1# RECORD 7409c1d2d6de0d63dfd17da30463090cc23beeee2# TITLE: Property Supply, Economic Factors and House Prices in Nairobi Kenya3# YEAR: 2024.0 VENUE: Journal of African Real Estate Research DOI: 10.15641/jarer.v9i2.14574# AUTHORS: D. Chirchir; Mirie Mwangi; Cyrus Iraya5# PAGES: 32 CANDIDATE TABLE PAGES (1-based): [18, 23]6# ABSTRACT: Residential real estate is an important asset class for both institutional and individual investors. The housing bubble in the US led to a global financial crisis in 2007 – 2008. As such, understanding the factors that influence property prices is central to the investment process and portfolio management. The objective of the paper was to first construct a Nairobi house price index and then determine the relationship among economic factors, property supply, and house prices. The study period was ten years (2011Q1 to 2020Q4). Autoregressive Distributed Lag (ARDL) model was adopted to estimate long run and short run relationships. The Nairobi real estate price index was developed using hedonic model based on the data collected on actual selling price and characteristics of a sample of houses in Nairobi spanning ten years. The paper finds that GDP and inflation have significant positive long run effect on house prices. Property supply has a significant negative long run effect on real estate prices. However, interest rates do not have significant effect on house prices. In the short run however, GDP had a significant negative effect on prices. The speed of adjustments towards equilibrium relationship was 71.9%. The real estate market seems to be efficient despite its illiquid nature. The paper contributes to knowledge through its construction of house price index in Nairobi. Also, it included property supply a departure from past studies. The findings may have implications for investors, banks, regulators of securities and policy makers. The government of Kenya may be motivated to commission an official house price index.789===== PAGE 1 (score 5, numeric-paren 0, table-words 0, var-words 5) =====1011Journal of African Real Estate Research Volume 9(2) 2024 12 13 14 155516Journal of African Real Estate Research 17Volume 9, Issue 2 18 www.journals.uct.ac.za/index.php/JARER/ 19 20Property Supply, Economic Factors and House Prices in Nairobi Kenya 21Dan Chirchir1* (https://orcid.org/0000-0003-0049-7566), Mirie Mwangi2 22(https://orcid.org/ 0000-0002-5947-4468) and Cyrus Iraya3 (https://orcid.org/ 0009-0007-232883-3580) 241 2 3 University of Nairobi 25To cite this article: Chirchir, Mwangi and Iraya (2024) Property Supply, Economic Factors 26and House Prices in Nairobi Kenya. Journal of African Real Estate Research, 9(2), pp.55-86. 27DOI: https://doi.org/10.15641/jarer.v9i2.1457 28 29 30Abstract 31 32Residential real estate is an important asset class for both institutional and individual investors. 33The housing bubble in the US led to a global financial crisis in 2007 – 2008. As such, 34understanding the factors that influence property prices is central to the investment process and 35portfolio management. The objective of the paper was to first construct a Nairobi house price 36index and then determine the relationship among economic factors, property supply, and house 37prices. The study period was ten years (2011 Q1 to 2020 Q4). Autoregressive Distributed Lag 38(ARDL) model was adopted to estimate long-run and short-run relationships. The Nairobi real 39estate price index was developed using a hedonic model based on the data collected on actual 