Formal self-insurance is the deliberate decision to pay for otherwise insurable losses out of one's own money.[citation needed] This can be done on a formal basis by establishing a separate fund into which funds are deposited on a periodic basis, or by simply forgoing the purchase of available insurance and paying out-of-pocket. Self-insurance is usually used to pay for high-frequency, low-severity losses. Such losses, if covered by conventional insurance, mean having to pay a premium that includes loadings for the company's general expenses, cost of putting the policy on the books, acquisition expenses, premium taxes, and contingencies. While this is true for all insurance, for small, frequent losses the transaction costs may exceed the benefit of volatility reduction that insurance otherwise affords.[citation needed]
Business insurance can take a number of different forms, such as the various kinds of professional liability insurance, also called professional indemnity (PI), which are discussed below under that name; and the business owner's policy (BOP), which packages into one policy many of the kinds of coverage that a business owner needs, in a way analogous to how homeowners' insurance packages the coverages that a homeowner needs.[31]

At the most basic level, initial ratemaking involves looking at the frequency and severity of insured perils and the expected average payout resulting from these perils. Thereafter an insurance company will collect historical loss data, bring the loss data to present value, and compare these prior losses to the premium collected in order to assess rate adequacy.[24] Loss ratios and expense loads are also used. Rating for different risk characteristics involves at the most basic level comparing the losses with "loss relativities"—a policy with twice as many losses would therefore be charged twice as much. More complex multivariate analyses are sometimes used when multiple characteristics are involved and a univariate analysis could produce confounded results. Other statistical methods may be used in assessing the probability of future losses.

Let's use the aftermath of a major storm to illustrate the differences between collision and comprehensive. Within that storm, let's consider two hypothetical events: First, a heavy telephone pole was blown down and fell on your truck, or second, you swerved to avoid a falling tree and wound up crashing into a guardrail. In the first event, you couldn't control when or why a tree fell on your car. This kind of accident would get reimbursed under your comprehensive policy. In the second situation, you were driving the car and ultimately swerved into the guardrail. This makes it a collision, and collision insurance pays for the damages.