Insure vs. Self-Insure
The micro-insurance trap: does the math justify the premium?
All figures are in today's money: every rate on this page is real, i.e. above inflation. Why?
The Item & Risk
The Insurance Policy
Your Financial Buffer
Max you can comfortably pay out of pocket today without financial stress.
| Metric | Self-Insure | Insured | Advantage |
|---|---|---|---|
| Expected Cost Probability-weighted average total cost over 3 years | €225 | €470 | €245 |
| Break-Even insurance wins (15.6% likely) | – | – | 2+ breaks |
How likely is each outcome?
Breaks over 3 years follow a binomial distribution. Insurance only pays off at 2+ breaks, which is about 15.6% likely. The 3+ bar is priced at exactly 3 breaks.
Percentages show how likely each break count is.
- Expected Value (EV): with a 25% annual risk over 3 years, you expect 0.8 incidents, costing €225 out of pocket.
- Total premiums over 3 years: €432. Even in the lucky scenario (0 breaks), you've already paid that.
- The Peace of Mind Tax (€245): insurers price premiums above EV to cover staff, offices and profit. The gap is your tax for certainty.
- Binomial risk: chance of 0 breaks is 42.2%, 1 break 42.2%, 2 breaks 14.1%, and 3+ breaks 1.6%.
- Insurance wins at 2+ breaks: P(K ≥ 2) = 15.6%. Below that, self-insuring is cheaper (or the incident cost exceeds your financial buffer, in which case buy anyway).
Expected cost, self-insured
evSelfInsured = expectedIncidents × 300 expectedIncidents = p × n
- p: probPerYear / 100
- n: lifespanYears, rounded to the nearest whole year
- incidentCost: cost of one incident
Expected cost, insured
evInsured = totalPremiums + expectedIncidents × min(50, incidentCost) totalPremiums = 12 × n × 12
- premiumMonthly: monthly premium
- deductible: deductible per claim (you never pay more than the damage itself)
Peace of mind tax
- evInsured: see Expected cost, insured above
- evSelfInsured: see Expected cost, self-insured above
Break-even incident count
- incidentCost − min(deductible, incidentCost): what insurance saves per incident; with nothing saved there is no break-even
Chance of exactly k incidents
- n: lifespanYears, rounded to the nearest whole year
- p: probPerYear / 100
Assumptions behind every figure: how this site models the market →
Every model leaves things out. Here is what this one does not see:
- Only one incident per year is possible in this binomial model. Real damage is closer to a Poisson process (a bad year can produce more than one claim), so the model understates real claim counts and, with them, the case for buying cover.
- The chance of an incident and its cost are held flat for every year of the item's life: no rising failure rate as it ages, no repair-cost inflation, and no chance it is lost or stolen rather than merely broken. Real items usually get more fragile with age, so this understates the true cost of self-insuring in the later years and overstates it early on.
- Premiums and payouts (in today's money) are simply summed, with no time value of money. A premium paid in year one is treated identically to one paid in year ten, and the model never asks whether that same premium money, left invested instead, would have grown while waiting for a claim. That understates self-insuring's real advantage.
- No claim denial, exclusion or insurer-side friction is modelled. The insured cost line assumes every valid claim pays out cleanly at premium plus deductible, when contested claims and policy small print routinely change the real payout. So the model is too generous to insurance, understating self-insuring's real relative advantage.