Model of the Influence of Air Pollution and Other Environmental Factors on the Real Estate Market in Warsaw in 2010–2022

Abstract

Air pollution has a significant impact on the housing market, both in terms of property prices and buyer preferences, as well as urban development. Below, we present the main aspects of this impact. These may include a decline in property values in polluted areas, a change in buyer preferences (more buyers are taking environmental factors into account when choosing a home, including air quality—both outdoor and indoor—which translates into increased demand in ‘green’ neighborhoods), the development of energy-efficient and environmentally friendly buildings, the impact on spatial planning and urban policy, health effects, and the rental market. The study showed that air pollution has a significant negative impact on housing prices in Warsaw, particularly in relation to two pollutants: nitrogen dioxide (NO₂) and particulate matter (PM_2.5). As their concentrations decreased, housing prices increased, with the highest price sensitivity observed for smaller flats on the secondary market. The analysis used GRM and OLS statistical models, which confirmed the significance of the relationship between the concentrations of these pollutants and housing prices (per m²). NO₂ had a significant impact on prices in the primary market and on the largest flats in the secondary market, while PM_2.5 affected prices of smaller flats in the secondary market. No significant impact of other pollutants, meteorological factors, or their interaction on housing prices was detected. The study also showed that the primary and secondary markets differ significantly, requiring separate analyses. Attempts to combine them do not allow for the precise identification of key price-determining factors.

1. Introduction

1.1. Theoretical Background

Ensuring adequate air quality is particularly critical in urban areas, where a high concentration of emission sources is confined within relatively small spaces. As the proportion of populations residing in large cities steadily increases, maintaining clean air is expected to become a significant public health challenge, thereby prompting intensified research in this area. A growing number of evidence indicates that prospective homebuyers are increasingly considering environmental aspects in their decision-making processes, with particular emphasis on both outdoor and indoor air quality. This trend has resulted in heightened demand for environmentally sustainable neighborhoods and has stimulated the advancement of energy-efficient, eco-conscious building practices [1].

Extensive studies have shown that air pollutants are linked to a wide range of adverse health outcomes, from mild respiratory irritation to severe systemic conditions such as hypertension, and even premature mortality due to stroke or sudden cardiac arrest [2,3]. Moreover, an expanding body of research underscores that atmospheric pollutants can not only induce physical health disorders but also contribute to psychological conditions [4,5]. These pollutants have also been identified as potential contributors to cognitive decline and accelerated aging, as they promote neurodegenerative processes [3,6,7,8]. Further studies suggest that urban areas with low levels of fragmentation and limited urban sprawl experience lower concentrations of PM_2.5 pollution compared to cities with more dispersed, fragmented, and complex urban layouts [9,10].

Cárdenas, Dupont-Courtade, and Oueslati demonstrated that a higher degree of urban fragmentation correlates with increased population density and poorer air quality in urban areas [11]. Their findings indicate that dense, compact, and minimally dispersed urban structures tend to exhibit better air quality in developed countries. In areas with low road density, the development of transportation infrastructure generally facilitates a decrease in PM2.5 concentrations. In contrast, when traffic intensity reaches a high threshold, increased roadway congestion contributes to elevated PM_2.5 levels.

Excluding meteorological and geographic factors, cities with advanced urban development generally exert a smaller impact on PM_2.5 concentrations. Sun, Chen, and Zhang recommend that such cities adopt moderately decentralized, polycentric development models [12]. However, the connection between urbanization, increased traffic demand, higher energy consumption, and deteriorating air quality remains evident [13].

A key challenge for contemporary society is minimizing transportation-induced air pollution and its associated health risks by reducing harmful emissions. Air pollutants, particularly nitrogen oxides (NO_x) and particulate matter (PM), which are excessively emitted by road traffic, have consistently been identified as major contributors to urban air quality degradation.

In the Warsaw metropolitan area, for instance, a significant portion of total emissions arises from vehicle traffic and fuel combustion. Traffic pollution, in the form of dust, is primarily generated through the abrasion of brakes, tires, and road surfaces, as well as the drift of pollutants from these surfaces, while nitrogen oxides are emitted from vehicle exhaust pipes.

Improving air quality in Warsaw would require expanding infrastructure to redirect traffic away from urban agglomerations, phasing out diesel-powered vehicles, and encouraging the adoption of hybrid and electric cars [14]. Investments in expanding the green public transport fleet and further developing metro lines represent additional strategies to improve air quality [14].

The principles of sustainable development aim to balance economic, ecological, and social dimensions. Achieving this balance requires concerted efforts to improve the natural environment, particularly through sustainable mobility solutions. Sustainable transportation promotes environmentally friendly mobility and encourages behavioral shifts and adaptive strategies across all economic sectors [15,16,17,18,19].

