1# RECORD 8885ddb1a206684210159fea8394d2ea9c6111112# TITLE: Natural Environmental Amenities and House Rentals: A Hedonic Analysis for Integrated Planning3# YEAR: 2018.0 VENUE: Journal of African Real Estate Research DOI: 10.15641/jarer.v0i0.4824# AUTHORS: F. Famuyiwa5# PAGES: 19 CANDIDATE TABLE PAGES (1-based): [11, 13]6# ABSTRACT: Environmental valuation techniques have been used in quantifying the influences of environmental amenities on house prices. The hedonic approach is used to evaluate contributions of environmental amenities to rents in this study. Though the theme of this study has attracted pedagogical attention, the comparison between man-made and natural environmental amenities in housing values is a nouvelle approach in identifying and explaining effective demand and value systems of urban and natural amenities through market evidence. It extends existing body of literature on hedonic studies in Nigeria by contextually assessing these two broad categories of environmental amenities on the same scale. Results reveal the variations in demand and significance of physical infrastructure and natural amenities in rents. This /study's significance lies in bringing to light, the realities of market demand for natural amenities in comparison with physical infrastructure789===== PAGE 1 (score 7, numeric-paren 0, table-words 0, var-words 7) =====10111 ffamuyiwa@unilag.edu.ng 12 13Journal of African Real Estate Research 14Volume 3, Issue 2 15www.journals.uct.ac.za/index.php/JARER/index 16 17 18Natural Environmental Amenities and House Prices - A Hedonic Analysis 19for Integrated Planning 20 21Funlola Famuyiwa1 22 23 241 Department of Estate Management, Faculty of Environmental Sciences, University of 25Lagos, Nigeria 26 27 28To cite this article: Famuyiwa, F. (2018). Natural Environmental Amenities and House 29Prices - A Hedonic Analysis for Integrated Planning. Journal of African Real Estate 30Research, 3(2), pp.44-62. DOI: 10.15641/jarer.v0i0.482. 31 32Abstract 33 34Environmental valuation techniques have been used to quantify the influences of 35environmental amenities on house prices. Although this theme has attracted pedagogical 36attention internationally, a Nigerian perspective relating to effective demand and value 37systems of natural environmental amenities is lacking significantly. The gap in local market 38evidence relating to this topic is apparent in comparison to other real estate attributes such as 39physical infrastructure. In this study, the hedonic pricing model (HPM) is used to evaluate the 40contributions of environmental amenities to the rental values of detached houses in a sub-41market in Lagos, Nigeria. This paper, therefore, extends existing body of literature on 42hedonic studies by contextually assessing the broad category of natural environmental 43amenities in Lagos. Using a sample of 111 properties, the HPM was used to reveal value 44inferences of various environmental characteristics on house rents. Results show that ‘green 45environment’ as a natural amenity in the housing market studied, produces the highest 46variation in rental values. This study’s significance lies in bringing to light the realities of 47market demand for natural amenities which, in turn, will help decision makers and public 48authorities in the strategic planning of the preservation and sustenance of natural amenities. 49 50Keywords: Environmental Valuation; Housing Market; Natural Amenities; Physical 51Infrastructure; Rental Values 52 53 541. Introduction 55 56There is a growing awareness of the importance of natural environmental 57amenities for public health, productivity, social and economic development, 58among others. For example, Littenberg et al. (2015) found that residing in 59areas with access to the natural environment is associated with better health, 60which increases the demand for specific housing attributes, such as 61proximity to green areas, water bodies, and other natural physical 626364===== PAGE 11 (score 17, numeric-paren 0, table-words 2, var-words 10) =====6566Journal of African Real Estate Research Volume 3(2) 2018 67 68 6954 70Wind Quality in Building (Ventilation) 71WINDQUA Good = 1; Poor= 0 72Landscape of Neighbourhood 73LSCAPE 74Good = 1; Poor= 0 75View of Navigable water body 76VIEWWAT Good = 1; Poor= 0 77 784. Findings and Discussion 79 80In Table 2 the descriptive statistics of the sampled residential properties in 81the study area are displayed. Apart from ‘net rental value’, ‘living room’, 82‘bedroom’, ‘bathroom’, and ‘plot size’, all other variables were estimated 83using binary measurements. The average net rental value (dependent 84variable) of detached buildings in the area is N1,604,879 while maximum 85and minimum values are N3,600,000 and N1,450,000 respectively. 