ECO 4190 University of Ottawa Impact of Externalities on The Demand and Supply of Healthcare PPT

Description

4 attachmentsSlide 1 of 4attachment_1attachment_1attachment_2attachment_2attachment_3attachment_3attachment_4attachment_4.slider-slide > img { width: 100%; display: block; }
.slider-slide > img:focus { margin: auto; }

Unformatted Attachment Preview

Faculté des sciences sociales | Faculty of Social Sciences
ECO 4190: Health Economics
Project Proposal, Fall 2021
The written component
The idea is to put together a proposal for an empirical project. Proposals are often
requested from supervisors for graduate theses/projects, as well as from granting
agencies.
For this proposal, you will first need to identify an unanswered (or insufficiently
answered) question in the existing literature. Then you will need to propose the
methodology and necessary data to address this question.
Specifics
? The subject matter should be related to some aspect of ‘health economics.’
? You will not have to carry out this project – unless you want to!
? The data set you propose to use must be a real data set, and the variables you
propose using need to exist in the data set.
? The proposal should be no more than 10 pages in length (double spaced, 12
point font). A suggested outline is provided below.
? Plagiarism is unacceptable.
? Due Date: November 24th, 2021 in class.
? Please bring 2 copies of the proposal (one for me and one for the discussant)
Other Thoughts
? This proposal is worth 30% of your final grade. It does not have to be very long
but what I am most interested in is seeing evidence of creativity and
independent thought.
? I will not give you a topic – learning to come up with interesting questions is a
valuable exercise in and of itself. (Of course, once you have an idea or two, I
would be happy to help finesse it or make some suggestions)
************************************************************************
A great reference for writing help:
McCloskey, Deidre N., 2000, “Economical Writing, Second Edition” Waveland Press
Inc.
This book is cheap, a quick read and very helpful.
************************************************************************
The oral components
Presentation:
? 10 minutes (exact time limit will depend on final enrollment and will be
announced in class)
? It is best to have some visual aids such as PowerPoint Slides
? Speak clearly and not too fast
? Slides should be clear – don’t overcrowd them and make the font size
sufficiently large (don’t use fancy fonts or colours that are difficult to see)
? It’s a short talk – keep your slides to a minimum. A good rule of thumb is
3 to 4 minutes per slide.
? Presentations will take place in the last 2 weeks of class.
Discussion/critique of a classmate’s proposal:
? One page document to be submitted by email the day of the classmate’s
presentation that summarizes the main strength(s) of their proposal and
describes concerns, possible limitations, possible extensions….
? Your grade for this component will depend on the quality/ originality/
creativity of your thoughts and arguments
? As the discussant, you will be required to ask the presenter one question
after the presentation.
Suggested Outline for the Written Component
Introduction
? Motivate the main question your paper would address. Why is it
interesting? Why is it relevant? What are the policy implications?
? Mini literature review. Introduce the main existing works in the area. Be
sure to describe the contributions these works have made so that the
reader can have a good understanding the contributions of your proposed
work.
? “This work will contribute to the existing literature in the following ways.”
Description of Methodology and Data to be used
? Name and description of the data set: How were the observations
chosen? From where? When? Are there other relevant aspects of this
data? What is the unit of observation?
? What type of analysis are you proposing? If it is regression analysis – be
sure to indicate the dependent variable(s) and key independent variables.
? Where is the identification coming from?
? Are there any anticipated empirical hurdles the project would have to
overcome (endogeneity, sample selection issues, measurement error
etc)? Do you have proposed solutions to these problems?
References
? Make sure all works cited in the proposal appear in the list of references
? Provide a complete reference, i.e. author, title, journal, etc. Follow the
style guide for one of the major journals.
Catherine Deri
What Lessons Can We Learn From the Body Mass of Immigrants?
Submitted for consideration to the U of T/York/Ryerson and the future Carleton, Ottawa and
Outaouais Local Research Data Centers
January 2005
Introduction
One of the most striking findings in the literature on immigrant health is the ‘Healthy Immigrant
Effect’ (HIE), the finding that recent immigrants are healthier than the average locally born resident,
but that over time this health advantage shrinks or even disappears. Evidence of this phenomenon
comes from Canada and elsewhere and spans a variety of health outcomes including self-reported
health, the incidence of various chronic conditions (such as heart disease, diabetes and cancer), mental
health, perinatal health and body mass.