The relationship between air pollution and real estate values has been the subject of research since the late 1960s. Ridker and Henning conducted one of the earliest studies in 1967, employing a hedonic pricing model to examine this relationship in the United States [20]. Their results demonstrated that reducing sulfate concentrations by 0.25 mg/day increased housing values by 84–245 USD. Since then, numerous studies have explored the link between air quality and both housing and land prices [21]. Boyle and Kiel’s 2001 review of 12 studies revealed mixed findings: some studies indicated statistically significant relationships between air quality and housing prices [20,22,23,24], while others found no significant correlation [25,26].

In the 21st century, the degradation of environmental conditions and increasing public awareness of health issues have spurred further investigations into this subject. Many studies rely on hedonic pricing models to assess the relationship between air quality and real estate values [27,28]. Commonly used spatial hedonic models include the spatial lag model (SLM) [29,30], spatial error model (SEM), spatial Durbin model (SDM) [31], geographically weighted regression (GWR) [32], quantile regression models (QRM) [33], and bootstrap autoregressive distributed lag (BARDL) models [34].

The objective of this study is to quantify the impact of air pollution on residential property transaction prices, as well as on transaction volumes in both the primary and secondary real estate markets in Poland. Further research is being conducted across regions differentiated by levels of air pollution, geographic characteristics, and real estate market dynamics. The initial study was conducted in the Gdańsk agglomeration, an area characterized by high air quality, to establish a baseline for comparative analysis with Warsaw and Kraków. This approach facilitates comparisons with regions exhibiting varying pollution levels, such as Warsaw (moderately polluted) and Kraków (heavily polluted), over an extended period (2010–2022, analyzed quarterly). Ensuring the comparability of findings across these areas is achieved through a consistent methodological framework, including the use of high-quality data and the same stochastic models. This paper represents the second phase of a comprehensive investigation into the relationship between air quality and real estate market dynamics.

1.2. Geographical Characteristics of Warsaw

Warsaw, the largest city in Poland and the capital of both the country and the Mazovian Voivodeship, has a population of 1,794,166. It is a significant scientific, cultural, political, and economic hub. The Mazovian Voivodeship lies within the temperate climate zone. Due to its central location in Europe, the climate of this region is influenced by both maritime and continental factors. The dispersion of pollutants in the lower layers of the atmosphere is determined by meteorological factors, including wind speed and direction, precipitation, air temperature, and the vertical dynamic structure of the atmospheric boundary layer. In this central region of the country, the average wind speed typically falls within the range of 5 to 10 m/s. In 2021, the annual mean air temperature across most of the Mazovian Voivodeship ranged between 8 °C and 9 °C, while annual total precipitation ranged from 500 mm to 700 mm, with high spatial variability across different months.

The Mazovian Voivodeship is characterized by a highly diverse industrial distribution, with most industrial activities concentrated in urban areas, particularly within the Warsaw agglomeration and its vicinity, as well as in Płock, Radom, Ostrołęka, Siedlce, and Ciechanów. Key sectors include energy, chemicals, food processing, machinery, and textiles, with the petrochemical industry being particularly prominent in Płock. The region plays a pivotal role in the country’s infrastructure systems, serving as a central point for road, rail, and air transport, urban communication, and energy networks.

Due to the above-mentioned factors, Warsaw has a moderate level of ambient air pollution among major agglomerations in Poland (Figure 1).

At the same time, residential property prices in Warsaw have consistently ranked as the highest in Poland during the studied period (Figure 2), demonstrating a high growth rate despite market fluctuations induced by the COVID-19 pandemic.

1.3. Air Quality Indicators in Warsaw

The primary source of air pollution in the Mazovian Voivodeship is anthropogenic emissions from the municipal and residential sectors, classified as surface emissions. Secondary contributors include emissions from transportation, categorized as linear emissions, and industrial activities, referred to as point emissions. Furthermore, a substantial share of air pollutant concentrations in the region can be attributed to transboundary inflows, both from other areas within Poland and from neighboring regions across Europe (Figure 3).

The main local sources of air pollution include emissions from chimneys of individually heated houses and road transportation, both of which significantly influence pollutant concentrations, particularly in areas adjacent to high-traffic roadways. Industrial activities within the Mazovian Voivodeship, primarily from the energy sector, contribute to pollution export beyond the region due to the elevated heights of industrial chimneys, which facilitate the dispersion of emissions over long distances.

The primary pollutant emitted by road transport and point sources in Warsaw is nitrogen dioxide (NO₂). The average annual concentration of nitrogen dioxide is measured at three reference air quality monitoring stations situated along three streets in Warsaw, as shown in Figure 4.

