perspectives on the economics of aging jul 2004

In addition, in recent years the re-annual retirement plan contribution rate, defined as retirement plan contri-butions as a percentage of National Income and Product Accounts NIPAperso

Perspectives on the Economics of Aging

A National Bureau

of Economic Research

Conference Report

National Bureau of Economic Research

Korea Development Institute

Chung-Hua Institution for Economic Research Tokyo Center for Economic Research

Hong Kong University of Science and Technology Productivity Commission, Australia

Perspectives on the Economics of Aging

The University of Chicago Press

Chicago and London

D  A W is the John F Stambaugh Professor of Political omy at the John F Kennedy School of Government, Harvard Univer- sity, and director of the NBER Program on Aging Among the many

Econ-titles he has edited in this area are the recent Themes in the Economics

of Aging and Advances in the Economics of Aging.

The University of Chicago Press, Chicago 60637

The University of Chicago Press, Ltd., London

© 2004 by the National Bureau of Economic Research

All rights reserved Published 2004

Printed in the United States of America

13 12 11 10 09 08 07 06 05 04 1 2 3 4 5

ISBN: 0-226-90305-2 (cloth)

Chapter 11 is reprinted from Journal of Econometrics, vol 112, Adams et

al., “Healthy, Wealthy, and Wise? Tests for Direct Causal Paths between Health and Socioeconomic Status,” pp 3–56, 2003, with permission from Elsevier.

Library of Congress Cataloging-in-Publication Data

Perspectives on the economics of aging / edited by David A Wise.

p cm.—(A National Bureau of Economic Research Conference report)

Includes bibliographical references and index.

ISBN 0-226-38680-5 (cloth : alk paper)

1 Aged—United States—Economic conditions 2 Retirement— Economic aspects—United States 3 Retirement income—United States 4 401(k) plans—United States 5 Individual retirement accounts—United States I Wise, David A II Series

National Bureau of Economic Research

Michael H Moskow, chairman Robert Mednick, treasurer

Elizabeth E Bailey, vice-chairman Kelly Horak, controller and assistant Martin Feldstein, president and chief corporate secretary

Susan Colligan, vice president for secretary

administration and budget and corporate

secretary

Directors at Large

Peter C Aldrich Jacob A Frenkel Michael H Moskow Elizabeth E Bailey Judith M Gueron Alicia H Munnell John H Biggs Robert S Hamada Rudolph A Oswald Andrew Brimmer George Hatsopoulos Robert T Parry

John S Clarkeson Karen N Horn Richard N Rosett Don R Conlan Judy C Lewent Marina v N Whitman

Martin Feldstein Laurence H Meyer

Directors by University Appointment

George Akerlof, California, Berkeley Joel Mokyr, Northwestern

Jagdish Bhagwati, Columbia Andrew Postlewaite, Pennsylvania Michael J Brennan, California, Los Angeles Uwe E Reinhardt, Princeton

Glen G Cain, Wisconsin Nathan Rosenberg, Stanford

Franklin Fisher, Massachusetts Institute Craig Swan, Minnesota

Saul H Hymans, Michigan Arnold Zellner, Chicago

Marjorie B McElroy, Duke

Directors by Appointment of Other Organizations

Mark Drabenstott, American Agricultural Richard D Rippe, National Association for

Gail D Fosler, The Conference Board John J Siegfried, American Economic

A Ronald Gallant, American Statistical Association

Richard C Green, American Finance Labor and Congress of Industrial

Robert Mednick, American Institute of Josh S Weston, Committee for Economic

Angelo Melino, Canadian Economics Gavin Wright, Economic History

Directors Emeriti

Carl F Christ Paul W McCracken Eli Shapiro

Lawrence R Klein Peter G Peterson

Franklin A Lindsay Bert Seidman

Since this volume is a record of conference proceedings, it has been exempted from the rules governing critical review of manuscripts by the Board of Directors of the National Bureau (resolution adopted 8 June 1948, as revised 21 November 1949 and 20 April 1968).

Relation of the Directors to the

Work and Publications of the

National Bureau of Economic Research

1 The object of the NBER is to ascertain and present to the economics profession, and to the public more generally, important economic facts and their interpretation in a scientific manner without policy recommendations The Board of Directors is charged with the respon- sibility of ensuring that the work of the NBER is carried on in strict conformity with this ob- ject.

2 The President shall establish an internal review process to ensure that book manuscripts proposed for publication DO NOT contain policy recommendations This shall apply both to the proceedings of conferences and to manuscripts by a single author or by one or more co- authors but shall not apply to authors of comments at NBER conferences who are not NBER

a ffiliates.

3 No book manuscript reporting research shall be published by the NBER until the dent has sent to each member of the Board a notice that a manuscript is recommended for pub- lication and that in the President’s opinion it is suitable for publication in accordance with the above principles of the NBER Such notification will include a table of contents and an ab- stract or summary of the manuscript’s content, a list of contributors if applicable, and a re- sponse form for use by Directors who desire a copy of the manuscript for review Each manu- script shall contain a summary drawing attention to the nature and treatment of the problem studied and the main conclusions reached.

Presi-4 No volume shall be published until forty-five days have elapsed from the above tion of intention to publish it During this period a copy shall be sent to any Director request- ing it, and if any Director objects to publication on the grounds that the manuscript contains policy recommendations, the objection will be presented to the author(s) or editor(s) In case

notifica-of dispute, all members notifica-of the Board shall be notified, and the President shall appoint an ad hoc committee of the Board to decide the matter; thirty days additional shall be granted for this purpose.

5 The President shall present annually to the Board a report describing the internal script review process, any objections made by Directors before publication or by anyone after publication, any disputes about such matters, and how they were handled

manu-6 Publications of the NBER issued for informational purposes concerning the work of the Bureau, or issued to inform the public of the activities at the Bureau, including but not limited

to the NBER Digest and Reporter, shall be consistent with the object stated in paragraph 1 They shall contain a specific disclaimer noting that they have not passed through the review procedures required in this resolution The Executive Committee of the Board is charged with the review of all such publications from time to time.

7 NBER working papers and manuscripts distributed on the Bureau’s web site are not deemed to be publications for the purpose of this resolution, but they shall be consistent with the object stated in paragraph 1 Working papers shall contain a specific disclaimer noting that they have not passed through the review procedures required in this resolution The NBER’s web site shall contain a similar disclaimer The President shall establish an internal review pro- cess to ensure that the working papers and the web site do not contain policy recommenda- tions, and shall report annually to the Board on this process and any concerns raised in con- nection with it.