A careful understanding of this effect and its magnitude is important for several reasons. It can
promote understanding of any social burden immigrants’ pose on the Canadian population 1. Further, as
immigrants represent over one sixth of the population, measures of the average health of Canadians –
often cited and important statistics in international comparisons – will be affected2. Finally, since
employability is one of the main characteristics by which immigrants gain admittance to Canada and
since health is essential for productivity in the workforce, a decline in the health of immigrants with
tenure may undermine one of the main goals of Canadian immigration policy3.
Until recently, only cross section estimates of the HIE were found in the literature. The problem with
drawing conclusions about immigrant health dynamics from a cross section alone is that a researcher
cannot separately identify changes in health with each year that an immigrant spends in Canada from
fixed differences between entry cohorts. That is, a researcher cannot disentangle the story of declining
health following immigration from either of the following: a story of increasing ‘health quality’ of
immigrants that have immigrated to Canada over time or a story of selective emigration such that the
healthier immigrants emigrate again from Canada resulting in a declining average health of the
immigrants that remain.
Only a few very recent studies have addressed this problem. Deri (2004) uses the first three cycles of
the Canadian National Population Health Survey (NPHS) to document the HIE using four measures of
health including both subjective measures of health (self assessed health and the self reports of activity
limitations) and objective measures (the incidence of chronic conditions and body mass index).
McDonald and Kennedy (2004) use data from the NPHS and the first cycle of the Canadian
Community Health Survey and find evidence of the HIE for the subjective measures of health
considered by Deri (2004). While not directly addressing the question of the HIE, Antecol and Bedard
1
The concern is the potential for the average cost of immigrants’ use of publicly provided health services to exceed the
average contribution immigrants make through taxes for these services.
2
In 1996, one sixth of the population was foreign born. One in five children was an immigrant or a child of an immigrant
(Kinnon 1999).
3
Canadian immigration policy requires that the economic class of immigrant applicants be screened through a ‘points
system’ that puts heavy emphasis on the migrant’s intended employment. In 2001, 61 percent of all immigrant applicants
belonged to this class. In 1992, 43 of the 70 points required for immigration could be obtained from the migrant’s intended
occupation. (See Grant 1999, and references therein).
1
Catherine Deri
(2003) use data from the American National Health Interview surveys from 1989-96 to examine the
assimilation profile of immigrants’ weights to those of the American born. They find that both men
and women enter the US with obesity rates on average seven to ten percentage points lower that white
American born men and women, but the gap closes entirely within 15 years for women, and half closes
for men in the same period.
However, these works do little more than verify the existence of the HIE and provide general estimates
of its magnitude. This proposed work seeks to investigate this phenomenon further. More than
verifying once again the existence of the HIE, I will exploit a larger, richer data set to highlight the
mechanisms through which the immigrants may be losing their relative health advantage and to
identify mitigating and aggravating factors. As well as providing insights in the HIE, they may also
provide insight into the determinants of obesity more generally.
There are three main characteristics of the proposed work I wish to emphasize. Firstly, this work will
focus exclusively on the body mass index (BMI) as the indicator for health status. BMI is an indicator
of total body fat or obesity. It is calculated as (weight in kilograms)/(height in meters)2. Obesity has
been shown to exacerbate many chronic conditions, have negative effects on patient’s health related
quality of life, to be associated with deteriorating functional status and increasing levels of distress,
and is a strong predictor of mortality4,5 By focusing on body mass as a measure of health, I circumvent
the following major criticism of the existing estimates of the HIE derived from subjective, self reported
measures of health. The concern is that, instead of health declining following immigration, it could be
that new immigrants are reluctant to reveal their poor health when they first arrive, but over time, they
become willing to reveal their problems. In such a case, a researcher using the standard methodologies
and data sets would find evidence of the HIE, when in fact no such deterioration of health over time
exists6. However, this mechanism is highly unlikely to affect BMI, as this measure is derived from
self-reports of the respondent’s weight and height.
Second, the data set will span a longer period that any of the existing works. This will allow for better
identification of the HIE for two reasons. First, for each cohort of arrival (say for the cohort of
immigrants arriving between 1986 and 1990) a longer data set would allow the researcher to observe
this cohort for more periods of time. Second, as it has been shown (Deri 2004, and McDonald and
Kennedy 2004) that the majority of the decline in health associated with the HIE occurs in the first few
years in the country, a data set spanning more years would permit the researcher to observe more
cohorts during their initial years in Canada.