The station named MzWarAlNiep is traffic-oriented, located in close proximity to a road with very high traffic volumes. This indicates that the above-limit concentration of NO₂ is directly linked to emissions from road transport. The other two stations are urban stations. A noticeable decline in NO₂ concentration at the traffic monitoring station in 2020 can be attributed to the onset of the COVID-19 pandemic. During this time, quarantine, mobility restrictions, remote learning, and significant limitations on sectors such as tourism, gastronomy, and retail led to a reduction in car traffic in Warsaw.

Additionally, compact, low-rise buildings, along with associated heating processes in the municipal and residential sectors (surface emissions), significantly contributed to elevated concentrations of particulate matter PM₁₀ and PM_2.5, especially from the traffic station (Figure 5 and Figure 6).

According to research conducted by the city hall for the purposes of the Environmental Protection Programme for the capital city of Warsaw (2021–2024), the highest impact on the concentrations of PM₁₀ and PM_2.5 in Warsaw (over 40%) is attributed to transboundary pollution from outside the city. These pollutants primarily originate from individual heating systems using solid fuel boilers (coal and wood) and road traffic.

Analysis of the data clearly reveals a consistent downward trend in pollution levels for each of the studied compounds. This decline can be attributed to various factors. Notably, since 2017, new legal regulations, such as the so-called anti-smog resolution, were introduced, which set requirements for heating devices and deadlines for their replacement with more environmentally friendly alternatives. Additionally, bans were imposed on the use of low-quality fuels.

The opening of the second metro line in 2018 contributed to a reduction in car transportation, leading to a decrease in air pollution levels. Since 2017, the Warsaw City Hall has also been running a financial support program to assist residents in upgrading their heating systems. Moreover, social campaigns have been promoting activities aimed at improving air quality. In 2019, the Office for Air Protection and Climate Policy was established to coordinate efforts towards the development of a cleaner and greener Warsaw, with the goal of achieving climate neutrality by 2050.

In 2020, a new Air Protection Programme for the Mazovia region was introduced, outlining corrective actions aimed at improving air quality. These measures include the inventory of heat sources, the establishment of low-emission zones, and the modernization of public transportation fleets, among others.

1.4. Economical Characteristics of Warsaw

The prices of residential properties are determined by the interplay of demand and supply dynamics, as well as production costs. Demand and supply are influenced by social and macroeconomic factors, and key variables affecting housing prices include Gross Domestic Product (GDP), inflation, interest rates, wage levels, and unemployment rates. It is notable that the strength and impact of these factors can vary significantly even within different districts of the same city, reinforcing the highly location-dependent nature of residential property prices [35].

Social factors that influence fluctuating demand in the residential property market include migration patterns, demographic trends, settlement preferences, and broader patterns of social behavior. The unique characteristics of the real estate market are further shaped by the rigidity of supply due to limited land availability, the finite number of available properties in the short term, and regulatory restrictions [36]. Additionally, the interaction between demand and supply factors varies in intensity over time, contributing to the seasonal nature of the housing market.

Warsaw, as the capital of a country with a population of 37 million and the largest business, transportation, and academic hub, experiences continuous growth in demand for real estate, both in the sales and rental markets. To analyze the dynamics of real estate prices in Warsaw during the studied period, it is essential to examine the local determinants of price changes. Over the last decade, favorable economic conditions have significantly contributed to the city’s development. Stable income growth (Figure 7), combined with low unemployment rates (Figure 8) and historically low interest rates (Figure 9), has led to a sharp increase in demand for residential properties. Interest rates experienced a sudden increase in 2022 because of the post-pandemic economic crisis and the geopolitical instability caused by the Russian invasion of Ukraine.

This economic stability and continuous development, combined with the permanent influx of the working-age population (Figure 10), has positioned Warsaw as an attractive market for residential real estate. However, this trend was disrupted in 2020 due to the COVID-19 pandemic. Positive net migration—defined as the inflow of population for permanent residence—is an additional factor driving demand for residential properties.

During the analyzed period, the growing demand was met and supported by an increasing number of completed dwellings. This growth can be attributed to the recovery from the post-crisis downturn observed internationally in previous years and the microeconomic situation in Poland (Figure 11).

2. Data

The study utilized selected data on the real estate market and the results of measurements of selected pollutant concentrations that are crucial to the issue at hand. Both data repositories were combined into a single aggregate for quarterly periods by summing and weighting the variables.

