# RECORD 63d3e41634a76fe68a3edb5f056b73b7098ad523 # TITLE: The Impact of Water Body Restoration on the Sales and Rental Markets in Beijing # YEAR: 2023.0 VENUE: Land Economics DOI: 10.3368/le.99.4.041322-0030r1 # AUTHORS: Ying-Dan Mei; P. Liu; Li Gao # PAGES: 36 CANDIDATE TABLE PAGES (1-based): [30, 31] # ABSTRACT: Black and odorous water bodies (BOWBs) are increasingly common in cities worldwide. This article evaluates the impact of an urban water bodies restoration program in Beijing, China, based on detailed housing sale and rental transaction data. We implement a hedonic price model with a difference-in-difference and a triple-difference approach and find that listing and restoring a BOWB significantly increase house prices by 2.2%–3.9% and 2.3%–4.9%, respectively, and increases rents by 11.3%–13.7% and 13.4%–16.9%, respectively. ===== PAGE 1 (score 3, numeric-paren 0, table-words 0, var-words 3) ===== 1 The Impact of Water Body Restoration on the Sales and Rental Markets in Beijing Yingdan Mei, Associate Professor, School of Applied Economics, Renmin University of China, Beijing, PR China; meiyingdan@vip.sina.com Pengfei Liu, Assistant Professor, Department of Environmental and Resource Economics, University of Rhode Island, Kingston; pengfei_liu@uri.edu Li Gao, Associate Professor, School of Economics and Management, China University of Petroleum Beijing, Beijing, PR China; gaolicup@126.com Abstract: Black and odorous water bodies (BOWBs) are increasingly common in cities worldwide. This paper evaluates the impact of an urban water bodies restoration program in Beijing, China, based on detailed housing sale and rental transaction data. We implement a hedonic price model with a difference-in-difference and a triple difference approach and find that listing and restoration of a BOWB significantly increase house prices by 2.2-3.9 percent and 2.3-4.9 percent, respectively, while at the same time increase rents by 11.3-13.7 percent and 13.4-16.9 percent, respectively. Keywords: hedonic price model; urban water body restoration; rent; sales price doi:10.3368/le.99.4.041322-0030R1 by guest on October 20, 2026. Copyright 2023 Downloaded from ===== PAGE 30 (score 83, numeric-paren 53, table-words 11, var-words 0) ===== 2 Table 2. The effect of restoration of BOWBs on housing values: Difference-in-Difference results (1) (2) (3) (4) (5) (6) (7) (8) 0-500m 0-600m 0-700m 0-800m 0-900m 0-1,000m 0-1,100m 0-1,200m Panel A: Sales market Interim × Near 0.0309*** 0.0193* 0.0133 0.0117 0.0128 0.0103 0.0104 0.0091 (0.0109) (0.0102) (0.0100) (0.0093) (0.0091) (0.0085) (0.0083) (0.0079) Post × Near 0.0384** 0.0272* 0.0253 0.0235 0.0233* 0.0195 0.0175 0.0149 (0.0155) (0.0154) (0.0154) (0.0141) (0.0138) (0.0132) (0.0132) (0.0129) Control variables × × × × × × × × Month FE × × × × × × × × BOWB FE × × × × × × × × Ring-road FE × × × × × × × × District × year FE × × × × × × × × Observations 70521 73702 75582 77308 78815 80924 82574 84547 Adj R-squared 0.8651 0.8642 0.8639 0.8642 0.8638 0.8637 0.8640 0.8645 Panel B: Rental market Interim × Near 0.0593*** 0.0415** 0.0336* 0.0302* 0.0292* 0.0312** 0.0286** 0.0296** (0.0179) (0.0190) (0.0179) (0.0164) (0.0152) (0.0135) (0.0127) (0.0124) Post × Near 0.0808*** 0.0547** 0.0480** 0.0429* 0.0390* 0.0387** 0.0354** 0.0361** (0.0223) (0.0244) (0.0237) (0.0215) (0.0201) (0.0182) (0.0169) (0.0163) Control