8 Unless otherwise determined by the Board or exempted by the terms of paragraphs 6 and

7, a copy of this resolution shall be printed in each NBER publication as described in graph 2 above

1 The Transition to Personal Accounts and Increasing

Retirement Wealth: Macro- and Microevidence 17James M Poterba, Steven F Venti, and

David A Wise

Comment: Sylvester J Schieber

2 For Better or for Worse: Default E ffects and

James J Choi, David Laibson, Brigitte C Madrian, and Andrew Metrick

Comment: James M Poterba

Steven F Venti and David A Wise

Comment: Jonathan Skinner

Jeffrey R Brown and Scott J Weisbenner

Comment: Alan J Auerbach

James Banks, Richard Blundell, and James P Smith

Comment: John B Shoven

6 Mortality, Income, and Income Inequality over

Angus Deaton and Christina Paxson

Comment: James Banks

Anne Case

Comment: Robert T Jensen

8 Socioeconomic Status, Nutrition, and Health

Robert T Jensen

Comment: David M Cutler

9 Changes in the Age Distribution of Mortality

David M Cutler and Ellen Meara

10 Area Di fferences in Utilization of Medical Care

Victor R Fuchs, Mark McClellan, and Jonathan Skinner

Comment: Joseph P Newhouse

11 Healthy, Wealthy, and Wise? Tests for

Direct Causal Paths between Health and

Peter Adams, Michael D Hurd, Daniel McFadden, Angela Merrill, and Tiago Ribeiro

Comment: James M Poterba

viii Contents

ix

This volume consists of papers presented at a conference held at Carefree,Arizona in May 2001 Most of the research was conducted as part of theProgram on the Economics of Aging at the National Bureau of EconomicResearch The majority of the work was sponsored by the U.S Department

of Health and Human Services, through National Institute on Aging grantsP01-AG05842 and P30-AG12810 to the National Bureau of Economic Re-search Any other funding sources are noted in individual papers

Any opinions expressed in this volume are those of the respective authorsand do not necessarily reflect the views of the National Bureau of EconomicResearch or the sponsoring organizations

David A Wise

1

This is the ninth in a series of volumes on the economics of aging The

pre-vious ones were The Economics of Aging, Issues in the Economics of Aging, Topics in the Economics of Aging, Studies in the Economics of Aging, Ad- vances in the Economics of Aging, Inquiries in the Economics of Aging, Fron- tiers in the Economics of Aging, and Themes in the Economics of Aging.

Most of the papers in this volume pursue areas of research begun in lier volumes For example, the work in this volume emphasizes the spread

ear-of personal retirement accounts and macrodata on the implications ear-of the

diffusion of these accounts Prior work emphasized the net saving effects

of personal retirement accounts based on microdata Work in this volumealso revisits the implications of housing wealth for the financing of generalconsumption as households age The current work confirms previous find-ings that households typically do not withdraw equity from housing to fi-nance general consumption, but such withdrawal is more likely when aspouse dies or enters a nursing home In addition to discussion of the per-sonal retirement plans and home equity, the first five papers in this volumeconsider other aspects of wealth accumulation and compare asset accu-mulation in the United States and the United Kingdom

Trying to understand the explanation for the very strong relationship tween health and wealth is one of the most challenging research issues inthe economics of aging Perhaps the most vexing issue that arises in thisanalysis is the direction of causality Is it from health to wealth or fromwealth to health, or perhaps both? There have been several papers on this

be-David A Wise is the John F Stambaugh Professor of Political Economy at the John F Kennedy School of Government, Harvard University, and the director for Health and Re- tirement Programs at the National Bureau of Economic Research.

issue in past volumes The next six papers in this volume continue analysis

of several aspects of the relationship between health and wealth, with thegeneral aim of advancing our understanding of the reasons for the rela-tionship The analyses include consideration of the decline in mortality inthe United States and the United Kingdom and the relationship betweenmedical care and mortality The relationship between pension income andhealth in South Africa also contributes to our understanding of thehealth–wealth relationship This volume also includes more formal discus-sion of econometric methodology to determine the direction of causality.This introduction provides a summary of the papers and draws heavily onthe authors’ own language

Personal Retirement Plans

Implications of the Transition to Personal Accounts

Retirement saving in the United States has changed dramatically overthe last two decades There has been a shift from employer-managed de-fined benefit pensions to defined contribution retirement saving plans thatare largely controlled by employees In 1980, 92 percent of private retire-ment saving contributions were to employer-based plans, and 64 percent ofthese contributions were to defined benefit plans Today, about 85 percent

of private contributions are to plans in which individuals decide how much

to contribute to the plan, how to invest plan assets, and how and when towithdraw money from the plan In “The Transition to Personal Accountsand Increasing Retirement Wealth: Macro- and Microevidence,” James M.Poterba, Steven F Venti, and I use both macro- and microdata to describethe change in retirement assets and in retirement saving We give particu-lar attention to the possible substitution of pension assets in one plan forassets in another plan, such as the substitution of 401(k) assets for definedbenefit plan assets

Aggregate data show that between 1975 and 1999 assets to support tirement increased about fivefold relative to wage and salary income Thisincrease suggests large increases in the wealth of future retirees The enor-mous increase in defined contribution plan assets dwarfed any potentialdisplacement of defined benefit plan assets In addition, in recent years the

re-annual retirement plan contribution rate, defined as retirement plan

contri-butions as a percentage of National Income and Product Accounts (NIPA)personal income, has been over 5 percent This is much higher than theNIPA total personal saving rate, which has been close to zero

Retirement saving as a share of personal income today would likely be atleast one percentage point greater had it not been for legislation in the1980s that limited employer contributions to defined benefit pension plansand the reduction in defined benefit plan contributions associated with the

2 David A Wise

rising stock market of the 1990s It is also likely that the retirement plancontribution rate would be much higher today if it were not for the 1986 re-trenchment of the Individual Retirement Account (IRA) program.Rising retirement plan contributions, as well as favorable rates of return

on retirement plan assets in the 1990s, explain the large increase in these sets relative to income Employee retirement saving under a defined con-tribution plan is easily measured and quite transparent to the employee

as-On the other hand, annual employee saving under a defined benefit plan ismore difficult to measure It is also less likely to be clearly understood byemployees The average annual saving rate under a typical 401(k) plan isroughly twice as high as the average saving rate under a typical defined ben-efit plan, when properly measured In addition, the early retirement incen-tives inherent in the provisions of most defined benefit plans will tend toreduce the aggregate accumulation of defined benefit retirement assetsrelative to defined contribution assets because defined contribution partic-ipants are likely to work longer and contribute for more years

The microdata show no evidence that the accumulation of 401(k) assetshas been offset by a reduction in defined benefit assets Because annual sav-ing is much greater under 401(k) than under defined benefit plans, assets atretirement after lifetime employment under a 401(k) plan typically would

be much higher than under a defined benefit plan In addition, a large tion of new 401(k) enrollees retained defined benefit coverage, which prob-ably further increased their retirement saving

frac-The Importance of Plan Features

In the last several years, dozens of employers have automatically rolled new employees in the company 401(k) plan Employees can opt out

en-of the 401(k), but few choose to do so In “For Better or for Worse: Default

Effects and 401(k) Savings Behavior,” James J Choi, David Laibson,Brigitte C Madrian and Andrew Metrick analyze three years of 401(k)data from two firms that have experimented with automatic enrollment.They find that automatic enrollment has a dramatic impact on retirementsavings behavior

Under automatic enrollment (also called negative election), employees

are automatically enrolled in their company’s 401(k) plan unless the ployee elects to opt out of plan participation This contrasts with the usualarrangement in which employees actively elect to participate in their em-ployer’s 401(k) plan

em-The institution of automatic enrollment manipulates the way the savingsdecision is framed In theory, the existence of automatic enrollment shouldnot influence the employee’s decision; automatic enrollment doesn’tchange the economic fundamentals of the planning problem But auto-matic enrollment nonetheless increases participation in 401(k) plans Opt-ing out requires workers to actively make a decision and then act on that

decision It is easier to follow the path of least resistance and passively cept the default.

