reserves at fractional durations

Reserve formulas give a clean value at each policy anniversary — end of year 1, year 2, and so on. But a company closes its books on December 31, which for most policies falls partway through a policy year. So you constantly need the reserve not at a tidy anniversary but at some fraction of a year in between. Reserves at fractional durations are the methods for estimating that in-between value.

Precisely, suppose you need the reserve at time t + s where t is the last anniversary and s is a fraction of a year (say s = 0.4). A common practical approach interpolates between the reserve already built up and what is coming. One standard rule says the fractional reserve roughly equals the linear interpolation between the reserve at t (after adding the premium just paid) and the reserve at t+1, then steps back the unearned portion of the premium for the rest of the year. A cleaner way is to integrate Thiele's equation forward from the last anniversary by the fraction s. The reason it is not simply linear is that the just-paid annual premium covers the whole coming year, so part of it is 'unearned' at a mid-year date and must be handled separately.

Fractional reserves matter at every valuation date, because anniversaries rarely line up with the accounting calendar. They also tie directly to the unearned premium concept and to how cash values are credited between anniversaries. The honest point is that the common interpolation formulas are approximations chosen for convenience and consistency; the exact answer comes from integrating the reserve dynamics, and the two can differ slightly, especially in the heavily front-loaded first policy year.