# RECORD 6663c2076db87ffe9baf2162abc65f7ffa450ab0
# TITLE: Evaluating the effects of supervised consumption sites on housing prices in Montreal, Canada using interrupted time series and hedonic price models
# YEAR: 2024.0  VENUE: Drug and Alcohol Dependence Reports  DOI: 10.1016/j.dadr.2024.100242
# AUTHORS: M. Schaefer; D. Panagiotoglou
# PAGES: 16  CANDIDATE TABLE PAGES (1-based): [13, 14, 15]
# ABSTRACT: Background In 2017, three brick and mortar supervised consumption sites (SCS) opened in Montreal, Canada. Opponents argued the sites would attract people who use drugs and reduce local real estate prices. Methods We used interrupted time series and hedonic price models to evaluate the effects of Montreal’s SCS on local real estate prices. We linked the Quebec Professional Association of Real Estate Brokers’ housing sales data provided by Centris Inc. with census tract data and gentrification scores. Homes sold within 200 m of the SCS locations between 1 January 2014 and 31 December 2021 were included. We adjusted for internal (e.g., number of bed/bathrooms, unit size) and external attributes (e.g., neighbourhood demographics), and included a spatio-temporal lag to account for correlation between sales. For sensitivity analysis we used site-specific dummy variables to better account for unmeasured neighbourhood differences, and repeated analyses using 500 m and 1000 m radii. Results We observed a price shock after the opening of the first two SCS in June 2017 (level effect: −10.5%, 95% CI: −19.1%, −1.1%) but prices rose faster month-to-month (trend effect: 1.1%, 95% CI: 0.7%, 1.6%) after implementation. Following the implementation of the third site in November 2017 there was no immediate impact (level effect: 2.4%, 95% CI: −10.4%, 17.0%) but once more prices roses faster (0.9%, 95% CI: 0.4%, 1.5%) thereafter. When we replaced neighbourhood attributes with a site-specific dummy variable, we observed the same pattern. Sales’ prices dropped (level effect: −9.6%, 95% CI: −15.0%, −3.8%) but rose faster month-to-month (trend effect: 0.9%, 95% CI: 0.6%, 1.2%) following June 2017’s SCS implementations, with no level effect (4.9%, 95% CI: −7.3%, 18.6%) and a positive trend (0.9%, 95% CI: 0.5%, 1.3%) after November 2017’s SCS opening. In most 500 m and 1000 m radii models, there were no immediate shocks following SCS opening, however, positive trend effects persisted in all models. Conclusion Our models suggest homes sold near SCS may experience a price shock immediately post-implementation, with evidence of market recovery in the months that follow.


===== PAGE 1 (score 6, numeric-paren 0, table-words 0, var-words 6) =====

Title: Evaluating the effects of supervised consumption sites on housing prices in Montreal, Canada 
using controlled interrupted time series and hedonic price models 
Authors: Maximilian Schaefer MScPH1, Dimitra Panagiotoglou PhD1 
1Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, 
Canada 
 
Corresponding author: 
Dimitra Panagiotoglou 
Department of Epidemiology, Biostatistics and Occupational Health 
McGill University 
1200-2001 McGill College Avenue 
Montreal, QC H3A 1G1 
Canada 
 
Contributors 
DP conceptualized the study, acquired funding, and obtained the data. All authors accessed and 
verified the underlying data. MS conducted the analyses. All authors interpreted the results. DP drafted 
the original version of the manuscript. MS compiled tables and figures. All authors reviewed and edited 
the manuscript. All authors approved the final version and accept responsibility for the decision to 
submit the manuscript for publication. 
 
Declaration of interests 
We declare no competing interests. 
 
Data sharing 
Due to commercial restrictions, supporting data is not available.  
 
Funding 
This study was supported by an Insights Development Grant from the Social Sciences and 
Humanities Research Council of Canada (Grant # 430-2022-00061) to DP. DP is also supported by a 
Junior 1 Scientist Award from the fonds de recherche du Québec – Santé (Award #297427). The funders 
of the study had no role in the study design, data collection, data analysis, data interpretation, or writing 
of the report. 
 
 
 
 
All rights reserved. No reuse allowed without permission. 
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 
The copyright holder for this preprint
this version posted June 12, 2023. 
; 
https://doi.org/10.1101/2023.06.09.23291209
doi: 
medRxiv preprint 
NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.


