Ratemaking Overview. Generally, for private passenger automobile and residential property insurance ratemaking, the ratemaking analysis is performed by coverage by state. That is, a rate filing contains a number of ratemaking analyses. For each major coverage, there is an analysis of the average statewide rate change.

Historical Premium. Written premium is the total premium generated from the sale of policies during a given time period. Earned premium is the amount of premium booked by an insurance company due to the passage of time and that would not be returned if the policy is cancelled. For instance, assume that a company sells a one-year homeowners policy for $100 on July 1, 1999. For 1999, its written premium would be $100 and its earned premium would be $50. For 2000, its written premium would be $0 and its earned premium would be $50. Historical premiums are adjusted to future premium in two steps: adjusting to current rate levels and premium trend.

Projecting Future Premium - Current Rate Level Factors. The first step is to bring historical premiums to current rate levels. For example, if the rates for a particular coverage increased by 10% on January 1, 1997, then the historical 1996 premium must be increased by 10% to reflect the premium that would have been collected at current rate levels. This is important because any rate change indication is applied to current rate levels.

Projecting Future Premium - Premium Trend. The estimation of future premiums may also require the application of a premium trend factor. Average premium per exposure may change for a variety of reasons, most of which affect physical damage coverages. Physical damage coverages are related to the value of the vehicle being insured because the coverage is for the actual cash value, not the replacement value, of the vehicle. As consumers trade in older cars for newer cars, insurance companies gain more premium, all other factors constant. Many insurers increase the amount of coverage in residential property policies automatically each year to reflect inflation in construction costs. Another factor affecting auto physical damage and residential property premiums is changes in deductibles chosen by consumers. As consumers move to higher deductibles, insurance companies collect less premium, all other factors constant. A factor affecting auto liability coverages is changing increased limits selected by consumers or required by law. A shift by consumers to higher limits means additional premium for the insurance company (as well as additional exposure). Factors affecting all coverages are shifts in the distribution of consumers among risk classifications, such as increasing or decreasing numbers of consumers in higher-rate rating territories or higher-rate driver classifications.

To account for expected changes in average premium per exposure, a premium trend factor may be applied to historical premium. Expected future premium is generally the result of premium trend factors applied to historical premiums at current rate levels.

Historical Losses. Paid losses are dollars actually paid out for claims during a particular calendar period. Incurred losses are paid losses plus changes in reserves. Loss reserves are estimates of future anticipated payouts for claims. Paid losses are typically paired with written premiums to provide a cash-flow picture of the insurance company's operations. Incurred losses are typically paired with earned premiums to provide a more accurate estimate of the insurance company's results for policies issued during a particular calendar period. Insurance company-specific incurred and paid losses and written and earned premiums by line of insurance are readily available and can give an indication of the insurance company's historical profitability. Additional information is needed to perform the prospective ratemaking analysis.

It should also be noted that insurers can dramatically over- or under-state reserves and, consequently, dramatically misstate historical losses. For example, insurance companies dramatically overstated private passenger automobile liability reserves in the early 1990's. As a result, the incurred-to-earned loss ratios reported by insurance companies for those years dramatically understated the insurance companies' actual profitability.

Description of Historical Losses. For ratemaking analyses, loss data are organized in three different ways with tradeoffs between the timeliness of the data (i.e., how quickly the data are available after a particular experience period) and how well the losses are matched to the premium and policies under which the losses were paid.

Calendar year data typically represents incurred losses (paid losses and changes in reserves) regardless of when the claim occurred or when the policy was issued. Calendar year data are typically financial data and generally do not effectively match losses with the premium and exposure of the policies under which the losses were paid. Calendar year data are generally not used for ratemaking analyses, but are sometimes used for certain short-tailed lines because the calendar year data may not be significantly different from accident year data. The benefit of calendar year data is that the data are available quickly after the end of the particular time period.

A short-tailed line is one in which claims arise and are paid soon after the policy is issued. A long-tailed line is one which claims may arise long after the policy is issued. Examples of short-tailed lines are dwelling coverage and automobile physical damage coverage. Examples of longer-tailed lines are auto bodily injury liability and medical malpractice. The longer the "tail," the longer the insurance company holds the policyholder's money in reserves and the greater the amount of investment income earned by the insurance company.

Accident year data track claims paid and reserves on accidents occurring within a particular year, regardless of when the claim occurred or when the policy was issued. Accident year data do a better job at matching losses with the premium of the policies under which the losses were paid. Accident year data are not available as quickly as calendar year data because time is needed for the accident year data to develop, i.e., time for claims occurring within a particular period to be reported and settled.

