As Healthcare Costs Rise, Insurance Companies Strengthen Data-Driven Underwriting

Bacajuga:

Beyond Borders: Rethinking Travel Healthcare in a Digitally Connected World

OJK Wajibkan Co-Payment 10 Persen untuk Klaim Asuransi Kesehatan Mulai 2026

When healthcare inflation runs seven times faster than the general price index, traditional underwriting doesn’t just underperform. It becomes structurally unsustainable. Indonesia’s projected medical trend rate of 17.8% for 2026, against a general inflation forecast of just 2.5%. Is not an anomaly. It is the sharpest expression of a global pattern forcing insurers to replace experience-based judgment with data-driven decision architecture.

The question is no longer whether the industry will adopt data-driven underwriting. It is whether it will do so fast enough to remain solvent.

The Cost Curve That Traditional Underwriting Cannot Outrun

Healthcare costs are rising everywhere, but the dynamics in emerging markets are particularly acute. In Indonesia, medical cost inflation has consistently ranged between 10% and 19%, driven by an aging population, rising prevalence of chronic diseases such as diabetes, hypertension, and cardiovascular conditions, increasingly expensive medical technology, and shifting lifestyle patterns. This is not a temporary spike. It is a structural repricing of health risk.

Traditional underwriting — built on manual review, historical tables, and individual underwriter judgment — was designed for an era when risk moved slowly. When the gap between premium adjustments and claims acceleration was measured in single digits, experienced underwriters could bridge it with intuition and conservative reserving.

At a 17.8% medical trend rate, that bridge collapses. Claims outpace premiums faster than annual renewal cycles can correct, and the compounding effect turns small pricing errors into portfolio-level losses.

The evidence is already visible globally. In the United States, health plan medical-expense ratios surged to 93.2% — up from 85.3% the prior year — and underwriting losses ballooned to $10.4 billion. The mechanism is the same everywhere: healthcare costs are accelerating faster than underwriting can price for them using conventional methods.

Indonesia’s Regulatory Push: From Experience to Evidence

Indonesia’s response has been notably direct. PERUJI (Perkumpulan Underwriter Jiwa Indonesia) is urging the industry to move beyond relying solely on individual experience and to integrate data quality, process discipline, and evidence-based decision-making into every stage of risk assessment.

The regulator is aligned. OJK (Otoritas Jasa Keuangan) is pushing underwriters to understand target-market and product alignment while leveraging data, digital technology, and AI — without sacrificing data accuracy.

The practical recommendations are specific:

Pricing tools based on experience studies: using actual claims data rather than market averages or historical tables to set rates
Digital intake and automated data collection: reducing manual entry errors and accelerating the time from submission to decision
Predictive analytics for risk segmentation: identifying high-risk cohorts earlier in the underwriting process, before they appear in the claims data
Continuous portfolio monitoring: shifting from annual reviews to real-time loss ratio tracking that triggers repricing actions when thresholds are breached

This is not aspirational guidance. It is an operational mandate from both the professional body and the regulator, aimed at an industry where the cost of inaction is already measurable.

The Global Convergence: Machine-First, Human-Governed

Indonesia’s push mirrors a global convergence toward what McKinsey describes as a “machine-first, human-governed” underwriting operating model. The core idea: automate data collection, risk scoring, and standard-case processing, while keeping human underwriters accountable for judgment calls, exceptions, and governance.

The adoption data supports this trajectory. According to recent industry surveys, 45% of life-and-health insurance organizations now use AI regularly in their underwriting processes — a figure that, while lower than the 60% adoption rate in broader insurance lines, represents a significant acceleration from near-zero just five years ago.

The applications span the full underwriting workflow:

Digital intake: OCR and NLP tools extracting structured data from medical records, lab results, and application forms
Risk scoring: machine learning models that assess individual risk profiles against portfolio-level patterns, flagging outliers for human review
Group health underwriting: automated pricing engines that can process employer-group renewals in minutes rather than days
Underwriter copilots: AI systems that draft underwriting memos, surface relevant precedents, and recommend terms, with the underwriter reviewing and signing off rather than building from scratch
Straight-through processing: standard-risk cases that pass automated checks are approved without human intervention, freeing underwriters for complex cases

The efficiency gains are real. Organizations implementing these systems report 40–60% reductions in case review times and corresponding increases in throughput, without adding headcount.

The Governance Question: Data-Driven Is Only as Good as the Data

Here is where the narrative needs a corrective. The dominant framing “AI will transform underwriting” skips the harder question: will the data support it?

Data-driven underwriting introduces risks that experience-based underwriting never had to manage:

Proxy discrimination: models trained on historical data can encode biases that result in unfair pricing for specific demographic groups, even when protected attributes are excluded from the input
Model opacity: complex ML models can produce accurate predictions without explainable reasoning, creating regulatory and ethical exposure
Data quality gaps: in markets like Indonesia, where digital health records are unevenly distributed and claims data varies significantly by region and provider, the raw inputs for data-driven models may be incomplete or inconsistent
Model drift: a model trained on pre-pandemic claims data may systematically misprice post-pandemic health risks unless continuously recalibrated

Regulators are responding. In the United States, the NAIC Model Bulletin on AI governance has been adopted by more than half of states as of September 2026, establishing standards for transparency, fairness, and accountability in AI-assisted underwriting. Indonesia’s OJK, for its part, has been deliberate in coupling its technology push with an explicit emphasis on data accuracy — recognizing that the tool is only as reliable as what it is trained on.

As one industry executive put it plainly: “AI speeds up underwriting, but the decision stays human.” The governance challenge is ensuring that the human in the loop has both the authority and the information to override the machine when it gets it wrong.

What This Means for the Industry

The shift to data-driven underwriting is not optional. At 17.8% medical inflation, the alternative is systematic underpricing, portfolio deterioration, and — for some carriers — insolvency.

But the shift is also not sufficient on its own. Technology without governance is just faster error-making. The insurers that will emerge strongest from this transition are the ones that build three capabilities simultaneously: automated data infrastructure that reduces manual error and accelerates processing; predictive models that improve risk segmentation and pricing accuracy; and governance frameworks that ensure fairness, transparency, and human accountability at every decision point.

Indonesia, with its acute cost pressures and proactive regulatory posture, may well become the proving ground for this model. The rest of the industry is watching.