Sub-Theme Economy (DOSM Pahang)

Malaysia measures household poverty through the Household Income and Expenditure Survey (HIES), which is conducted only two to three times every five years. At district level, DOSM has published results for just three years: 2019, 2022, and 2024.

Labour force statistics, by contrast, are published annually by district, and the Consumer Price Index is published monthly by state. This creates an information gap: between HIES rounds, state planners in Pahang have frequent employment data but no current picture of who is actually struggling.

The standard assumption is that employment data can stand in for wellbeing; if unemployment is low, hardship should be low. In Pahang, that assumption breaks down.

Analysis of DOSM's 2024 district-level data shows that employment and poverty in Pahang have decoupled: the districts with the most jobs are not the districts with the least poverty. The problem is therefore twofold. First, Pahang has a working-poor population that employment statistics cannot detect: residents are employed, but in work that does not lift the household above the Poverty Line Income, the monthly income threshold below which a household is officially counted as poor. Second, this population is invisible for years at a time, because the only instrument that measures it runs on a multi-year cycle.

Without a way to estimate household vulnerability between HIES rounds, district-level assistance subsidies, skills programmes, and targeted aid are allocated using data that may be up to three years old and using an indicator (unemployment) that demonstrably does not correlate with hardship in this state.

1. To quantify the relationship between employment indicators and household poverty across Pahang's 11 administrative districts and demonstrate statistically that unemployment is a poor stand-in for hardship; a district can have almost no unemployment and still have widespread poverty.
2. To develop a data-driven District Vulnerability Index that classifies districts by household economic risk using indicators available annually or monthly.
3. To build and validate a model that nowcasts estimates of present-day values before official figures exist for district median income and poverty rate for years in which no HIES survey was conducted.
4. To deliver an interactive dashboard that allows non-technical users in state government to see current estimated vulnerability by district.

All sources are open data from DOSM's OpenDOSM portal, licensed CC BY 4.0.

Supporting datasets for interpretation: crops_district_production (agricultural dependence), schools_district and enrolment_school_district (education access).

• Empirical evidence that unemployment rate is an inadequate targeting indicator for household assistance in Pahang.
• A public dashboard usable by state agencies, local councils, and NGOs to prioritise districts for intervention.

• The training set is small (33 district-year observations), so the model must be kept deliberately simple and its margin of error stated openly rather than hidden.
• DOSM's raw files contain inconsistent district naming across years (for example, "Cameron Highland" in 2018–2020 versus "Cameron Highlands" from 2021), requiring a documented cleaning step.
• Nowcast values are model estimates, not official statistics, and every interface element presenting them must label them as such.