Bank CIMB Niaga

1. 01Win the cheap money (the part it got right)

   Current accounts grew 47% (Rp70.7tn → Rp104.0tn) and savings 17% (Rp74.1tn → Rp86.4tn), while time deposits were allowed to shrink 2% (Rp82.8tn → Rp81.1tn). CASA therefore climbed 63.6% → 70.1%, helped by the OCTO digital platform pulling transactional balances.

   EconomicsA current account is transactional money that sits at near-zero cost because the customer keeps it there to operate, not to earn. Growing it is slow, unglamorous work and it is the most durable advantage a bank can build. Danamon began FY2022 with the same 63.6% CASA and ended at 42.5%, which is why these two banks are worth reading as a pair.

2. 02Pay far more for money anyway

   Interest expense rose 95% (Rp5.45tn → Rp10.65tn) while interest income rose 25% (Rp18.15tn → Rp22.78tn). Net interest income fell in every year: −1.6%, −2.8%, −0.2%. NIM went 4.69% → 3.97%.

   EconomicsThis is the lesson most readers miss. Niaga did the right thing on funding and still lost 72bp of margin, because when policy rates rise the whole deposit book reprices, not just the expensive part. The mix decides how MUCH you lose, not whether you lose. Compare the damage: Niaga −72bp with CASA rising, Danamon −71bp with CASA collapsing, MEGA −124bp with CASA under 30%. In FY2025 Niaga hands 47% of what its assets earn back to depositors, against 30% in FY2022.

3. 03Lend it out, but only moderately

   Loans grew 19.8% across the window (Rp144.2tn → Rp172.7tn, yearly 5.1%, 4.5%, 9.0%), securities grew Rp58.8tn → Rp73.3tn, and the loan-to-deposit ratio stayed in a narrow 85.6% to 89.3% band.

   EconomicsCompare Danamon, which grew loans 42% in the same window and earned no extra profit for it. Niaga grew half as fast and its profit rose 41.5%. Growth is not the variable that decides bank returns; the price of the funding and the cost of the risk are.

4. 04Harvest the credit cycle (where the profit actually came from)

   Gross NPL improved 2.80% → 1.81% and the impairment allowance on the balance sheet fell from Rp13.09tn to Rp8.05tn, with coverage still 258% of non-performing loans at FY2025. Pretax profit rose Rp2.39tn while net interest income fell Rp0.58tn, so roughly Rp3.0tn of improvement came from outside the margin line.

   EconomicsPost-pandemic, a large part of Indonesian bank earnings growth came from provisions falling rather than revenue rising, and Niaga is a textbook case. Note the honest limit of what these figures show: the OJK ratio table for this bank does not disclose cost of credit, so the exact provisioning charge is not visible here, only its direction through the allowance balance and the NPL ratio. Check the bank’s own income statement for the charge itself.

5. 05Keep enough capital, pay out the rest

   Equity grew Rp43.8tn → Rp55.8tn (+27.5%) and CAR 22.19% → 24.83%, while the bank still distributed about 57% of earnings (DPS Rp155.73).

   EconomicsPaying out more than half your profit and still lifting the capital ratio is only possible when the balance sheet grows slower than earnings, which is exactly the trade Niaga chose. It is the opposite of BRIS, which pays about 13% and grows financing 15% a year. Neither is right in the abstract, they are different answers to the same question: is another rupiah of loans worth more inside the bank or in the shareholder’s pocket?

Cost structureFunding first, operations second. Interest paid was Rp10.65tn against Rp22.78tn earned in FY2025, so 47% of what the assets produce goes back to depositors, up from 30% in FY2022, and that single shift is larger than anything on the operating-expense line. Operating efficiency itself is middling rather than poor: BOPO ran 74.1%, 71.5%, 74.0%, 72.4%, and cost-to-income is disclosed only for the last two years at 44.3% and 45.9%, against roughly 31% for BBCA and 34% for BMRI in their best years. A corporate-and-consumer universal bank carries branch, technology and relationship costs that a narrower franchise does not.