ac-Choi, Laibson, Madrian, and Metrick find that automatic enrollmenthas a dramatic impact on participation rates Under automatic enrollment,401(k) participation rates exceed 85 percent regardless of the tenure of theemployee In the absence of automatic enrollment, employees exhibit par-ticipation rates of only 25 percent after six months of tenure and 60 per-cent after three years at the firm

They also find that automatic enrollment has a dramatic impact on ings rates and asset allocation choices Under automatic enrollment, ap-proximately 80 percent of plan participants save at the default saving rateand invest exclusively in the default fund This percentage declines slowlyover time, falling to 50 percent after two or three years of tenure

sav-Automatic enrollment encourages participation, but it anchors pants at a low savings rate and in a conservative investment vehicle Higherparticipation raises average wealth accumulation, but low savings ratesand conservative investments undercut accumulation In the sample, thetwo effects are roughly offsetting Controlling for income and tenure, theycompare total 401(k) balances for employees who joined the firm beforeautomatic enrollment and employees who joined the firm after automaticenrollment They find that automatic enrollment has a positive impact onlong-run wealth accumulation in one of the firms and no impact on long-run wealth accumulation in the other firm

partici-Although automatic enrollment does not have a dramatic impact on themean level of balances, it does have a large impact on the distribution ofbalances The high participation rate associated with automatic enroll-ment drastically reduces the fraction of employees with zero balances,thereby thinning out the bottom tail of the distribution of employee bal-ances In addition, anchoring on low savings rates and conservative invest-ments sometimes shrinks the upper tail of the distribution of balances.Hence, automatic enrollment reduces the variance of wealth accumulationacross all employees

The Financial Implications of Housing Equity as People Age

Home equity is the principle asset of a large fraction of elderly cans and can have important implications for the well-being of elderlyhouseholds In “Aging and Housing Equity: Another Look,” Steven F.Venti and I have used Health and Retirement Study (HRS) and Asset andHealth Dynamics Among the Oldest Old (AHEAD) panel data, as well asSurvey of Income and Program Participation (SIPP) data, to understandthe change in the home equity of households as they age We give particu-lar attention to the relationship between changes in home equity andchanges in household structure

Ameri-There are two ways for households to change home equity: by

discon-4 David A Wise

tinuing home ownership or by selling and moving to another home Wefind that, overall, households are unlikely to discontinue home ownership.Ownership terminations are most likely to occur following the death of aspouse or the entry of a family member into a nursing home But, even inthese circumstances, selling the home is the exception and not the rule Inthe absence of a precipitating shock, it is much more likely that a family willsell and buy a new home than discontinue ownership And households thatsell and buy again tend to increase rather than reduce home equity That is,assets are transferred to housing.

Overall—combining the effects of discontinuing ownership and moving

to another home—we find that housing equity of HRS households creases with age, and the equity of AHEAD households declines some-what The overall decline in the housing equity of the older AHEADhouseholds is about 1.76 percent per year, which is accounted for prima-rily by a 7.84 percent decline among households experiencing precipitat-ing shocks to family status Families that remain intact reduce housing eq-uity very little, only 0.11 percent per year for two-person households and1.15 percent per year for one-person households

in-We use two approaches to determine whether households wish to reducehome equity as they age One approach is to compare the change in thehome equity of movers with the change for stayers If households withdrawequity when they sell and move to a new home, the reduction in the equity

of the movers typically will be greater than the change for stayers Thesecomparisons, however, are confounded by the tendency of the self-assessedhome values to exceed actual values, as measured by selling prices A com-parison of the selling prices of homes with the prior self-assessment ofhome values shows that home values reported prior to a sale far exceed re-alized sales prices Comparing the change in the home equity of moversand stayers, but accounting for this bias, we conclude that families who selland buy a new home increase home equity on average

The second approach is based on the comparison of the selling price ofthe old home (minus the mortgage on the home) with the reported equityvalue in the newly purchased home We believe that these are the most re-liable data on the change in home equity when families move from one

home to another Based on these sale price data, we find that on average

households increase home equity when they move to a new house We alsofind, however, that equity-rich and income-poor families tend to reducehome values when they sell and buy a new house, whereas equity-poor andincome-rich families tend to increase home equity For continuing two-person HRS households, for example, we estimate that the between-wavereduction for those at the 80th equity quantile and at the 20th incomequantile is –$15,422 On the other hand, we estimate that households at the 20th equity quantile and the 80th income quantile increase equity by

$54,778

These results suggest that in considering whether families have savedenough to maintain their preretirement standard of living after retirement,housing equity should not, in general, be counted on to support nonhous-ing consumption Families apparently do not intend to finance general re-tirement consumption by saving through investment in housing, as theymight through a 401(k) plan or through some other financial form of sav-ing Rather, we believe the findings here, as well as our earlier findings, sug-gest that families purchase homes to provide an environment in which tolive, even as they age through retirement years In this case, the typical ag-ing household is unlikely to seek a reverse annuity mortgage to withdrawassets from home equity It may be appropriate, however, to think of hous-ing as a reserve or buffer that can be used in catastrophic circumstancesthat result in a change in household structure In this case, having used thehome equity along the way—through a reverse mortgage, for example—would defeat the purpose of saving home equity for a rainy day.

Although these results are based largely on new HRS and AHEAD datafiles and are based on different methods of analysis, the findings corre-spond closely to the conclusions we reached in our earlier papers, based on

different data sources These conclusions also correspond closely to thefindings of a recent survey of older households sponsored by the AmericanAssociation of Retired Persons (AARP), showing that the preponderance

of older families agree with the statement: What I’d really like to do is stay

in my current residence as long as possible Like our findings, the results ofthe AARP survey also imply that most households do not intend to liqui-date housing equity to support general nonhousing retirement consump-tion as they age

How Is Wealth Accumulated?

Economists generally agree that there are two possible sources of hold wealth: households can engage in life-cycle saving by not consumingall of their income, or they can receive bequests or inter vivos transfersfrom individuals outside of their household For at least two decades, how-ever, there has been an ongoing debate about the relative magnitude ofthese two sources of wealth

house-In “house-Intergenerational Transfers and Savings Behavior,” Jeffrey R Brownand Scott J Weisbenner present further evidence of the importance ofthese two routes to wealth accumulation

The source of household wealth is important for many reasons The havioral effects of many government programs, such as Social Security, thetaxation of savings, and targeted savings programs, likely depend uponthe source of wealth Debates about the fairness of the wealth distribution

be-in the United States, and the extent to which there is be-intergenerationalmobility across this distribution, depend on whether wealth is primarilyearned or inherited

6 David A Wise

Brown and Weisbenner reach three conclusions First, using the 1998Survey of Consumer Finances, they provide evidence suggesting that trans-fer wealth accounts for approximately 20–25 percent of current householdnet worth, suggesting a large role for life-cycle savings.

