How long does a redemption window take to resolve?
Minnesota's statutory redemption period is six months from the sheriff's sale.
Half of tracked windows were still unresolved at eighteen months.
That gap between the statute and the observed timing is the finding, and it matters to anyone holding capital in a sheriff's certificate on the assumption that six months means six months.
The curves#
Across 1,336 tracked Minnesota mortgage foreclosure windows, followed from the sheriff's sale date through recorded deeds and county ownership records:
| Time from sheriff's sale | Reached a foreclosure sale | Owner sold during the window |
|---|---|---|
| 3 months | 1.2% | 4.4% |
| 6 months | 3.2% | 11.7% |
| 9 months | 11.0% | 13.0% |
| 12 months | 33.0% | 15.0% |
| 18 months | 51.0% | 15.0% |
Two different shapes, and the difference is the useful part.
Owner sales cluster early and then stop#
15.0% at one year. 15.0% at eighteen months.
The curve is flat across that entire second half. An owner who sells during the redemption window does it inside the first year, and the mechanism is straightforward: the statutory period has expired long before, and what is happening after month twelve is a process being extended by other means — bankruptcy, litigation, negotiation — in which a voluntary sale is no longer the live option.
The practical reading for anyone in a redemption window: the option to sell is a first-year option. It does not stay available while the process drags.
Foreclosure resolutions come later, and slowly#
3.2% at six months, against a six-month statutory period.
That is the number that surprises people. At the point the statute says the window closes, fewer than one in twenty-five had actually resolved as a foreclosure sale.
Then it accelerates: 11.0% at nine months, 33.0% at twelve, 51.0% at eighteen.
What is happening in between:
Sale postponements. Common, and each one restarts a schedule.
Bankruptcy filings. An automatic stay halts everything immediately, and the window does not resume until the stay lifts.
Loss mitigation. Servicers negotiating modifications, forbearance or short sales while the foreclosure runs in parallel.
Litigation. Contested foreclosures, title defects, service problems.
Recording lag. The event happens before the document appears.
None of these is unusual. Collectively they mean the statutory period describes when the owner's right expires, not when the matter concludes.
What this means for capital#
A buyer at a Minnesota sheriff's sale is committing money for longer than the statute implies.
| Assumption | Reality |
|---|---|
| Six months, then take possession | 3.2% resolved at six months |
| A year at the outside | 33.0% resolved at one year |
| Certainly resolved by eighteen months | 51.0% — half still open |
And the holding cost is real. Property taxes advanced, insurance on an interest rather than an asset, and capital unavailable for other purchases throughout. On a certificate bought expecting a six-month turn and still open at month twenty, the opportunity cost is the largest line and it appears on no statement.
Combined with the redemption rate, the picture is sharper still. Of 326 resolved windows, 33.4% ended with the owner redeeming — so a third of certificates return capital plus statutory interest rather than a property, after a wait considerably longer than six months.
The correction that produced these numbers#
This section describes an error in our own earlier figures, because the correction is instructive and because publishing only the corrected number without the correction would be dishonest about how the estimate was reached.
The timing table was previously computed over 1,671 windows. The extra 335 were tax forfeiture windows under Minn. Stat. ch. 281 — a three-to-five year clock from judgment — pooled with ch. 580/582 mortgage redemption windows on a six-month clock.
Every one of those forfeiture windows was censored and always would be. The outcome checker detects mortgage foreclosure outcomes. A tax forfeiture window cannot reach a foreclosure sale, so it entered the risk set, remained there for an average of 921 days, and never experienced the event.
In a Kaplan-Meier estimate that depresses the hazard at every step, because the risk set is inflated by observations that cannot fail.
Removing them moved "reached a foreclosure sale within one year" from 18.8% to 33.0%.
The lesson is general: two statutes, two clocks, one estimate produces a confident number that is wrong by a wide margin. Pooling populations that cannot experience the same event is the most consequential error available in this kind of analysis, and it does not announce itself.
Reading a survival curve properly#
Worth a short note, because these figures are frequently misread in ways that change the conclusion.
The percentages are cumulative, not periodic. 33.0% at one year means a third of all windows had reached a foreclosure sale by that point, including the 11.0% that had done so by nine months. It is not 33% during the twelfth month.
