SEDAP (Social and Economic Dimensions of an Aging Population) is a multidisciplinary research program studying a wide range of aging-related issues and is funded by the Social Sciences and Humanities Research Council of Canada. SEDAP is centred at McMaster University and involves researchers from that institution as well as from the University of British Columbia, Université de Montréal, Queen's and the University of Toronto.
I. Statistics Canada Research Data Centre
The pilot Research Data Centre at McMaster University, first announced in the Winter 1999/2000 SEDAP Bulletin, is now in operation. The Centre was officially opened on December 19, 2000 by representatives from Statistics Canada, the Social Sciences and Humanities Research Council of Canada, and McMaster University.
Above (left to right): Byron Spencer (centre director), Cindy Cook (Statistics Canada research data analyst), Mike Sheridan (assistant chief statistician for Social, Institutions and Labour Statistics) Photo: Ron Scheffler
The creation of this data centre at McMaster means that researchers will no longer have to travel to Statistics Canada in Ottawa to use many Statistics Canada data files. For details on which survey files are currently available at the McMaster centre, which software is available for working with the files, and how to apply to become a user of the centre, please see the SEDAP website (then click on "Related Links", then on "Statistics Canada Research Data Centre at McMaster").
SEDAP director Byron Spencer (Economics, McMaster) is the director of the Research Data Centre. SEDAP member Michael Veall (Economics, McMaster) will serve as associate director.
II. SEDAP Conference
SEDAP hosted a conference at McMaster University on "Macroeconomic Aspects of an Aging Population" on December 8, 2000. The conference program as well as downloadable versions of some of the papers presented are available on our website.
III. SEDAP Research Papers
SEDAP Research Papers are available on the SEDAP website at no cost. Paper copies may be obtained for a nominal charge. Please contact Mrs. Gail Kalika, Department of Economics, KTH-426, McMaster University, Hamilton, Ont., Canada, L8S 4M4.
Brief descriptions follow of the most recently released papers in the SEDAP series.
SEDAP Research Paper No. 28:
The Life Cycle Model of Consumption and Saving
Martin Browning (Economics, University of Copenhagen) and Thomas F. Crossley (Economics, York University)
The life cycle model is widely used in economics and suggests that households "smooth" expenditures over time. Such smoothing may occur, for example, over a year, over the working life or across the stages of life. This model has come under increasing criticism in recent years and the authors attempt to evaluate in this paper whether such criticism is justified.
In addition to a detailed examination of the considerable empirical literature in this area, Browning and Crossley also make use of two household data sets in their investigation. The U.K. Family Expenditure Survey (FES) provides a time series (1968-95) of cross- section information on family expenditures, income and demographics. The FES is run continuously with about 7000 households each year keeping two-week diaries of their expenditures on all goods. The Canadian Out of Employment Panel (COEP) follows respondents for two years after a job loss and collects detailed information on household income, expenditures, assets and debts.
The authors acknowledge that the standard life cycle model has its failures. For example, the elderly do not dissave as much in retirement as the model would predict. Browning and Crossley conclude, however, that the model has had more successes than failures. Where researchers have found deviations from the model's predictions, these appear to involve small welfare costs. They note that in many of the areas in which the standard life cycle model fails, it is only very recently that good micro data, particularly panel data, have become available.
SEDAP Research Paper No. 29:
Population Change and the Requirements for Physicians: The Case of Ontario
Frank T. Denton (Economics, McMaster University), Amiram Gafni (Clinical Epidemiology and Biostatistics, McMaster University) and Byron G. Spencer (Economics, McMaster University)
Population aging is occurring in all regions of Canada, in the United States, and in many other countries. Although the implications of aging for the health care system, and in particular for the demand for physician services, have been much discussed, the authors of this paper feel there has not been careful examination of "how much" and "when". They attempt to answer such questions with data for Ontario.
Using unpublished OHIP data, the authors construct ratios of fee payments to population ("utilization rates") for males and females in 19 age groups, for 19 physician specialties. The data are for fiscal year 1995-96 and indicate that utilization rates generally tend to rise with age. However, with the exception of General Practice, utilization rates generally begin to fall in the very oldest age groups. That is, the "old old" make somewhat less use per capita of physician services than do the "younger old".