The results of air pollution concentration measurements in Warsaw from 2010 to 2022 come from reference devices measuring air quality within the network of the Chief Inspectorate for Environmental Protection. Air quality monitoring stations in Poland, operating within the State Environmental Monitoring system, are located according to representativeness criteria outlined in national and European legal acts (e.g., Directive 2008/50/EC of the European Parliament and Council, 21 May 2008, on ambient air quality and cleaner air for Europe [37]). The study used four variables representing the results of hourly measurements of air pollutant concentrations in mg/m³ (PM_2.5, PM₁₀, NO₂, and O₃), aggregated to quarterly periods. The raw data, prior to aggregation, underwent a detailed quality analysis, including interpolation of missing data, removal of data by cases or pairs, and a detailed, multi-stage analysis of anomalies based on robust estimators (DFITS, Cook’s distance, Mahalanobis distance, and classic Grubb and Rosner tests) [38].

The variability in the obtained pollutant concentrations (Figure 12) indicates strong seasonality in annual cycles, particularly for ozone and particulate matter PM_2.5, and less so for particulate matter PM₁₀, especially since 2018. Changes in nitrogen dioxide concentrations do not show clear seasonal patterns.

The second group of variables used in the study consisted of data from the real estate market in Warsaw, namely the number of residential properties sold (variables named ‘Prem’) and transaction prices (‘Prices’) per square meter of residential properties in Warsaw, divided into the secondary market, primary market, and total market, adjusted for inflation. Data on sales volumes and prices came from the most reliable source in Poland—the Central Statistical Office (GUS)—and were compiled based on information obtained from the Real Estate Price Register (RCN) maintained by district authorities. The register contains data from notarial deeds. The data was segmented according to the size of the flats: flats up to 40 m², from 40 to 60 m², from 60 to 80 m², over 80 m², and all flats without division by size.

The total number of residential properties sold in Warsaw (Figure 13) from 2010 to 2013 remained at 2500–3000 per quarter, and from 2013 to 2016 it increased to around 5500–6500 transactions per quarter. This level remained relatively stable until the end of the observed period. The total number of residential properties sold (‘Prem_Tot’) refers to the primary and secondary markets combined. The growing demand for residential properties in Warsaw is evidenced by the steady increase in the number of units sold (Figure 13), which, as previously indicated, is characterized by significant seasonality. Residential properties in Warsaw are purchased for both private use and investment purposes, such as resale or rental (capital investment).

During the analyzed period, the share of total residential property sales varied across segments. The largest share of sales was accounted for by flats with an area of 40 to 60 m², while the smallest share was accounted for by the largest flats, with an area exceeding 80 m². Over the past 12 years, both the secondary and primary real estate markets in Warsaw have experienced substantial growth, particularly after 2017, which was attributed, among other factors, to the completion of subsidies under the governmental program ‘Apartment for Young People’. In most of the years observed, a decline in the number of transactions in the last quarters of the year can be seen, which indicates seasonality in the real estate market. This is particularly evident in 2017 and 2021, when sharp declines in the number of flats sold in the fourth quarters of the above-mentioned years were the result of record increases in the first, second, and third quarters of 2017 and 2021, further reinforced by seasonal effects (Figure 14).

The volatility of the number of flats sold on the primary market (‘Prim’) in 2010–2012, in total and by segment, is presented in Figure 15. From 2010 to 2014, the volume of flat sales on the primary market grew steadily from approximately 500 to around 1500 units per quarter. From the end of 2014, there was a period of intense growth, which peaked in mid-2016 at 3700 units. The following 18 months saw a period of intense declines in sales in the primary market, and by the end of 2017, the level of 2011 was reached below 800 units. The subsequent years saw a recovery in sales volume to another record level of over 3400 units in the second quarter of 2021, followed by the sharpest decline in the analyzed decade, to 460 units in the fourth quarter of 2021. Throughout the analyzed period, residential properties with an area of 40–60 m² accounted for the largest share of sales, while the smallest share was held by the largest flats, i.e., those with an area exceeding 80 m², from 2015 onwards.

On the secondary market (‘Sec’), similarly to the primary market, the total number of residential properties sold in the analyzed period showed an upward trend, although the rate of change was more moderate and the number significantly higher (Figure 16). From 2010 to 2014, the total volume of residential property sales on the secondary market grew from approximately 1750 to approximately 2600 units per quarter. Since 2015, the rate of change has accelerated, continuing the upward trend. In the fourth quarters of 2016, 2017, 2020, and 2021, there were sharp but short-lived declines in sales and a rapid return to levels between 3000 and 4000 units per quarter. The structure of apartment sales by segment remained relatively consistent throughout the entire period analyzed. The largest share was held by residential properties with an area of 40–60 m², followed by those up to 40 m², while the smallest share was held by the largest apartments, i.e., those over 80 m².

