69% in the 1950s).
How we checked. Holding the house fixed: resold small old homes bought in 2012-19 earned 6.0% a year after inflation against 4.3% for other homes, and 1.3 points a year more inside the same ZIP code and purchase year. Homes cheap at both purchase and resale beat homes dear at both by about 3 points a year, in all 37 counties with enough sales, so the premium is not just price noise.
Your county. A small old home now costs 3.2 years of median household income in Franklin County, 2.9 in Hamilton and Clermont, 1.8 in Belmont. Company share of small old homes: Franklin 14.6%, Hamilton 14.2%, Montgomery 13.7%; lowest Morgan 3.5%, Monroe 4.0%.
What it does not show. "Company" means any owner name with a legal form or trade word, including banks holding foreclosures, builders and a family's own LLC, and it is today's owner. No deed records are held for Cuyahoga, Lucas, Summit and 25 other counties.
Sources: county auditor records (3.95 million homes) and deed histories in our property database.
A company-owned house gets less than half the roof, furnace and plumbing permits of the owner-occupied house next door
We followed 291,857 single-family houses in six permit offices from 2020 through 2025 and matched every permit to its parcel. In Columbus, owner-occupied houses drew 4.94 improvement permits (roofs, siding, remodels, additions and so on) per 100 houses a year; company-owned houses 1.84. Compared only with owner-occupied houses of the same ZIP, decade and size, companies run at 0.43 of the owner rate, and 0.33 for houses a company held the whole six years. Clermont County 0.43, Kettering 0.40, Medina 0.58, Xenia 0.18. Furnaces and plumbing: also about 0.43.
Improvement permits on company-owned houses as a share of the rate on owner-occupied houses in the same ZIP, decade built and size (1.0 = the same), 2020-2025
Why it happens. For a landlord a roof is a cost against rent; for an owner it is the house they live in. The pattern fits deferred maintenance on long-held rentals, and also work done without a permit. The data cannot tell those apart, and both matter to tenants and neighbors.
How we checked. A second record that shares nothing with the permit office agrees: the county auditor's own "year remodeled" field shows company houses remodeled less often (0.83 of the owner rate in Franklin, 0.65 in Medina). Cincinnati is the exception (1.31): the extra there is remodels by companies that bought during the period, renovation for resale; its long-held company houses are at 1.06 and its furnace and plumbing permits at 0.67.
Your county. Only six offices have permits joined to parcels: Columbus (Franklin), Cincinnati (Hamilton), Clermont, Medina, Kettering (Montgomery) and Xenia (Greene).
What it does not show. A permit is a floor on work, not work: a landlord who re-roofs without a permit is invisible, and that is part of what this measures. Vacancy is not observed.
Sources: municipal and county building permits (our permit database); county auditor records.
What gets permitted follows what the house is worth, and the small jobs got dearer: an Ohio roof now costs about half again what it did in 2016
In Columbus the dearest fifth of houses draws twice the permits of the cheapest fifth and five times the declared dollars ($1,961 a house per year against $386): 0.51% of the house's value against 0.32%. Their roof costs $13,900 on paper, the cheapest fifth's $8,000. Within every decade built, the dearest third of houses gets about twice the permits of the cheapest. Meanwhile the median declared roof rose from $7,188 (2016-17) to $11,000 (2023-24) in Columbus, +53% after inflation, and +44% in Cleveland; roofs declared under $5,000 went from 45% to 16% and 10%.
Columbus single-family houses: declared improvement dollars per house per year, 2020-2025, by fifth of the house's value
Show the numbers
| Value fifth | Roof permits per 100 houses a year | Median declared roof | Median declared remodel | Share with any permit in six years |
|---|---|---|---|---|
| Cheapest | 2.11 | $8,000 | $12,000 | 26% |
| Second | 2.5 | $9,090 | $13,360 | 32% |
| Middle | 2.8 | $10,295 | $9,401 | 36% |
| Fourth | 3.04 | $11,509 | $12,934 | 39% |
| Dearest | 3.65 | $13,938 | $30,000 | 49% |
Why it happens. Cash and ownership: the owner of a $386,000 house spends a larger share of a larger number, and a quarter of the cheapest fifth is company-owned. The cheapest roof jobs rose fastest, so the floor price of a crew visit moved, not a few big jobs.
How we checked. The same value-by-age order appears in five separate permit offices. The roof rise appears in two cities with different software and staff. The worry that early Columbus values were understated was tested against sale prices: new houses were declared at 0.88 of their sale price in 2016-17 and 0.90 in 2023-24, so the rise is real.
Your county. Typical declared costs, 2016-2026 (2024 dollars): roof $9,700 (Columbus) and $9,000 (Cleveland); deck about $8,500; addition $51,000-$87,000; new house $257,000-$312,000 in the three big cities, $507,000 in Delaware County.
What it does not show. A declared value is the applicant's figure, not a receipt, and offices collect it differently: the same new house is declared at 92% of its sale price in Columbus but 46% in Cincinnati, so permit values cannot be compared between cities. Columbus cannot separate a roof from siding or windows.
Sources: building permits from 20 Ohio offices (716,938 residential); county auditor appraisals.
Permitted work comes back at the sale at about 90 cents on the dollar
We matched permits to 24,688 homes bought since 2016 and resold, and measured each home's return beyond its ZIP code's price index over the same months. A remodel or addition between the two sales goes with 3.0 extra points a year; a roof or exterior job 1.0; a new furnace or plumbing job 0.5. In Columbus each declared dollar of work comes back as about $0.87 of extra sale price ($0.92 for homes held two years or more).