86 87Table 2: Descriptive Statistics of Regression Variables 88 89 90MINIMUM 91MAXIMUM 92STD 93MEAN 94RENT 951,450,000 963600000 97238030 981,604,879 99STRQUAL 100.00 1011.00 102.24085 103.92333 104REPAIR 105.00 1061.00 107.32048 108.81775 109LIVIN 1101.00 1112.00 112.04822 1131.61098 114BED 1153.00 1165.00 117.28616 1184.00723 119BATH 1202.00 1215.00 122.48250 1233.59881 124STORE 125.00 1261.00 127.49983 128.555641 129BCNY 130.00 1311.00 132.22081 133.880378 134PLOT 135592.73 1361,406.91 137193.68204 138736.75291 139NTLIGHT 140.00 1411.00 142.00985 143.68080 144DISCBD 145.00 1461.00 147.39225 148.61046 149DISRECR 150.00 1511.00 152.40391 153.42907 154SECRTY 155.00 1561.00 157.36119 158.64816 159SERENE 160.00 1611.00 162.00839 163.32832 164ROAD 165.00 1661.00 167.40071 168.63784 169WATER 170.00 1711.00 172.23641 173.60120 174WASTE 175.00 1761.00 177.21093 178.76059 179SRTLIGHT 180.00 1811.00 182.38658 183.32199 184DRAIN 185.00 1861.00 187.47550 188.66493 189GREEN 190.00 1911.00 192.10993 193.78036 194GARDEN 195.00 1961.00 197.09001 198.29732 199WINDQUA 200.00 2011.00 202.33029 203.62435 204LSCAPE 205.00 2061.00 207.30723 208.57446 209VIEWWAT 210.00 2111.00 212.00033 213.00778 214Source: Field Study, 2016. 215 216217218===== PAGE 12 (score 29, numeric-paren 1, table-words 10, var-words 7) =====219220Journal of African Real Estate Research Volume 3(2) 2018 221 222 22355 224From Table 2, it can be seen that the number of bedrooms ranges from 3 to 2255 with a standard deviation of 0.28616, implying a low dispersion around a 226mean of 4.00723. The mean for ‘natural lighting’ is 0.68080. Sixty-five per 227cent (65%) have ‘good neighbourhood security’, while the conditions of 228road surfaces of about 64% of these properties are good. Only about 60% of 229the properties have access to public pipe-borne water, and about 76% 230dispose of their domestic waste using government waste disposal services. 231Only about 32% of the properties have streetlights around their premises. 232Other variables can be interpreted from the mean values and standard 233deviations in the table accordingly. 234 2354.1 Model Evaluation 236 237There is little guidance from economic theory about the proper functional 238relationship between house price and its attributes (Selim, 2008). The 239estimation and selection of appropriate HPM have therefore been of 240significant concern. Widely used functional forms include the ‘linear' form, 241the ‘semi-logarithmic' form (log-linear) and the ‘log-log'. Consequently, this 242study analysed field data using these three models. The three models 243utilised, depict the preponderance of these methods in similar studies. All 244three models used, (only the log-linear model shown) yielded significant 245coefficients and good model statistics. The log-linear model performed most 246satisfactorily, in terms of its ability to predict property value influences in 247the study area as indicated by its R² and Adjusted R² statistics of 0.790123 248and 0.723652 respectively. An R2 of 0.790123 infers that about 79% of 249variance in rental values is accounted for by the joint influence of all 250explanatory variables. The adjusted R2 explains only the variation accounted 251for by the significant variables. The individual relationship of each of the 252attributes to value is shown in the second column of Table 3 (tagged 253‘coefficients’). Babawale and Johnson (2012), remark that high values of R² 254do not necessarily ensure the accuracy of models. Hence, tests were carried 255out to check for ‘multicollinearity’ and ‘heteroskedasticity’. 256 257Multicollinearity is a statistical phenomenon when two or more independent 258variables in a multiple regression model are highly correlated (Xiao, 2017). 259This is undesirable in statistics as the estimate of one variable's impact on 260the dependent variable while controlling for the others tends to be less 261precise. ‘Heteroskedasticity’ on the other hand refers to a statistical 262phenomenon where the variance of the dependent variable varies across the 263data. Babawale and Johnson (2010) imply that the variance of the errors of 264the data increases or decreases. The White’s test and Variance Inflation 265Factor (VIF) statistics were used to check for heteroskedasticity and 266multicollinearity respectively. In the White’s Test, the F-statistic of 2671.682110, the Obs*R-squared (51.481659), and p-values indicate constancy 268of the random error. The P-value of 0.043974 is the probability that random 269error is not heteroskedastic (not higher than 5 per cent). Table 5 shows a 270maximum VIF value of 2.99256, which is low, desirable and indicative of 271the absence of multicollinearity in the model. If the VIF of a variable 272exceeds 10, that variable is said to be highly collinear and will pose a 273problem in the analysis (Hair et al., 2006 cited in Hill & Throshani, 2009). 