40selling prices and characteristics of a sample of houses in Nairobi spanning ten years. The paper 41finds that GDP and inflation have significant positive long-run effects on house prices. Property 42supply has a significant negative long-run effect on real estate prices. However, interest rates 43do not have a significant effect on house prices. In the short run, however, GDP had a 44significant negative effect on prices. The speed of adjustments towards the equilibrium 45relationship was 71.9%. The real estate market seems to be efficient despite its illiquid nature. 46The paper contributes to knowledge by including property supply, a departure from past 47studies. The findings may have implications for investors, banks, regulators of securities and 48policymakers. The government of Kenya may be motivated to commission an official house 49price index. 50 51Keywords: ARDL, ECM, Cointegration, Hedonic 52 53 54*Corresponding author’s email address: dchirchir@uonbi.ac.ke 55©2024 The Author(s). Published by UCT Library. This is an open-access article under the CC BY 4.0 license. 56https://creativecommons.org/licenses/by/4 575859===== PAGE 18 (score 14, numeric-paren 0, table-words 5, var-words 2) =====6061Journal of African Real Estate Research Volume 9(2) 2024 62 63 64 657266Table 4: Diagnostic test results 67Test 68Method 69Statistic Prob 70Autocorrelation 71Breusch Godfrey LM test 722.62 0.6232 73Heteroscedasticity Breusch-Pagan test 7422.80 0.2985 75Normality 76Jarque-Bera 770.24 0.8854 78Source: Author (2023) 79 80Table 5: Multicollinearity test 81Variable 82 VIF 83 Inflation Index 842.592 85 Interest 862.448 87 GDP 881.833 89 LN Value of Approved Plans 901.516 91 Mean VIF 922.097 93Source: Author (2023) 94The Augmented Dickey-Fuller and Phillips-Perron unit root tests were used to test for 95stationarity with outcomes depicted in Table 6 and Table 7 respectively. 96 Table 6: Stationarity test (Augmented Dickey-Fuller) 97Variable 98Test 99statistic 100P-101value 102I (0) 103Test 104statistic 105P-106value 107I (1) 108Price Index 109-2.162 1100.220 111-5.517 1120.000 113 114Interest 115-2.014 1160.281 117-4.019 1180.001 119 120GDP 121-3.077 1220.028 123 124 125 126Value of Approved Plans 127-3.451 1280.009 129 130 131 132Inflation Index 1330.157 1340.969 135-6.56 1360.000 137 138Source: Author (2023) 139 140141142===== PAGE 19 (score 12, numeric-paren 0, table-words 4, var-words 3) =====143144Journal of African Real Estate Research Volume 9(2) 2024 145 146 147 14873149Table 7: Stationarity test (Philips-Perron) 150Variable 151Test 152statistic 153P-154value 155I (0) 156Test 157statistic 158P-159value 160I (1) 161Price Index 162-2.564 1630.101 164-9.012 1650.000 166 167Interest 168-1.207 1690.671 170-3.577 1710.006 172 173GDP 174-3.276 1750.016 176 177 178 179Value of Approved Plans 180-4.286 1810.001 182 183 184 185Inflation Index 186-0.326 1870.922 188 189-5.910 1900.000 191 192Source: Author (2023) 193GDP and log of the value of approved plans were stationary at levels [(I (0)]. On the other 194hand, the price index, interest, and inflation index were not stationary at levels. However, the 195series became stationary at first difference [1 (1)]. The stationarity test resulted in mixed 196stationarity. The study therefore adopted the ARDL model. The ARDL model was 197implemented in five steps and the results are presented as follows. The ARDL optimal lags 198were determined as (1 1 0 0 0) using BIC as shown in Table 8. 