Extra yearly return at resale over the neighborhood's home-price index, percentage points, when this work was permitted between the two sales (± one standard error), 24,688 resales
Why it happens. Buyers pay for work at roughly what it was declared to cost, and no more: upkeep keeps a house's value rather than creating a premium, and a new furnace is barely priced at all. Quick flips return only $0.29 per declared dollar: their gain is mostly the discount at purchase.
How we checked. The design could have shown nothing, since a hot neighborhood lifts permitted and unpermitted houses alike; it did not. The remodel premium appears at every holding period, and Medina and Kettering replicate it within error.
Your county. Measured in Franklin (Columbus), Clermont, Medina, Montgomery (Kettering) and Greene (Xenia). Hamilton cannot be tested: its auditor file keeps one sale per home.
What it does not show. Owners who pull remodel permits may also be owners who present a house well, so 3 points is an upper bound on what the work itself does. Declared dollars are not receipts.
Sources: building permits; county deed records; Zillow ZIP home-value index.
When you bought mattered far more than where: the typical Ohio home did not beat inflation from 2000 to 2020
In today's dollars the typical Ohio home was worth $198,200 in 2000, $152,400 in 2012 and $188,600 in 2019, still below 2000; nearly all of the gain since 2000 (98%) came in 2019-22. 81 of 86 counties were below their 2000 value in 2019. Homes bought in 2003-07 and resold lost a median 9.7% after inflation, two in three lost value, and 38% sold for fewer dollars than they cost; homes bought in 2012-19 gained 25%. Fewer homes change hands now: 2.9 of every 100 older homes sold in 2025, against 4.5 in 2005.
Median yearly return after inflation on resold Ohio homes, by year of purchase (48 counties, 692,502 resales)
Why it happens. The whole state moved as one wave: a flat 2000-06, a bust that spared no county (all 88 lost 8% to 34%), then one repricing in 2019-22. When the market moves together, the year you bought dwarfs the street: purchase and sale years explain six times as much of a resold home's return as its ZIP code (18% against 3%).
How we checked. Three sources with no shared inputs agree on the 2006-2024 change after inflation: the Census survey +14.0%, the federal FHFA index +14.3%, Zillow +11.1%. On the slowdown in sales we corrected our first explanation: buyers who borrowed at 6-7% in 2022-23 resell as rarely as those who locked in 3% in 2020-21, so this is a market-wide freeze, not owners clinging to cheap loans.
Your county. Still below their 2006 value after inflation (2024): Lucas -8.8%, Lake -8.6%, Geauga -5.1%, Sandusky, Huron, Gallia, Henry, Ottawa, Williams, Wood, Ashtabula, Cuyahoga, Lorain, Erie; 13 of these 14 are in Ohio's northern tier. Furthest ahead: Monroe +59%, Franklin +30%, Harrison and Licking +28%.
What it does not show. A price index is not an owner's return: no taxes, repairs, interest or rent saved. Only resold homes are counted, so patient holders are missing.
Sources: Zillow Home Value Index; FHFA county house price indexes; ACS; county deed records.
Ohio builds fewer than half the houses it did in 2003; the comeback is the Columbus ring, and Cleveland razes three homes for every new unit
Ohio permitted 40,019 single-family homes a year in 2000-05 (peak 42,703 in 2003), 9,320 in 2011 and 17,889 a year in 2020-25: 45% of the old pace. The ten Columbus-area counties went from 21% of Ohio's permits in 2006 to 33% in 2021-25, and Delaware County alone now issues 10% of the state's. At the other end, Cleveland's demolition permits name about 12,300 homes razed in 2016-25, against 3,774 new units authorized.
Single-family homes authorized by building permit in Ohio each year (Census)
Why it happens. Building follows people and jobs, and both moved to the Columbus metro. The bust was statewide; the recovery was not. Cleveland's razing falls on its cheapest neighborhoods (rank r -0.79 against ZIP home value), the same places that hold the starter-home stock.
How we checked. Two sources that share no inputs, Census permits and the auditors' "year built" field, draw the same curve year by year. Of 75 counties issuing 20 or more permits a year in 2006-07, only eight build at or above that pace now: Pickaway, Union, Madison, Delaware, Miami, Morgan, Fairfield and Mercer.
Your county. Against their own 2006-07 pace (2021-25): Franklin 0.82, Hamilton 0.74, Montgomery 0.72, Cuyahoga 0.62, Summit and Lucas 0.46. Heaviest razing in Cleveland ZIPs 44127, 44104, 44103 and 44108.
What it does not show. Permits are not completions, and a demolition permit is not a finished demolition. The 2006-07 baseline is already a bust year.
Sources: Census Building Permits Survey; county auditor records; Cleveland demolition permits.
Half the houses built in the late 1970s heat with electricity, and they sell at no discount
In the eleven counties whose auditors record heating fuel, 54% of homes built in 1975-79 are electric-heated, against 5% of 1950s homes and 7% of 2020s homes. Despite the running-cost reputation, 117,364 sales in 2021-25, compared with gas-heated homes of the same size, age, features and ZIP code, show no discount: +0.3% (within error).
Why it happens. The timing fits the 1970s all-electric-home promotions and the late-1970s natural-gas shortage that limited new gas hook-ups. Why no discount: the all-electric house of that era was often the better-equipped house on its street, and many have since been converted (15% now carry a heat-pump code).
How we checked. Two outcomes (sale prices and auditor appraisals) and two methods agree, in each of the eleven counties. Every county auditor shows more electric heat in 1965-84 homes than in older ones.
