ICICI Lombard Lost ₹318 Crore on Insurance and Still Earned ₹946 Crore. Here Is How

The idea in one sentence

A general insurer has two sources of profit, the insurance itself and the income on money it holds before paying claims; in ICICI Lombard's audited September-quarter accounts the first lost ₹318 crore while investment income was ₹1,568 crore, of which ₹1,161 crore came on policyholders' funds and ₹407 crore on shareholders' funds (our arithmetic for the total); and a rule on counting long-term policies gives the quarter two profit figures, ₹946 crore and ₹1,031 crore.

The two results side by side

ICICI Lombard filed its results for July to September 2026 on 14 October [RES].

Source: [RES], Annexure-I, lines 3, 12, 13, 4, 18(b), 25 and 27, and the analytical ratios. Small items are left out, so the rows do not add exactly. The press release gives investment income of ₹1,554 crore; the two audited lines add to ₹1,568 crore (our arithmetic), and the filings do not reconcile the two. The joint auditors' report has no qualification; as an other matter, they relied on the appointed actuary's certificate for reserves estimated by statistical methods, such as claims incurred but not yet reported.

Plain definitions. Premium earned is the part of premium that covers the period already passed. The underwriting result is premium earned less claims, commission and running costs: the result of the insurance alone. The combined ratio adds claims as a share of premium earned to commission and costs as a share of premium written (sold) in the period [PR, press release notes]. Above 100% means the insurance lost money. The company says the quarter includes ₹118 crore of catastrophe losses; without them its 1/n combined ratio was 105.3%, still above 100% [PR, slide 15]. For the quarter: 72.7% plus 34.3% (our arithmetic).

The insurance lost money in the quarter, a year earlier and in FY26. The company's own ten-year chart shows the same for the industry: its combined ratio was above 100% in every year from FY17 to FY26, and ICICI Lombard's in nine of the ten [DD, slide 4]. The company made a profit in all three periods. Investment income on policyholders' funds alone was 3.6, 5.0 and 3.2 times the underwriting loss (our arithmetic); with income on shareholders' funds added it was 4.9, 6.8 and 4.3 times. These lines include capital gains, which were ₹453 crore in the latest quarter [PR].

One policy, followed

Take a motor policy. Indian general insurance, the presentation notes, works on a "cash before cover" model [PR, slide 23]. The premium is paid on day one. Claims come later, sometimes much later.

Motor third-party claims, for injury, death or property damage caused to others, are among the slowest to settle in the company's tables. For FY2017 accidents it held ₹1,683 crore of these claims unpaid at the end of that year; at March 2026, nine years later, ₹391 crore was still unpaid [PR, slide 30]. IRDAI counts reserves like these within policyholders' funds, which must be backed by investments [IRDAI, Section IV, 3.1].

Some policies are sold for more than a year at once; the company names long-term third-party motor cover and long-term health [Q3C-25]. The customer pays up front, but IRDAI's rule counts only one year's share as premium written each year: the whole premium times 1/n, where n is the number of years. The rest is held as "Premium Deposit" or "Advance Premium", and commission is paid "only on the gross written premium recognised for the year" [IRDAI, Section IV, 5A and 5B]. With round numbers of our own: a three-year policy paid ₹30,000 up front counts as ₹10,000 of premium written in each year, and ₹20,000 waits as advance premium.

For motor this is not new. The company says long-term products have been counted this way since 1 October 2024 [PR, slide 17]; the circular, dated 17 May 2024, gives no date in that paragraph. In January 2025 the company also said that across the industry "new motor policies due to long-term TP were recognized on 1/n basis" already. The change fell on long-term health and other lines [Q3C-25].

The mechanism: two engines

Money held from policyholders until claims are paid or cover is given is often called float. The company's risk management slide speaks of "insurance float & value investing philosophy" [PR, slide 11].

At 30 September the company held ₹47,378.51 crore of investments against policyholders' funds and ₹16,953.17 crore against shareholders' funds [RES, balance sheet]. So about three-quarters of its investments are held against what it owes policyholders (our arithmetic). The split is made "on notional basis", and a shortfall in policyholders' investments "arising out of the loss in the Revenue Account" must be made good with securities from the shareholders' side [IRDAI, Section IV, 3.1 and 3.2]. Our reading: if the policyholders' account ends a period in loss after its investment income, shareholders' securities top it up. This quarter that account showed an operating profit of ₹873 crore [RES, lines 16 and 17], so the underwriting loss was covered by investment income on policyholders' funds. The cost of float is the underwriting loss; here the income on the float more than covered it.

