The main drawback of the data is that we do not observe whether an individual is a social assistance recipient. However, an additional advantage of testing for check
effects among IDUs in Vancouver is that we know from other sources that social assistance receipt is likely to be extremely high in our data. For example, in June
1999, of the 405 IDUs in the provincially funded Vancouver Injection Drug-User Program, 89 percent were welfare recipients Palepu et al. 2001. Unfortunately, it is
not possible to match our data with social assistance records from the Ministry of Human Resources due to confidentiality concerns.
We also conducted our own test of social assistance receipt. We randomly chose a week in June 2002 and asked a welfare caseload worker from the Ministry to assist us
in a survey of all IDUs that were admitted to the hospital on that day. We first col- lected information including detailed interviews on the IDUs admitted, and then
called the worker at the Ministry the following week to confirm the relevant social assistance information. Of the 25 IDUs admitted, 22 were receiving Assistance two
of three not receiving assistance indicated homelessness including four on basic, four on Disability I, and 14 on Disability II.
We also have information on housing status, which will be used in the analysis to provide a test of social assistance receipt. To collect social assistance on a consistent
basis, one must have a fixed address. Based on our discussions with Ministry staff, some recipients have been cut off from support because they were unable unwilling
to find shelter. Our own experience with this population is that the homeless IDUs do not keep the Ministry updated on information required for continuing benefits,
such as efforts to find a job or alternative income sources. Moreover, British Columbia law eliminates the “comfort money” component of the support allowance
payment for recipients defined as transients, which includes those without depend- ent children who have no fixed address thereby leaving such recipients with only the
food component.
11
V. Welfare Wednesday and the Distribution of Overdose Admissions
In this section, we examine the relationship between the distribution of welfare payments and the distribution of drug consumption. The dates of check
arrival were obtained from the British Columbia Ministry of Human Resources. The date of check arrival must fall on the Wednesday that has at least three working days
before the end of the month.
12
This typically amounts to the last Wednesday of the month with the exception of December, where check arrival is always the second to
last Wednesday of the month due to the holidays. Including December, welfare checks are released on the second to last Wednesday in about 40 percent of cases. The
Ministry releases a listing of Welfare Wednesdays posted at local offices and the web at the beginning of each year.
11. See Section 10 of Schedule A of the British Columbia Benefits Regulations. 12. This rule is because it takes three working days for checks to clear the bank, and thus there is time for
a check to be replaced if lost or stolen.
Riddell and Riddell 145
We use drug overdose admissions as a proxy for drug consumption, which includes all drug poisoning admissions.
13
Figure 1 presents the distribution of overdose admis- sions over time. A spike in overdose admissions is seen on Welfare Wednesday
through Welfare Friday; there is no evidence of a “Mardi-Gras Weekend” effect. To further explore the pattern in Figure 1 we estimate the following probit regression:
1 OVERDOSE
it
= f γ
+ γ
1
WDAYS
it
+ γ
2
X
it
+ γ
3
DAY
it
+ γ
4
WEEK
it
+ γ
5
MONTH
it
+ γ
6
YEAR
it
+ µ
it
where OVERDOSE equals one if the ith individual was admitted to the hospital on the tth admission with a principal diagnosis of a drug overdose, zero if any other diagnosis;
WDAYS is a vector of five welfare days Welfare Wednesday through Welfare Sunday corresponding to Figure 1 dummies; X is a vector of demographic variables age, gender,
homeless, HIV status; DAY, WEEK, MONTH, and YEAR are sets of six day-of-the week, three week-of-the-month
14
last week of month, penultimate week of month, and so forth, 11 calendar-month, and four fiscal-year dummy variables respectively; and
µ is an error. As noted above, Welfare Wednesday is not always during the last week of the
month. In fact, about 40 percent of checks over the sample period were distributed on an earlier Wednesday. Equation 1, and other regressions estimated later in the paper,
exploit this variation and allow us to address any general end-of-the-month effects unrelated to welfare checks per se.
Table 2 presents the results. The estimates are consistent with the pattern seen in Figure 1 although the Welfare Wednesday variable is just barely outside of conven-
tional significant levels. The interpretation from Column 1 is somewhat unwieldy. For instance, overdoses are about six percentage points more likely to occur on Welfare
Thursday relative to any other Thursday, and three 0.06 −0.03 percentage points
more likely to occur relative to Sunday the omitted day. Ceteris paribus, overdoses are less likely to occur during the week relative to the weekend. A test of the equality
of all day-of-the-week dummies can be rejected at the one percent level χ
2
= 15.4, but an equality test on the five weekday dummies cannot be rejected
χ
2
= 2.0 and so we alter the specification as given in the second column. The welfare day variables
now can be evaluated relative to any nonwelfare-week weekday. Overdoses admission are about five percentage points more likely to occur on a Welfare Thursday and about
four percentage points more likely to occur on the Welfare Friday. No statistically sig- nificant effects are found for any other day during the welfare week. The homeless are
more likely to overdose while women and HIV positive individuals are less likely.