Second, the authors examine the variation in the size of transfers ceived and expected They find that while in aggregate, transfer wealthdoes not appear to be as large as some prior estimates suggest; it is impor-tant for a small subset of the population They find that approximately one-fifth of households report receiving a transfer, and one-eighth expect asubstantial transfer in the future For those households that have receivedtransfers, transfer wealth accounts for, on average, half of current networth For lower-wealth households (those with less than $75,000), trans-fer wealth on average exceeds current wealth

re-Third, Brown and Weisbenner examine whether past transfers and pected future transfers cause people to save less from their labor income toreduce life-cycle saving They find evidence that the receipt of transfers re-duces life-cycle saving, with point estimates suggesting slightly less thandollar-for-dollar crowd-out But expected future transfers do not reducelife-cycle saving, perhaps suggesting that a bird in hand is indeed worthmore than a bird in the bush

ex-Wealth at Retirement in the United States and the United KingdomThe accumulation of personal wealth differs substantially across coun-tries So does the distribution of wealth among assets In “Wealth Portfo-lios in the United Kingdom and the United States,” James Banks, RichardBlundell, and James P Smith discuss a “housing equity puzzle”: why doyounger households in the U.K accumulate so much of their wealth inhousing equity compared to their U.S counterparts?

In trying to address this puzzle the authors have built up a detailed ture of housing choices and wealth accumulation in both countries Usingmicrodata sources, they document how the difference in housing equity hasevolved for different age groups, for different demographic groups, and for

pic-different education groups in both countries They show that young adults

in the United Kingdom leave their parental home later than in the UnitedStates, and when they do leave they are much more likely to accumulatewealth in housing equity rather than in other investment instruments.Why? Is it just the differential tax treatment of mortgages or the differ-ent institutional structures of the housing and stock markets in the twocountries? The authors argue that these differences explain some of the

difference in housing equity, but the higher volatility of house prices in theUnited Kingdom is the key reason They derive a modeling framework thatexplains the higher price volatility in the United Kingdom and use themodel to explain the differences in housing equity and stock holdingsacross the countries

The inefficient rental market, the authors say, places many more U.K.households in the owner-occupier sector at an earlier age than in theUnited States The higher volatility of house prices in the United Kingdomadds to this incentive because, for those expecting to move up the house-size ladder, housing equity is an efficient insurance vehicle for house-priceuncertainty The only way to invest in housing equity is to become anowner Once an owner, this insurance mechanism increases the incentive tohold a higher proportion of wealth in housing equity rather than in someother asset Where house prices are less volatile, as in the United States, thisincentive is much reduced Consequently, as households age and wish toaccumulate wealth, they will do this more through housing equity in theUnited Kingdom than in the United States.

Health and Wealth

A striking empirical regularity is the strong, positive relationship tween wealth and health Several papers discuss different aspects of this re-lationship and use different methods to understand the relationship Thelast chapter in the volume presents formal methods to test for causal linksbetween socioeconomic status and health

be-Mortality, Income, and Income Inequality

In both the United States and Britain, for both men and women and formost age groups, there has been a very substantial decline in mortalityrates since 1950 In “Mortality, Income, and Income Inequality over Time

in Britain and the United States,” Angus Deaton and Christina Paxson usecohort data from the two countries to understand the relationship betweenincome and the decline in mortality

The comparison between the two countries is interesting in part because

of the different systems of health care—one country with universal age and the other with private provision until Medicare coverage at agesixty-five Comparative analysis is also useful because there are both simi-larities and differences in patterns of income in the two countries Al-though changes in income inequality are similar in Britain and the UnitedStates, patterns of income growth are not According to purchasing powerparity estimates, incomes are higher in the United States than in Britain,but, in recent years, real incomes have been growing more rapidly inBritain Both countries experienced historically large increases in incomeinequality in the 1980s

cover-In both Britain and the United States, for men and for women and formost age groups, there has been a very substantial decline in mortalityrates since 1950 The authors’ examination of these rates, by sex and agegroup and in relation to the evolution of incomes and income inequality,

8 David A Wise

does not suggest any simple relationship between income growth and thedecline in mortality, or between income inequality and mortality rates Inthe United States, the period of slowest income growth saw substantial ac-celerations in the rate of mortality decline, particularly among middle-aged and older men and women In both the United States and Britain, theincrease in income inequality took place at the same time as a deceleration

in mortality decline at the younger ages, including infant mortality Butthere are previously slow rates of decline when nothing was happening toincome inequality, and the later rise in income inequality was associatedwith the acceleration in mortality decline among middle-aged and olderadults in both countries Deaton and Paxson conclude that a more plaus-ible account of the data is that, over time, declines in mortality are driven

by technological advances or by the emergence of new infectious diseases,such as AIDS These advances and retreats are associated with specificconditions and specific treatments, and so affect men and women differ-ently and different age groups differently They also happen first in theUnited States, with the British experience following with a lag of severalyears The authors say that this hypothesis needs a great deal more investi-gation, for example, by looking at more countries

Deaton and Paxson then compare these results with their prior analysis,suggesting that if changes in mortality over time are driven by technologyand not by income, there must be some doubt as to whether their previousanalysis came to the correct conclusions about the role of cohort incomes

in the decline of cohort mortality Their prior results cannot be replicated

on the British data They suspect—but have not been able to demonstratedecisively—that the cohort analysis is flawed by the necessity to make thealmost certainly invalid assumption that age effects in mortality are con-stant through time This is contradicted, for example, by the spread ofAIDS, which has almost certainly raised the early life relative to later lifemortality rates among recently born men and women compared with theirseniors If this is a serious problem, the cohort method may not be useful

in this context, or it will at least require substantial modifications in order

to give sound results

The authors conclude that this comparative international work is a ductive direction for future research Even so, they say, there remains a ma-jor puzzle about the role of income Income growth seems to play little role

pro-in declpro-ine of mortality at the national level At the cohort level the same ispossibly true as they argue in this paper Yet in the individual-level datafrom the National Longitudinal Mortality Study, as from many other datasets, income is protective against mortality, even when education and othersocioeconomic variables are controlled Why there should be such a con-trast between the individual and national effects of income is a topic thatrequires a good deal of further thought and analysis, the authors conclude

Money and Health in South Africa

In “Does Money Protect Health Status? Evidence from South AfricanPensions,” Anne Case approaches the relationship between health andwealth by considering the effect on health status of an exogenous increase

in income

Case quantifies the impact on health status of the large increase in come associated with the South African state old age pension Elderlyblack and colored men and women who did not anticipate receiving largepensions in their lifetimes, and who did pay into a pension system, are cur-rently receiving more than twice median black income per capita Theseelderly men and women generally live in large (three, four, or five genera-tion) households, and this paper documents the effect of the pension on thepensioners, on other adult members of their households, and on the chil-dren who live with them She finds, in households that pool income, thatthe pension protects the health of all household members, working in part

in-to protect the nutritional status of household members, in part of improveliving conditions, and in part to reduce the stress under which the adulthousehold members negotiate day-to-day life The health effects of deliv-ering cash provide a benchmark against which other health-related inter-ventions can be evaluated, Case concludes

Socioeconomic Status, Nutrition, and Health

Robert Jensen explores the relationship between “Socioeconomic tus, Nutrition, and Health among the Elderly.”