variables × × × × × × × × Month FE × × × × × × × × BOWB FE × × × × × × × × Ring-road FE × × × × × × × × District × year FE × × × × × × × × Observations 63283 65502 67124 68711 70282 71757 73338 75001 Adj R-squared 0.8799 0.8794 0.8795 0.8794 0.8790 0.8784 0.8791 0.8795 Note: This table reports regression coefficients from 16 separate regressions, 8 per panel. The dependent variable is Log(sales prices) in Panel A, and Log(monthly rent) in Panel B. Both the regression sample and the buffer distance denoted by “Near” change as one moves across the columns, indicated by the column headings. Properties outside the buffer zone are all located within 1,200-3000 meters of a BOWB for all regressions. All standard errors in the parentheses are clustered at the BOWB level. ***Significant at the 1 percent level. **Significant at the 5 percent level. *Significant at the 10 percent level. by guest on October 20, 2026. Copyright 2023 Downloaded from ===== PAGE 31 (score 91, numeric-paren 61, table-words 11, var-words 0) ===== 3 Table 3. The effect of restoration of BOWBs on housing values: Triple difference results (1) (2) (3) (4) (5) (6) (7) (8) 0-500m 0-600m 0-700m 0-800m 0-900m 0-1,000m 0-1,100m 0-1,200m Panel A: Sales market Treat × Interim × Near 0.0388** 0.0301* 0.0261 0.0247 0.0297** 0.0244* 0.0239* 0.0219* (0.0182) (0.0163) (0.0164) (0.0159) (0.0143) (0.0136) (0.0124) (0.0119) Treat × Post × Near 0.0486** 0.0353* 0.0353* 0.0333* 0.0350** 0.0306* 0.0275* 0.0230 (0.0190) (0.0179) (0.0185) (0.0170) (0.0162) (0.0160) (0.0154) (0.0152) Control variables × × × × × × × × Month FE × × × × × × × × BOWB FE × × × × × × × × Ring-road FE × × × × × × × × District × year FE × × × × × × × × Observations 101542 108110 113477 118147 122778 127301 130953 134272 Adj R-squared 0.8478 0.8476 0.8476 0.8497 0.8499 0.8499 0.8513 0.8520 Panel B: Rental market Treat × Interim × Near 0.1369*** 0.1139*** 0.1167*** 0.1189*** 0.1199*** 0.1191*** 0.1164*** 0.1133*** (0.0364) (0.0354) (0.0334) (0.0339) (0.0343) (0.0334) (0.0326) (0.0318) Treat × Post × Near 0.1685*** 0.1399*** 0.1453*** 0.1448*** 0.1439*** 0.1416*** 0.1375*** 0.1338*** (0.0457) (0.0446) (0.0429) (0.0434) (0.0442) (0.0430) (0.0416) (0.0405) Control variables × × × × × × × × Month FE × × × × × × × × BOWB FE × × × × × × × × Ring-road FE × × × × × × × × District × year FE × × × × × × × × Observations 93166 98950 104009 108356 113663 117542 121283 124185 Adj R-squared 0.8550 0.8539 0.8542 0.8558 0.8567 0.8548 0.8558 0.8563 Note: This table reports regression coefficients from 16 separate regressions, 8 per panel. The dependent variable is Log(sales prices) in Panel A, and Log(monthly rent) in Panel B. Both the regression sample and the buffer distance denoted by “Near” change as one moves across the columns, indicated by the column headings. Properties outside the buffer zone are all located within 1,200-3000 meters of a BOWB for all regressions. All standard errors in the parentheses are clustered at the BOWB level. ***Significant at the 1 percent level. **Significant at the 5 percent level. *Significant at the 10 percent level. by guest on October 20, 2026. Copyright 2023 Downloaded from ===== PAGE 32 (score 0, numeric-paren 0, table-words 0, var-words 0) ===== 4 Figure 1. The timeline of restoration for urban BOWBs in Beijing by guest on October 20, 2026. Copyright 2023 Downloaded from