274All these imply that the data fits well into the regression line and the model 275is adequate in explaining changes in rental values for the category of 276277278===== PAGE 13 (score 19, numeric-paren 0, table-words 6, var-words 4) =====279280Journal of African Real Estate Research Volume 3(2) 2018 281 282 28356 284property in the study area. Consequently, the model appears adequate for 285predicting rental values of detached buildings in the study area. 286 287Table 3: Regression Results 288 289 290Log-linear Model 291Variables 292Coefficients 293T- statistics 294STRQUAL 295.191025 2961.733654 297REPAIR 298.081544 2991.380998 300LIVIN 301.084432 3023.994811 303BED 304.072237 3053.410468 306BATH 307.061126 3084.736524 309STORE 310.090283 3113.028461 312BCNY 313.152038 3143.842652 315PLOT 316.008826 3173.947261 318NTLIGHT 319.009435 3206.884632 321DISCBD 322.280001 3231.920375 324SECRTY 325.000983 3261.677463 327ROAD 328.120567 3291.814028 330WATER 331.097635 3322.376251 333WASTE 334-.039012 335-3.429712 336SRTLIGHT 337.050754 3381.092836 339DRAIN 340.009127 3412.348283 342GREEN 343.170362 3442.736555 345WINDQUA 346.011275 3471.983463 348LSCAPE 349.093552 3505.012254 351VIEWWAT 352.034900 3534.773652 354Constant 35514.994821 35637.839204 357R2 358.790123 359 360Adjusted R2 361.723652 362 363F-statistic 36419.336092 365 366Prob 367.050000 368 369Source: Field Study, 2016 370 371Table 5: Collinearity Statistics 372 373 374VIF 375 376STRQUAL 3771.73635 378 379REPAIR 3802.00483 381 382LIVIN 3831.04877 384 385BED 3861.94763 387 388BATH 3891.04987 390 391STORE 3921.77465 393 394BCNY 3952.53684 396 397PLOT 3982.99256 399 400NTLIGHT 4011.02736 402 403 404Table 4: White’s Test 405 406F-Statistic 4071.682110 408 Probability 0.043974 409Obs*R-squared 41051.481659 Probability 0.408321 411412413===== PAGE 14 (score 26, numeric-paren 0, table-words 7, var-words 11) =====414415Journal of African Real Estate Research Volume 3(2) 2018 416 417 41857 419DISCBD 4201.99715 421 422SECRTY 4231.37541 424 425ROAD 4262.55301 427 428WATER 4291.37599 430 431WASTE 4321.90548 433 434SRTLIGHT 4352.40918 436 437DRAIN 4381.28285 439 440GREEN 4412.48635 442 443WINDQUA 4442.34975 445 446LSCAPE 4471.11906 448 449VIEWWAT 4501.65421 451 452 4534.2 Interpretation of Independent Variables 454 455The results from the log-linear model as seen in Table 3 indicate real estate 456attributes (variables) that predict rental values as described in Table 1. 457Variables with negative coefficients imply an inverse relationship with 458rental values and vice versa. The coefficients in the log-linear model are 459interpreted as variations in rental values when there is a change (per 460measurement unit) in a given variable. Accordingly, ‘good structural 461quality’ contributes 19% to rental values. For every additional living room, 462bedroom and bathroom, rental values increase by 8.4%, 7.2% and 6% 463respectively. Other independent variables in the table can be interpreted in 464the same way. For variables categorised as ‘natural environmental 465attributes’, the study suggests that ‘green environment in neighbourhood’ 466accounts for 17% of variation in rental values, while ‘good landscape’ 467contributes up to 9.3% to rental values. 468 469In terms of physical infrastructure; ‘government waste disposal services' 470possesses an adverse effect on rental values implying that houses serviced 471by government waste management agencies will likely experience a 3.9% 472decline in rental value, thereby contradicting existing theory of the positive 473effect which infrastructure has on property values. Further investigation into 474this suggests that this is because the erratic and unreliable state of 475government services has given rise to aesthetic issues with waste littering 476the streets due to delays in waste collection. This thereby constitutes a 477negative externality and impacts negatively on rental values. 478 479Although the distance to Central Business District (CBD) possesses the 480highest coefficient value (confirming location as one of the most significant 481factors to property value), its significance level is low. Natural 482environmental amenities performed almost at par with human-made 483amenities from the coefficient values. However, the strongest unique 484contribution (asides distance to CBD) explaining rental values is ‘green 485environment in neighbourhood' with the highest coefficient value of 4860.170362 (17%). This implies that this particular variable is highly valued. 487Natural environmental amenities like ‘view of navigable water body', 488account for only about 3.4% of variation in rental values in the study area. 489Some independent variables which are excluded in the regression table are 490due to their insignificance, e.g. ‘distance to recreational centres 491(DISRECR)'. The variables ‘plot size (PLOT)' and ‘private garden space 492(GARDEN)' were found to highly correlate, thus introducing the problem of 493