Sta-In his paper, Jensen uses data from a nationally representative hold-level survey to explore the relationship between health and socioeco-nomic status (SES) for the elderly in Russia

house-Jensen has two main objectives: first, to explore the basic relationship,which is valuable because there has been little evidence on the health-SESrelationship for transition economies; and second, to present evidencefrom a variety of measures of health status, including measurements ofblood pressure, weight and height conducted by trained enumerators, aswell as nutrient intake, derived from twenty-four- and forty-eight-hourfood intake diaries Jensen uses these data to show that the relationship be-tween health and SES in Russia can’t be adequately described by simplestatements, such as the poor are less healthy than the rich; although on netthe rich are healthier than the poor in some overall sense, there are impor-tant ways in which the rich face greater health risks

In the study of the relationship between health and income, the biggestchallenge, Jensen says, is to decompose the health differentials into the rootcauses There are numerous channels through which the two could be linked.Jensen narrows the focus to one particular mechanism—nutrition—through which SES may affect health The role of nutrition as a factor in

10 David A Wise

the differential health status between rich and poor is often overlookedwhen examining middle- and upper-income countries, because widespreadhunger and starvation, even among the poorest, have largely been elimi-nated and, in fact, widespread obesity is considered a greater public healthconcern However, nutrition must be viewed as more complex than hunger

or simply sufficient caloric intake In particular, there are important nutrients beyond calories that are important for good health, especially forthe elderly And the intake of these nutrients may be sensitive to income, asthe lowest cost staple foods in most countries (for example, bread or rice)may yield sufficient ‘bulk’ or calories, but (unless fortified) may have lowlevels of vitamins, minerals, and protein On the other hand, these foodstend to be low in fat, cholesterol, and sodium, compared with foods thatmay be more expensive and eaten in larger quantities by the rich, for ex-ample, meat Therefore, Jensen concludes, it is quite possible that nutri-tion plays a role in the relationship between health and SES, even in coun-tries where calorie malnutrition is scarce and obesity is widespread.Jensen uses data on food intake to provide a detailed analysis of nutrientintake for the elderly, how it varies with income, the consequences of nu-tritional intake for health, and the relationship between health and SES

micro-He does this by exploring differences in the diets of the rich and poor, how

differences in diet translate into differences in nutrient intake, and the pact of nutrient intake on health

im-Mortality

Mortality and Changes over the Twentieth Century

Mortality rates declined extremely rapidly in the United States over thetwentieth century, as they did in all developed countries In 1900, 1 in 42Americans died annually On an age-adjusted basis, the share in 1998 was

1 in 125 people, for a cumulative decline of 67 percent Given such a stantial improvement in mortality, it is natural to ask how we achieved suchgains in health and which innovations or policies contributed most to thesegains David M Cutler and Ellen Meara do this in “Changes in the AgeDistribution of Mortality over the Twentieth Century.”

sub-Cutler and Meara start by considering major trends in mortality overthe century, noting how mortality declines differ by age and cause of death

By providing detailed information on which demographic groups enced the largest mortality improvements and for what causes of death,these analyses motivate hypotheses to explain the overall improvement inmortality in the twentieth century

experi-The mortality decline is approximately linear over the time period tality decreased at a relatively constant rate of 1 percent per year between

Mor-1900 and 1940 There was then a period of rapid decline from 1940 to 1955

in which mortality declined 2 percent per year, followed by essentially flatmortality rates until 1965 Since 1965, mortality rates have fallen atroughly 1 to 1.5 percent per year This relative constancy of mortality de-cline suggests that perhaps a single factor can explain the trend in longerlife.

But the aggregate trends mask as much as they reveal While mortalitydeclines have been relatively continuous over the twentieth century, the agedistribution of mortality decline has not Cutler and Meara start by high-lighting a basic fact about mortality declines in the past century: mortalityreduction used to be concentrated at younger ages but is increasingly con-centrated among the aged In the first four decades of the century, 80 per-cent of life expectancy improvements resulted from reduced mortality forthose below age forty-five, with the bulk of this for infants and children Inthe next two decades, life expectancy improvements were split relativelyevenly by age In the latter four decades, about two-thirds of life expec-tancy improvements resulted from mortality reductions for those over ageforty-five; only one-third was from the younger population

This change has been accompanied by several important cal trends Throughout the first half of the twentieth century, infectiousdiseases were the leading cause of death Changes in the ability to avoidand withstand infection were the prime factors in reduced mortality in thefirst part of the century This disease-fighting ability was not predomi-nantly medical Nutrition and public health measures were vastly moreimportant in reduced mortality over this time period than were medicalinterventions, as substantial research documents Nutrition and publichealth were particularly important for the young, and so mortality reduc-tion was concentrated at younger ages

epidemiologi-Between 1940 and 1960, infectious diseases continued to decline, butwas due more to medical factors Antibiotics, including penicillin and sulfadrugs, became important contributors to mortality reduction in this era.Antibiotics help the elderly as well as the young, and so mortality reduc-tions became more widespread across the age distribution

Since 1960, mortality reductions have been associated with two new tors: the conquest of cardiovascular disease in the elderly and the preven-tion of death due to low birth weight infants While it is not entirely clearwhat factors account for the reduction in cardiovascular disease mortality,the traditional roles of nutrition, public health, and antibiotics are cer-tainly less important, the authors conclude Taking their place are factorsrelated to individual behaviors, such as smoking, diet, and high-tech med-

fac-ical treatment Cutler and Meara term this change the medfac-icalization of

death: Increasingly, mortality reductions are attributed to medical careand not social or environmental improvements

The medicalization of death does not imply that medicine is the only tor influencing mortality For several important causes of death, income

fac-12 David A Wise

improvements and social programs have had and continue to have a largeimpact on mortality For example, Medicare likely has a direct impact onmortality by increasing elderly access to medical care, but it also may haveimportant income effects since it reduced out of pocket spending by theelderly for medical care Social security and civil rights programs may also

be important in better health The authors do not quantify the role of icine, income, social programs, and other factors in improved mortality inthe last half-century, but they do show examples where each is important

med-as a first step in their ongoing research

Mortality and Age Differences in Medical Care

Perhaps the two most important, most enduring questions in health nomics are “What are the determinants of expenditures?” and “What arethe determinants of health?” Extensive research over the last thirty-fiveyears has produced a variety of answers to these questions, depending inlarge part on the specific context within which the questions are posed Onecrucial distinction is between explaining changes over time and explainingcross-sectional differences at a given time With regard to secular changes

eco-in the United States eco-in recent decades, most health economists now believethat advances in medical technology provide the major explanation forboth increases in expenditures and improvements in health With regard tocross-sectional differences, there is less agreement Victor R Fuchs, MarkMcClellan, and Jonathan Skinner consider cross-sectional differences in

“Area Differences in Utilization of Medical Care and Mortality amongU.S Elderly.” By exploiting a rich body of data from the Centers forMedicare and Medicaid Services [formerly the Health Care Financing Ad-ministration (HCFA)], the U.S Census of Population, and other sources,the authors hope to narrow that disagreement, at least with respect to area

differences in utilization of care and mortality of the elderly

Their focus on the elderly is motivated in part by the fact that the elderlyaccount for a disproportionate share of national health care expendituresand an even greater share of government health care expenditures More-over, the elderly experience the bulk of the major health problems of thepopulation Approximately one-half of all deaths occur between agessixty-five to eighty-four, and another one-fourth occur at ages eighty-fiveand above These shares are based on the current age distribution of theU.S population For a stationary population experiencing current age-specific mortality rates, deaths at ages sixty-five to eighty-four would stillaccount for almost one-half the total; the share at age eighty-five and abovewould rise to one-third The focus on the elderly is facilitated by the factthat the Medicare program generates a large, detailed body of data on uti-lization and mortality

One reason for focusing on area differences is that the large number ofmetropolitan and nonmetropolitan areas in the United States provides a

convenient framework for aggregating individual data in the search forvariables that may be related to utilization and mortality Moreover, manyhealth policy analysts believe that an understanding of area differencesmay suggest opportunities to limit expenditures and/or improve health.The paper has two main sections: utilization and mortality In most mar-kets an interest in expenditures would require attention to prices as well asquantities, but given universal insurance coverage through Medicare andadministrative price setting by HCFA, utilization is a natural subject forstudy Mortality is only one of many possible measures of health, but thereare several reasons to concentrate on it First, mortality is by far the mostobjective measure Secondly, it is, for most people, the most importanthealth outcome Thirdly, it is probably significantly correlated with mor-bidity since most deaths are preceded by illness.