Your county. Share of 1965-84 homes heated electrically: Clermont 65%, Gallia 59%, Seneca 53%, Fairfield 52%; Stark 10%. The other 77 counties do not record fuel.
What it does not show. "No discount" is not "no cost": utility bills are not in the record.
Sources: county auditor records and deeds, eleven counties.
Part
Money: paychecks, bills and businesses
What Ohioans earn, what they pay to keep the lights on, and what kind of businesses they start.
Ohio's new-business boom is a boom in one-person ventures, and half its growth since 2012 is people driving
Ohioans filed 168,460 applications to start a business in 2025, 77% more than in 2019. Applications planning to pay wages rose only 7%, and are 43% below 2006. Nearly all of the extra filings (98.8%) plan no payroll. What actually opened matches: Ohio had 909,227 businesses with no employees in 2023, up 58% since 1997, while businesses with a payroll fell 3.7%. More than half of the one-person growth since 2012 is three kinds of driving: couriers and messengers (7,124 to 57,676), ride services (4,622 to 29,259) and trucking.
Applications to start a business in Ohio each year, all and those planning to pay wages
Show the numbers
| 2006 | 2019 | 2025 | |
|---|---|---|---|
| All applications | 66,860 | 95,240 | 168,460 |
| Planning to pay wages | 23,910 | 12,800 | 13,680 |
| One-person courier businesses | 19,614 | 57,676 (2023) |
Why it happens. App-based delivery and ride services: ride-service businesses were flat until 2013 and then took off; couriers were flat until 2016 and then multiplied. A one-person courier business took in $11,928 in 2023, against $28,788 in 2012. The boom is a metro and suburban one: it grew fastest where paychecks are biggest.
How we checked. Federal payroll records (BLS) confirm payroll businesses did grow after 2019, just far less than filings. Of the roughly 190,800 extra filings in 2020-23, 48% show up as new no-employee businesses and 5% as extra openings with a payroll.
Your county. Couriers and ride services as a share of one-person businesses: Franklin 14%, Montgomery 12.5%, Butler 12.4%, Cuyahoga 11.6%; Noble 1%, Monroe 1.8%. Fastest one-person growth 2012-23: Union, Warren, Franklin, Delaware.
What it does not show. An application is an intent, not a business. A no-employee business is a tax filer, not a person: one driver on two apps can be one filer or two.
Sources: Census Business Formation Statistics; Census Nonemployer Statistics; BLS Business Employment Dynamics.
Ohio has one thin-paycheck map: the same counties lean on benefits, face the dearest power and the heaviest water bills
Government benefits (Social Security, Medicare, Medicaid and the rest) are 21% of Ohioans' income, up from 15% in 2001; 28 counties now get 30% or more. Those same counties have the most households below a bare survival budget, the dearest electricity and the heaviest water bills. In Adams and Athens counties a basic survival budget costs 120% of the median household's income; in Delaware, 53%. Electricity, gas, water and sewer take 10.7% of the median income in Meigs County and 3.0% in Delaware.
How strongly each measure belongs to the one "thin paycheck" pattern found across 75 counties (loading on the main shared axis, 0 = not at all)
Why it happens. One driver: thin earnings. The utility burden varies across counties because incomes do, not because bills do: incomes vary twice as much between counties as utility bills. On top of that sits one supply fact: the dearest electricity in Ohio (AEP Ohio, Buckeye Rural, South Central Power) is sold where incomes are lowest, 17.0 cents per kilowatt-hour in the highest-benefit quarter of counties against 15.5 in the lowest. Across counties the benefit share tracks residents on disability Medicare (rank r 0.91) far more than age (0.39).
How we checked. Seven sources were combined (BEA, United For ALICE, CMS, HUD, Ohio EPA, FDIC, EIA). One shared pattern carries 47% of their variation. Housing cost burden is a separate, big-city pattern. Bank-branch loss belongs to neither (0.00): the closings are not where the thin paychecks are.
Your county. Highest benefit share: Scioto 37.7%, Gallia 37.0%, Vinton 36.4%, Adams 36.3%; lowest Delaware 9.3%, Holmes and Union 11.7%. Utilities as a share of median income: Meigs 10.7%, Vinton 8.6%, Mahoning and Jackson 8.4%; Delaware 3.0%, Warren 3.2%, Union 3.7%.
What it does not show. County medians say nothing about any one household. BEA counts Medicare and Medicaid payments as residents' income even though the money goes to doctors and hospitals. Utility territories were drawn decades ago; no utility chose its customers.
Sources: BEA county personal income; United For ALICE; CMS Medicare enrollment; HUD CHAS; Ohio EPA water and sewer rate survey with ACS income; EIA-861.
Utility bills are lower after inflation than in 1984 and the gas bill has halved, so a fixed charge now carries most of a small home's gas bill
The average Midwest household spent $5,232 a year on utilities in 1984 and $4,376 in 2024, in today's dollars (-16%); natural gas fell by half. Shelter rose 46%. Housing, utilities and insurance together have taken a steady quarter of income since 1990. But as use fell, cost moved into the fixed part of the bill: Columbia Gas of Ohio's fixed monthly charge rose from $36.57 in 2023 to $54.01 in September 2026, now 54% of an average home's yearly gas bill and 74% of an apartment's.
Midwest household spending a year, 2024 dollars, 1984-2024
Why it happens. A utility's pipes cost the same whether a home burns a little gas or a lot, so as use falls the cost moves from the per-unit charge to the per-account charge. The average bill fell, and the part a household can still cut by using less shrank: cutting a typical year's use by a fifth now saves a Columbia customer about 9%.
