# RECORD ddac707c3201e8fd8ee39f4a8176310e5a638976 # TITLE: Potential impact of property management on the market value of dwellings in multi-owned housing # YEAR: 2023.0 VENUE: Journal of Housing and the Built Environment DOI: 10.1007/s10901-023-10032-2 # AUTHORS: Andrzej Muczyński # PAGES: 21 CANDIDATE TABLE PAGES (1-based): [13, 15] # ABSTRACT: The expanding role of property management services and the increasing share of multi-owned housing in many housing markets raises issues concerning the impact of such services on property value in this sector. The presented study investigated the potential impact of property management on the market value of dwellings in multi-owned housing developments. This impact was identified based on estimated implicit values of the extracted common property attributes dependent on property management services. An original research methodology tailored to the institutional arrangement in multi-owned housing developments was used, involving the estimation of hedonic regression models. Empirical research was conducted on representative samples of multi-owned housing developments in the housing market of the city of Olsztyn, Poland. The results showed that common property attributes such as building condition, building accessibility, cleanliness and orderliness, and land development significantly influenced dwelling prices and thus their market value. In extreme cases, the cumulative potential impact of property management services estimated on this basis can reach up to ca. 30% of the dwellings’ value. These findings have far-reaching policy and practical implications for urban development policies, including the built environment, land use and local housing, as well as for housing market investors and developers, homeowners associations, dwelling owners and property managers. Hence, it is recommended to expand research on this topic to other housing markets functioning in different institutional arrangements. ===== PAGE 1 (score 5, numeric-paren 0, table-words 0, var-words 5) ===== ARTICLE Received: 15 December 2021 / Accepted: 28 March 2023 / Published online: 25 April 2023 © The Author(s) 2023 Andrzej Muczyński amucz@uwm.edu.pl 1 Faculty of Geoengineering, Institute of Spatial Management and Geography, University of Warmia and Mazury in Olsztyn, Prawocheńskiego 15, 10-720 Olsztyn, Poland Potential impact of property management on the market value of dwellings in multi-owned housing Andrzej Muczyński1 Journal of Housing and the Built Environment (2023) 38:2207–2227 https://doi.org/10.1007/s10901-023-10032-2 Abstract The expanding role of property management services and the increasing share of multi- owned housing in many housing markets raises issues concerning the impact of such services on property value in this sector. The presented study investigated the potential impact of property management on the market value of dwellings in multi-owned housing developments. This impact was identified based on estimated implicit values of the ex­ tracted common property attributes dependent on property management services. An origi­ nal research methodology tailored to the institutional arrangement in multi-owned housing developments was used, involving the estimation of hedonic regression models. Empirical research was conducted on representative samples of multi-owned housing developments in the housing market of the city of Olsztyn, Poland. The results showed that common property attributes such as building condition, building accessibility, cleanliness and or­ derliness, and land development significantly influenced dwelling prices and thus their market value. In extreme cases, the cumulative potential impact of property management services estimated on this basis can reach up to ca. 30% of the dwellings’ value. These findings have far-reaching policy and practical implications for urban development poli­ cies, including the built environment, land use and local housing, as well as for housing market investors and developers, homeowners associations, dwelling owners and property managers. Hence, it is recommended to expand research on this topic to other housing markets functioning in different institutional arrangements. Keywords  Property management · Property market value · Multi-owned housing · Hedonic pricing models · Poland 1 3 ===== PAGE 13 (score 43, numeric-paren 19, table-words 8, var-words 2) ===== Potential impact of property management on the market value of… ever, this resulted from positive effects