The presentation puts total investment assets, net of borrowings, at 3.58 times net worth, which is share capital plus reserves [PR, slide 18]. Our reading: that ratio, with the return earned, is how a small underwriting loss can sit beside a profit. Each ₹100 of net worth carries about ₹358 of investments (our arithmetic); if they earn more than the insurance loses, the shareholder gains.

The company's ten-year chart footnotes its combined ratios as "BAGI adjusted numbers", without saying what is adjusted. Its FY25 and FY26 ratios (102.8% and 103.4%) match the 1/n ratios in the results presentation [PR, slide 17]; we cannot tell whether earlier years are the audited figures. On it, the company's ratio ranged from 99.4% in FY19 to 108.8% in FY22, and its return on equity, profit as a share of shareholders' money, from 14.0% to 21.2%. The industry's ratio ran from 112% to 120.3%, and its return on equity from minus 4.0% to 11.9% [DD, slide 4]. The industry has float too. Our reading: float pays only if the insurance loses less than the float earns. A ratio near 103% leaves room for that; one near 118% mostly does not. So the test of a moat here is not the float, which every insurer has, but underwriting that keeps its cost low. The test is ours, not the company's or the regulator's.

On the July call the MD and CEO put the industry's FY2026 combined ratio at 117.8% against the company's 103.4%, "a gap of over 14 percentage points" [Q1C]. Its ratio has been higher since: 107.2% in April to June and 107.0% in July to September [PR, slide 15]. We read no industry figure for those quarters.

Why the quarter has two profits

The audited figures match the company's 1/n figures: profit after tax of ₹946.40 crore, up 15.5%, and a combined ratio of 107.0% [RES] [PR, slide 15]. The press release reports profit on the n basis: ₹1,031 crore, up 27.5% [PR]. The presentation shows both.

[PR], slides 14 and 15.

The 1/n basis is IRDAI's rule, described above. In the company's words, "Long-term Products are accounted on a 1/n basis, as mandated by IRDAI" [PR, slide 17]. The n basis is the company's own term; the circular does not use it. In January 2025 the company labelled the two "Excluding 1/n Impact" and "With 1/n Impact" [Q3-25, slide 15], and said that before the rule "The premiums under long-term health and other line of business were recognized in the year of underwriting itself" [Q3C-25]. Our reading: the n basis counts such a premium in full in the year of sale.

In the quarter the n basis counts ₹492 crore more gross direct premium (our arithmetic). A year earlier the gap was ₹361 crore (our arithmetic from [PR] slides 14 and 15). Our reading: the growth of the long-term health book is part of the reason; on the July call it was 53.4% of new retail health business, against 31.8% a year earlier [Q1C].

The profit difference is harder to pin down, and the company has said so. In January 2025 its CFO said 1/n means "whatever acquisition costs that you incur, you defer it over the contract period", and the same for reinsurance commission received. The outcome, he said, "could be either reverse" [Q3C-25]. On 1/n, profit was higher than on n in FY25 (₹2,508 crore against ₹2,419 crore) and FY26 (₹2,772 crore against ₹2,761 crore), and lower in both quarters of FY27 [PR, slides 14 to 17] [Q1, slide 15]. Neither figure is consistently the higher one.

Where this breaks

The n basis may be the better guide to this year's business. The company said the change "does not impact the economic value" [Q3C-25]: the cash and risk are the same; only the year premium is counted differs. The argument that follows is ours. While the long-term book grows fast, 1/n pushes premium into later years and can make a growing insurer look smaller. The 1/n combined ratio was the higher one in every period shown: 102.8% against 102.6% in FY25, 103.4% against 102.4% in FY26, 107.0% against 104.5% this quarter [PR, slides 14 to 17]. 1/n lowers premium written, the divisor for costs, but also defers commission, so the net effect depends on the mix. The filings do not say whether the two bases will come closer; in January 2025 the CFO said the difference could go either way [Q3C-25].

The 1/n basis is what the regulator requires, and the audited figures match it. The audit report does not name the basis. Both this quarter and the year-ago quarter are on 1/n, so the 15.5% growth compares like with like. Our reading: 1/n also matches premium to the years of cover.