In this paper, Fuchs, McClellan, and Skinner focus on whites ages five to eighty-four, or more specifically, those not identified as AfricanAmerican They exclude blacks because at those ages both utilization andmortality of blacks are higher than for whites, and the percentage of blacks

sixty-in an area is correlated with other variables of sixty-interest Moreover, other search suggests that the relationship between those other variables and uti-lization and mortality may be significantly different for blacks than forwhites They exclude anyone aged eighty-five and over because it is more

re-difficult to obtain accurate measures for self-reported variables, such as ucation and income About one-half of the population aged eighty-five andover suffer from some form of dementia and about one-fifth are in nursinghomes where measurement of income is particularly problematic More-over, most nursing home utilization is not covered by Medicare, the source

ed-of the data on utilization

The authors find wide variation in the utilization of health servicesacross regions It is not simply that some regions are higher along all di-mensions of care, but that in some regions there is much more diagnostictesting, even while per capita inpatient services are comparable to the na-tional average In general, utilization is strongly positively associated withmortality across areas—in other words, areas with more sick elderly usemore health care, other things being equal There remains, however, sub-stantial variation in utilization after controlling for factors such as educa-tion, income, and mortality

Cross-area variations in mortality rates among this elderly group are not

as large as variations in utilization, but they are still substantial The 10percent of metropolitan statistical areas (MSAs) with the highest mortal-ity (age-sex adjusted) have an average death rate 38 percent greater thanthe 10 percent of MSAs with the lowest mortality The comparable differ-ential between the high and low utilization areas is 49 percent

Education, real income, cigarettes, obesity, air pollution, and the centage of blacks account for more than half of the variation in mortality

per-14 David A Wise

across areas, but there are still substantial differences across regions plained by these variables Florida, in particular, has death rates signifi-cantly below the national average; the differential is particularly large forareas in the southern portion of the state The final section of the paper ex-plores two puzzles—why Florida is so different from the rest of the coun-try with respect to utilization and mortality, and why the presence of moreblacks in an area should be associated with higher mortality among elderlywhites The authors consider several possible solutions to these puzzles, in-cluding differential migration patterns of the elderly, but ultimately theyarrive at conjectures rather than robust explanations.

unex-Health and Wealth—Econometric Analysis of Causal Links

In their paper “Healthy, Wealthy, and Wise? Tests for Direct CausalPaths between Health and Socioeconomic Status.” Peter Adams, Michael

D Hurd, Daniel McFadden, Angela Merrill, and Tiago Ribeiro test for theabsence of causal links from socioeconomic status to health and mortality,and from health to wealth

This paper is in large part a technical discussion of how econometricstest for causal links The authors use innovations in health conditions and

in wealth in the AHEAD panel to carry out tests for causality from SES tohealth, and from health conditions to wealth By advancing beyond the de-tection of association to a framework in which there is some possibility ofdetecting the absence of causal links, this paper is an advance on much ofthe literature on this subject

The authors conclude that for mortality and for accidents, the

hypothe-sis of no causal link from SES is accepted, and for incidence of mental

problems the hypothesis is rejected The results for chronic and tive diseases are not definitive, but using the preferred test procedure, thehypothesis is marginally rejected in both cases The hypothesis appears to

degenera-be accepted for acute conditions, but the necessary invariance propertyfails, so the possibility that this is an artifact is not ruled out The pattern

of results suggests that incidence of acute, sudden onset health conditionsdoes not exhibit a significant SES gradient, whereas incidence of chronic,mental, and degenerative conditions appears to have an association to SESdue to some combination of direct causal links and common unobservedbehavioral or genetic factors Specifically, there may be an SES gradient inseeking treatment for the second class of conditions that may influence de-tection, or for maintaining preventative regimens that may maintain someconditions below the reporting thresholds Adams, Hurd, McFadden,Merrill and Ribeiro’s findings are not inconsistent with the possibility thatfor mental and chronic illnesses where the acute care procedures covered

by Medicare are often inapplicable, ability to pay may be a causal factor inseeking and receiving treatment

The technique put forth in this chapter has been applied to the Whitehalldata in the United Kingdom with results very similar to the results foundusing U.S AHEAD data This chapter includes not only the technical pa-per but comments from many experts in econometric analysis as well Theauthors are pursuing their line of investigation based on alternative mod-els and using subsequent waves of AHEAD.

16 David A Wise

The transition from employer managed defined benefit pensions to ment saving plans that are largely managed and controlled by employeeshas been the most striking change in retirement saving over the last twodecades Individual managed and controlled retirement accounts, particu-larly 401(k) plans but also 403(b) plans for nonprofit organizations, 457plans for state and local employees, the Thrift Savings Plan for federal em-ployees, Keogh plans for self-employed workers, and Individual Retire-ment Accounts (IRAs), have grown enormously Employer-provided de-fined benefit (DB) pension plans have declined in importance In 1980, 92percent of private retirement saving contributions were to employer-basedplans; 64 percent of these contributions were to DB plans In 1999, about

retire-85 percent of private contributions were to accounts in which individuals

17

1

The Transition to Personal Accounts and Increasing Retirement Wealth

Macro- and MicroevidenceJames M Poterba, Steven F Venti, and David A Wise

James M Poterba is the Mitsui Professor of Economics and associate head of the nomics department at the Massachusetts Institute of Technology, and the director of the Public Economics Research Program at the National Bureau of Economic Research Ste- ven F Venti is the DeWalt Ankeny Professor of Economic Policy at Dartmouth College and

eco-a reseeco-arch eco-associeco-ate eco-at the Neco-ationeco-al Bureeco-au of Economic Reseeco-arch Deco-avid A Wise is the John F Stambaugh Professor of Political Economy at the John F Kennedy School of Gov- ernment, Harvard University, and the director for Health and Retirement Programs at the National Bureau of Economic Research.

This paper was written while Poterba was a visiting fellow at the Hoover Institution We are grateful to the National Institute on Aging, the Hoover Institution, and the National Science Foundation for financial support, and to Gary Engelhardt, Bill Gale, Al Gustman, Syl Scheiber, Jon Skinner, Tom Steinmeier and participants in the Berkeley Department Seminar, the Stanford Public Economics workshop, and two NBER meetings for helpful comments We also thank Dan Beller at the Pension and Welfare Benefits Administration of the United States Department of Labor for generous help in understanding the Form 5500 data.

controlled how much to contribute to the plan, how to invest plan assets,and how and when to withdraw money from the plans.