During the analyzed period, prices per square meter of residential properties on both the secondary and primary markets showed significant volatility, primarily due to macroeconomic factors such as GDP, inflation, interest rates, and government housing policy. In 2010–2013, both markets and all segments recorded a decline in prices per square meter of residential properties (Figure 17 and Figure 18), which was due to the deterioration of the country’s economic situation following the global crisis of 2008–2009. However, since 2014, there has been a gradual increase in prices per square meter of residential space in both markets and across all segments, apart from the largest dwellings—those with an area exceeding 80 square meters in the primary market. For this segment, prices exhibited significant volatility, ultimately forming a downward trend throughout the analyzed period. In contrast, in the other segments, both on the primary and secondary markets, long-term trend lines show an upward trajectory.

In the analyzed period, both on the primary and secondary markets, the highest prices per square meter were recorded for the largest flats, i.e., those with an area exceeding 80 square meters, apart from the period from the third quarter of 2020 to the third quarter of 2021. During this period, the prices for the smallest flats, i.e., those under 40 square meters, were the highest in the primary market.

The changes described in both the primary and secondary markets are reflected in similar dynamics across the entire market (Figure 19). In 2010–2013, there was a decline in prices per square meter of residential properties on both markets, while since 2014, a clear upward trend has been observed. The trends for all residential segments, when combining both markets, exhibit a positive slope, reflecting a long-term increase in prices per square meter of residential properties.

As seen in both markets, the highest price per square meter of residential properties was recorded in nearly all periods for the largest properties. Only at the turn of 2020 and 2021 were higher prices recorded for the smallest properties.

3. Methods and Models

To achieve the objective of this study, this report uses data on transaction prices and the number of residential properties sold, obtained from the Central Statistical Office. Data on air pollution concentrations are the results of measurements taken by the state air quality monitoring system.

Both data sets were used to ensure the reliability of the sources and the conclusions derived from the stochastic cause-and-effect models. The selected models for identifying real estate market factors related to the selected air pollution parameters allow for the precise identification of the impact of these factors on any measurement scale, both in terms of the individual factors themselves, as well as the combined impact of the factors and their interactions.

Currently, no statistical models are more effective in identifying both factors and their interactions over such a broad spectrum. Ensuring the highest quality of data at this methodological stage and clarifying the links between the models should ultimately ensure reliable conclusions.

To identify and characterize statistically significant effects of individual variables and their interactions on property prices, statistical models from the OLS (Ordinary Least Squares) and GLM/GRM (Generalized Linear Regression Models) families were used.

The OLS model is a classic cause-and-effect model, enabling the assessment of relationships between phenomena or characteristics. It establishes a relationship between variables described as causes (independent, explanatory variables) and a variable representing the effect (dependent, explained variable). This influence is formally recorded as regression functions, which determine how the values of the dependent variable correspond to specific values of the independent variable. OLS models allow for the identification of linear relationships, that is, the strongest relationships, between independent variables (explanatory variables) and dependent variables. The OLS model can take the following structural form:

$$y_t={\beta }_0+{\beta }_1{x}_{t1}+{\beta }_2{x}_{t2}+{\beta }_3{x}_{t3}{+\cdots +{\beta }_k{x}_{tk}+\epsilon }_t$$

where

• $\beta$—are unknown, constant parameters over time,
• ${\beta }_0$—is the structural parameter of the intercept,
• ${\beta }_i$—are the structural parameters of variables, which reflect the strength and direction of the explanatory variable’s influence on the dependent variable,
• x_t1—NO₂, x_t2—PM₁₀, x_t3—PM_2.5,
• ${\epsilon }_t$—random variable,
• k—is the number of explanatory variables in the model.

Due to the possibility of non-linear relationships between variables or their collinearity, GRM (Generalized Regression Models) were also used in the study. GRM allows for the examination of complex experimental systems and can include both qualitative and quantitative variables. This is particularly important when analyzing data expressed in different measurement scales. Another important advantage of GRM is its ability to describe curvilinear relationships between variables. It is possible to apply appropriate predictor transformations, as well as substitution methods using a standardized variable and a series of selected linear variable transformations, as well as a non-linear linking function. GRM models were used to represent both air pollution and the real estate market. The identification of interactions is a particularly crucial element of the presented research. GRM models and the identification process have been described in detail in previous works [14,38,39].

Unlike other models, the GRM model is not a model in the strict sense, but a model path encompassing many classes of models and estimation methods. The GRM model combines these models and allows the identification of a cause-and-effect relationship, regardless of the measurement scale of the independent variables.