199Table 8: Optimal lags selection 200Model 201N 202ll (null) 203ll (model) 204Df 205BIC 206. 20736 208-145.1738 209-127.8625 2107 211280.8096 212Price Index 2131 214GDP 2151 216Inflation Index 2170 218Interest 2190 220LN of Value of Approved Plans 221 222 223 224 2250 226Source: Author (2023) 227The ARDL (1,1,0,0,0) was specified as follows: 228𝑃𝑟𝑖𝑐𝑒𝐼𝑛𝑑𝑒𝑥𝑡 = 𝐶0 + ∅𝑃𝑟𝑖𝑐𝑒𝐼𝑛𝑑𝑒𝑥𝑡−1 + 𝛽1𝐺𝐷𝑃𝑡+ 𝛽2𝐺𝐷𝑃𝑡−1 + 𝛽3𝐼𝑛𝑓𝑙𝑎𝑡𝑖𝑜𝑛𝐼𝑛𝑑𝑒𝑥𝑡+229𝛽4𝐼𝑛𝑡𝑒𝑟𝑒𝑠𝑡𝑡+ 𝛽5𝐿𝑛 𝑉𝑎𝑙𝑢𝑒𝑜𝑓𝐴𝑝𝑝𝑟𝑜𝑣𝑒𝑑𝑃𝑙𝑎𝑛𝑠𝑡+ 𝑢𝑡…………………………………..… [4] 230 231The ARDL was run to test for the model’s goodness of fit. Table 9 summarises the result. 232 233234235===== PAGE 23 (score 19, numeric-paren 0, table-words 7, var-words 2) =====236237Journal of African Real Estate Research Volume 9(2) 2024 238 239 240 24177242 243Table 12: ECM Regression results 244ARDL (1,1,0,0,0) regression 245Sample: 2011q2 thru 2020q4 Number of obs = 39 246 R-squared = 0.5599 247 Adj R-squared = 0.4774 248Log likelihood = -138.18548 Root MSE = 9.2368 249D. Price Index 250Coefficient 251 Std. err. 252T 253 P>t 254ADJ 255Price Index 256L1. 257 -0.719 258 0.140 259 -5.150 260 0.000 261LR 262GDP 263 484.340 264 176.530 265 2.740 266 0.010 267Inflation Index 268 0.706 269 0.145 270 4.880 271 0.000 272Interest 273 242.373 274 163.963 275 1.480 276 0.149 277LN Value of Approved Plans 278 -18.722 279 8.042 280 -2.330 281 0.026 282 283SR 284GDP 285D1. 286 -221.910 287 92.540 288 -2.400 289 0.022 290 291_cons 292 88.785 293 66.075 294 1.340 295 0.188 296Source: Author (2023) 297The results indicate that economic factors and property supply explain 55.99% of the variation 298in house prices. Inflation and GDP have a positive significant effect on house prices while the 299value of approved plans has a negative effect. However, interest rates had no significant long-300run effect (p>0.05) on house prices. 301The adjustment factor of negative 0.719 is significant (t= -5.15; p<0.05). This indicates that 30271.9% of the deviations from the long-run equilibrium in a particular quarter will be corrected 303in the subsequent period. The high speed of adjustment may point to the efficiency of the 304housing market as it experiences correction from shocks. This tends to agree with the stock-305flow model and EMH. 306307308===== PAGE 24 (score 5, numeric-paren 0, table-words 2, var-words 1) =====309310Journal of African Real Estate Research Volume 9(2) 2024 311 312 313 31478315In the short run, changes in GDP (t = -2.40; p<0.05) have a significant effect on prices. A 1% 316increase in GDP will lead to a 2.21% decline in real estate prices in the short run. This seems 317out of step with the theoretically expected relationship. 318The validity of the outcome of the hypothesis test was subject to post-estimation tests. The 319model passed the four tests as shown in Table 13, Figure 6 and Figure 7. Specifically, there 320was no autocorrelation and heteroscedasticity on the fitted model. The errors were also 321normally distributed. 322Table 13: Summary of post-estimation tests 323Test 324Method 325Statistic Prob 326Autocorrelation 327Breusch Godfrey LM test 3284.021 0.4032 329Heteroscedasticity Breusch-Pagan test 33030.33 0.2994 331Normality 332Jarque-Bera 3330.395 0.8206 334Source: Author (2023) 335The fitted model assumes that the estimated parameters will remain stable over time. 336Cumulative sum (Cusum) and Cumulative sum of squares (CUSUMSQ) tests for parameter 337stability were used (Brooks, 2019). The fitted model was found to be stable as shown in Figure 3386 and Figure 7. 339 340 341 342 343 344 345 346 347 348 349 350