How we checked. Ohio's own price records agree with the long fall and the recent turn: the residential electricity price (recomputed from federal filings to match the published figure) and the median town water and sewer bill both lagged inflation lately. Households under $30,000 spend 6.1% of their budget on utilities, those over $150,000 2.9%.
Your county. Columbia Gas serves 55 counties; the fixed-charge arithmetic is the same in all of them, so the deciding factor is home size. Electricity and gas together take 6.1% of median income in Meigs and 1.9% in Delaware.
What it does not show. The spending survey is the Midwest region, not Ohio, and its "utilities" includes phone service. Only Columbia's fixed charge is held, not Duke's, CenterPoint's or Enbridge's. The 2028 cap of about $56 is as reported in the press.
Sources: BLS Consumer Expenditure Survey (Midwest) 1984-2024; PUCO bill surveys; Columbia Gas tariffs; EIA RECS.
Ohio produces 20% more per job than in 2001 but pays 13% more, and in the gas counties output tripled while paychecks stood still
From 2001 to 2022 Ohio's real output per job rose 19.6% and the real wage per job 12.8%. Employee pay and benefits were 58.8 cents of each dollar of Ohio's output in 2001 and 51.1 cents in 2024. The extreme case is the Utica shale gas counties: Harrison's real output rose 345%, Carroll's 303%, Monroe's 174%, yet residents' real earnings per person rose 22% in Harrison and fell 3% in Monroe, their populations fell, and the share of their income from government benefits rose.
Ohio, index 2001 = 100
Show the numbers
| county | real output growth % 2001-24 | real per-capita personal income growth % | real net earnings per person growth % | transfer share 2001 | transfer share 2024 | pop growth % |
|---|---|---|---|---|---|---|
| Harrison | 345.0 | 36.1 | 22.0 | 23.6 | 31.2 | -11.1 |
| Carroll | 303.0 | 34.7 | 17.3 | 18.2 | 26.7 | -8.6 |
| Delaware | 262.0 | 26.9 | 18.0 | 6.4 | 9.3 | 100.6 |
| Guernsey | 180.0 | 35.0 | 16.2 | 23.5 | 32.1 | -6.5 |
| Monroe | 174.0 | 33.8 | -3.3 | 24.3 | 35.9 | -13.6 |
| Warren | 154.0 | 42.8 | 24.7 | 9.9 | 12.8 | 52.6 |
| Mercer | 144.0 | 55.0 | 54.4 | 14.6 | 16.6 | 4.5 |
| Noble | 117.0 | 3.2 | -16.0 | 19.8 | 30.3 | 3.0 |
Why it happens. Gas output is counted where the well is, and little of it is pay: mining wages are 3.5% of mining output in Carroll County. Drilling jobs peaked in 2012 and left. What stays local arrives as royalties, which lifted Harrison from 68th to 44th of 88 counties on income per person but left Monroe and Noble at the bottom. Statewide, factories make 10% more with 27% fewer workers.
How we checked. Employer tax records (QCEW), independent of the federal output estimates, show county pay growth only loosely following output growth (rank r 0.41). Ohio's output totals match the published state figure within 0.1%. The fall in pay's share is national too (57.0 to 51.2 cents) and survives counting the self-employed.
Your county. Real output growth 2001-24 and income per person: Harrison +345% / +36%; Carroll +303% / +35%; Delaware +262% / +27%; Mercer +144% / +55%, the one county where output and pay grew together.
What it does not show. County output is an estimate built from state data, not a survey. Royalties paid to landowners living elsewhere never appear in the county. Nothing here says the wells made anyone poorer; it says they did not raise local paychecks.
Sources: BEA county GDP and personal income 2001-2024; BLS QCEW.
Ohioans on city-owned electric pay 16% less than investor-owned customers and 25% less than co-op members, a gap that opened after 2021
In 2024 the median price was 12.80 cents per kilowatt-hour at city-owned utilities, 15.26 at investor-owned ones and 17.05 at co-ops: about $251 and $433 a year more than city-owned power for a typical home. In 2019 there was no gap between city-owned and investor-owned power. City-owned systems also lose power far less: about 40 minutes per customer a year, against about 116 to 119.
Median residential electricity price by type of utility, cents per kilowatt-hour, 2024 dollars (utilities filing every year)
Why it happens. Timing points to wholesale market prices: the state's standard-offer bills from investor-owned utilities fell or held until 2021, then rose 43% to 114% by September 2026, while city-owned prices held. Outages track how densely customers are packed, not price.
How we checked. Recomputing Ohio's average residential price from the same federal filings reproduces the published figure every year 2015-2024. 2024 comes before the June 2025 capacity price step, so the gap today is probably wider.
Your county. Mean 2024 price of the utilities serving the county: Athens, Gallia, Jackson, Lawrence and Meigs 19.3 cents; Trumbull 13.3, Warren 13.4, Mercer 13.6. City systems run from 9.1 cents (Celina) to 15.4.
What it does not show. City-owned power serves only those towns. A county average is not what any household pays. Outage reporting is voluntary.
Sources: EIA-861 2015-2024; PUCO bill surveys.
Towns with old combined sewers paid less for sewer service in 1990 and about a third more by 2019
In towns whose sewers carry rain and sewage in one pipe, the median sewer bill was $449 a year in 1990 against $538 elsewhere (today's dollars). By 2019 it was $1,071 against $818. Sewer bills more than doubled after inflation in those towns (+113%) and rose 50% elsewhere. Statewide, the median water and sewer bill rose 61% above inflation from 1990 to 2019, then paused, falling 5% behind inflation from 2019 to 2023.