We consider the changes in the magnitude and the composition of ing for retirement over the last two decades We begin with an analysis ofaggregate data on retirement plan contributions We then turn to micro-data, describe patterns in these data, and try to reconcile these patternswith the aggregate data We document the changes in aggregate retirementsaving over the past twenty-five years and describe how these changes arerelated to the shift from employer-sponsored defined benefit plans to indi-vidual-controlled retirement saving We then investigate whether the shifttoward individual retirement saving, and the accumulation of retirementassets in these accounts, has been offset by a reduction in the assets in otherretirement saving plans

sav-In a series of earlier papers, summarized in Poterba, Venti, and Wise(hereafter PVW, 1996, 1998b), we found large net saving effects of IRAs and401(k)s We emphasized the potential offsets between saving in self-directedretirement accounts, other forms of financial asset saving, and the accumu-lation of home equity On balance we found little, if any, offset in these cases.More recently, Benjamin (2003) and Pence (2001) have also found little or

no offset between 401(k) contributions and non-401(k) financial asset ing, although the latter study also found little evidence that 401(k)s in-creased total wealth Recent work by Engen and Gale (2000) finds little

sav-offset among low earners, but more substantial offsets among high earners.Much less attention has been directed to the possible offset of personalretirement assets by a reduction in the assets in DB pension plans Engen,Gale, and Scholz (1994) found a negative relationship between participa-tion in DB pension plans and 401(k) plan assets in the 1987 and 1991 Sur-veys of Income and Program Participation (SIPP) Papke (1999) concludedthat between 1985 and 1992 about one-fifth of ongoing sponsors of DBplans terminated their plans and adopted or retained a conventional de-fined contribution (DC) or a 401(k) plan It is not clear from her analysis,however, whether the growth in 401(k) plans displaced DB plans Papke,Petersen, and Poterba (1996) surveyed firms with 401(k)s and found thatvery few had terminated a preexisting DB plan when they instituted their401(k) plan Their sample, however, may not have been representative ofthe broader population of firms

More recently, Ippolito and Thompson (2000) combined Form 5500data with information from the Pension Benefit Guarantee Corporation(PBGC) to study within-firm changes in plans over time They found littlefirm-level displacement of DB plans by 401(k) plans, and concluded thatthe replacement of a DB plan by a 401(k) is rare Engelhardt (2000), bas-ing his findings on data from the Health and Retirement Study (HRS), con-cludes that households eligible for a 401(k) have higher non-DB assets than

18 James M Poterba, Steven F Venti, and David A Wise

households not eligible for a 401(k), but have the same level of assets when

DB wealth is included He interprets this as evidence of firm-level tution of 401(k)s for DB pensions However, as we explain later, the HRSdoes not allow accurate categorization of individuals into 401(k) eligibleand noneligible status

substi-Most recently, LeBlanc (2001) has estimated the reduction in tions to the Registered Retirement Saving Program (RRSP) in Canadawhen persons are newly covered by an employer-provided DB plan Based

contribu-on a lcontribu-ongitudinal panel of individual tax data, and using a di

fference-in-difference estimation procedure, he finds that for a dollar of DB plan ing, RRSP contributions are reduced by only about $0.15

sav-Our analysis of these issues is divided into six sections In section 1.1 weconsider aggregate data on the total stock of retirement wealth The verylarge increase in total retirement assets relative to income over the pasttwenty-five years strongly suggests that the enormous growth in individualretirement assets has more than offset any displacement of asset growth intraditional DB pension plans

In section 1.2 we show that the “retirement plan contribution rate” ismuch greater than the personal saving rate reported in the National In-come and Product Accounts (NIPA) in recent years Our retirement plancontribution rate is determined by the retirement saving of current em-ployees The NIPA saving rate, in contrast, depends on the saving and con-sumption patterns of retirees as well as those who are currently working

We document the substantial growth over time in contributions to directed retirement saving programs, such as 401(k) plans We also suggestthat the retirement plan contribution rate was reduced by legislation re-stricting contributions to DB pension plans, as well as by the strong stockmarket performance of the late 1980s and 1990s and the associated reduc-tion in required DB plan contributions

self-In section 1.3, we distinguish between retirement saving from the point of an employee, and employer contributions to retirement savingplans We argue that from the perspective of the employee, 401(k) retire-ment saving is likely to be much greater than traditional DB plan saving atmost ages We use data on accruing DB plan liabilities to compare 401(k)and DB plan saving rates, and conclude that the saving rate under a typi-cal 401(k) plan is about twice that under a typical DB plan

stand-In section 1.4 we begin to explore the possible substitution between

different types of retirement plans We use data from both the Department

of Labor Form 5500 filings, and from the SIPP We find no evidence ofstrong substitution patterns between 401(k) participation and other retire-ment plans Section 1.5 shows that further analysis of substitution, usingdata from the HRS, supports the results in section 1.4

A brief conclusion summarizes our findings

1.1 Aggregate Data on Assets in Retirement Saving Plans

1.1.1 Retirement Account Assets

While it is not possible to link particular assets with particular motivesfor saving, for many households assets in retirement saving accounts arethe best single indicator of the amount that they have saved for retirement

A number of factors are likely to contribute to variation in retirement sets For example, one would expect that households with higher earningswould have more retirement assets For a given level of aggregate earnings,

as-a las-arger shas-are of the working populas-ation neas-ar retirement as-age is likely to beassociated with greater retirement assets Variation in life expectancy and

in the typical retirement age can also affect the stock of retirement assets.The “adequacy” of any given level of assets depends on the years of sup-port the assets are expected to provide

Our analysis begins with measures of aggregate retirement assets thatare not adjusted for demographic trends We then explain the likely effect

of adjustment for demographic changes Figure 1.1 shows the ratio of sets in all private retirement accounts—including DB plans, 401(k), other

as-DC plans, IRAs, 403(b) plans, and Keogh plans—to private wage andsalary earnings.1This ratio increased more than fivefold between 1975 and

1998, from 0.39 to 2.02 The figure shows modest growth in the ratio of

re-20 James M Poterba, Steven F Venti, and David A Wise

1 Appendix A describes all of our data sources.

Fig 1.1 Ratio of private and total retirement assets to wage and salary earnings

tirement assets to earnings through 1981; more rapid growth between 1982and 1994, after the introduction of IRAs and 401(k) plans and during a pe-riod of positive stock market returns; and rapidly accelerated growth be-ginning in 1995, corresponding to large increases in equity market returns.Figure 1.1 also shows the ratio of assets in all retirement plans, the privateplans as well as public sector plans, to wages and salaries The trend is verysimilar to that for private plan assets alone.

Figure 1.2 shows private retirement assets disaggregated into severalcomponents It shows that assets in DB plans continued to grow after theintroduction of 401(k) and IRA plans, but that the bulk of the gain was inindividual accounts In this figure, 401(k) assets are included with other

DC plans There is no evidence of a decline in the assets in conventionalemployer-provided plans during the time period when assets in individualaccounts were growing most rapidly

The foregoing data alone cannot rule out the possibility of substitution,because we do not have data on the time path that other retirement planassets would have followed in the absence of the growth in DC assets To

place the growth in DC assets into perspective, however, we note that if all

contributions to personal retirement accounts between 1985 and 1998 hadcome at the expense of DB contributions, DB assets would have grown by

a factor of 8.4 instead of 2.7

The private retirement assets in figure 1.2 exclude assets in federal, state,

Fig 1.2 Private retirement assets

and local retirement plans, and assets held by life insurance companies inretirement plans that are also part of the retirement asset pool.2Figure 1.3shows the assets in private plans as well as the assets in these other plans.