on prices of relatively expensive and small dwellings located in the peripheral city zones in newer (sub-developer) buildings being in better con­ dition. Based on the results of the correlation analysis, the initial set of dwelling attributes as potential independent variables explaining their prices was reduced in all research samples by those variables for which both correlation coefficients proved to be insignificant at the accepted level of statistical significance. It should be noted that positively verified inde­ pendent variables by the correlation analysis were correlated with each other much weaker than with the dependent variable. Thus, the principle that a regression model should include variables that are highly correlated with the response (dependent) variable and at the same time relatively weakly correlated with each other was fulfilled. Using the basic analytical form of the regression function (1) and the reduced sets of explanatory variables, hedonic pricing models were estimated for the main sample S1 (model M1) and the subsamples S2, S3 and S4 (models M2, M3 and M4, respectively) of MOH developments in the studied local housing market. Estimation of the structural param­ eters of the models was performed using the ordinary least squares (OLS) technique. Since outlier observations can significantly bias the regression equations, the estimation proce­ dures for each of the models were conducted in two steps. First, the originally estimated models were subjected to residuals analysis, which allowed to eliminate the most outlier observations from the originally formed observation sets S1-S4 based on the Cook’s dis­ tance criterion (Cook & Weisberg, 1982). On the outlier-cleaned samples, the final regres­ sion models M1-M4 were then estimated. These models showed a better fit to empirical data, higher parameter stability, and lower estimation errors compared to the originally esti­ mated models. Standard statistical tests (following Aczel 2006; Goryl et al., 2009), such as Fisher-Snedecor (F) statistics, Student’s tests, and Durbin-Watson (DW) statistics, were used to statistically verify the finally estimated models. In addition, coefficients of deter­ mination (R-squared and adjusted R-squared) and standard errors of estimation (Se) were calculated. The hedonic regression results in the estimated models are presented in Table 3. Based on these results, it was stated that the estimated models meet standard statistical postulates as the regression coefficients obtained were statistically significant (at p < 0.05), the residuals were symmetric, random in nature and revealed no autocorrelation. Moreover, given complex dependencies and conditions concerning property transactions in the hous­ ing market, it can be considered that the estimated models are characterized by the required (satisfying) adaptation to empirical data. This is evidenced by the relatively high values Table 2  Correlation results between dwelling price (P) as the dependent variable and the pre-selected dwelling attributes (variables) in the sampled MOH developments. (source: own elaboration) Research sample S1 S2 S3 S4 Attribute Variable rS τ rS τ rS τ rS τ OL X1 − 0.587 − 0.457 − 0.053* − 0.044* − 0.052* − 0.041* − 0.539 − 0.421 NL X2 0.080* 0.062* 0.519 0.411 0.262* 0.212 0.436 0.352 FL X3 0.142* 0.109 0.334 0.258 0.117* 0.090* 0.093* 0.078* DA X4 0.364 0.287 0.210* 0.163* 0.499 0.399 0.294 0.227 BC X5 0.441 0.357 0.444 0.356 0.237* 0.196 0.550 0.446 BA X6 0.386 0.307 0.109* 0.090* 0.307 0.246 0.237* 0.187* CO X7 0.556 0.458 0.464 0.387 0.403 0.333 0.686 0.568 LD X8 0.418 0.337 0.428 0.350 0.413 0.337 0.573 0.458 Note: (*) Correlation coefficients not significant at p < 0.05 1 3 2219 ===== PAGE 14 (score 41, numeric-paren 0, table-words 15, var-words 0) ===== A. Muczyński of R-squared coefficients (63.4–73.2%) and low values of standard errors of estimation Se (0.0447 to 0.0997), which, in relative terms, indicate the accuracy of estimating dwelling values with these models ranges from ± 0.6% to ± 1.2%. The obtained results, therefore, confirmed the statistical validity of the estimated models, including the correctness of the choice of the analytical form of the regression function and the sets of explanatory variables that significantly influence the dependent variable. Thus, the models provide a reasonable Table 3  Hedonic regression estimation results for the dependent variable ln (Price). (source: own elaboration) Variable Regression coefficient Standard