Calendar year effect

On February 27, 2017, the UK Lord Chancellor announced that the Ogden rate would move from 2.5% to minus 0.75%. Every open large bodily injury claim whose future losses will be settled as a lump sum took on its new value overnight, whether the claim dated from 2009 or 2016. In the triangle, that shock follows neither a row nor a column: it hits a diagonal, the accounting year in which it lands. That is the calendar year effect. Inflation gives the slow version. In the UK, consumer prices rose 11.1% year on year in October 2022, and parts and labor for motor repairs followed. Payments made in 2022 cost more for the 2019 accident year as much as for 2022. Chain ladder assumes development depends only on age. So it reads a diagonal shock as a rise in the factors of a few columns, averaged with clean diagonals. It keeps a fraction of it that is never the right one, and says nothing about the next diagonal. Practitioners spot it by setting incremental amounts against their expected values and lining them up by diagonal, or in the residuals of a three-dimensional model: accident, development and calendar. Taylor published a separation method in 1977 that extracts one index per calendar year. Diagnosis is easy. Treatment is much harder, because someone has to decide whether the last diagonal's index carries forward, and that assumption about future inflation is nowhere in the triangle. Models that estimate all three trends also run into an identity, calendar equals accident plus age, which leaves only two free trends. Not to be confused with claims inflation, which is one cause among others, or with Taylor's separation method, which is one way to measure it.