In this study, the coefficients of determination and adjusted coefficients of determination were primarily used as measures of model fit, i.e., the agreement between empirical data and the values calculated based on the model. These values are interpreted as the degree to which the constructed model explains the variability of the explained variable, taking values from the range of 0 to 1. The higher the values of both coefficients, the greater the proportion of the variability of the explained variable that can be explained by the model. The root mean square error (RMSE) was also used as an auxiliary measure in the analysis. RMSE can take any non-negative value, and the lower its value, the lower the average error, meaning the model can be considered more accurate. This measure is resistant to the non-linearity of models and allows the goodness of the model to be assessed under all conditions. Detailed information on model fit measures can be found in sources, such as [40,41].

4. Analysis Results

In the analysis assessing the impact of air pollution and meteorological conditions on the real estate market in Warsaw, all possible models were constructed for all combinations of independent variables and their first-order interactions. The models were constructed for the primary market, the secondary market, and the housing market in Warsaw as a whole. Two groups of dependent variables were used to characterize the real estate market: prices per square meter of residential premises and the number of flats sold for both the primary and secondary markets. Prices per square meter of flats on the aforementioned markets were divided into segments according to the size of the flats: up to 40 m², from 40 to 60 m², from 60 to 80 m², over 80 m², and flats without division by size. This segmentation resulted in the construction of over 3000 price models. Only those models that met all statistical assumptions—i.e., had all relevant structural parameters and confirmed linearity—were used for further analysis.

The basic parameters of the distributions of all analyzed variables are presented in

Table 1. Descriptive Statistics of key variables.

Variable	Valid N	% Valid obs.	Mean	Median	Minimum	Maximum	Std.Dev.	Coef.Var.	Skewness
WAW.NO2	56	100.0	31.8	31.9	22.0	41.1	5.4	17.0	0.0
WAW.PM10	56	100.0	31.6	29.3	19.5	53.7	8.7	27.5	0.7
WAW.PM2.5	56	100.0	21.9	19.3	9.6	40.2	9.2	42.1	0.7
Price1mPrim40real	52	92.9	7475.4	7400.3	6041.2	9994.4	711.3	9.5	1.0
Price1mPrim40_60real	52	92.9	7270.8	7163.6	6478.9	8709.1	536.3	7.4	0.8
Price1mPrim60_80real	52	92.9	7231.7	7235.4	6337.9	8921.0	476.4	6.6	0.8
Price1mPrim80real	52	92.9	8372.1	8255.8	6786.6	11132.1	963.8	11.5	0.8
Price1m2 Prim totalreal	52	92.9	7652.2	7454.3	6631.0	9455.0	632.7	8.3	0.7
Price1mSec40real	52	92.9	7964.4	7774.5	6598.9	9912.2	1019.8	12.8	0.5
Price1mSec40_60real	52	92.9	7531.8	7397.9	6271.1	9269.1	891.9	11.8	0.4
Price1mSec60_80real	52	92.9	7676.8	7552.3	6438.6	9774.4	939.7	12.2	0.6
Price1mSec80real	52	92.9	8794.6	8519.4	7193.2	11194.4	1055.7	12.0	0.6
Price1m2 Secon totalreal	52	92.9	7923.3	7728.7	6621.8	9857.4	938.0	11.8	0.5
PremPrim40	56	100.0	333.3	346.0	0.0	783.0	245.8	73.7	0.2
PremPrim_40_60	56	100.0	715.1	672.0	0.0	1995.0	491.2	68.7	0.6
PremPrim_60_80	56	100.0	419.1	391.5	0.0	935.0	255.0	60.8	0.1
PremPrim_80	56	100.0	282.3	284.5	0.0	672.0	137.9	48.9	0.2
PremPrimTotal	56	100.0	1749.9	1680.5	0.0	3731.0	1064.4	60.8	0.1
PremSec40	56	100.0	758.1	747.0	0.0	1204.0	294.5	38.8	−1.0
PremSec_40_60	56	100.0	1059.7	1010.5	0.0	1926.0	471.0	44.4	−0.4
PremSec_60_80	56	100.0	423.9	395.5	0.0	824.0	204.5	48.2	−0.2
PremSec_80	56	100.0	292.6	282.0	0.0	572.0	138.9	47.5	−0.2
PremSecTotal	56	100.0	2534.4	2433.0	0.0	4299.0	1091.8	43.1	−0.6

Source: own work based on collected data.

. All statistically significant (p ≤ 0.05) price models for flats built for the primary market are presented in

Table 2. Primary market—results of estimation of model parameters for the price of 1 m² of apartment and diagnostic statistics for these models, p ≤ 0.05.