Median yearly sewer bill, 2024 dollars, towns with combined sewers vs other towns
Why it happens. Combined sewers overflow when it rains, and federal orders made cities like Cleveland, Cincinnati, Akron, Toledo, Youngstown and Columbus build tunnels and separate their pipes, paid for by ratepayers. Size, income and population change held fixed, a combined-sewer flag adds about 25% to a sewer bill. In shrinking towns water bills weigh more mainly because incomes are lower, not because bills are higher.
How we checked. The flags come from federal permits, the bills from Ohio EPA surveys: separate sources. A second survey year (2023) gives the same gap (+30%).
Your county. Biggest real sewer rises 1990-2019 among combined-sewer towns: Forest (Hardin) $204 to $1,107, Delphos (Allen) $300 to $1,569, Hicksville (Defiance) $302 to $1,520; Youngstown $478 to $1,736, Cincinnati $408 to $1,185.
What it does not show. The flag is today's, not 1990's; Akron has a federal order but no flag. Debt and grants are not in the data.
Sources: Ohio EPA Sewer and Water Rate Surveys 1990-2023; EPA permit records.
Ohio has 994 fewer bank branches than in 2010, and 113 towns that had a bank in 1994 have none
Ohio had 4,082 bank branches at its 2007 peak and 3,021 in 2025, a quarter fewer than in 2010. Of 825 places with a branch in 1994, 113 have no bank office today, among them Mingo Junction, East Cleveland, Leetonia, Sheffield Lake and South Point. Ohio-based banks with Ohio branches fell from 428 to 162, and out-of-state banks now run a quarter of Ohio's branches, up from 3%.
Bank branches in Ohio, all banks, 1994-2025
Correction. While doing this work we found that one bank-branch table in our database holds only the branches of Ohio-headquartered banks, not every Ohio branch, so it leaves out PNC, Citizens and the other out-of-state banks. Figures here use the complete table. Any county figure built on the incomplete one is being checked and will be corrected.
Why it happens. Chain-wide pruning rather than local conditions: losses are no greater in rural counties than in cities once size is held fixed, and the biggest drops are old industrial counties (Marion -50%, Ashtabula and Trumbull -46%, Montgomery -41%). The towns that lost their only bank, though, are shrinking ones (-9.9% population since 1990, against +4.1% for towns that kept one).
How we checked. The popular claim that closings starve new business found no strong link: counties that lost more branches started only slightly fewer businesses afterward (weak, rank r -0.30 per resident). Counties with the most branches per person have long been the slowest to start businesses, a pattern already present in 1994.
Your county. Biggest losses 2010-2025: Marion, Ashtabula, Trumbull, Henry, Auglaize, Montgomery. Fewest branches per 10,000 residents: Union, Greene, Morrow; most: Ottawa, Harrison, Putnam, Mercer.
What it does not show. Net counts only. Credit unions and online banks are not in this data, so a town with no bank may still have a credit union.
Source: FDIC Summary of Deposits (all banks by branch location) and BankFind.
Two more verified facts: gambling and job-credit promises
Ohio's licensed gambling took in $3.5 billion in 2025, about $377 for every adult: casinos $1.03 billion, racinos $1.42 billion and sports betting $1.04 billion on $10.3 billion wagered. Sports betting, legal since January 2023, supplied nine of every ten new dollars, and casino revenue barely dipped when it arrived. Separately, the state's job-creation tax credits promised 207,182 jobs across 1,610 deals from 2008 to 2022; one in four deals that has run its course was cancelled, and where the promises landed shows no link with where jobs later grew (rank r 0.05).
How we checked. Every gambling total re-derives to the dollar from the Casino Control Commission workbooks; every job-credit count re-derives exactly from the state file. The job file records promises, not jobs delivered, so a null across counties does not show the credits did nothing.
Sources: Ohio Casino Control Commission; Ohio Lottery; Ohio Department of Development job-creation tax credit file.
Part
Water, weather and roads
What comes out of the tap, what falls from the sky, and who dies on the road.
Ohio's warming winters show up on the road in one place: motorcycle deaths
A month that runs 6°F or more colder than normal carries about half the usual number of motorcycle deaths; a month more than 6°F warmer carries 1.77 times as many. Car and truck occupants barely move. Across the years, deaths of riders in November through March went from 14.3 a year (1976-85) to 18.5 a year (2015-24) as those months warmed by about 4.4°F. The warmest cold season on record, 2024, had 34 rider deaths, the most ever; the two coldest, 1977 and 1978, had 9 and 10.
Road deaths in a month, relative to what that year and season would predict (1.0 = as expected), by how warm the month was against normal. Ohio, 1975-2024
Show the numbers
| Decade | Cold-season (Nov-Mar) rider deaths a year | Warm-season rider deaths a year | Cold-season mean temperature, °F |
|---|---|---|---|
| 1976-1985 | 14.3 | 186.0 | 31.9 |
| 1985-1994 | 13.4 | 146.1 | 34.5 |
| 1995-2004 | 8.4 | 114.0 | 34.9 |
| 2005-2014 | 13.7 | 153.2 | 34.4 |
| 2015-2024 | 18.5 | 175.8 | 36.3 |
Why it happens. Exposure. Riders ride when it is warm, and in a cold or snowy month everyone drives less and slower, so fatal crashes fall even when fender-benders rise. Motorcyclists are now about one Ohio road death in five (240 in 2023), so milder winters matter more than they used to.
