Have you had difficulty finding housing as a student in Rotterdam? Chances are that part of this is the result of high prices. It is common knowledge that the increasing demand for housing alongside the limited supply in recent years have driven up prices, making it difficult for people to find affordable accommodation. Furthermore, some homes sell for millions while others struggle to find buyers, raising the question of what truly determines housing prices. Econometrics, the application of statistical and mathematical models to economic data, provides a framework for analyzing these factors and predicting home values. By using econometric techniques, buyers, sellers, and policymakers can make informed decisions about pricing, investment, and urban development.
One of the most commonly used models in real estate to estimate housing price levels is the Hedonic Pricing Model (HPM). This model predicts the value of a house or apartment by decomposing the price of previously sold homes into various contributing factors, allowing economists and statisticians to estimate how much different characteristics influence the value of a property. Structural attributes such as the size of the home, number of bedrooms and bathrooms, age, and any renovations are possible components. Location factors like proximity to good schools, low crime rates, and access to public transportation may also be applied. Finally, broader market conditions, such as interest rates, inflation, and economic growth, can influence the housing market significantly, and may hence be applied to the model. A simple example of such a model is the following:
Price = β₀ + β₁(Size) + β₂(Bedrooms) + β₃(Distance to City) + β₄(Crime Rate) + ε
Here, each β coefficient measures the influence of a specific factor, while ε captures unobservable elements like buyer preferences. The HPM helps to predict why certain homes are priced higher, giving both buyers and sellers some data-backed insights into value. It should be noted that the Hedonic Pricing Model is basically just a standard linear regression, with Price as the dependent variable and pricing attributes as the independent variables.
One of the most noteworthy benefits of the Hedonic Pricing Model is the buyer's ability to reduce asymmetric information, a situation where one party (often the seller) knows significantly more about the value of a home than the other (the buyer). In the past, buyers had to rely mainly on intuition, subjective appraisals, or the word of the seller, which may lead to overpaying. By contrast, the HPM offers a more objective, data-backed estimate of what a home should be worth based on its features and market context. Hence, sellers are less able to overcharge for homes, while buyers gain more bargaining power. In this way, while relatively simple, the model brings more transparency and fairness to the housing market, reducing market imperfections often arising from asymmetric information.
While beneficial due to its simplicity and ease of implementation, Econometricians have come up with better models for housing price predictions. Specifically, they combine basic regressions with Machine Learning (ML) models. Machine learning models are computer programs that are used to recognize patterns in data or make predictions. Examples of these are random forests and decision trees, neural networks, and time series forecasting. While more complex in terms of application, they make more realistic assumptions such as non-linear prices, complex interactions between terms, and time trends. Furthermore, these models need to be trained with previous data and keep learning as more data is added, resulting in better predictions over time.
One example of the application for ML algorithms to predict housing prices is Zillow’s Automated Valuation Model (AVM). Zillow, which is a leading real estate platform based in the US, uses econometrics and machine learning models to generate a housing price estimator called a Zestimate. This tool uses ML models to predict home values based on millions of data points, improving accuracy over time. As the model receives more and more data points over time, it is able to improve its accuracy and predict more accurate housing estimates.
However, one must be careful when relying heavily on such tools, as no model is perfect. The results should be used as an estimate, not as a completely accurate valuation. Zillow faced losses when its model overestimated housing demand, proving that even advanced models have risks. According to the Zillow website, the Zestimate’s accuracy may depend on the location of the property and the availability of data in an area. The Zestimate’s median error rate for on-market homes in the US is 3.2%, meaning Zestimates for half of all on-market homes are within 3.2% of the ultimate sale price, and half are outside that range. For off-market homes, the median error rate is 7.52%. Hence, considering homes are often priced in the hundreds of thousands to the millions, these ranges in estimator accuracy are high from a monetary perspective.
From a real-world implications perspective, the impact of econometrics in real estate is far-reaching. For homeowners and realtors, econometric models provide guidance on setting listing prices that reflect both property features and market dynamics, a challenge when people have to assess housing prices subjectively. Investors, on the other hand, may use these tools to spot undervalued properties or assess long-term return potential, reducing some of the risk of investment. Policymakers and urban planners also rely on housing price models to evaluate affordability and plan infrastructure investments, indicating that housing prices have macroeconomic implications as well.
In a housing market affected by rising prices and limited availability, especially in cities like Rotterdam, understanding what drives home values is crucial to reduce overpaying. Simple tools such as the Hedonic Pricing Model offer a clear, understandable way to analyze the many factors influencing property prices, reducing asymmetric information. While simple models provide some insights, the integration of machine learning techniques has largely improved the accuracy of econometric predictions. While not without limitations, the existence of such algorithms creates a canvas for future development. As data continues to grow in both quantity and quality, these models will become even more valuable in guiding real estate decisions. Ultimately, whether you're a student looking for a room, an investor seeking an opportunity, or a policymaker determining urban development, econometrics offers powerful tools for navigating the complex world of real estate