Finally, and perhaps most excitingly, the household survey data used will be linked to more detailed
information about the individual’s environment. This will provide insight into mitigating and
aggravating environmental factors. It has been documented that lower income neighborhoods present
barriers to healthy lifestyle. Cities with fewer local parks will have fewer places for children to run
around and play. There may be fewer grocery stores, or more fast food restaurants. The information
available through Geographic Information Systems (GIS) will provide information on number of
4
See Hakim, Wolf and Garrison (2002) and Costa (1996) and references therein.
The limits of the BMI are that it will overestimate body fat in athletes and those with muscular builds and underestimate
body fat in older individuals and those who have lost muscle mass. See
http://www.nhlbi.nih.gov/health/public/heart/obesity/lose_wt/risk.htm#limitations for additional details on BMI.
6
There is some evidence for example that there are some cultural differences in the interpretation of mental health
questions and in the willingness to report symptoms of depression of alcohol dependence (Noh, Speechley, Kaspar et al.,
1992)
5
2
Catherine Deri
parks, grocery stores, meat and fish markets, fruit and vegetable markets and restaurants of various
kinds by small geographic regions. Though multivariate regression, I will be able to decipher how
much of the decline in health following immigration can be associated with such factors. As well, I
will consider whether social networks can mitigate the negative impact on body mass with tenure in
Canada. Using the subsample of immigrants whose mother tongue is neither English nor French, I will
proxy for the immigrants’ social network with a variable based on the density of their language group
living in the same geographic region7. This information will be obtained from the Canadian Censuses.
The motivation here is that, the more densely populated an ethnic group, the more access there is likely
to be to foods and restaurants from the home country. The less assimilation there is to Canadian norms,
the less assimilation there should be to Canadian weights.
Research Questions
This project will allow me to address the following questions:
1. Does the initial documented immigrant weight gap (the difference in weight between otherwise
comparable immigrants and Candian born residents) diminish with time in Canada?
2. Does this weight trajectory vary by source country (not available in the work by Antecol and
Bedard (2003))? Age at immigration? Sex? Socio-economic status? Measure of weight (BMI,
overweight, obesity)? Data sets used?
3. Are the estimates of weight assimilation comparable to those found for immigrants to the
United States by Antecol and Bedard (2003)?
4. Considering the subsample of those whose mother tongue is not English or French, does living
in a neighborhood more densely populated by others of his/her language group lessen the effect
of additional years in Canada on weight? The motivation here is that, the more densely
populated an ethnic group, the more access there is likely to be to foods and restaurants from
the home country. The less assimilation there is to Canadian norms, the less assimilation there
should be to Canadian weights. Put another way; is more integration into one’s culture
protective of one’s weight? Integration into one’s culture will be proxied for with controls for
the number of ethnic restaurants in the geographic neighborhood of the individual, as well as
the density of those of the individual’s language group living in the same neighborhood.
5. Does the number of local parks, grocery stores, meat and fish markets, fruit and vegetable
markets in the neighborhood affect changes in body weight of immigrants? Of Canadians more
generally?
Construction of the Data
The proposed data set will link Census and GIS data to the neighborhoods of the respondents from the
National Population Health Survey 1994-95, 1996-97, and 1998-99 and the first two cycles of the
Canadian Community Health Survey 2000-01 and 2003. For matching purposes, the Master Files of
the surveys, the version of the surveys which identifies the geographic location of the respondents, will
need to be used.
Both the Census and GIS data are available through the University of Ottawa’s Geographic
Information Services and I have confirmed by access to these data and the necessary variables.
7
A similar methodology was used in Deri (2003).
3
Catherine Deri
Methodology
With this new data set, I will apply multivariate regression techniques to examine the dynamics of
immigrants’ body mass with tenure in Canada. I will pool the five surveys together to create a quasipanel which will allow me to separately identify cohort effects from true changes in the body mass of
immigrants with tenure. I will use the existing literature to identify the other determinants of obesity
that need to be controlled for in my models. I will address problems of reverse causality for variables
such as household income. The necessary empirical strategies are similar to those I have used for my
doctoral thesis and other research endeavors.
The main specifications that will address the dynamics of immigrant body weight will be:
(1)
Body Weight ? ? 0 ? ?1YSM ? ? 2YSM 2 ? ? 3 X ? ? ? j Cohort j ? ? ? k Country of Birthk ?? ? l Cycle l ?
j
k
l
? ?? mCountry of Birth * YSM m ??? nCountry of Birth * YSM ??