===== PAGE 13 (score 72, numeric-paren 40, table-words 3, var-words 11) =====

Table 1: Comparison of house and neighborhood features of treated and control units pre-intervention (1 January 2014 – 31 July 2017) 
 
Primary analysis 
Subset analysis 
 
Control units 
N=1,537 
Treated units 
N=765 
p-value 
Control units 
N=830 
Treated units 
N=714 
p-value 
Closing price, mean (SD) 
$419,001.00 
($339,096.23) 
$271,473.73 
($123,347.53) 
<0.001 
462,517.94 
(405,532.79) 
271,662.04 
(125,255.45) 
<0.001 
Housing features, mean (SD): 
 
  No. of bathrooms 
1.26 (0.47) 
1.09 (0.30) 
<0.001 
1.32 (0.51) 
1.09 (0.30) 
<0.001 
  No. of bedrooms  
1.60 (0.70) 
1.87 (0.71) 
<0.001 
1.67 (0.72) 
1.86 (0.71) 
<0.001 
  No. of extra rooms 
1.96 (1.19) 
2.27 (1.14) 
<0.001 
1.92 (1.26) 
2.26 (1.14) 
<0.001 
  Floor size (in m2) 
80.85 (40.57) 
83.87 (27.15) 
0.063 
86.81 (44.49) 
83.71 (27.10) 
0.105 
  Distance to closest 
shelter/SCS 
684.38 (239.35) 
607.41 (276.54) 
<0.001 
632.41 (259.01) 
605.04 (282.36) 
0.047 
Neighbourhood demographics, mean (SD): 
 
  Age of the population  
32.91 (4.53) 
35.3 (2.12) 
<0.001 
31.53 (3.96) 
35.80 (2.05) 
<0.001 
  Household income  
$36,615.76 
($17,124.33) 
$28,663.55 
($4,515.64) 
<0.001 
$26,304.72 
($15,998.87) 
$28,520.07 
($3,833.68) 
<0.001 
  Household size 
1.68 (0.18) 
1.77 (0.10) 
<0.001 
1.69 (0.18) 
1.77 (0.10) 
<0.001 
Proportion of population, %:  
 
 
 
 
 
 
    Indigenousb 
0.41 
1.08 
<0.001 
0.25 
1.03 
<0.001 
    Visible minoritiesc 
41.87 
15.65 
<0.001 
49.76 
15.57 
<0.001 
    Without secondary school 
completedd 
2.23 
12.04 
<0.001 
2.70 
12.07 
<0.001 
    With postsecondary 
education 
89.20 
73.75 
<0.001 
87.83 
73.75 
<0.001 
    Unemployed (rate) 
8.49 
8.40 
0.555 
10.39 
8.33 
<0.001 
Proximity on scale from 0 to 1, mean (SD): 
 
  Employmente 
 
 
 
0.37 (0.08) 
0.18 (0.05) 
<0.001 
  Pharmacyf 
 
 
 
0.26 (0.17) 
0.14 (0.09) 
<0.001 
  Childcare facilityg  
 
 
 
0.35 (0.08) 
0.34 (0.11) 
0.017 
  Healthcare facilityh 
 
 
 
0.11 (0.05) 
0.04 (0.05) 
<0.001 
  Grocery storef 
 
 
 
0.27 (0.11) 
0.19 (0.11) 
<0.001 
  Primary schoolg 
 
 
 
0.05 (0.05) 
0.27 (0.10) 
<0.001 
All rights reserved. No reuse allowed without permission. 
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 
The copyright holder for this preprint
this version posted June 12, 2023. 
; 
https://doi.org/10.1101/2023.06.09.23291209
doi: 
medRxiv preprint 


===== PAGE 14 (score 18, numeric-paren 12, table-words 0, var-words 3) =====

  Neighbourhood parkf  
 
 
 
0.25 (0.10) 
0.23 (0.15) 
0.015 
  Public transitf  
 
 
 