error t-statistic p-value Model M1 Intercept term β0 7.9240 0.0638 124.1523 0.0000 X1 β1 − 0.0684 0.0121 − 5.6485 0.0000 X3 β3 0.0211 0.0097 2.1740 0.0314 X4 β4 0.0516 0.0111 4.6560 0.0000 X5 β5 0.0518 0.0152 3.3989 0.0009 X6 β6 0.0329 0.0138 2.3773 0.0188 X7 β7 0.0495 0.0092 5.3782 0.0000  F - statistic 40.224 DW - statistic 1.8404 R - squared 0.6345 Adjusted R - squared 0.6188 Se - standard error of estimation 0.0997  N - number of observations 146 Model M2 Intercept term β0 7.8061 0.0318 245.2907 0.0000 X2 β2 0.0347 0.0095 3.6467 0.0007 X3 β3 0.0190 0.0076 2.4687 0.0178 X5 β5 0.0217 0.0098 2.2129 0.0325 X7 β7 0.0354 0.0122 2.8845 0.0062 X8 β8 0.0293 0.0123 2.3769 0.0222  F - statistic 17.043 DW- statistic 1.7901 R - squared 0.6752 Adjusted R - squared 0.6355 Se - standard error of estimation 0.0470  N - number of observations 47 Model M3 Intercept term β0 8.0291 0.0406 197.8903 0,0000 X4 β4 0.0454 0.0106 4.2847 0.0001 X6 β6 0.0328 0.0141 2.3329 0.0249 X7 β7 0.0456 0.0141 3.2439 0.0024 X8 β8 0.0330 0.0158 2.0799 0.0441  F - statistic 18.0742 DW - statistic 1.7479 R - squared 0.6496 Adjusted R - squared 0.6136 Se - standard error of estimation 0.0460  N - number of observations 46 Model M4 Intercept term β0 7.8071 0.0568 137.4793 0,0000 X2 β2 0.0597 0.0234 2.5526 0.0145 X4 β4 0.0529 0.0141 3.7544 0.0005 X5 β5 0.0461 0.0190 2.4304 0.0195 X7 β7 0.0570 0.0264 2.1616 0.0365 X8 β8 0.0447 0.0176 2.5360 0.0151  F - statistic 22.367 DW - statistic 2.2045 R - squared 0.7317 Adjusted R - squared 0.6990 Se - standard error of estimation 0.0707  N - number of observations 47 1 3 2220 ===== PAGE 15 (score 14, numeric-paren 0, table-words 5, var-words 2) ===== Potential impact of property management on the market value of… tool to assess the influence of the individual explanatory variables ceteris paribus on the dwelling value in each sample of MOH developments. Within substantive verification of the estimated models, their consistency with theoreti­ cal and practical knowledge of price formation and market value determination of dwell­ ings in MOH developments was assessed. It should be noted that the regression analysis in models M1, M3, and M4 resulted in the further reduction in the sets of explanatory variables previously verified by correlation analysis. In consequence, the following variables were eliminated: X8 (land development) in model M1, X2 (neighborhood location) and X5 (build­ ing condition) in model M3, and X1 (overall location) in model M4, which can be justified by the relatively high correlation of variable X8 with variables X6 and X7 in sample S1 and the low correlation of variables X2 and X5 with the response variable in sample S3, or the location specificity of MOH buildings in sample S4. Secondary reduction of the explana­ tory variables was, therefore, a logical extension of the correlation analysis. The strength and direction of the influence of extracted explanatory variables on the dependent variable were taken as the main criteria of substantive assessment of the models. This was carried out by assessing the consistency of signs and values of the regression coefficients located at individual variables. On the one hand, given the adopted way of variable description, the signs of the estimated regression coefficients were found to be consistent with a priori expectations. This is due to the fact the direction of influence of the individual explanatory variables on the dependent variable was generally positive. It should be emphasized that this regularity occurred for all common property variables (X5 -X8) in each of the estimated models. The only exception was variable X1 (overall location) in model M1, whose influ­ ence proved to be negative again, which was expected given the correlation results and was explained above. On the other hand, taking into account the presented interpretation of regression coefficients in log-linear models, the values of the estimated coefficients were considered to be reasonable, as they were consistent with the theory of property valuation and corresponded to the practical experience of real estate appraisers on the analyzed hous­ ing market. The estimated values of these coefficients indicate that the average strength of the influence of a unit change in particular common property variables ranged, ceteris paribus, respectively: 3.3% for X5(building condition), 3.8% for X6(building accessibility), 4.7% for X7(cleanliness and orderliness), and 3.6% for X8 (land development). Noteworthy is the