2
n
m
n
Where Body Weight will be one of four measures of body weight: BMI, ln(BMI), a dichotomous
indicator for being overweight (BMI ? 25) and a dichotomous indicator of obesity (BMI ? 30). YSM
and YSM2 are the number of years since immigration (equal to 0 if Canadian born) and its square. X is
a vector of personal characteristics which include age, age2, income, sex, marital status, household
size, educational attainment, province of residence, work status and various controls for other health
related behaviours such as smoking and alcohol consumption. Cohort are dummy variables for the
entry cohorts of the immigrants and Cycle are a set of dummy variables for the survey and cycle the
observation came from. To attempt to account for the large heterogeneity that exists across immigrants
from different parts of the world, dummy variables for country of birth are created. Further I allow for
flexibility in the YSM profiles by country of birth by including interaction terms with Country and
YSM.
To examine whether more integration into one’s culture is protective of one’s weight, I will create
measures for such integration using information on the number of ethnic restaurants in the geographic
neighborhood of the individual, as well as the density of those of the individual’s language group
living in the same neighborhood. This measure will be similar to the network variables used in Deri
(2003) and Bertrand et al (2000) and combines both quantity and quality characteristics of networks. It
will be constructed as follows
? density of
? ? number of ethnic ?
?
? ?
?
Netw jk ? ? language group k ? x? restaurants of
? .
? in area j
? ? language group k ?
?
? jk ?
?k
More specifically, the first term, the measure of quantity, measures contact availability – the number of
people who live in close proximity that speak the same language as the respondent. This information
will come from the 1996 Census. The area groups I will consider are the 1996 Census Sub-Divisions
and Census Consolidated Subdivisions.
The second term, the quality measure, will capture the accessibility to “like home” food available
locally. Information on the number of ethnic restaurants will be obtained from the GIS data. Using the
4
Catherine Deri
sample of immigrants whose mother tongue is not English or French, and whose language group is
coded in the relevant survey, I will estimate a model similar to (1) above including the Network
variable.
To address the final research question, I will add additional controls for the number of local parks,
grocery stores, meat and fish markets, fruit and vegetable markets in the neighborhood to the
regressions.
It is my hope that this analysis will provide new insights and evidence of the HIE but go further than
the existing works by examining various neighborhood characteristics that can mitigate or aggravate
the decline in the health.
Data and Computer Requirements:
This project would require access to the 1994, 1996 and 1998 master files of the National Population
Health Survey, as well as 2001 and 2003 master files for the Canadian Community Health Survey. As
explained above, the master files are necessary as only they contain the geographic identifiers with
which I can link the Census and GIS data.
The statistical software required is Stata SE. Any reasonable computer hardware and operating system
will suffice for the analysis.
Proposed Period of Research
March 2005 – December 2008
References
Antecol, Heather and Kelly Bedard, 2003, “Unhealthy Assimilation: Do Immigrants Converge to
American Weights?,” unpublished manuscript.
Bertrand, M., Luttmer, E., and Mullainathan, S., 2000. Network Effects and Welfare Cultures. The
Quarterly Journal of Economics, 1019-1055.
Costa, Dora, 1996, “Health and Labor Force Participation of Older Men, 1990-1991,” The Journal of
Economic History,” 56(1), p62-89.
Deri, Catherine, 2004, “Essays in Immigrant Health and Health Service Utilization in Canada,” PhD
Dissertation, University of Toronto, Department of Economics.
Deri, Catherine, 2003, “Social Networks and Health Service Utilization,” Unpublished manuscript.
Grant, Mary L., 1999, “Immigration in Canada: Context, Assimilation and Labour Market Impacts,”
PhD Dissertation, University of Toronto, Department of Economics.
Hakim, Zafar, Anne Wolf and Louis P. Garrison, 2002, “Estimating the Effect of Changes in Body
Mass Index on Health State Preferences,” Pharmacoeconomics, 20(6), p393-404.
Kinnon, Dianne, 1999, “Canadian Research on Immigration and Health,” Health Canada, Cat. N o H21149/1999E.
McDonald, James Ted and Steven Kennedy, 2004, “Insights into the ‘healthy immigrant effect’: health
status and health service use of immigrants to Canada,” Social Science and Medicine, 59 (8),
p1613-1627.
Noh, Samuel, Mark Speechley, Violet Kaspar and Zheng Wu, 1992, “Depression in Korean
Immigrants in Canada: 1. Method of the study and prevalence of depression,” Journal of Nervous
and Mental Disease, 180, p573-577.
