The industry and where the money goes
The regressivity finding is one of the most replicated results in public finance. The common misstatement of it — that poor people play more — is false.
A lottery is not legally a tax. Nobody is compelled to buy a ticket. But roughly a third of every ticket is diverted to the state before any prize is calculated (where each pound or dollar goes), and economists have spent forty years asking who bears that burden.
The finding is remarkably consistent, and it is almost always described wrongly.
Lottery spending is roughly flat in absolute terms across income bands. Income is not.
If a household earning $15,000 and a household earning $150,000 both spend about $450 a year on tickets, the first has spent 3% of its income and the second 0.3%. Nobody needs to play more for the burden to be ten times heavier.
That is the whole argument, and the empirical work is about establishing the premise.
The most detailed public dataset is Clotfelter, Cook, Edell and Moore's State Lotteries at the Turn of the Century, the 1999 report to the National Gambling Impact Study Commission. Their Table 11 gives annual lottery play per household, adjusted for known under-reporting:
| Household income | Annual lottery play per household |
|---|---|
| Under $10,000 | $520 |
| $10,000–24,999 | $505 |
| $25,000–49,999 | $464 |
| $50,000–99,999 | $301 |
| Over $100,000 | $338 |
Their own summary of it:
"Income has little relationship to lottery play overall up to $50,000, and drops off sharply at higher incomes. Hence lottery expenditures represent a much larger burden on the household budget for those with low incomes than for those with high incomes."
Melissa Kearney found the same shape with different data in State lotteries and consumer behavior, Journal of Public Economics, volume 89, issues 11–12 (2005), pages 2269–2299 (NBER working paper 9330):
"Average annual lottery spending in dollar amounts is roughly equal across the lowest, middle, and highest income groups. Reported annual expenditures are $125, $113, and $145, respectively. This implies that on average, low-income households spend a larger percentage of their wealth on lottery tickets than other households."
Kearney also found where the money comes from: households in the lowest income third funded lottery play through a 2.7% reduction in non-gambling consumption, against 0.5% in the middle third. Her conclusion is that "low-income households are financing their lottery gambling completely by a decline in consumption."
Take the 1999 per-household figures and divide by the midpoint of each income band. (The midpoints are our assumption, not the study's — the top and bottom bands are open-ended, so treat those two rows as indicative.)
| Household income | Assumed midpoint | Annual play | Play as % of income | Implicit tax at 50% takeout | Effective rate |
|---|---|---|---|---|---|
| Under $10,000 | $7,500 | $520 | 6.9% | $260 | 3.5% |
| $10,000–24,999 | $17,500 | $505 | 2.9% | $253 | 1.4% |
| $25,000–49,999 | $37,500 | $464 | 1.2% | $232 | 0.6% |
| $50,000–99,999 | $75,000 | $301 | 0.40% | $151 | 0.20% |
| Over $100,000 | $150,000 | $338 | 0.23% | $169 | 0.11% |
The last column is the number that matters. Treating the takeout as an implicit excise, the effective rate falls from about 3.5% of income to about 0.11% — a factor of roughly 31. No formal tax in any developed economy is that steeply regressive.
Clotfelter and Cook established the result in Implicit Taxation in Lottery Finance, National Tax Journal, volume 40, issue 4 (1987), pages 533–546: "We find that the implicit tax is regressive in virtually all cases." They developed it at book length in Selling Hope: State Lotteries in America (Harvard University Press, 1989) and summarised the economics in On the Economics of State Lotteries, Journal of Economic Perspectives, volume 4, issue 4 (1990), pages 105–119.
Since then:
| Study | Data | Finding |
|---|---|---|
| Hansen, Public Finance Quarterly 23(3):385–398 (1995) | Colorado instant games, county level | Regressive; instants behave as an inferior good |
| Price & Novak, National Tax Journal 52(4):741–751 (1999) | Three Texas games, ZIP level | All three "highly regressive"; the instant game an inferior good |
| Oster, National Tax Journal 57(2.1):179–187 (2004) | Powerball sales by jackpot size | "Significantly less regressive at higher jackpot sizes" |
| Combs, Kim & Spry, Applied Economics 40(1):35–39 (2008) | Seven Minnesota games, ZIP level | Every product regressive on the Suits Index; the instant scratch product the most regressive |
| Ghent & Grant, National Tax Journal 63(2):253–268 (2010) | South Carolina, three game types | All three regressive, but "may not be as regressive as suggested by the earlier literature" |
Clotfelter and Cook's own summary in 2007 is about as flat as academic writing gets: "Study after study confirms that expenditures on lotteries represent a larger share of the incomes of low-income households… the regressivity charge sticks because the evidence to support it is overwhelming" (Ends and Means in State Lotteries).
Note also what varies: which game matters enormously. Instants are consistently the most regressive product and big-jackpot draw games the least. That maps directly onto the payout structures set out in return to player by game.
The regressivity finding is not that poor people are more likely to play. That claim is false, and it is repeated constantly.
Gallup's 2016 survey found lottery participation rising with income: 40% of adults in households under $36,000 had bought a state lottery ticket in the past year, against 56% at $36,000–89,999 and 53% at $90,000 or more (Gallup).
Clotfelter and colleagues found the same in 1999: participation was 48.5% for households under $10,000 and 61.2% for $50,000–99,999. But average spend among players went the other way — $597 a year in the lowest band against $225 in the $50,000–99,999 band.
So the correct statement is: lower-income adults are somewhat less likely to play, and those who do play spend considerably more, both absolutely and as a share of income. The burden is concentrated in a minority of heavy players. Clotfelter's Table 12 describes the heaviest 20% of purchasers: 61.4% male against 48.5% of adults, 20.3% high-school dropouts against 12.3%, and 9.7% from households under $10,000 against 5.0%.
The argument that lotteries are not really a regressive tax rests on three points, and they are not frivolous.
1. It is voluntary. A tax is compulsory; a lottery ticket is a purchase. Someone who buys nothing pays nothing. Applying the language of tax incidence to a discretionary consumer good is a choice, and you could equally analyse cinema tickets this way.
2. Consumers get something. The counter-argument is that the product is entertainment, plus the option value of a large win. On this view the "implicit tax" is just a price, and prices for many goods take a larger share of small incomes without anyone calling them regressive taxes.
3. Operators are not targeting the poor. This one is important because it is so often assumed. Clotfelter and Cook state that the claim "that lottery agencies direct their marketing at the poor… generally does not hold up to scrutiny." Their finding is about spending patterns, not about predatory placement.
4. Big jackpots recruit richer players. Oster's 2004 result — the game is "significantly less regressive at higher jackpot sizes" — means the incidence is not fixed. A game whose peak jackpots keep growing draws in occasional, better-off players. That is one genuine side effect of jackpot inflation.
The empirical claim is settled and the normative one is not.
Settled: absolute lottery spending is close to flat across income bands, so the implicit tax embedded in the takeout falls far more heavily, as a share of income, on poorer households. Every study since 1987 finds this, and instant games are the most regressive product.
Not settled: whether a voluntary purchase should be analysed as a tax at all, and whether the entertainment value offsets the transfer. That is a question about values, and no dataset will resolve it.
What can be said without argument is that the money is real, it is large, and it moves in one direction. The next questions are who is actually playing — who plays the lottery — and what the money buys once it arrives, in the good causes claim examined.
If you want to know what your own play costs over time, the lifetime spend calculator and the invest-instead calculator do that arithmetic directly.
Last verified: 2026-08-29