Apartment Area	Significant Factor	$\mathbf{Parameter} {\mathit{\beta }}_1$ (Slope) Estimator	$\mathbf{Adjusted} {\overline{\mathit{R}}}^2$	F-Statistics	Parameter p
apartments <40 m2	NO2	−174.69	37.4%	31.521	0.0000
apartments 40–60 m2	NO2	−148.80	38.0%	32.213	0.0000
apartments 60–80 m2	NO2	−113.93	40.2%	35.214	0.0000
apartments >80 m2	NO2	−117.10	23.2%	16.376	0.00001
all apartments	NO2	−133.70	39.0%	33.657	0.0000

Source: own work based on collected data.

. A factor significantly related to flat prices in this market is the concentration of nitrogen dioxide (NO₂). In all cases, the estimate of the parameter responsible for the slope of the regression is negative, indicating an inverse relationship between price and NO₂ concentration. The estimated parameter values, ranging from nearly −114 to −175, suggest that for every 10 µg/m³ increase in NO₂ concentration, the price per square meter in individual market segments decreases by an average of PLN 1140 to PLN 1750.

The highest estimated values of the coefficients, indicating the greatest impact on prices in monetary terms, were observed for the smallest flats, under 40 m². In the case of the largest flats, this relationship was somewhat weaker but still evident. According to data on the number of transactions for individual segments (Figure 13), the highest turnover of flats in each quarter of the period under review concerned the segments 40–60 m² and below 40 m², which together accounted for 66% of all transactions. These are also the segments for which the increase in NO₂ concentrations had the most significant impact on the decline in prices per square meter of flats. Nitrogen oxide concentrations in the air are directly linked to transport emissions. Therefore, it can be concluded that the highest turnover concerned properties located near major transport arteries. Moreover, the decline in prices that occurs with an increase in NO₂ concentrations suggests that, for buyers in the capital, a location away from traffic pollution may be advantageous.

Table 3. Secondary market—results of estimation of model parameters for the price of 1 m² of apartment and diagnostic statistics for these models, p ≤ 0.05.

Apartment Area	Significant Factor	$\mathbf{Parameter} {\mathit{\beta }}_1$ (Slope) Estimator	$\mathbf{Adjusted} {\overline{\mathit{R}}}^2$	F-Statistics	Parameter p
apartments <40 m2	PM2.5	−142.53	51.9%	56.069	0.0000
apartments 40–60 m2	PM2.5	−82.74	18.6%	12.686	0.0008
apartments 60–80 m2	none	-	-	-	-
apartments >80 m2	NO2	−247.34	41.2%	36.716	0.0000
all apartments	NO2	−230.21	46.2%	44.713	0.0000

Source: own work based on collected data.

presents the statistically significant impact of air pollution and meteorological factors on the prices of flats in the secondary market in Warsaw. All relevant models demonstrating the impact of these factors on flat prices are presented. The analysis shows that for each flat segment, most factors had a significant impact, with one factor in particular having the greatest influence. For the smallest flats, the concentration of PM_2.5 particulate matter had a significant impact, while for the largest flats, it was nitrogen dioxide (NO₂).

In each of the models obtained, the estimation of the directional coefficients of the models (parameter β₁) showed a negative value. In the case of the smallest flats, there was a negative correlation between the price per square meter of a flat and the concentration of PM_2.5 particulate matter. With a 10% increase in PM_2.5 concentration, the price per square meter of flats with an area of up to 40 square meters decreased by an average of PLN 1425, while for flats with an area of 40–60 square meters, the decrease was PLN 827 on average. Conversely, for the largest flats (those over 80 m²), a negative correlation was found between the price per square meter and the concentration of NO₂. With a 10% increase in NO₂ concentration, the price per square meter of flats in this segment decreased by an average of PLN 2473. These flats had the highest prices per square meter (Figure 18).

In the analysis of all flats, without segmentation by size, the impact of nitrogen dioxide proved to be significant. As in other cases, the estimate of the structural parameter was negative, indicating that as NO₂ concentration increases, the price per square meter of flats on the secondary market decreases.

The second group of dependent variables was the number of housing sales in the primary and secondary markets, broken down into corresponding segments in terms of size.

Table 4. Primary market—results of estimation of model parameters for number of sales of apartments and diagnostic statistics for these models; final models without nonsignificant variables, p ≤ 0.05.