5
Catherine Deri Armstrong
Long Term Effects of Maternal Labour Force Participation:
Evidence from the LAD
Submitted for consideration to the IRND (New Directions) Grant Committee
Catherine Deri Armstrong
Department of Economics
April 18, 2006
This proposal seeks funding to begin a joint research initiative with Ross Finnie (Queen’s
University and Statistics Canada) and Jennifer Stewart (Carleton University) 1. The goal
of this project is to investigate the long-term effects of maternal labour force participation
on daughters’ future labour force participation, earnings, single parent status, social
assistance participation, low income status and participation in post secondary studies.
The majority of the existing work in this area investigates how current maternal labour
force participation affects current child outcomes, including scholastic, behavioural and
health outcomes. The literature thus focuses on children still living at home. The
Longitudinal Administrative Databank (LAD) began following a 20 percent sample of
Canadian tax filers in 1982. This data permits a unique opportunity to link information
regarding a mother’s labour force participation from the birth of child onward, to the
daughter’s own labour force participation, earnings, single parent status, social assistance
participation, low income status and participation in post secondary studies in adulthood.
This would be the first Canadian study to investigate the long-term effects of maternal
labour force participation.
The money requested in this grant would be used to access the LAD data and to begin to
investigate the key relationships described above. This information would be used to
develop a SSHRC grant application (Standard Research Grants).
This work would constitute a new direction in my research agenda. Much of my recent
work has focused on issues surrounding immigrant health, health care utilization and
assimilation, including a paper called “Social Network and Health Service Utilization”
published in 2005 in the Journal of Health Economics, and ongoing work funded in part
by an IRND Grant in December, 2004 entitled: “What Can We Learn from the Body
Mass of Immigrants?”
Funds Requested: $9749.98
Introduction
Canada has seen an increase in female labour force participation similar to the increases
found in other developed countries. In 2005, 77 percent of females aged 15 to 44
participated in the labour force. This represents a 39 percent increase from 1977 when
only 55 percent of females in this age group participated in the labour force (Statistics
Canada, 2005). Labour force participation by mothers has followed a similar trend. These
1
Please note that this will be an equal joint effort between the three researchers with respect to the
planning, the programming and the writing of the results. Ross Finnie is a Research Fellow and Adjunct
Professor at Queen’s University’s School of Policy Studies and an Analyst at Statistics Canada. Jennifer
Stewart is an Assistant Professor at Carleton’s School of Public Policy and Administration.
Catherine Deri Armstrong
numbers are influential in the current debate surrounding child-care policy in Canada, but
also of concern is the relationship between mother’s labour force participation and child
outcomes.
Research has been conducted in the US and the UK using older longitudinal data sets
than has been available in Canada. These studies generally focus on children at young
ages and are based on smaller sample sizes than the LAD, available to us. Many studies
use an American data set; the National Longitudinal Survey of Youth – Child Supplement
(Blau and Grossberg (1990); Ruhm (2000); Baum (2003); Anderson, Butcher, and Levine
(2003)). The outcomes studied vary from test scores to child weight status. These studies
consistently find a negative effect of maternal employment early in the child’s life (i.e.,
the first year), but do not find a consistent impact of subsequent maternal employment.
One paper that examines the effect of maternal employment on child outcomes at an
older age is Ermisch and Francesconi (2002). These authours use the British Household
Panel Survey to estimate the impact of parental employment on the child’s educational
attainment as a young adult. They find that maternal employment when the child was less
than 6 years old has a negative impact on educational attainment, as does paternal
employment although the effect is much smaller.
Research using Canadian data has recently been conducted with the release of the
National Longitudinal Survey of Children and Youth (NLSCY). Lefebvre and Merrigan
(2002) use data from the first cycle of the NLSCY to study the outcomes of children less
than 6 years old. They find that the number of weeks worked full-time by the mother has
a positive relationship with Motor and Social Development scores for children aged 1 to
3 and with the Peabody Picture Vocabulary Test (PPVT) scores for children aged 4.
Gagne (2002) examines the same questions as Lefebvre and Merrigan, but uses data from
all three cycles of the NLSCY. However, she restricts her sample to 3½ to 5 year olds and
only examines the PPVT score. Unlike Lefebvre and Merrigan, her results indicate that
maternal labour supply has little effect on PPVT scores, although the effect varies with
education and parental education. Roberts (2002) uses the NLSCY to examine the effect
of maternal employment on behavioural-emotional scores. He uses a sample of children
aged 6 to 11, an older sample than the previous studies. He finds that the number of
periods of maternal unemployment longer than 4 weeks is associated with worse scores.
