2. Scope¶
2.1. Research Framework¶
2.1.1 The Research Question¶
Does proximity to high-quality primary schools generate a significant price premium in the HDB resale market? Specifically, we test whether this premium is concentrated at the 1km Phase 2C registration boundary or declines gradually with distance, after controlling for housing attributes and time-varying amenities.
2.1.2 Significance and Policy Impact¶
If access to “Good Schools” is concentrated within the 1km boundary, educational opportunity becomes tied to housing wealth, disadvantaging lower-income families. For MND and MOE, measuring this effect is important for assessing whether distance-based admission rules unintentionally reinforce socioeconomic inequality.
2.1.3 Causal Identification Strategy¶
Simple price comparisons are inadequate because popular schools are often located in mature estates with stronger amenities. To isolate the school effect, we use four complementary methods: Hedonic Regression, Regression Discontinuity Design (RDD), Spatial Error Model, and Town-level Heterogeneity Analysis.
2.2. Success Criteria¶
This project will be considered successful if it delivers the following:
2.2.1 Credible causal estimates of the school proximity premium.¶
The hedonic OLS and RDD models should produce stable and broadly consistent estimates across robustness checks. Where they differ, the analysis should explain the source of divergence.
2.2.2 Elimination of look-ahead bias.¶
By incorporating temporal join logic for MRT stations and shopping malls, the model should identify the extent of data leakage in prior research. Success is achieved if school-exposure estimates are more accurate when infrastructure is matched by actual opening dates.
2.2.3 Actionable affordability insights for MND.¶
The project should quantify the extent to which educational demand contributes to localized HDB price surges, providing evidence for more targeted housing or cooling policies aimed at preserving spatial equity.
2.2.4 Evaluation of admission policy impact for MOE.¶
By estimating price jumps at the 1km and 2km boundaries, the study should provide empirical evidence on the market effects of the current tie-breaker system, including whether it creates exclusive housing zones that link educational access to household wealth.
2.3. Assumptions¶
2.3.1 Geospatial Completeness¶
We assume the OneMap geocoding API provides accurate coordinates for most HDB addresses, and that non-geocodable records are missing at random. For polygon-based properties, centroid coordinates are treated as sufficient spatial representations.
2.3.2 Popularity as a Proxy for Demand¶
The school ranking, constructed from Phase 2B/2C balloting outcomes and characteristics such as GEP, SAP, and CCA diversity, is assumed to capture the key dimensions of school desirability influencing parental housing decisions.
2.3.3 Methodological Sufficiency¶
We assume that Hedonic Pricing, Regression Discontinuity Design, and Spatial Regression jointly provide a robust framework for isolating the school effect in the HDB resale market.
2.3.4 Boundary Invariance¶
We assume no major general equilibrium effects within the narrow RDD bandwidth; specifically, that the 1km admission boundary does not materially alter housing supply or land-use decisions in ways that confound the estimated price discontinuity.