The authors make of number of population projections for Ontario, based on different assumptions regarding mortality rates, fertility rates and immigration levels. Assuming that the population changes but that the male/female age-utilization rates for 1995-96 remain constant, projections of the requirements for physician services are made for 5-year intervals to 2020. These projections should not be interpreted as predictions of what the actual requirements will be (as utilization rates, medical technology and so forth have been held constant), but rather what they would be if only population changes were taken into account. These physician services projections are also broken down to account separately for the effects of population growth versus population aging.
Among the findings of this paper are:
SEDAP Research Paper Nos. 30 and 31:
Nonparametric Identification of Latent Competing Risks
and Roy Duration Models
Simplified Estimation of Multivariate Duration Models with Unobserved Heterogeneity
Gordana Colby (SEDAP Post-doctoral Research Fellow, Economics, McMaster University) and Paul Rilstone (Economics, York University)
In these technical papers, the authors consider some advanced econometric techniques which may be applied to certain issues in the area of aging and retirement.
SEDAP Research Paper No. 32:
Structural Estimation of Psychiatric Hospital Stays
Gordana Colby (SEDAP Post-doctoral Research Fellow, Economics, McMaster University) and Paul Rilstone (Economics, York University)
Since objective measures of mental health are problematic, a popular measure of outcomes is the patient's length of stay (LOS) in hospital. However, research in this area has not considered that, in addition to the caregiver, the patient directly or indirectly influences the LOS decision. Thus this paper develops a model of length of stay which is consistent with joint decision making and applies new econometric techniques to estimate the parameters of the model.
The main source of data in this study is the Northwestern Ontario Data Linkage System, a large Canadian data set on psychiatric hospital admissions, which covers the period from October 1971 to February 1997. Variables in the data set include age, gender, ethnic status, employment status on admission, marital status as well as patient diagnosis data. To this is added data on local unemployment rates, seasonal indicators, hospital occupancy rates and number of beds.
Among Colby and Rilstone's finding are:
SEDAP Research Paper No. 33:
Have 401(k)s Raised Household Saving? Evidence from the Health and Retirement Study
Gary V. Engelhardt (Dept. of Economics and Center for Policy Research, Syracuse University)
The most popular tax subsidy to household saving in the United States is the 401(k) plan. 401(k)s subsidize saving through income tax deferral on contributions and on investment earnings. They differ from Canadian RRSPs in that only employees at firms with a plan are eligible to participate. There is much debate in the literature on household saving as to whether 401(k) contributions represent new saving or are offset by declines in other saving, and this paper examines in some detail some of the prominent studies in this area. The author also conducts a study of his own and compares his results to previous findings in this area.
To investigate whether 401(k)s represent new saving, the saving behaviour of those eligible to participate in 401(k) plans is compared with that of those not eligible. However, Engelhardt finds that in the data set he uses here, the 1992 Health and Retirement Study, there is significant measurement error regarding 401(k) eligibility when comparing self-reported and firm-reported responses. He finds that lower-to-middle income households understate 401(k) eligibility and that high-income households overstate it. This means that any analysis using data on self-reported eligibility (which is all of the previous literature) would yield estimated saving effects that are biased upward significantly. The author also finds significant discrepancies in reported pension assets.
Using firm-reported 401(k) eligibility and pension assets instead of self-reported data, Engelhardt finds that 401(k)s generate large and sometimes statistically significant household saving effects for lower-to-middle income households. However, these effects diminish with income and are not statistically different from zero for middle-to- upper income households. Overall, it is estimated that 38 cents of the average dollar of 401(k) wealth represents new household saving. Using integrated firm-reported and self-reported data on pension assets, a similar pattern emerges of large saving effects that decline with income, with the overall impact on household saving estimated to be minus 8 cents and not statistically different from zero.