How we checked. The pattern holds in 1975-99 and again in 2000-24, and county by county (57 of 88 counties). A separate weather record, the National Weather Service storm log, agrees: winter months with a recorded snow or ice storm ran at 0.965 of expected road deaths, other winter months at 1.023. Snow is not what kills. For all road users together, warmer winters move the yearly total very little; the change is on two wheels.
Your county. Among counties with 400 or more deaths, warm months run furthest above cold months in Fairfield (1.20 vs 0.79), Allen (1.20 vs 0.85), Belmont (1.16 vs 0.86) and Erie (1.00 vs 0.70); Franklin 1.08 vs 0.95, Cuyahoga 1.09 vs 0.99. In Ashland, Athens, Guernsey, Licking, Hancock and Logan cold months are no safer.
What it does not show. These are deaths, not deaths per mile driven: no monthly traffic counts are held, so part of a warm month's extra deaths is extra driving. Temperature is a county-month average, not the weather at any crash.
Sources: NHTSA Fatality Analysis Reporting System, Ohio 1975-2024 (71,543 deaths; 8,367 motorcyclists); NOAA nClimDiv county temperatures; NOAA Storm Events 2000-2024.
A six-inch month is when your county floods, and Ohio counties now get more of them
Months with 6 inches of rain or more are only 6.8% of county-months, yet they hold 47% of Ohio's federal flood-insurance claims and 58% of the dollars paid. A month with 6 to 8 inches brings 20 times the claims of a dry month; 8 inches or more, 80 times. Ohio counties logged 2,385 six-inch months in 1996-2025, against 1,525 in 1951-80, up 56%, and 86 of 88 counties had more.
Flood-insurance claims per county per month, by that month's rain in the county. Ohio, 1978-2025 (times the dry-month rate in brackets)
Show the numbers
| Rain in the county-month | Claims per county-month | Share of county-months | Share of claims | Share of dollars paid |
|---|---|---|---|---|
| Under 2 inches | 0.1222 | 21.9% | 4.9% | 1.7% |
| 2 to 3 inches | 0.1549 | 24.8% | 7.0% | 3.4% |
| 3 to 4 inches | 0.2596 | 22.7% | 10.7% | 6.9% |
| 4 to 5 inches | 0.5703 | 15.7% | 16.2% | 14.6% |
| 5 to 6 inches | 0.9337 | 8.1% | 13.7% | 15.0% |
| 6 to 8 inches | 2.3958 | 5.5% | 23.8% | 25.1% |
| 8 inches or more | 9.7242 | 1.3% | 23.6% | 33.4% |
Share of Ohio county-months with 6 inches or more of rain, by decade
Why it happens. A six-inch month is rarely one storm. It is saturated ground plus a multi-day rain, and that is when basements fill and creeks leave their banks. The 2010s were the wettest decade in this record for big months (2011, 2018 and 2019 carry it), and they were also the decade of the claims.
How we checked. Two records that share nothing agree on the threshold: FEMA's paid claims and the National Weather Service flood reports. Of 4,793 Weather Service flood events in Ohio counties since 2000, 42% fell in the 7.5% of county-months at 6 inches or more. Claims in dry months are mostly river or lake floods from somewhere else, such as the Lake Erie storm that hit Ottawa County in April 1982.
Your county. Most six-inch months in 1996-2025: Adams 49, Brown 48, Clermont 46, Highland 46, Hamilton 42, Pike 40, Clinton 40. Fewest: Ottawa 9, Hardin 12, Wood 13, Erie 14, Sandusky 14. Claims per 1,000 homes since 1978: Hancock 61 (August 2007 alone brought 863), Ottawa 48, Washington 46.
What it does not show. A claim needs a flood policy, and policy counts are not held, so county claim rates mix risk with how many people insure. 2020-25 is back to the old rate so far: this is a record, not a forecast.
Sources: NOAA nClimDiv county precipitation 1895-2025; FEMA NFIP claims 1978-2025 (27,894 claims); NOAA Storm Events 2000-2025.
Ohio's winters have warmed more than twice as much as its summers, and the hottest summer is still 1934
Ohio is about 2°F warmer than in 1901-1930, but the change is lopsided: winters are 3.0°F warmer, summers 1.2°F. Twelve of the twenty warmest winters since 1896 came in 1990 or later; only two of the twenty coldest did (2014, and 2015, which ties for twentieth). The two coldest winters on record are 1977 and 1978, the Blizzard of '78 years. 2024 was the warmest year on record, yet the hottest summer is still 1934.
Ohio winter and summer temperature, 10-year average, as degrees F above or below the 1901-1930 average
Show the numbers
| 1901-1930 | 1951-1980 | 1996-2025 | Change since 1901-1930 | |
|---|---|---|---|---|
| Winter (Dec-Feb) | 28.0°F | 28.0°F | 31.0°F | +3.0°F |
| Summer (Jun-Aug) | 70.5°F | 70.7°F | 71.8°F | +1.2°F |
| Whole year | 50.0°F | 50.1°F | 52.0°F | +2.0°F |
Correction. Our first draft said winters "warmed 0.9°F a decade since 1970". The deep check found that 1970 is the start year that gives the steepest slope, because it sits just before the coldest winters on record. Measured from 1950 it is 0.48°F a decade, and over the full record 0.23. We now quote only the change between 30-year periods.
Why it happens. The county records cannot say why the cold half of the year changed most. Two local explanations we tested did not hold: Lake Erie (warming shows no clear link with distance from the lake) and city growth (some of the fastest-warming counties since 1951, Mahoning and Trumbull, lost people).
