The tax sale pays 12%. Unless you bid.

In October 2025, someone bid $105,000 at the Oconee County tax sale for a parcel that owed $3,656.60 in back taxes.

On January 6, 2026, the owner redeemed it. They paid $6,806.60.

Not $105,000. Not anything close. The owner got the property back for the taxes plus $3,150, and the bidder got their money returned with three percent for their trouble. The parcel was never for sale. It was collateral on a loan the bidder did not know they were making.

Winning a parcel does not mean owning it. It starts a twelve-month redemption clock, during which the owner can pay what they owe plus interest and keep the property. Only if that clock runs out do you get a deed.

So we looked at every parcel in Oconee and Anderson counties where somebody competed, meaning the winning bid came in above the opening. 2,131 parcels, and for every one of them the county itself told us whether the owner redeemed. Not our guess. Their record.

It climbs the whole way. Bid under $500 and the owner takes it back about a quarter of the time. Bid twenty-five thousand or more and it is four times in five.

Those are two different counties, and they are not overlapping much: Oconee's middling bid is around $750 and Anderson's is around $9,000, so Oconee is carrying the cheap end of that table and Anderson the expensive end. But look at the one place they meet. Between $5,000 and $10,000, Oconee redeems 66.9% and Anderson 67.0%. Two counties, two separate records requests, the same answer to two significant figures. That is why we are comfortable putting them in one table.

The reason is not complicated once you see it. A high bid means a valuable parcel. A valuable parcel means an owner with equity worth defending, or a lienholder with a reason to step in and defend it for them. The parcels worth having are the parcels somebody is already fighting for.

Which means the thing you most want to win is the thing you are least likely to keep.

A third county, measured a completely different way, says the same thing. Pickens does not report its redemption outcomes, so we reconstructed them by tracing deed records instead, and that method climbs from 55% in the cheapest quarter to 91% in the dearest. Three counties, two methods, one direction. We keep Pickens out of the table above anyway, and the data section explains why.

Here is where it turns from disappointing into something worse.

South Carolina sets redemption interest on a schedule. Redeem in the first three months and the owner pays 3% of the bid. Months four through six, 6%. Months seven through nine, 9%. The final quarter, 12%. That 12% is the number you see advertised.

But Section 12-51-90 caps it. In every redemption, the interest owed cannot exceed the opening bid, and under Section 12-51-55 the opening bid is only the delinquent taxes, penalties, and costs.

Read that again, because it is the whole game. The interest is a percentage of your bid. The cap is a fixed dollar amount set by the taxes owed. Bid a little over the taxes and the percentage governs. Bid far over the taxes and the cap governs, and your rate collapses toward nothing.

We checked it against 1,059 real redemptions in Oconee, using the amounts the county actually collected.

The return peaks in the one-to-five-thousand-dollar band, at 11.18%, close to the statutory maximum. Above that it falls off a cliff. Investors bidding six figures earned 5.41%, less than half the advertised rate, because seven out of ten of them ran into the cap.

Go back to that $105,000 parcel. The bidder earned $3,150. Had they bid $4,000 on the same parcel and it redeemed on the same day, they would have earned $120 on $4,000, which is the same 3%. The extra $101,000 they put up bought them nothing at all. It sat with the county for three months and came back with nothing to show for it.

The under-$1,000 band earns less than the band above it for a different reason: those parcels tend to redeem early, and early redemption pays 3%, not 12%. That is the clock, not the cap.

Line those two up and there is no bid size at which South Carolina hands you a valuable property cheaply. Bid big and you probably get 5% instead of the property. Bid small and your rate looks better, but the first table shows what happens at the bottom of the market: the owner walks away more than half the time, and you are going to end up owning things.

The statute and the market close from both sides. That is not a loophole waiting to be found. Tax sales exist to collect taxes, not to transfer real estate to investors, and they are good at their job.

Separate out the sales where the winning bid was exactly the opening bid, meaning nobody competed at all, and the pattern breaks completely.

Flat. Low. Price stops predicting anything. These parcels are acquired 79 to 88% of the time, and that is where essentially all of the real estate in a South Carolina tax sale ends up.