This study is unique in its focus on current maternal unemployment. Most of the studies
mentioned above focus on younger children or the impact of maternal employment when
the child was young.
Basic theoretical models about the relationship between maternal employment and child
outcomes are based on two ideas; “Value Formation” and “Production Function.” The
first idea argues that a mother who works sets an example for her children and creates
specific values about working and attachment to the labour market. The second idea
argues that child outcomes are produced at home using a combination of maternal time
and market goods. Maternal employment will decrease the time available for producing
child outcomes, but will increase the amount of market goods that can be purchased. A
priori, the theoretical models do not make a consistent prediction about the impact of
maternal employment on child outcomes.
Research Questions
This project will allow us to address the following questions:
?
?
Catherine Deri Armstrong
How does maternal employment when a child is 0-16 years affect a daughter’s
future labour force participation, earnings, single parent status, social assistance
participation, low income status and participation in post secondary studies when
the daughters are adults, controlling for various demographic and neighborhood
characteristics?
Does the age of the child matter? For example, does maternal labour force
participation during the early years of life, or during the latter years of high school
have the larger effects?
The Longitudinal Administrative Databank (LAD)
The LAD is an ongoing longitudinal dataset that began in 1982. Constructed from the
Canada Revenue Agency’s tax files, it comprises a 20 percent sample of Canadian tax
filers selected using a random number generator. Individuals in the LAD are matched into
family units on an annual basis. This provides information on sources of income, taxes
and demographic information at both individual and family levels. The most recent year
of the LAD available is 2003. Because the rate of tax filing is very high in Canada
(unlike the United States for example), the LAD’s coverage of the adult population is
estimated to cover over 95 percent of the target population in any given year. (Finnie and
Bernard, 2005)
There are three characteristics of the LAD that make it well suited for this proposed
work. First, the LAD will allow us to identify the families of origin for young women
who filed taxes while still living at home (or who used their parents’ address for filing
purposes). Once the family of origin link has been made, for those whose mothers are
also in the LAD, we will be able to follow the mothers back in time and identify their
labour force participation at different stages of the daughters’ childhood. (A discussion of
these linkages is provided in the following section)
Second, the extended period covered by the LAD will allow us to not only look back at
daughters’ family characteristics in childhood (back to the age of zero in some cases), but
will also allow us to look forward at these women as adults to as late as 37 years of age.
It is important to note that the individual records will be unbalanced in this regard. For
example, women who are generally older over the period covered by the LAD are
followed for more years as adults but are not tracked as far back into their childhood,
while younger women are not followed as far into adulthood but are tracked back to
younger ages in childhood.
Finally, because the LAD contains a very large number of observations (nearly six
million in 2000), we will be able to construct very large samples of young women and
identify the parameters attached to the effects of their mothers’ labour force participation
on their own outcomes as adults. The geographic identifiers in the LAD will also allow us
to link Census Tract level neighbourhood characteristics from Canadian censuses to be
used as important control variables.
Methodology
From the sample of nearly six million tax filers in the LAD, we will restrict our sample of
“daughters” to the females aged 20-37 in the 1987-2003 years of the LAD who 1) filed a
tax return when living at home with their parents between the ages of 14 and 19 and 2)
whose mother also appears in the LAD2. The former link, linking the daughter to the
2
We restrict our sample to women aged 20-37 for the following reasons. The lower bound is necessary as
Catherine Deri Armstrong
family of origin, has been shown to be successful in 65 percent of the cases (Finnie and
Bernard, 2005)3. The latter link, to the mother, is made through the family identifier on
the file. Because the LAD is a true 20 percent random sample of Canadian tax filers and
because of the high rate of tax filing, this daughter-mother link has been shown to be
successfully made in almost exactly 20 percent of the cases (Finnie and Bernard, 2005).
For each observation we will construct six dependent/outcome variables: dichotomous
indicators for whether the daughter is currently working, is a single parent, has low
income status, has ever participated in post secondary studies, receives income from
social assistance and a measure of her current earnings. Participation in post secondary
education will be identified using reported tuition deductions.
For each observation, we will track the daughter’s mother back in time and create
indicators for whether the mother had any employment income (in excess of $1,000) as
declared in her taxes when the daughter was 0-1, 2-3, 4-5, 6-10, 11-14 and 15-16 years
old. This will allow us to estimate the impact of maternal labour forc