How we checked. Everything here was recomputed by a second, independent analysis to within 0.01°F, and the pattern holds when measured from 1951-1980, when Ohio had many more weather stations. Winters swing a lot from decade to decade (the chart shows mild 1930s-50s and harsh 1960s-70s), but the latest 30 years are among the mildest 30-year stretches on record (about 31.0°F; the mildest earlier stretch, 1928-1957, averaged 30.0°F), and 2016-2025 averaged 33.0°F, 2°F milder than the mild 1948-57 decade.
Your county. Warming since 1951-80: Stark +2.6°F, Summit +2.5, Portage +2.5, Mahoning +2.4; least in Gallia and Meigs (+1.4). County figures from the early 1900s rest on few stations, so lean on the statewide numbers.
What it does not show. Average temperature, not extremes: this record has no daily highs or lows.
Source: NOAA nClimDiv county series (v1.0.0, file of 2026-09-04), area-weighted over 88 counties.
All of the fall in Ohio's road deaths since the 1970s came from people under 25
Ohio road deaths peaked at 2,281 in 1979. Deaths of people under 25 fell from 939 a year in 1975-79 to 232 a year in 2019-23, down 75%. Deaths of people 25 and over were 1,026 a year then and 1,018 now. Teen drivers killed went from 189 a year to 48; drivers 65 and over killed from 89 to 166. The median age of a person killed went from 25 to 43, and motorcyclists rose from 8% of deaths (1994) to 19% (2023).
Ohio road deaths a year, by age of the person killed, 1975-2024
Show the numbers
| Years | Under 25, a year | 25 and over, a year |
|---|---|---|
| 1975-79 | 939 | 1,026 |
| 2019-23 | 232 | 1,018 |
Why it happens. The timing fits seat-belt laws (1986), the 21 drinking age (1987), graduated teen licensing (1999) and safer cars, but the records give the timing, not the cause. Drinking has not moved since 2000: about a third of tested drivers killed were at 0.08 or above then and now.
How we checked. Ages are recorded for every death in every era, so the split is clean. Per Ohioan of all ages the 25-plus death rate did fall 9%, because there are more adults now; the count did not. Death certificates (County Health Rankings) agree with these federal crash records county by county.
Your county. Deaths per 100,000 residents, 2015-24: lowest Lake 4.7, Medina 6.7, Warren 7.0, Delaware 7.4; highest Harrison 31.2, Morgan 27.6, Vinton 27.2, Henry 25.6.
What it does not show. "Same count" is not "same risk". Deaths are counted where the crash happened, not where the person lived.
Source: NHTSA FARS, Ohio 1975-2024 (re-checked against the raw 2019 file: 36,355 U.S. deaths, NHTSA's published total).
The same road is deadlier in the country, and it is not drinking or seat belts
Rural roads carry 31% of Ohio's driving and 41% of its road deaths. Interstates are equally safe in town and country (0.48 deaths per 100 million miles), but a rural collector road is 1.8 times as deadly as the same class of road in a city, and a rural local road 2.8 times. By county, Morgan has 4.2 deaths per 100 million miles and Lake 0.53.
Deaths per 100 million miles driven, by type of road, rural and urban Ohio, 2022-24 deaths over 2023 miles
Why it happens. Speed and distance. 85% of rural fatal crashes are on roads posted 55 or more (31% in town), and the median time from crash to hospital is 51 minutes in the country against 31 in town. The extra rural danger holds for sober drivers and belted occupants alike, and rural Ohioans killed were less often unbelted (52%) than urban ones (57%).
How we checked. The county ranking repeats from one five-year half to the next (rank r 0.79) and survives re-assigning the miles driven with a different county method (0.995). We also tested whether one "beyond the water mains" rural Ohio explains it; it does not: private-well share has no link with road deaths per mile.
Your county. Most deaths per mile, 2015-24: Morgan, Vinton, Harrison, Meigs, Perry, Coshocton, Highland, Hardin, Adams, Champaign. Fewest: Lake, Medina, Union, Warren, Erie, Summit, Delaware. Longest median crash-to-hospital time: Brown 74 minutes, Pike 71, Henry 67.
What it does not show. County miles driven are a 2025 snapshot applied to ten years of deaths; small counties have wide ranges (Morgan's rate rests on 39 deaths).
Sources: NHTSA FARS; FHWA Highway Statistics 2023 (Table VM-2); ODOT 2025 traffic counts; hospital inventory (HIFLD).
Where your water is drawn, not your town, decides what is in it, and wet weather reaches the tap through the lake and the reservoirs
Raw water tested positive for algae toxin in 21% of samples at inland reservoirs, 10% on Lake Erie, 6% on inland rivers and under 1% on the Ohio River; Celina's reservoir in 757 of 767 samples, Cleveland's lake intake in 2 of 1,068. Treatment holds: 18 detections in 21,035 treated samples. A farm herbicide, atrazine, reached the treated water of 47 of 103 surface-water systems but 5 of 749 well systems. And after a wetter than usual season, chlorine byproducts rise at Lake Erie and reservoir systems, not at river or well systems.