variable X7, which revealed a relatively high influence on the market value of dwell­ ings in all estimated models, and which remains under the direct control of the property manager. To enable easier comparison of regression coefficients in the estimated models, they are listed separately in Table 4. Regression coefficient Model M1 Model M2 Model M3 Model M4 β0 7.9240 7.8061 8.0291 7.8071 β1 − 0.0684 - - - β2 - 0.0347 - 0.0597 β3 0.0211 0.0190 - - β4 0.0516 - 0.0454 0.0529 β5 0.0518 0.0217 - 0.0461 β6 0.0329 - 0.0328 - β7 0.0495 0.0354 0.0456 0.0570 β8 - 0.0293 0.0330 0.0447 Table 4  Listing of regression co­ efficients in the estimated mod­ els. (source: own elaboration) 1 3 2221 ===== PAGE 16 (score 1, numeric-paren 0, table-words 0, var-words 1) ===== A. Muczyński The obtained findings have therefore confirmed that there are statistically significant positive relationships between observable common property attributes dependent on PM services and dwelling prices in MOH developments in the studied housing market. This makes it possible to conclude that the research hypothesis posed in the introduction was positively verified by empirical research. In other words, the extracted common property attributes were actually perceived by market participants as value drivers, as evidenced by their significant impact on dwelling prices in that market. Since these attributes are depen­ dent on the intensity and quality of PM services, the estimated influence of them on dwell­ ing prices identifies the potential impact of PM services on the market value of dwellings in MOH developments. This means that if the property manager (or owner) made decisions and actions to introduce the new and/or maintain the existing state of the considered com­ mon property attributes as part of the PM services, it was statistically proven that this had an impact on the market value of dwellings in MOH developments in the studied housing market. This is a potential impact because for it to manifest itself, it is necessary to include active influence on individual common property attributes in the scope of PM services, and furthermore, it is difficult to conclusively determine to what extent managerial and to what extent executive activities have contributed to changes in these attributes, which, as a result, have had an impact on the market value of dwellings. The magnitude and structure of this impact are also determined by the estimated implicit values of the common property attributes. Determining the real impact of PM services on the market value of dwellings in the presented methodology requires further in-depth object-oriented research to assess the factual contribution of these services in shaping the value-creating attributes related to both common and individual properties. 5  Conclusions Given the expanding role and importance of property management in many housing mar­ kets, the impact of these services on property market value is undoubtedly a topical research problem of a global scope and a complex nature. This problem takes on special relevance in MOH developments, various forms of which are increasingly proliferating in different parts of the world, currently comprising a large proportion of the world’s housing stock. Although many authors have emphasized the impact of PM on the market value of a man­ aged property, there have been relatively few efforts to measure it quantitatively, and there is still no consensus among researchers and professionals on the issue. This is particularly true for empirical studies in MOH developments, where the issue is largely ignored for various reasons. In an attempt to contribute to filling the gap, in this preliminary study, the potential impact of PM services on the market value of dwellings in MOH developments existing under Polish conditions using a selected local housing market was investigated. The methodological achievement of this paper is the conceptualization and operationaliza­ tion of an original research methodology based on the triangulation of research methods tailored to the institutional arrangement in MOH developments. It assumes that PM services in such housing properties are mainly focused on the common property, but their potential impact on the property market value can only be manifested in dwelling prices since the common property, as such, cannot be traded on the market. Therefore, the said impact of PM services was estimated from dwelling prices, however, not directly, but in an indirect 1 3 2222