Self-insurance is the practice of retaining a risk rather than transferring it to an insurer: the organization pays losses out of its own resources when they occur, instead of paying premiums to an insurance company. It is common for large employers' health plans, workers' compensation, vehicle fleets, product warranties and property damage below a certain size. A company with thousands of employees or vehicles experiences losses frequently enough that their total each year is fairly predictable, so it may be cheaper to pay them directly than to pay an insurer's premiums, which include profit, overheads and taxes.
Self-insurance is part of a wider risk management strategy. It ranges from simply having no insurance for minor risks, to formal programs with dedicated funds, third-party claims administrators and backup cover, to owning a captive insurance company that insures the parent group.
How self-insurance programs are structured
- Retention level: the business decides how much of each loss, and of total losses each year, it will bear itself. A high deductible on a commercial policy is a partial form of self-insurance.
- Stop-loss or excess insurance: most programs buy cover for losses above the retention, either per claim, such as any single medical claim over $250,000, or in aggregate, such as total claims above 125 percent of expected. This works in a similar way to stop-loss reinsurance for insurers.
- Administration: a third-party administrator often processes claims, while the employer funds them.
- Regulation: in the United States, self-funded employee health plans are generally governed by federal ERISA rules administered by the Department of Labor rather than by state insurance law. Self-insuring workers' compensation usually requires state approval and proof of financial strength, such as a bond or letter of credit.
Accounting for self-insured losses
Under US GAAP, self-insurance is accounted for under the loss contingency rules of ASC 450. A company may not record a general reserve for possible future losses simply because it is uninsured. Instead, it records a contingent liability and expense when it is probable that a loss has been incurred by the balance sheet date and the amount can be reasonably estimated. For programs with many claims, that includes claims already reported and an estimate for incurred but not reported claims, often prepared by actuaries using claims history. Losses that are only reasonably possible are disclosed but not accrued. IFRS reaches a similar result through the provisions standard, IAS 37.
Worked example
A retailer with 3,000 employees self-insures its health plan. It expects claims of about $36 million a year and buys stop-loss cover for any individual's claims above $300,000. During December, employees receive treatment worth an estimated $3.1 million, of which only $1.4 million of claims have been submitted by 31 December. Based on past lag patterns, the actuary estimates another $1.7 million of December treatment will be billed in the coming months.
At year end the retailer records the full $3.1 million as health insurance expense for December, with a liability, often described as a claims reserve, for amounts not yet paid. It does not record anything for illnesses that might occur next year. If one employee's claims reach $450,000 for the year, the retailer records a receivable of $150,000 from the stop-loss insurer.
Advantages and risks
Self-insurance can lower total costs, improve cash flow because money stays in the business until claims are paid, give the employer access to claims data and encourage safety programs. The risks are volatility and catastrophe: a bad year, a pandemic or one very large claim can produce losses far above expectations, which is why stop-loss cover is usually essential. Self-insurance is generally unsuitable for small businesses, whose losses are too unpredictable to budget for, and for risks that could threaten the company's survival, where transferring risk to an insurer remains the better choice. Self-insured losses also create earnings swings that analysts may need to adjust for when assessing contingent losses and underlying performance.