HOUSING TYPES, POPULATION ATTRIBUTES AND THE DEVELOPMENT PATH OF HOUSING IN LARGE CITIES – A CASE STUDY OF ZHENGZHOU, CHINA

Zhifeng Shen, Jing Li, Ahsan Siraj, Shilpa Taneja, Zhihui LiInternational Journal of Strategic Property Management 2023 · DOI

TL;DR

This paper argues that the housing crisis in large Chinese cities isn't just about a lack of supply, but a fundamental mismatch between the types of housing built and the specific attributes of the population. By applying ridge regression to a decade of data from Zhengzhou, the authors forecast housing needs through 2035, providing a roadmap for urban planners to move away from speculative "financialized" real estate and toward a more stable, demand-matched housing market.

What Problem Does This Paper Address?

In many of China’s "mega-cities," housing has transitioned from a basic consumer good to a speculative investment product. This "financialization" has led to skyrocketing prices that outpace local incomes, creating a paradox: a surplus of high-end commercial housing exists alongside a severe shortage of affordable options for the people who actually drive the city's economy.

The authors identify a critical gap in urban planning: most models look at total population size but ignore population attributes—such as whether residents are migrant workers, university students, or registered "hukou" holders. Without understanding who is moving in, cities continue to build the wrong types of units, leading to social inequality, long commutes, and housing bubbles.

Methods

The researchers conducted a case study of Zhengzhou, a major transportation hub in China. Their approach involved three main steps:

  1. Variable Selection: They identified five independent population variables (Permanent residents, Registered population, Migrant workers, University graduates, and College students) and three dependent housing variables (Commercial housing, Talent apartments, and Affordable housing).
  2. Ridge Regression: Standard linear regression often fails with demographic data because population categories overlap (a problem called multicollinearity). To solve this, the authors used Ridge Regression, a technique that adds a degree of bias to the estimates to make the model more stable and reliable when variables are highly correlated.
  3. Forecasting: Using SPSS, they first predicted population trends for Zhengzhou from 2021 to 2035. They then plugged these predictions into their regression equations to forecast the necessary scale of each housing type.

Datasets

The study utilized official government data from 2010 to 2019, specifically:

  • Henan Statistical Yearbook (HAPBS): For population demographics and general economic indicators.
  • Zhengzhou Housing Security and Real Estate Bureau Annual Reports: For specific data on the floor area of new commercial, talent, and affordable housing projects.

No new benchmarks were introduced, but the integration of these specific administrative datasets for predictive modeling at the city level is a key contribution.

Key Insights and Ideas

  • The "Registered" Driver: The registered population (those with local hukou) has the strongest impact on the demand for commercial housing. This suggests that official residency status remains the primary gateway to traditional homeownership in China.
  • The Student-Affordability Link: Interestingly, college students and the registered population are the strongest correlates for affordable housing. As students graduate but remain in a "transitional" economic phase, they rely heavily on subsidized or low-cost options.
  • The Talent Apartment Mismatch: The data shows that migrant workers and the general resident population influence the demand for "talent apartments" more than graduates do. This suggests that current "talent" housing might be serving a broader demographic of mobile workers than originally intended.
  • Inverse Correlation for Permanent Residents: The model found an inverse relationship between the permanent resident population and certain housing types. The authors attribute this to "exorbitant housing prices and the exclusionary effect"—as the city grows, the most stable residents may actually be priced out of new developments that gravitate toward investment capital rather than local needs.

Key Results

The authors’ forecast for 2035 highlights a massive required scale-up in housing:

  • Commercial Housing: Predicted to grow from roughly 3,442 units (in 10,000 $m^2$) in 2021 to over 10,690 by 2035.
  • Affordable Housing: Expected to nearly quadruple in demand, rising from 439 to 1,584 (in 10,000 $m^2$) over the same period.
  • Talent Apartments: While the smallest sector, it shows the steepest relative growth, jumping from a forecast of 13 to 112 (in 10,000 $m^2$).

Quantitatively, the Ridge Regression models showed high reliability, with $R^2$ values (a measure of how well the model fits the data) ranging from 0.957 to 0.990 across the three housing types.

Limitations and Future Work

The authors acknowledge several limitations:

  • Geographic Scope: The study is limited to Zhengzhou. While it serves as a representative "national central city," the dynamics might differ in "first-tier" cities like Beijing or "third-tier" regional towns.
  • Missing Income Data: The study lacks specific income data for different demographic groups. The authors note that while they use "graduates" as a proxy for a certain income bracket, direct financial data would make the model much more robust.
  • Demographic Shifts: The model may not fully account for the long-term "sluggish" population growth and aging trends currently facing China, which might decelerate the rate of new construction more than the linear model suggests.

Future research plans include expanding the dataset to multiple cities for a comparative analysis and incorporating more granular income and age-structure data.

Summary updated

Abstract

Large-city housing costs are commonly attributed to excessive investment or financialization of the property market. The primary cause of housing issues is the imbalance between housing supply and demand. In the context of the urban housing market, the dynamic matching of housing supply and population type plays a crucial role in promoting equilibrium, maximizing the effectiveness of the market’s long-term mechanism, and preserving the coordination and stability of the market. To illustrate this point, the paper analyzes the impact of different attributes of population on housing demand from the perspective of dynamic matching between population attributes and housing types. Taking Zhengzhou City in Henan Province as an example, we consider the relationship between population attributes and housing types in the past, construct a theoretical model of population and housing type and anticipate the future supply scale of various housing kinds. Ultimately, our findings provide a theoretical justification for curbing excessive financialization and housing bubbles in large cities. Furthermore, we propose a practical method for urban housing to return to the essence of life from the perspective of matching housing supply and demand.
BibTeX
@article{shen2023housing,
  author = {Zhifeng Shen and Jing Li and Ahsan Siraj and Shilpa Taneja and Zhihui Li},
  title = {{HOUSING TYPES, POPULATION ATTRIBUTES AND THE DEVELOPMENT PATH OF HOUSING IN LARGE CITIES – A CASE STUDY OF ZHENGZHOU, CHINA}},
  journal = {International Journal of Strategic Property Management},
  year = {2023},
  doi = {10.3846/ijspm.2023.20425},
}
Figure