Research
Publications
Tan, K. M. (2025). Testing for loss severity: Impact on loss reduction. Risk Management and Insurance Review
Working papers
Genetic testing: Assessing severity risk and implications for health insurance (with Helmut Gründl)
We theoretically analyze how genetic testing to determine risk severity impacts health insurance purchases under various information regimes. If claims payments can exceed actual loss, individuals get tested and choose appropriate contracts, resulting in efficient outcomes. If insurers only offer a contract covering average loss, testing is forgone, leading to inefficiencies: high-risk individuals are under-insured, and low-risk individuals are over-insured. When contracts limit payments to actual losses and insurers can observe test results, people avoid testing and purchase full coverage, causing high-risk individuals to underpay and low-risk individuals to overpay for insurance. If test results are unobservable, adverse selection requires contract designs to eliminate it, achieving efficiency. Our study suggests encouraging the sharing of genetic test results for contracts exceeding actual loss to ensure adequate coverage. For contracts limited to actual loss, privacy protections are essential to avoid discrimination and ensure fair pricing.
How does information on individual risk affect the demand for insurance (with Volker Benndorf)
In our experiment, we examine how ambiguity and the cognitive load required to process information affects accuracy and insurance decisions. With specific information on loss probabilities, individuals are accurate but least willing to pay for insurance. To address this, we introduced ambiguity by only sharing average loss probabilities, lowering accuracy but not affecting willingness to pay. We then incorporated information frictions, requiring individuals to exert effort in assessing risk factors. This reduced accuracy without increasing willingness to pay for insurance. We then restricted the amount of cognitive load by requiring individuals to only assess factors which increased risk. This improved accuracy of guesses and significantly increased the willingness to pay. Providing specific information on loss probabilities can help prevent over-insurance while requiring individuals to assess factors which increase risk can help prevent under-insurance.
Policy papers
Behind the Blackboard: How basic indicators mask gaps in quality of education (with Alyssa Chua Lee-Yen, Alia Muhammad Radzi and Rachel Gong)
This paper seeks to understand why Malaysia’s improving performance in basic education indicators does not translate into good performance outcomes in the international TIMSS assessment. We begin by briefly describing Malaysia’s performance in the TIMSS assessment over the years and across the Asia Pacific region. We find that Malaysia’s performance has deteriorated from 1999 until 2011, and only improved slightly in 2015. Compared to other countries, Malaysia consistently does worse. This could be due to the quality of teaching and the level of parental involvement in Malaysia. While there are enough teachers in Malaysia, they spend relatively little time in the classroom per day teaching and are not sufficiently equipped with the skills to effectively deliver mathematics and science content. Additionally, while parents are generally committed to spending appropriately on their children’s education, their direct involvement, for example, helping with their child’s homework is limited. This is further exacerbated by the proliferation of tutoring, which creates a disincentive for teachers to teach effectively during formal hours and restricts social mobility since only better-off households can afford better quality tutors.
More or less equal? Accounting for missing top and bottom incomes in measurement of income inequality in Malaysia (with Allen Ng)
Measurement of income inequality derived from household surveys typically suffers from the problem of under-sampling and/or under- reporting of households in the upper-end of the income distribution. This ‘missing income’ could lead to underestimation of income inequality. In this note, we try to account for the ‘missing income’ in both the upper and lower end of the income distribution in Malaysia for the period from 2012 to 2016. To account for the upper end of the income distribution, we calculate ‘missing income’ using the difference between household private final consumption component of national accounts and from the gross household income from the household surveys, and add this to the top decile of households. For the lower end of the income distribution, we use income earned by foreign workers, fit those educated up to the primary level of education in the bottom decile, and those educated up to the secondary level in the second lowest decile. We find that both these adjustments increase the level of income inequality. We then use Pareto imputation to stretch the top decile having accounted for ‘missing income’ to estimate the change in income shares of the top 1%. We find that the top 1% of the population saw declining income shares from 2012 to 2016.