Capital gains are lumpy. Gains on investments sold were ₹453 crore in the quarter against ₹236 crore a year earlier [PR], and ₹183 crore in April to June against ₹380 crore [Q1]. At 30 September the net unrealised gain, not yet in profit, was ₹2,402 crore: ₹2,805 crore on equity and ₹77 crore on mutual funds and AT1 bonds, less an unrealised loss of ₹480 crore on investments other than equity [PR, slide 18].

The underwriting result is partly an estimate. Claims include reserves for those not yet settled. The reserving tables show estimates falling over time for every past accident year, by 7.0% for FY2017 [PR, slide 28]; the company says such favourable run-offs reflect prudence [PR, slide 11]. Reserves can also move the other way. Profit after tax fell 46.0% in April to June on the 1/n basis. The company put this down to two large fire losses of ₹63 crore in all and a Supreme Court judgment that raised its motor third-party reserves by ₹165 crore, adding 2.8 points to the combined ratio; without them, it said, profit fell 23.0% [Q1].

Float income depends on markets and on timing. In the half year the policyholders' fair value change account, which holds gains or losses on investments not yet sold (it was a loss of ₹562 crore at March), rose by ₹2,630 crore (our arithmetic) [RES, balance sheet]. Float also needs claims to be paid later than premium arrives. In April to September claims paid rose 19.7% and premium received 5.5% (our arithmetic) [RES, receipts and payments account]; investments held against policyholders' funds still rose, from ₹43,492.57 crore at March to ₹47,378.51 crore, a rise of ₹3,885.94 crore (our arithmetic), of which ₹2,630 crore is the revaluation above [RES, balance sheet]. Net cash from operating activities was an outflow of ₹383.66 crore in the half year, against an inflow of ₹459.46 crore a year earlier [RES, receipts and payments account].

Our sample is one company. The industry figures are the company's own; we read no peer's accounts.

What follows from the documents

Three things a reader can track in the filings:

• The combined ratio against the industry's. The company points to the gap, over 14 points for FY2026 [Q1C].
• Investment income without capital gains, which the company reports separately.
• Both profit figures, and the gap between them, as the long-term health book grows.

Back to the motor policy

A long-term motor policy pays the company on day one and counts as premium a year at a time. Some of its third-party claims may stay unpaid for nine years or more. In between, the money is invested. That is the second engine; the first decides how much it is worth. At 107% the insurance cost more than it brought in, but income on investments held against policyholders' funds was more than three times that loss (our arithmetic). The industry also holds policyholders' money before paying claims, though the chart does not say how much. At combined ratios of 112% and above its return on equity ranged from minus 4.0% to 11.9%, against 14.0% to 21.2% for the company [DD, slide 4]. Our reading: float helps; the combined ratio decides how much.

Related: IRDAI's proposed distribution reforms, which would cap commissions, and The Insurer Whose Superpower Is Saying No, on writing less business rather than underwriting at a poor price.

Sources

• [RES] ICICI Lombard, audited results for the quarter and half year ended 30 September 2026, with joint auditors' report (NSE)
• [PR] ICICI Lombard, press release and investor presentation, 14 October 2026 (NSE)
• [Q1] ICICI Lombard, press release and investor presentation, 15 July 2026 (NSE)
• [Q1C] ICICI Lombard, transcript of the earnings call of 15 July 2026 (NSE)
• [DD] ICICI Lombard, analyst Digital Day presentation, 29 September 2026 (NSE)
• [Q3-25] ICICI Lombard, investor presentation, 17 January 2025 (NSE)
• [Q3C-25] ICICI Lombard, transcript of the earnings call of 17 January 2025 (NSE)
• [IRDAI] IRDAI, Master Circular on Actuarial, Finance and Investment Functions of Insurers, 17 May 2024 (copy hosted by the Institute of Actuaries of India)

Educational research, not investment advice. Every figure above is drawn from ICICI Lombard's own filings and from IRDAI's master circular, read from the primary documents on 15 October 2026; sums, differences, shares and ratios marked "our arithmetic" are ours. The audited results match the 1/n basis the regulator requires; "n basis" figures are the company's own presentation figures and are not part of the audited results. The ten-year industry figures are as the company presents them. Nothing here says which profit figure is right, or why the company presents either. No buy/sell recommendations, no price targets, no share prices. Moat & Margin is not a SEBI-registered Research Analyst or Investment Adviser.