13. The primary diagnosis is the “most responsible” diagnosis. Formally, the drug overdose category includes “poisoning by heroin,” “poisoning by cocaine,” and some other miscellaneous drug poisonings. Our
drug overdose approach is thus much cleaner than in most studies where broad and generally hard-to- diagnose drug-related diagnoses are considered in particular, drug-related psychological diagnoses. For our
analysis that uses drug overdoses results discussed in Section V and Section VII, we get smaller effects if we include these psychological diagnoses as overdose cases.
14. We compute these based on seven-day intervals starting from the last day of the month as opposed to Monday to Sunday intervals. In this manner, the “last week of month” always includes the last Wednesday
welfare Wednesday 60 percent of the time and the “third week of the month” always includes the penulti- mate Wednesday welfare Wednesday 40 percent of the time. We use three week-of-the-month dummies in
the regressions with the omitted category being the first week. The omitted category includes the residual of a few days that normally remains—for instance, three days for 31 day months—but the results are unaf-
fected if we drop these residual days.
The Journal of Human Resources 146
Riddell and Riddell 147
2 4
6 8
12 14
16 18
10 M onday before
Tuesday before Wednesday before
Thursday before F riday before
Saturday before Sunday before
Welfare M onday Welfare Tuesday
Welfare Wednesday Welfare Thursday
Welfare F riday Welfare Saturday
Welfare Sunday M onday after
Tuesday after Wednesday after
Thursday after F riday after
Saturday after Sunday after
Figur e 1
Distrib ution of Drug Over
dose Admissions, FY1996–2000
The Journal of Human Resources 148
Table 2 Estimates of the Change in Probability of a Drug Overdose Admission
Variable Specification
[1] [2]
Welfare Wednesday 0.028
0.028 0.019
0.017 Welfare Thursday
0.057 0.048
0.028 0.022
Welfare Friday 0.039
0.041 0.023
0.021 Welfare Saturday
−0.029 −0.027
0.012 0.013
Welfare Sunday 0.002
−0.003 0.022
0.019 Homeless
0.026 0.025
0.011 0.012
Downtown Eastside postal code 0.007
0.007 0.009
0.009 Other downtown Vancouver postal code
−0.001 −0.001
0.009 0.009
Female −0.017
−0.017 0.006
0.006 HIV positive
−0.027 −0.027
0.006 0.006
Monday −0.028
— 0.009
Tuesday −0.017
— 0.010
Wednesday −0.023
— 0.010
Thursday −0.027
— 0.010
Friday −0.022
— 0.010
Saturday 0.011
— 0.017
Weekend —
0.041 0.012
Log likelihood −897.6
−898.9 χ
2
100.5 97.9
Number observations 4,760
Notes: Huber-White standard errors are in parentheses. Statistical significance is denoted by for 1 per- cent level, for 5 percent level, and for 10 percent level. The dependent variable equals one if the prin-
cipal diagnosis of the admission was a drug overdose drug poisoning, zero for any other diagnosis, and has a mean of 0.051. All regressions also include controls for: age and its square, three week-of-month, 11 cal-
endar-month, and four year dummies. All regressions are estimated by probit. All estimates are presented as marginal effects, and are evaluated at the mean of the relevant covariate. The source is a census of hospital
admissions of injection drug-users admitted over fiscal years 1996 to 2000 at St. Paul’s Hospital in Vancouver.
There are two types of overdose cases that we do not observe in our data: a fatal over- doses, and b nonfatal overdoses that do not result in a hospital admission.
15
One of the current views—with considerable empirical support—in the literature on drug overdoses
is that most fatal drug overdoses could be prevented because fellow users are unlikely to call for an ambulance Warner-Smith et al. 2001. For the current analysis, unobserved
overdoses are problematic if the distribution of overdoses that result in a hospital admis- sion differs from the distribution of unobserved overdoses. Overall, we believe this is
unlikely; however, one possibility is that the spike in overdose cases in Figure 1 is due to a greater police presence in the Downtown Eastside during the welfare week and thus a
higher probability of the police finding an individual who has overdosed and calling paramedics resulting in a hospital admission that would not normally occur. However,
the argument could go the other way—with a greater police presence in what is a very small geographic area, users may be even less likely to call for an ambulance.
There appears to be no previous evidence on the link between welfare day and either drug overdose hospital admissions or fatal drug overdoses. Phillips, Christenfeld, and
Ryan 1999 examine U.S. death certificates from 1973 to 1988 and find an overall increase of 1 percent in the number of deaths in the first week of the month the U.S. wel-
fare week relative to the last week of the previous month, and a 14 percent increase in substance abuse-related deaths which includes a wide variety of causes of death. From
Canada, there is only evidence from 1993 for British Columbia where Verheul, Singer, and Christenson 1997 find a 50 percent increase in coroner-reported deaths on welfare day.
The authors also find increases in detox center admissions and 911 calls on welfare day.
VI. Welfare Wednesday and the Distribution of Hospital Discharges