Cash cycleFunding cost → margin → credit cost → capital. Niaga strengthened link one (CASA to 70.1%), still lost link two (NIM to 3.97%), and won decisively on link three as the post-pandemic credit cycle turned (NPL 2.80% → 1.81%, allowance Rp13.09tn → Rp8.05tn). Link four then works properly, which is the contrast with Danamon: profit accumulates as capital here (equity +27.5%, CAR 22.19% → 24.83%) even while 57% is paid out. One disclosure wrinkle worth knowing: the engine’s implied retention for Niaga is 61%, higher than the 43% you would infer from a 57% payout, because equity growth also carries other-comprehensive-income marks on the Rp73.3tn securities book. The engine documents that approximation rather than hiding it.

Unit economicsFor BRIS and MEGA the sharpest unit test was margin minus cost of credit, but that subtraction cannot be run here because cost of credit is not disclosed for this bank. So use the identity that still closes: return on assets rose 2.16% → 2.43% in the same four years that NIM fell 4.69% → 3.97%. If the margin per rupiah of assets shrank and the return per rupiah of assets grew, the difference must have come from lines other than the margin, and the only one visibly improving is credit. That reasoning, using what IS disclosed to bound what is not, is the most transferable habit on this page.

Sound, with one composition question a reader should hold in mind. The credit improvement is real and conservatively carried: gross NPL fell in three of four years to 1.81% and coverage still ended at 258% of non-performing loans, so this is not a bank that emptied its reserves to make a number. But the composition of the profit growth matters. Net interest income fell every single year (Rp12.70tn → Rp12.12tn) while pretax profit rose Rp2.39tn, so approximately Rp3.0tn of the improvement came from outside the margin, and the credit cycle is the visible source. That lever is finite by nature: the allowance balance has already been drawn down 38% (Rp13.09tn → Rp8.05tn) and coverage has come off its 401% peak in FY2023. Two disclosure limits belong in the same breath, because honest analysis names what it cannot see: the OJK table for this bank does not disclose cost of credit in any year, and discloses cost-to-income only for FY2024 and FY2025. The direction is unambiguous from the balance sheet, the magnitude of the provisioning charge is not, so verify it in the bank’s own income statement before relying on it.

Checked: no material one-offs found in the reviewed window.

Cash conversionThe profit-to-capital link works cleanly, which is exactly what Danamon failed at over the same four years. Equity grew Rp43.8tn → Rp55.8tn while assets grew 20%, so CAR still improved 22.19% → 24.83% with 57% of earnings distributed. Liquidity is unstretched: a loan-to-deposit ratio of 86.8%, Rp73.3tn of securities and Rp22.9tn placed at the central bank. Nothing about this balance sheet forces the bank to grow expensively.

A deliberately balanced allocator, and the balance is the point. Niaga grew loans a modest 19.8% over four years, distributed about 57% of earnings, and still lifted both equity (+27.5%) and CAR (22.19% → 24.83%). That combination is only available to a bank whose balance sheet grows slower than its profit, and it is the mirror of both extremes in this roster: BRIS pays 13% and grows financing 15% a year, Danamon grew loans 42% and ended with less book value than it started with. Judged on the only test that matters, whether the retained rupiah earns more than it costs, Niaga passes: ROE 13.03% against a cost of equity of 11.21%.

DeploymentIn: loans +Rp28.5tn and securities +Rp14.6tn. Funded by: CASA +Rp45.6tn (current accounts +Rp33.3tn, savings +Rp12.3tn) with time deposits −Rp1.7tn, which is the cheapest possible way to fund an expansion. Out: dividends at about 57% of earnings (DPS Rp155.73). Left behind: Rp12.0tn of additional equity and a capital ratio 264bp higher than four years earlier. There is no capital raise anywhere in this record, and no hidden leverage either, since the asset growth of 20% roughly matched the equity growth of 27.5%.

Returns trendROE 12.59% → 15.02% → 14.34% → 13.03% against a cost of equity of 11.21% (risk-free 7.26% plus a 0.59 emerging-market money-centre bank beta on a 6.69% equity risk premium). The spread was positive in all four years, peaking near 3.8pp in FY2023 and narrowing to 1.82pp as the margin compressed. The excess-return model therefore values the bank above book at 1.34×, while the market pays 0.73×, roughly 54% of modeled value. Put plainly, the price implies the spread over the cost of equity disappears; the record says it has survived four years, a full rate cycle and a 72bp margin squeeze. That is the disagreement to research, and the honest counter-case is on the record’s side only if the credit-cycle tailwind is replaced by something durable, because net interest income has now fallen three years running.