Which sounds like the answer until you ask the obvious question: why did nobody bid?

Sometimes nobody was looking. Often the parcel is landlocked, or sits in a floodway, or carries a structure that has to come down, or has a title problem that costs more to fix than the land is worth. And when nobody bids at all, the county's own Forfeited Land Commission takes it. A meaningful share of this population is not a missed opportunity. It is inventory the county got stuck with.

The tax sale reliably hands out property, and the property it reliably hands out is the property the market already declined.

We are not saying tax sales are a bad business. We are saying the business is not the one that gets advertised.

Every rate in the second table is what you earn if the parcel redeems. If you came for the interest, that is only half a number. The other half is how often you actually get paid instead of getting a parcel. Multiply them together and you get what a bid is really worth as a loan. Here is that math, for contested parcels in Oconee.

There is no row that pays 12%. The best expected cash return on the board is 5.66%, and you reach it by bidding a few thousand dollars on a parcel somebody else also wants. Bid less and the rate looks better but you rarely collect it, because at the bottom of the market the owner walks away and hands you the parcel instead. Bid more and the cap eats you.

That last column treats a parcel you did not want as worth nothing, which is not quite fair. It might be worth a great deal. But that is the point: the moment your return depends on what the parcel is worth, you are not lending anymore. You are buying, and you had better know what you bought.

And if you are buying, you are not competing on price. You are competing on whether you can tell which of the parcels nobody wanted is a genuine sleeper and which is a trap you will spend three years regretting. That is a diligence problem, not a bidding problem, and no amount of capital solves it. The sale calendar will tell you what is coming. It will not tell you which ones are worth having.

The 12% is real. It is just not for sale to the highest bidder. It is close to the exact opposite.

Everything above comes from parcel-level records for two South Carolina counties. More of what we have found is on the findings hub.

Three counties, and not all evidence is the same kind. Oconee (2015 to 2023) and Anderson (2023) both answered records requests: their Delinquent Tax Collectors stated, parcel by parcel, whether the owner redeemed. Every row has an answer and it came from the people who took the money. Those two counties, and only those two, are in the tables above. Oconee also reported the amounts collected, which is why the return analysis is Oconee alone. Nobody else's file carries what the owner actually paid.

Pickens is corroboration, not evidence we mix in. Pickens does not report outcomes, so we reconstructed them from deed records: no tax deed in the chain, so presumably redeemed. That resolves only 604 of 1,037 parcels and is biased upward by construction, because a deed we failed to find reads as a redemption. It agrees on the direction, which is what corroboration is for, and it is pooled into nothing.

The counties look wildly different, and that is the point. Oconee redeems 36.7% of its sold parcels. Anderson redeems 65.2%. Pickens 73.0%. Someone reading only those three numbers would conclude Oconee is where you go to actually get property. What is really going on is that the counties sell different things, and the redemption rate follows the price exactly as the rest of this article says it should:

Oconee sells $750 parcels and redeems 37%. Anderson sells $9,000 parcels and redeems 65%. Pickens sells $10,000 parcels and redeems 73%. The county rate tracks what is on the block, not some local habit of paying or not paying taxes. This is also the cleanest read we have on how far to trust Pickens: Anderson sits at almost the same price point on a county-reported number and comes in about eight points lower, which is roughly the size of the upward bias you would expect from counting missing deeds as redemptions, plus whatever is genuinely different about the two places. We cannot split those two, and we are not going to average three counties into a single South Carolina rate that would be true of none of them.

What we cut and why. Sale years are included only once the twelve-month redemption window plus recording lag has fully run, so recent sales are excluded rather than counted as unredeemed. That is why Anderson contributes one year and not three. Parcels settled before a bid was struck have no bid, so they cannot sit on a price curve and are not on one.

Three counties is not a state. South Carolina has 46. The shape has now held three times, in three counties, under two different measurement methods, and the two counties that overlap on price agree to within a tenth of a point. That is why we believe it. It is not proof that it holds in the other 43, and we will say so when we have them rather than pretending we already do.

Data as of 2026-07-16.