Keeping Up With the County

Money raises how Americans rate their lives — that much is old news. The subtler question, the one Erzo Luttmer pressed in 2005, is whether other people's money lowers it. If you judge your life partly by comparison, then being surrounded by wealthier neighbors should sting, even when your own bank balance hasn't moved. Luttmer found exactly that: holding a household's own earnings fixed, higher average earnings in its local area predicted lower reported well-being. Neighbors, in his phrase, can be negatives.

This piece runs that test inside the United States, on a single evaluative question asked of millions of people: the Cantril ladder, where 0 is the worst possible life and 10 the best. The reference group is the county — specifically its median household income, a fixed profile of the place attached to every respondent who lives there. The trick is to hold each person's own income constant and ask what the surrounding affluence adds or subtracts.

The honest answer is that there is no single answer. The comparison penalty is real, but it lands almost entirely on the middle. The rich barely register it. And the poor, if anything, do slightly better among wealthier neighbors — the amenities and opportunity of a richer place outrunning any sting of comparison. One income, three different stories.

Read the lines from left to right and the disagreement is the whole story. The low-income line drifts gently upward: poorer Americans report a slightly higher ladder in richer counties, a rise of about six hundredths of a point from the poorest tenth of counties to the richest. The middle-income line does the opposite, sliding down by about thirteen hundredths — small in absolute terms, but the steepest and most orderly of the three. The high-income line barely moves at all.

These are modest gradients. Thirteen hundredths of a ladder point is not a life upended; it is a faint, consistent tilt visible only because there are two million people standing on the line. The point is not the size of any one gap but the sign: two of the three lines lean the way Luttmer predicted, and one leans the other way. The averages by tercile tell the same story more coarsely — middle earners fall from 7.14 in the poorest third of counties to 7.07 in the richest, while low earners rise from 6.35 to 6.40.

Hold the paycheck still and the question becomes: do the neighbors lift you or shrink you? For the middle, they shrink you. For the poor, they lift.

02The comparison effect, net of everything

A raw average can hide a confound. Richer counties have older residents, more college graduates, different racial composition, and they skew metropolitan — any of which could move the ladder on its own. So the real test is a regression that holds those constant and asks what is left for county affluence to explain. The model predicts the ladder from county median income (logged, then standardized) alongside a finer six-bracket measure of the person's own income, plus age, education, race, sex, and an urban-rural code. Run once for everyone, then separately inside each income band.

Overall, the coefficient is negative and easily significant: each additional standard deviation of log county income — roughly the distance from a $43,000-median county to a $71,000-median one — is associated with about seven thousandths of a ladder point lower, holding own income and demographics fixed. That is the comparison effect, surviving the controls. But the average buries the split.

Now the divergence is sharp. For middle earners the coefficient is about −0.027, with a tight interval far from zero — the clearest comparison penalty in the data, and consistent with the downward slope in the first chart. For high earners it is essentially zero (−0.000, not distinguishable from no effect). And for low earners the controlled estimate is a faint positive +0.003 — not significant on its own, but pointing the opposite way and matching their upward raw slope.

This is the reference-group hypothesis with a boundary drawn through it. The penalty is not universal; it is a property of where you sit in the distribution. The middle is the group most exposed to the comparison — high enough to feel the gap with wealthier neighbors, not high enough to be those neighbors. The rich are already at the top of the local ladder and shrug. The poor are far enough below everyone that one richer street over registers more as amenities — better services, jobs, schools — than as a wound.

03Does inequality matter on its own?

County average income is one thing; how unequally that income is spread is another. Alesina, Di Tella and MacCulloch found that inequality itself depresses happiness — more so in Europe than America, and uneven across the income ladder. Adding the county Gini coefficient to the model, alongside mean income, lets each one compete.

Inequality carries an independent, negative association overall — about four thousandths of a ladder point per standard deviation, net of county mean income. But it splits by income just as the mean did, and arguably more starkly. For middle earners more local inequality reads negative (−0.025); for low earners it flips firmly positive (+0.016), the strongest sign-reversal in the analysis. Toggle the first chart to county inequality and the low-income line climbs visibly with the Gini, while the middle line goes flat. The poor, it seems, are not made worse off by living somewhere unequal; they may be living where the opportunity — and the higher-income jobs — actually are.

04What this does, and doesn't, show

Several cautions hold the finding in place. First and largest: county median income is entangled with the cost of living. Expensive places have high median incomes almost by definition, and a dollar buys less there; some of the middle's comparison penalty could be a budget squeeze wearing a sociologist's costume. The data here cannot separate the two, so every number is an association, not a causal effect.

Second, people sort. Richer households move to richer counties, so within any own-income band the residents of expensive places differ from those of cheap ones in ways no survey fully captures. Holding own income fixed narrows that gap but does not close it. Third, the county profile is a static snapshot — a single ~2018 vintage joined to interviews stretching back to 2008 — so it describes a place's character, never its trajectory; nothing here is a change over time. The county assignment itself comes from an approximate ZIP-to-county mapping. And the Gallup sample carries no survey weight, so every figure is unweighted and describes patterns in the sample rather than calibrated national levels.

Finally, the effects are small. A tenth of a ladder point will not show up in any one person's life. What the two million interviews buy is confidence in the direction and in who bends which way — a middle that quietly measures itself against its neighbors, a top that doesn't bother, and a bottom for whom a richer county is, on balance, a better place to be.