The two columns are competing outcomes, not independent. A window that ends in an owner sale can no longer end in a foreclosure sale. Adding the columns gives the share resolved one way or the other; it does not give a probability of either in isolation.
The remainder is not "still in redemption". At eighteen months, 51.0% had reached a foreclosure sale and 15.0% an owner sale, leaving roughly a third unaccounted for. Those are windows we had not observed resolving at the time of the last check — some redeemed, some are still running, and some resolved in a way the detection did not catch.
Later points are less precise. Each estimate rests on the windows still at risk at that moment, and by eighteen months the risk set is much smaller than at three months. The eighteen-month figure carries a wider uncertainty than the six-month one, and the curves should be read as steadily less confident from left to right.
A flat segment means the event stopped happening, not that data stopped arriving. The owner-sale curve flat from twelve to eighteen months is a real finding: windows were still being observed during that period and owner sales were not occurring in them.
What would change these numbers#
Stating the sensitivities is part of publishing an estimate.
Working the pending backlog. Windows currently unresolved will resolve, and each resolution updates the curve. The direction is not predictable in advance — if pending windows resolve as foreclosure sales the curves rise, and if they turn out to have been redeemed they do not.
Better detection. The checker finds mortgage foreclosure outcomes through owner-of-record changes and recorded sales. Outcomes it misses appear as censored observations and depress the curves. Every improvement in detection has so far moved figures up rather than down, which suggests the current curves are if anything conservative.
County coverage. Counties are checked at different levels of completeness. A county checked less thoroughly contributes proportionally more censored observations.
Statutory change. Minnesota's redemption framework has been stable, but a change to the period or to abandonment provisions would break comparability with this series.
More counties. Broader coverage would improve precision and might shift the central estimate if the counties added differ systematically from those already tracked.
We will republish when the figures move materially, with the previous version and the reason for the change, rather than quietly updating a number others may have cited.
Method#
Data. Sheriff sale notices from county sheriff feeds and published legal notices, joined to a parcel spine built from county assessor and GIS records across Minnesota counties.
Window construction. Opens at the sheriff's sale date. Expires at the statutory period — six months ordinarily, twelve for certain agricultural property and older mortgages, five weeks where formally declared abandoned.
Events. Two, analysed separately because they are competing outcomes: foreclosure sale, meaning the property transferred to a lender, servicer or third party; and owner exit, meaning the pre-foreclosure owner sold during the window.
Censoring. Windows with no observed event are censored at the last date the record was checked. Kaplan-Meier estimation handles this without discarding the observation or assuming an outcome.
Exclusions. Tax forfeiture windows, per the correction above. Superseded and duplicate tracker rows.
Cross-check. The estimates were computed twice with independent
implementations, agreeing to four decimal places — owner_exit survival at 365
days 0.8504, foreclosure_sale 0.6697.
Limitations#
Minnesota only, under a six-month statutory redemption. Not transferable.
Right-censoring is heavy. Half the windows are unresolved at eighteen months, so the later portion of each curve rests on a smaller risk set and is correspondingly less precise.
Detection lags reality. A recorded deed appears after the event, so the curves are shifted slightly later than the underlying truth.
Ownership re-checks run on a schedule, not continuously, so event dates carry granularity of whatever the check interval was.
Redemption timing is not measurable at all. Redemption is inferred from the absence of a post-expiry transfer, and an absence has no date. The curves above measure time to foreclosure sale and time to owner sale — both of which leave a positive record — and cannot measure time to redemption.
Why the curves rather than a model#
A survival model was fitted on this data and is not published as a predictive tool.
Concordance index of roughly 0.66 to 0.70 after the statutory tracks were separated, down from 0.80 before — most of the pre-correction performance came from the artefact described above rather than from signal.
At 0.66, the model ranks individual windows barely better than chance. A published curve with a stated sample size answers the same question honestly and does not imply a precision the data does not support.
A model will be published here when one demonstrably beats the curve it would replace, and the comparison will be shown.
Citation#
These figures may be cited freely. Please include the sample size — a rate without its count is not a finding.
Suggested form: Govire, observed Minnesota redemption window timing, n=1,336 tracked windows, August 2026.
If you are a researcher or journalist and need a cut not published here, the underlying windows exist and the question is usually answerable. Access for research and reporting use is free.