0.09 (0.03) 
0.04 (0.02) 
<0.001 
Gentrification measure 
 
  Gentrifiable in 2006i  
0.00 (0.00) 
0.62 (0.49) 
<0.001 
0.00 (0.00) 
0.60 (0.49) 
<0.001 
  Gentrified in 2016i  
0.00 (0.00) 
0.18 (0.39) 
<0.001 
0.00 (0.00) 
0.19 (0.40) 
<0.001 
aNaturalized Canadian citizens, permanent residents, temporary residents 
bFirst Nations, Métis, Inuk and/or Registered or Treaty Indians and/or membership in a First Nation or Indian band 
cPersons, other than Indigenous persons, who are non-Caucasian 
dNo certificate, diploma, or degree 
eWithin driving distance of 10 km 
fWithin walking distance of 1 km 
gWithin walking distance of 1.5 km 
hWithin driving distance of 3 km 
iUsing the Grube-Cavers indicator 
 
All rights reserved. No reuse allowed without permission. 
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 
The copyright holder for this preprint
this version posted June 12, 2023. 
; 
https://doi.org/10.1101/2023.06.09.23291209
doi: 
medRxiv preprint 


===== PAGE 15 (score 15, numeric-paren 0, table-words 2, var-words 2) =====

Table 2: Results of variations of controlled interrupted time series adjusting for housing and 
neighborhood attributes, without and with proximity measures (models 1 and 2 vs. 3 and 4, 
respectively), and restricted to pre-March 2020 sales (models 2 and 4). 
 
Model 1a 
(95% CI) 
Model 2b 
(95% CI) 
Model 3c 
(95% CI) 
Model 4d 
(95% CI) 
Intercept ( 
$44,602 
($34,419, $57,797) 
$44,238 
($32,288, $60,611) 
$93,330 
($64,537, $134,968) 
$70,231 
($44,636, $110,504) 
Time ( 
1.002 
(1.001, 1.002) 
1.001 
(1.000, 1.002) 
1.003 
(1.002, 1.004) 
1.002 
(1.001, 1.004) 
Group ( 
0.732 
(0.701, 0.765) 
0.761  
(0.725, 0.798) 
0.722 
(0.680, 0.766) 
0.716 
(0.668, 0.766) 
Group*Time ( 
0.998 
(0.997, 1.000) 
0.999  
(0.998, 1.000) 
0.997 
(0.995, 0.998) 
0.997  
(0.996, 0.999) 
Level ( 
1.077 
(1.054, 1.101) 
1.001  
(0.976, 1.027) 
1.045 
(1.015, 1.076) 
0.986  
(0.953, 1.021) 
Trend ( 
1.003 
(1.002, 1.004) 
1.007  
(1.005, 1.008) 
1.000 
(0.999, 1.001) 
1.003  
(1.001, 1.004) 
SCS level ( 
0.948 
(0.912, 0.986) 
1.000  
(0.955, 1.047) 
0.967 
(0.925, 1.011) 
0.997  
(0.945, 1.051) 
SCS trend ( 
1.006 
(1.004, 1.007) 
1.003  
(1.000, 1.005) 
1.009 
(1.007, 1.010) 
1.008  
(1.005, 1.010) 
Pseudo-R2 
0.837 
0.842 
0.851 
0.857 
n 
6,903 
4,801 
4,815 
3,313 
aControlling for housing and neighbourhood attributes with spatio-temporal price lag, excluding proximity scores 
bControlling for housing and neighbourhood attributes with spatio-temporal price lag; restricted to sales pre-
March 2020, excluding proximity scores 
cControlling for housing and neighbourhood attributes with spatio-temporal price lag, including proximity scores 
dControlling for housing and neighbourhood attributes with spatio-temporal price lag; restricted to sales pre-
March 2020, including proximity scores 
bold indicates statistical significance 
 
 
 
All rights reserved. No reuse allowed without permission. 
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 
The copyright holder for this preprint
this version posted June 12, 2023. 
; 
https://doi.org/10.1101/2023.06.09.23291209
doi: 
medRxiv preprint 


===== PAGE 16 (score 2, numeric-paren 0, table-words 0, var-words 2) =====

 
 
 
 
Figure 2: Comparison of price in treated vs. control sales pre- vs. post-intervention implementation 
controlling for house and neighborhood amenities, with and without proximity scores in models 
 
All rights reserved. No reuse allowed without permission. 
(which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 
The copyright holder for this preprint
this version posted June 12, 2023. 
; 
https://doi.org/10.1101/2023.06.09.23291209
doi: 
medRxiv preprint