How Statistical Models Predict China's 2015 Housing Demand

Problem Restatement

The real‑estate sector is a pillar of the national economy and closely tied to people's lives; its development influences overall economic trends and living standards. Rapid growth in recent years has contributed to economic progress and improved housing conditions, but the industry also faces serious challenges. Accurate quantitative analysis and effective regulation are therefore essential.

Model Assumptions

Assumption 1: Data are not affected by policy or other human interventions.

Assumption 2: Natural disasters and social factors are ignored in predictions.

Assumption 3: Collected data have a reasonable degree of accuracy.

Assumption 4: Geographic location, transportation, regional density, and community maturity are omitted.

Assumption 5: Effects of demolition, household splitting, and re‑formation are not considered.

Housing Demand Model

Influencing Factors

The model predicts total urban housing demand using panel data. The dependent variable Y is the per‑capita built‑up area (m²). Independent variables include: X1: Per‑capita GDP (yuan) X2: Total sales value of commercial housing (100 million yuan) X3: Urban per‑capita disposable income (yuan) X4: Savings (100 million yuan) X5: Urban population (10 000 persons) X6: Urban employment (10 000 persons)

Data Collection

All data are sourced from the National Bureau of Statistics.

Correlation Analysis (SPSS)

The Pearson correlation matrix shows that all seven indicators have coefficients above 0.8 with p‑values near zero, indicating strong positive relationships.

Regression Fitting (SPSS)

The multiple R and R‑squared values demonstrate that the regression equation is highly significant.

ANOVA results indicate a regression sum of squares of 732.789, df = 6, mean square = 122.132, F = 1237.976, and a significance level of 0.000, confirming the model’s overall significance.

Y = -14.931 - 7.502e-5*X1 - 3.016e-5*X2 - 1.054e-5*X4 - 0.001*X5 + 0.003*X6

Using 2015 data (X1=49820, X2=83920.43, X3=27998.1, X4=48101.7, X5=77230, X6=41707) yields Y ≈ 35.9 m², giving a total housing floor‑area demand of 2,772,557 m².

Time‑Series Modeling (EViews)

The plot shows a linear relationship between Y and time T.

The regression model derived is:

Y = 1.047218*T + 14.58421

Forecast

The model predicts a 2015 per‑capita floor area of 36.58 m², corresponding to a total housing demand of approximately 2,824,748 m², which closely matches the SPSS prediction.