What I examine in a bank that lends to small businesses

I manage a small-business loan book on the assumption that the next downturn has already begun somewhere inside it. In Indonesia a loan moves through five collectability grades, from current to loss, and the grade often lags the borrower's real condition by months, particularly after a restructuring. I therefore watch loan at risk, which combines restructured current loans, special-mention loans and non-performing loans, alongside vintage curves cut by product, province and loan officer. I want field visits rather than telephone calls before a restructured loan is extended a second time, and I want our expected credit loss provisions to reflect what our collectors believe as well as what the model last said. At Kartadana I also tie the dividend recommendation to loan-at-risk and capital thresholds set a year ahead, so that a weak year never becomes a negotiation between management and the controlling shareholder.

A mid-sized Indonesian bank competes for deposits with larger banks that have stronger brands and state banks that hold captive payrolls. The easy answer is to pay more for time deposits, including special rates above the rate LPS uses to decide which deposits it insures. I treat those balances as market funding in disguise: the depositor has chosen yield over protection and will move for a few basis points. At Kartadana we have built current and savings accounts through the businesses we already lend to, using QRIS merchant acquiring, payroll for the workers in our MSME clusters and supplier payments for distributors. The share of current and savings accounts rose from 31% to 46% of third-party funds between 2021 and June 2026. Our balance sheet grew more slowly as a result, and I set that cost before the commissioners every year beside the steadier deposit base it bought.

I do not think the choice between branches and digital channels is the right question for a bank serving small businesses across Indonesia. Our customers in Jakarta and Surabaya now do most of their banking on their phones, while a rice miller in South Sulawesi still pays harvest labour in cash and wants a loan officer who has seen the mill. I plan the network district by district, using transaction mix, cash volumes, loan demand and the distance to the next point of service. Urban offices that duplicate each other are closed or merged. Rural cash offices may become lending posts staffed by loan officers, with cash handled by agents under OJK's Laku Pandai branchless-banking programme. I judge the result by deposits retained, loans originated and complaints received, not only by the cost removed, and every closure needs a named alternative for the customers it affects.

Social-engineering scams changed the fraud problem for Indonesian banks. The customer authorises the transfer, on their own device, after being deceived, and conventional fraud rules see nothing unusual because technically nothing is. I approach the problem from three sides. At account opening we match identity biometrically against population-registry data and check devices, so that fewer mule accounts open with us. In the app we add friction where scams concentrate, such as the first large transfer to a new beneficiary from a newly registered phone. After a loss we apply a published framework for reimbursing customers when our own controls failed. I work with OJK, the police and other banks on mule-account data, and I report scam losses, reimbursements and the complaints our controls generate to the Risk Monitoring Committee every quarter, so that directors see the inconvenience as well as the protection.

An Indonesian bank answers to several authorities with distinct roles. OJK licenses and supervises banks, assesses the fitness and propriety of directors and commissioners and examines risk management. Bank Indonesia sets monetary and macroprudential policy and runs the payment system, including QRIS and fast transfers. LPS insures deposits and would resolve a failing bank. I try to make each relationship predictable. Our supervisors hear about a material problem from us before they find it, with the facts we hold and those we are still establishing. I attend the meetings that matter myself, and I do not allow a response to an examination finding to be drafted to close the finding rather than the weakness behind it. In my experience supervisors are more patient with an honest timetable than with a confident one that slips, and they remember which banks told them early.

Many mid-sized Indonesian banks have a controlling shareholder, and many of those shareholders are families with other businesses. The two-tier structure helps: the Board of Directors runs the bank and a Board of Commissioners, with a substantial independent membership, supervises it. Structure alone does not settle the hard cases, which usually concern related-party credit, dividends and senior appointments. At Kartadana I have pressed for written rules that apply before any request arrives. Related-party transactions above Rp 25 billion are decided by independent commissioners with family commissioners recused; the dividend recommendation follows published credit and capital thresholds; and succession for the President Director runs through a nomination committee led by independents. I value the family's long horizon and its readiness to put in capital during a crisis. Minority shareholders are owed evidence that the family is treated as a shareholder, not as a preferred borrower.