Wording

  • Trying to explain double counting effect of exposures

    Pure premium method doesn't account for exposure correlation, thus when taking unadjusted premiums, class 1 has more of territory A exposures, which has relativity half of that of Z, while class 99 has exposures of territory Z which has twice pure premium relative to territory A. Thus, the effect of territory got double counted in class 99, and we got a higher relativity

  • You can instead write

    Disproportionate exposures are concentrated in higher and lower cost groups. When determining relativities, our expectation is 50-50. Instead 80% of class 99 exposures is in high-cost territory Z….
    > Thus class 99 looks relatively higher in cost than it actually is.

  • GLMs will model pick up (or reflect) the noise component in addition to true signal.
  • Overestimation and underestimation wording due to change in claims process

    Multiplying older development factors by current lower levels of paid/reported losses

  • ii. Case O/S observed are predictive of unobserved Case reserves at later maturities.

    ii. Case O/S observed are predictive of future claim development.

  • Replace "at later maturities" with "future" (simple)
  • Most of the time when you want to intuitively say "results", you either mean
    • "Projections"
    • "Age-to-age factors"
    • or something else, NAME IT! BE SPECIFIC
    • Or be general when appropriate and so call it "results".