Apartment Area	Significant Factor	$\mathbf{Parameter} {\mathit{\beta }}_1$ (Slope) Estimator	$\mathbf{Adjusted} {\overline{\mathit{R}}}^2$	F-Statistics	Parameter p
apartments <40 m2	PM2.5	−8.42	42.6%	5.95	0.01
	PM10	−5.19		1.89	0.00
apartments 40–60 m2	PM2.5	−7.232	25.9%	17.19	0.00
apartments 60–80 m2	PM2.5	−6.759	21.7%	13.72	0.00
	PM10	−3.467		10.64	0.00
apartments >80 m2	PM2.5	−0.035	26.9%	11.86	0.00
	PM10	−0.044		6.77	0.01
all apartments	PM2.5	−24.577	24.5%	16.78	0.00
	PM10	−9.022		14.00	0.00

Source: own work based on collected data.

presents all statistically significant (p ≤ 0.05) models for the number of flats sold on the primary market. The factors significantly related to the number of flats sold are the concentration levels of PM₁₀ and PM_2.5 particulate matter. In all cases, the estimate of the parameter responsible for the regression slope was negative, indicating an inverse relationship between the number of flats sold and the concentration of PM₁₀ and PM_2.5. The estimated parameter values for PM₁₀, ranging from −3.4 to −9.0, indicate that for every 10 µg/m³ increase in PM₁₀ concentration, the number of flats sold in the given segments and for the entire primary market decreased by an average of 34 to 90 units per quarter. On the other hand, the estimated parameters for PM_2.5, ranging from −7.2 to −24.5, indicate that with a 10% increase in PM_2.5 concentration, the number of flats sold in the given segments and for the entire primary market decreased by an average of 72–245 units per quarter. The impact of PM_2.5 concentrations is therefore stronger than that of PM₁₀.

Table 5. Secondary market—results of estimation of model parameters for number of sales of apartments and diagnostic statistics for these models p ≤ 0.05.

Apartment Area	Significant Factor	$\mathbf{Parameter} {\mathit{\beta }}_1$ (Slope) Estimator	$\mathbf{Adjusted} {\overline{\mathit{R}}}^2$	F-Statistics	Parameter p
apartments <40 m2	PM2.5	−2.45	32.2%	24.52	0.00
apartments 40–60 m2	PM2.5	−7.58	35.2%	28.73	0.00
apartments 60–80 m2	PM2.5	−48.63	40.4%	35.74	0.00
apartments >80 m2	PM2.5	−118.079	53.3%	9.65	0.00
all apartments	PM2.5	−873.82	43.7%	8.39	0.00

Source: own work based on collected data.

presents the results for the secondary market. There is a negative correlation between the number of flats sold and the level of PM2.5 dust concentrations. The estimated parameters for PM2.5, ranging from −2.4 to −118.0, indicate that with a 10% increase in PM2.5 concentration, the number of flats sold in the specified segments of the secondary market fell by an average of 24–1180 per quarter. For the market as a whole, the average decline is 8738.

One characteristic of the real estate market is its seasonal fluctuations [42], consisting, among other factors, of an increase in the number of transactions in the spring and autumn (March–June and September–November). Emissions of PM₁₀ and PM_2.5 particulate matter predominantly come from the municipal and residential sectors and are primarily associated with the heating season from October to April. The relationship between an increase in particulate matter concentrations and a decrease in the number of transactions is confirmed by the study. This relationship is stronger in the secondary market than in the primary market, since in the primary market, the transaction date is mainly determined by the developer’s construction cycle, rather than by seasonal increases in buyer activity.

5. Conclusions from the Study

Environmental pollution, including air pollution, has a significant impact on many aspects of human life. One area where this relationship is particularly evident is the real estate market. Air pollution is linked to the demand for new and secondary market flats, and above all, to the prices of these flats. The study has shown that housing prices in Warsaw are strongly and statistically significantly related to the concentrations of two air pollutants: nitrogen dioxide (NO₂) and particulate matter (PM_2.5). Similar findings have been reported in earlier studies on real estate prices in Kraków and Gdańsk [14,39].

The study developed GRM and OLS models—stochastic models of the impact of air pollution and meteorological factors on apartment prices in Warsaw per square meter. The final models retained pollution concentrations that were statistically significantly related to real estate market variables. Two air pollutants proved to be significant: nitrogen dioxide (NO₂) and particulate matter (PM_2.5). Nitrogen dioxide concentrations were associated with the prices of all flats in the primary market and the prices of the largest flats in the secondary market. Particulate matter concentrations were associated with the prices of smaller flats in the secondary market.

In all cases, the estimates of the structural parameters of the models responsible for the slope of the variable were negative. As pollutant concentrations decreased, apartment prices rose. The strength of this impact varied, and it can be said to have been stronger for smaller apartments.

No significant impact of other air pollutants or meteorological factors was found. No impact of the interaction between these factors on housing prices was found either. A significant impact of periodic fluctuations in the annual cycle on the concentrations of most pollutants and housing prices was observed. However, the authors suggest that this is likely a random effect caused by factors other than environmental ones.

The study also shows that it is not possible to identify factors significantly related to the prices of all properties in Warsaw without dividing them into market segments. It was found that the primary and secondary property markets differ so significantly from one another that it is not possible to describe them together and identify factors significantly related to these markets as a whole.