A general insurer's earnings engine has two cylinders:
Pre-tax profit ≈ underwriting result + investment income (on float + shareholders' funds)
Float here ≈ loss reserves (outstanding + IBNR) + unearned premium reserves − premium receivables and deferred acquisition assets. Indian general insurers commonly run investment leverage (investment assets ÷ net worth) of 3 to 4×. So a modest yield on a big float can dwarf the underwriting line.
Meghdoot continued. Float is ₹12,000 cr and net worth ₹5,000 cr. Investment assets of ₹17,000 cr yield 7.2%, which gives investment income of ₹1,224 cr. Pre-tax profit = −304 + 1,224 = ₹920 cr. After 25% tax, PAT is ₹690 cr, so ROE = 13.8% despite a 102.5% CR. Now the float's cost: 304 ÷ 12,000 = 2.5%. Meghdoot "borrows" ₹12,000 cr at 2.5% when the government pays ~7% to borrow (as of mid-2026). That's cheap leverage, and the business is fine.
When is a combined ratio above 100% acceptable?
So when is a CR > 100% acceptable? Run four tests:
Advertisement
- Is the float big and long? Cost-of-float logic only works with high float-to-equity and a long tail, as in liability and motor third-party. A short-tail insurer (crop, most health) holds premiums for months, so a CR > 100% there is just a loss with no float to make up for it.
- Is the CR honest? A 102.5% built on adequate reserves is fine. A reported 102.5% that's really 108% before prior-year reserve releases is a lie that compounds, as the reserve-development section shows.
- Is the investment income boring? Float belongs in high-grade bonds. If the profit bridge depends on equity capital gains or on reaching for credit yield, underwriting losses are being papered over with market risk. That's the brief's third red flag, a CR > 100% masked by investment gains. To test for it, recompute ROE using only interest/dividend income at a G-sec-like yield. If ROE collapses, so does the thesis.
- Does the arithmetic clear the hurdle? A quick screen is
sustainable ROE ≈ [investment yield × (float + equity) − underwriting loss] ÷ equity, after tax. For Meghdoot: (0.072 × 17,000 − 304) × 0.75 ÷ 5,000 = (1,224 − 304) × 0.75 ÷ 5,000 = 13.8%. That's above the cost of equity, so it's fine. Now keep the same CR but give the insurer a float of only ₹3,000 cr, a short-tail book with investment assets of ₹8,000 cr: (576 − 304) × 0.75 ÷ 5,000 = 4.1%. It's the identical combined ratio, and now value is being destroyed. The combined ratio is meaningless without the float attached to it.
And when rates rise, remember that the bond book marks down. Float that fuels ROE also carries duration risk. What SVB taught you about banks applies to insurers' available-for-sale books too, softened by the fact that policyholders, unlike depositors, can't run.
How do you value a general insurer?
Valuation lens for general insurers: P/B against ROE, the same machinery you used on banks. Book is real here, and one-year contracts need no EV. Sanity-check it with P/E for mature underwriters, and always adjust for quality, meaning the CR's honesty and the float's durability. A franchise that compounds float at negative cost deserves multiples of book. A serial 115% underwriter deserves a discount to book, whatever this year's investment gains say.