I was trying to calculate the incurred claims ratio for the New Assurance Assurance Co. (an Indian insurance company) for health insurance from the company's public disclosure information. This seems like it should be a simple thing, but it doesn't agree with the values published by India's Insurance Regulatory and Development Authority (IRDAI). This is just an example. I would also like to know this statistic for other insurance companies.
The definition of Incurred Claims Ratio is
Net Claims Incurred / Net Earned Premium
The IRDAI calculation in on IRDAI_English_Annual_Report_2018-19.pdf, which is at https://www.irdai.gov.in/admincms/cms/uploadedfiles/annual%20reports/IRDAI%20English%20Annual%20Report%202018-19.pdf
The calculation is on pg 164 of pamphlet, Pgs 188 of PDF (Statement 14), first table, second row.
2018/2019
Net earned premium (crores) Claims incurred (crores) Incurred Claims Ratio (%)
7448.10 7726.84 103.74
But this doesn't agree with the values given in the New India Assurance FY 2018-2019 public disclosure documents.
It suffices to look at the Q4 FY 2018-2019 public disclosure documents, which can be found at https://www.newindia.co.in/portal/readMore/PublicDisclosure or by searching for "new india assurance public disclosure".
The direct URL is https://www.newindia.co.in/cms/96659ada-2d42-43a2-9ecb-2b71f9e369a2/Public%20Disclosures%20March%202019%20Final.pdf
First the premium earned value is in the NL-4 document, specifically the value at the bottom row ("Premium Earned (Net)") of the column:
HEALTH -> UPTO THE QTR ENDED 31.03.2019
is 69384056. This is in thousands, so converting to crores (ten millions), gives 6938.4056 crores.
Similarly, the claims incurred value is in the NL-5 document, specifically the value at the bottom row ("Total Claims Incurred") of the column:
HEALTH -> UPTO THE QTR ENDED 31.03.2019
is 69050967 (in thousands), which is 6905.0967 crores.
HEALTH -> UPTO THE QTR ENDED 31.03.2019
The ratio is 6905.0967/6938.4056 = 0.9952. I.e. 99.52%.
I'd like to resolve the discrepancy. Am I doing the calculation that uses the public disclosure information wrong?