SEDAP Research Paper No. 34:
Health and Residential Mobility in Later Life: A New Analytical Technique to Address an Old Problem
Lynda M. Hayward (SSHRCC Post-doctoral Fellow, Centre for Gerontological Studies, McMaster University)
In this paper, the author looks at the relationship between health (in terms of functional mobility) and residential mobility. The literature in this area shows that this relationship may not be a straightforward one. While good health can facilitate amenity or life- style migration upon retirement, poor health is thought to be the main motivation for support-seeking moves in later life.
Hayward finds that logistic regression models, as are used in much of the literature in this area, give contradictory or ambiguous results, particularly when used with cross-sectional data. She proposes instead a proportional hazards model and uses a longitudinal data set.
The data set used in this paper, the Ontario Longitudinal Study of Aging, began in 1959 with a stratified quota sample of 2000 employed 45-year-old men. The survey was conducted every year (except in 1977) until 1978, at which time participants were 64 years old. In 1990, a follow-up survey was conducted of the 545 remaining respondents (now 76 years old), 49 proxies and 276 survivors.
The author runs both logistic regression and proportional hazards models, using a sample of 594 in the logistic regression case and 1063 (those who were still in the study upon retirement or age 65) in the proportional hazards case. The dependent variable in the logistic regressions is whether or not a move was made after retirement or upon turning age 65 (whichever came first); in the proportional hazards model, it was the number of years to the first move after retirement/age 65. In both models, the independent variables are dummy variables on health status which divide respondents into categories of good health, later poor health and early poor health, and dummy variables on proportion of years of declining health (categorized as low, moderate (17-22% of mid-life with declining health) and high (over 22%)).
Hayward's logistic regression results show that those with an early onset of poor health were more likely to move after retirement/age 65 than were those in good health. She finds no statistically significant relationship between the proportion of years with declining health and the likelihood of moving after retirement/age 65. The proportional hazards results also show that those with an early onset of poor health are more likely to move in later life. However, they show that those with a relatively high proportion of years with declining health are least likely to move shortly after retirement.
SEDAP Research Paper No. 35:
2 1/2 Proposals to Save Social Security
Deborah Fretz (Economics, McMaster University) and Michael R. Veall (Economics, McMaster University)
In the United States, the trustees of the major public pension plan (popularly known as Social Security) have been forecasting for about 15 years that the plan as structured could not meet its obligations through a 75-year window. Each of the books reviewed deals in some way with this problem.
Schieber and Shoven give a substantial history of the Social Security system, discuss a number of recent reform proposals, and give a proposal of their own. They propose requiring current workers to contribute an additional 2.5% of payroll (i.e., above current Social Security contributions) to private investment vehicles that they could not draw upon until retirement. They also propose moving to a retirement age of 67 at a faster rate than currently legislated, with the retirement age thereafter indexed to life expectancy.
Graetz and Mashaw undertake a comprehensive discussion of all U.S. social insurance programs, including Social Security, unemployment insurance and health care. Their proposals for Social Security reform include supplementary, mandatory personal savings accounts (which would also be drawn upon in times of unemployment). They also propose a lower level of pension inflation adjustment than is currently the case under the Consumer Price Index, and finding higher private sector rates of return for the Social Security trust fund.
Baker and Weisbrot argue that the foreseen shortfall in Social Security funding may be overstated, and thus there may not be a problem. (This accounts for the "«" a proposal to save Social Security in the title of this working paper: no pejorative connotation is intended.) They argue that economic growth assumptions used in forecasting the shortfall may be too pessimistic but that in any case, the U.S. will be a richer country in the future, making any required increase in payroll taxes easier to bear. They argue against any cut in current Social Security benefits, including any changes in adjusting for inflation or extending the normal retirement age.
Among Fretz and Veall's conclusions from these three volumes is that the recommendations of Schieber-Shoven and Graetz-Mashaw may simply reflect different values than those of Baker-Weisbrot. The first two proposals would tend to benefit future generations at the expense of those who will retire over the next thirty years, while Baker-Weisbrot would prevent these changes and thus preserve the position of the current generation of workers and retirees. As well, differing views of the role of government are apparent in proposals to increase funding through the use of private accounts rather than through existing Social Security arrangements.