Policy year data track claims arising from policies issued in the year, regardless of when the accident occurred or when the claim was reported. Policy year data does the best job of matching losses with the premium and exposures of the policies under which the losses were paid. Policy year data take the longest time to develop.

Fiscal Accident versus Calendar Accident Years - Calendar accident year data refer to accident year data for a given calendar year, i.e., the accident year from January 1 through December 31. Fiscal Accident year data refer to accident year data for a twelve-month period other than from January 1 through December 31. For example, a fiscal accident year may be the accident year data for the period July 1 through June 30.

Historical Losses Project Future Losses - Loss Development. Historical losses are subject to several adjustments to create ultimate projected losses. The pattern of claim occurrence, reporting and payment - loss development - occurs differently for different coverages. For property coverages, losses are generally reported and settled relatively soon after the accident occurs - short-tailed coverages. For other coverages, such as bodily injury liability, claims may not be made quickly after an accident and claims may take years to settle - long-tailed coverages. The most recent year of loss experience, for example, may not fully reflect the number of claims and amount of losses the insurer will eventually pay for coverage in a particular year. The loss development analysis adjusts historical losses for future development. Note that loss development will be minimal for older historical experience; however, that older experience will not be as reflective of current circumstances as more current experience which may not be fully developed.

Using Historical Losses to Project Future Losses - Loss Trend. A second adjustment to historical losses is the loss trend. Loss trend attempts to capture past and prospective changes in claim costs, claim frequencies, and pure premium (average loss per exposure). Loss trends may also capture many of the same changes in an insurer's risk profile that are reflected in premium trend. Loss trend data typically consist of earned exposures, paid losses, and paid claims by calendar quarter. These data are all easily obtained quickly after the end of the calendar quarter. However, the paid losses and paid claims are likely to be associated with policies (and earned premium) from earlier periods. If there are no significant changes in the volume of an insurance company's business, then the use of paid losses and paid claims matched to earned premiums will reasonably approximate the actual relationship of losses associated with policies in force during a particular calendar period. The loss trend data is analyzed to determine if changes in claim frequency (the number of claims per exposure) and/or claim severity (the average claim size) are occurring. These historical changes are typically applied to the historical loss data to not only adjust historical loss levels to estimated present loss levels, but also to adjust to estimated levels in the future.

Using Historical Losses to Project Future Losses - Adding ULAE. Historical private passenger automobile losses are generally reported inclusive of ALAE. The third adjustment to historical losses is to add ULAE. Historical losses adjusted for loss development, loss trend, and ULAE are known as ultimate projected losses.

Expenses. Variable expenses are generally estimated by calculating the average percentage of these expenses - commissions, taxes, and other acquisition expenses - to premium over the most recent two or three years.

Fixed expenses are generally estimated in the same manner as variable expenses with an additional step or two. In some cases, insurance companies apply an expense trend, or inflation factor, to fixed expenses. Such an adjustment, however, is unnecessary if the number of insured automobiles is increasing or if rates are increasing. If either one of these situations exists, the insurer is getting more fixed expense without any expense trend added and, therefore, no expense trend is necessary.

Profit Provision. The development of the profit provision starts with a target return on capital. This target return is offset by anticipated investment gains on surplus. The target return on capital less investment gains on capital is converted to a return on premium. The figure is further reduced, or offset, by investment gains on policyholder-supplied funds. The result is a profit provision based upon total returns, expressed as a percentage of premium. Investment gains include interest income, dividends, realized capital gains and unrealized capital gains.

Rate and Risk Classifications. The ratemaking analysis first produces average statewide rate change indications by coverage. For example, the ratemaking analysis may initially produce a 5% average statewide increase for bodily injury liability. The insurer then selects the average statewide rate change by coverage it will use or proposes to use. It is common for insurance companies to select rate changes significantly different from the actuarially indicated rate changes. There is generally little or no explanation provided by insurance companies for their selection of rates significantly different from the actuarially indicated rates.

The statewide average rate change is then distributed to the various risk classifications, such as different driver classes, increased limits factors and rating territories. If some parts of the state (rating territories) have better than average loss experience for a particular coverage, these rating territories should get a lower rate change than the statewide average for that coverage. Of course, if one rating territory gets a lower than average rate change, another rating territory must get a higher-than-average rate change.