Explanation of PTC Index
We have modelled the Propensity-To-Claim using input features such as climate, economic factors, land use, and location alongside how “well does the grass grows”, We have determined a “separable score” for each farm in Australia.
The PTC Index is NOT a probability.
You must interpret the PTC index in the same way you would interpret a credit score, that is, as a ranking system.
Credit scores are typically made by modelling the propensity to default on a loan given a set of inputs (income, debt, etc.). These scores are then used as a comparison to current portfolios (of all our customers, where does their score lie), not as an isolated comparison.
There are rules of thumb to a single measure (be it credit score or a PTC Index), but these rules of thumb are developed through analysis of large customer bases, not the model by itself.
The best use of the PTC Index is to look at a single region and find the range and average of the PTC Indices in that specific region, this will then establish a baseline of comparison for new and old customers.
Moving forward, additional claims data and further analysis of claims and long-term effects on the propensity to claim will likely improve the PTC Index. Of course, we need to ensure we are not able to predict the future but instead can quantify our level of uncertainty around it.
If a starting rule of thumb is absolutely necessary (although unadvised as this loses nuance given location), below 0.2 is typically when claims become less prevalent, above 0.2 claims become more prevalent, and above 0.4 again more prevalent.