Disinfection byproducts (HAA5) in treated water after the driest and wettest fifth of quarters, micrograms per litre above or below each system's own normal, 2012-2019
Show the numbers
| Summer | Maumee basin rain, Mar-Jul (inches) | Western Lake Erie intakes: raw samples with algae toxin |
|---|---|---|
| 2016 | 15.8 | 25% |
| 2017 | 23.6 | 62% |
| 2018 | 17.6 | 42% |
| 2019 | 24.7 | 43% |
| 2020 | 15.6 | 17% |
| 2021 | 18.5 | 47% |
| 2022 | 18.0 | 37% |
| 2023 | 15.3 | 42% |
| 2024 | 19.4 | 42% |
| 2025 | 20.5 | 42% |
Why it happens. Shallow, runoff-fed water (the western basin of Lake Erie, small reservoirs) carries farm chemicals, algae and the organic matter that chlorine turns into byproducts, and a wet spell adds more. A free-flowing river is different: it flushes in wet weather and concentrates in dry. Wells see none of it.
How we checked. From 84,000 federal lab results (2012-2019), each system compared with its own seasonal normal and trend: after wet quarters byproducts rose at 87% of 69 lake systems and 81% of 57 reservoir systems; well systems show nothing, a built-in control. Western-basin toxin detections rise with spring rain over the Maumee basin (rank r 0.55 over ten summers: 62% of samples in 2017's wet spring, 17% in 2020's dry one). Spring rain there is up 9% since 1951-80.
Your county. Toxin-prone intakes: Celina (Mercer), Alliance (Stark), Cadiz (Harrison), Struthers (Mahoning), Lima (Allen), Toledo, Oregon and Carroll on the western basin. Byproduct violations cluster in Fulton, Defiance, Putnam and Henry counties (19 of Ohio's 72 violating systems).
What it does not show. A detection is a lab result at the reporting limit, not a health dose; byproduct shifts are a few micrograms against an 80 microgram limit. The toxin evidence for rain rests on ten summers, and on inland reservoirs wet months lower the odds instead.
Sources: Ohio EPA microcystin results 2016-2026; EPA Six-Year Review 4 sample files 2012-2019; EPA SDWIS; USGS 2018 pesticide use; NOAA nClimDiv.
Eight plants supply half of Ohio, one Ohioan in six drinks water bought from someone else, and the private-well belt is the suburbs
Ohio has 1,125 community water systems, but counting the towns that buy from them, eight plants reach half of everyone on public water: Columbus and Cleveland about 1.4 million each, Cincinnati 860,000. 283 systems buy their water instead of treating it, serving 1.7 million people. At the other end, 429 tiny systems (mobile-home parks, subdivisions, camps) serve under 1% of Ohioans. Two-thirds of Ohioans on public water drink treated surface water; Lake Erie alone serves 2.7 million.
People reached by each of the eight largest water plants, counting the towns that buy from them
Show the numbers
| Ultimate source | People on public water | Share |
|---|---|---|
| Groundwater (wells) | 3,357,565 | 32.1% |
| Lake Erie | 2,727,890 | 26.1% |
| Inland rivers | 2,224,906 | 21.3% |
| Inland reservoirs | 1,125,163 | 10.8% |
| Ohio River | 1,019,659 | 9.8% |
Why it happens. Water follows the plant and the pipe, not the city limit. And wells are where the mains stop, which in Ohio is the commuter ring more than the back country: Summit, Stark, Medina, Geauga and Licking counties have the most people on wells.
How we checked. Every figure was re-derived from the federal water-system inventory by walking each purchase chain to the plant that treats the water. Toledo's reach (450,381) matches the 440,552 people the CDC counted under the 2014 do-not-drink advisory; county groundwater shares agree with USGS's independent estimates (rank r 0.91).
Your county. Lake Erie is essentially all public water in Lake, Erie, Ottawa, Cuyahoga and Lucas counties; the Ohio River 90% of Hamilton's. On wells: Geauga about 3 in 4 residents, Holmes and Carroll over 70%.
What it does not show. Served counts are reported by the systems and sometimes stale. Nobody tests private wells routinely, so the 1.3-1.9 million is a count, not a water-quality result. Cleveland Water alone holds 54% of Ohio's known lead service lines (144,193), and one line in four statewide is still of unknown material.
Sources: EPA SDWIS inventory and service-line inventories; the atlas purchase-chain research; USGS 2015 county water use; CDC MMWR 65(35).
Notes
How this was made, and what it does not say
How this was made
- Twelve research lanes, one per subject, each pulled its numbers straight from Cracked Buckeye's own databases and tested its first ideas. Then a second, independent analysis (a more capable model, working from scratch) re-derived every headline number, looked for the data fault behind anything surprising, tested each explanation against a source that shares no inputs, and joined subjects to find connections no single lane could see.
- Nothing was newly fetched for this page. Every figure is computed from records the entity already held on 5 October 2026, and the few published figures it compares against are named where they appear.
- Dollars across years are in 2024 dollars (consumer price index) unless marked otherwise.
- "Rank r" is a Spearman rank correlation from -1 to 1. Across 88 counties anything under about 0.2 cannot be told from chance; we call 0.2-0.4 weak, 0.4-0.6 moderate, above 0.6 strong.
What it does not say
- County and town patterns are not facts about any one person. That a county with more of X has more of Y does not mean the people with X are the people with Y.
- A pattern is not a cause. Where we say "consistent with", we mean the data fit an explanation without proving it.
- We name companies, utilities, banks and water systems; we name no private individual.
Where the work lives
- Every figure was computed by scripts kept with their results in Cracked Buckeye's own records, and every headline number was re-derived by a second analysis. Ask through Correct this page for any table behind a chart.
Cite
Cite this page as:
Cracked Buckeye, "What the data says about Ohio", https://crackedbuckeye.com/findings/ohio-one-level-down/, built 2026-11-14, accessed [date].
How to cite · Report a correction · Built 2026-11-14