In 1999, about 40 percent of all retirement assets were in federal, state, andlocal and insurance plan funds

Retirement account assets support current retirees as well as future tirees Although we are unable to distinguish the assets held by current re-tirees from those held by the working-age population, we suspect that theincrease in these assets represents a large upward trend in the assets of fu-ture retirees

re-1.1.2 Housing and Other Nonretirement Assets

Aside from promised Social Security benefits, housing equity is the mostimportant asset of a large fraction of Americans Unlike the increase in re-tirement account assets, however, there has been no increase in housingequity relative to income over the past two and one-half decades Figure 1.4shows housing equity as a fraction of disposable income from 1975 to 1998.The ratio increased about 25 percent between 1975 and 1989, but by 1999

it was essentially at the same level as in 1975 The figure also shows

non-22 James M Poterba, Steven F Venti, and David A Wise

2 The Flow of Funds Accounts (FFA) defines the latter series as including “assets of private pension plans held at life insurance companies, such as guaranteed investment con- tracts and variable annuity plans, that are managed for the benefit of individuals who are not separately identified to the insurance companies.”

Fig 1.3 Public and private retirement assets

retirement, nonhome equity net worth as a share of disposable income.This ratio decreased and then increased between 1985 and 1999 The in-crease between 1975 and 1999, 27 percent, was not nearly as great as the in-crease in retirement assets over this period.

1.1.3 Retirement Assets and Demographic Trends

The growth of retirement assets relative to income may be explained by

a number of changes These include the advent of new retirement saving hicles as well as other factors, such as demographic change Changes inthree features of the population—demographic composition, mortalityrates, and labor force participation—have likely contributed to the rise inretirement assets relative to income We describe each of these changes, al-though we do not attempt a formal adjustment of retirement wealth to cor-rect for these changes

ve-The increase in life expectancy at retirement age is the first substantialchange that may have contributed to rising retirement assets In 1975, lifeexpectancy for a U.S man at age sixty-two was 15.5 years, while that for awoman was 20.3 years By 1997, male life expectancy at age sixty-two hadincreased to 17.6 years, while female life expectancy had risen to 21.4 years.For men, this implies a 13.5 percent increase in the number of years thatneed to be supported with retirement resources, beginning at age sixty-two.For women, the change was 5.4 percent These proportional changes pro-vide a crude measure—crude, because they do not reflect the potential role

of risk and the prospect of drawing down resources too quickly—of the crease in retirement resources that would be needed to offset improved

Fig 1.4 Ratio of home equity and other assets to disposable income

longevity These changes might account for an increase in resources ofroughly 10 percent, much less than the actual growth of retirement assetsrelative to income.

The second important demographic change that might have contributed

to rising retirement assets was the aging of the labor force Translating formation on the age structure of the population into predictions about thewealth-income ratio requires detailed information on saving by age, yetthere is no agreement on the relative importance of life-cycle, precaution-ary, and other factors in saving decisions In 1975, the average age of thoseover the age of twenty in the U.S population was 44.6 years For men, theaverage age was 43.9 years Between 1975 and 1985, the average age ofthose over twenty actually declined to 44.3 years for the entire populationand 43.5 years for the male population This reflected the entry of the

in-“baby boom” cohorts into the 20-plus age group By 1998, the working agepopulation had grown older, the average age of all 20-plus persons was 45.5years, and that of 20-plus men was 44.8 years Thus between 1985 and

1998, the average age of the adult population rose by just over one year.Similarly, the average age of those in the labor force in 1985 was 38.5 years,whereas in 1998, it was 40.3 years

These data on the population and labor force age structure suggest that

by the late 1990s, those who were in their earning years were older and hadfewer remaining years of work to accumulate assets for retirement thanthose in the working population in the 1970s and early 1980s This also mayhave induced a rise in retirement assets

The final change that may have affected retirement assets is the shiftingage of retirement in the U.S population During the 1980s and 1990s,these changes were modest by comparison to earlier decades Burtlessand Quinn (2000) present detailed information on age-specific laborforce participation rates for U.S men in 1970, 1984–85, and 1998–89.Their data show a sharp decline in labor force participation rates between

1970 and 1984–85, but relatively little decline subsequently The pation rates for 1998–99 were virtually identical to those in 1984–85 Atages above sixty-five, the labor force participation rate in the late 1990swas greater than that in the mid-1980s There is no systematic difference

partici-in labor force participation rates at younger ages Labor force tion rates for women in their early sixties increased between the mid-1980s and the late 1990s as cohorts of women with greater labor forceparticipation rates when they were younger entered the retirement-agecohort

participa-Changes in retirement ages are therefore not likely to account for stantial changes in retirement wealth relative to income during the last twodecades Demographic factors—shifting age structure and lengthening lifeexpectancy—seem likely to account for modest increases in retirement as-sets, but are unlikely to account for more than a small fraction of the largechanges we observe

sub-24 James M Poterba, Steven F Venti, and David A Wise

The Transition to Personal Accounts and Increasing Retirement Wealth 25

A

B

Fig 1.5 A, Private pension contributions; B, all pension contributions

1.2 Plan Contributions and the Retirement Plan Contribution Rate

The accumulation of retirement assets depends on the inflow of butions, the payout of benefits, and the return on invested assets Panel A

contri-of figure 1.5 shows private pension plan contributions, which increased most sixfold between 1975 and 1999, while panel B of figure 1.5 shows con-

al-tributions to all retirement plans Neither of the series includes tions to privately held pension plans administered by insurance compa-nies, which hold about 9 percent of the assets in all pension plans Privateplans include self-directed plans such as 401(k) plans and IRAs IRA con-tributions exclude rollovers, while IRA assets include assets rolled in tothese accounts.

contribu-The pronounced “hump” in retirement plan contributions between 1982and 1986 corresponds to the beginning and subsequent retrenchment ofthe IRA program The pattern strongly suggests that IRA contributionsduring this period were not offset by a reduction in other forms of retire-ment saving Indeed, the rate of increase of non-IRA retirement saving wasthe same in the 1982–85 period as in prior years This pattern suggests thatthe total pool of assets in retirement plans likely would be much greater to-day if the IRA program had not been limited in 1986

Panel A of figure 1.6 shows both private and total retirement plan tributions scaled by disposable income Panel B of figure 1.6 shows plancontributions over wage and salary earnings In both figures, private con-tributions are scaled by private earnings, while all contributions are scaled

con-by all wage and salary earnings We define these ratios as retirement plan contribution rates They measure the proportion of current earnings that is

saved in retirement accounts by current employees The contribution rates

do not account for retirement plan earnings on existing assets, or for drawals from these plans In the following, we compare retirement plancontribution rates to NIPA national saving rates

with-Panels A and B of figure 1.6 show that retirement plan contribution ratesare remarkably stable over most of the period Scaled by personal dispos-able income, the private plan contribution rate was about 3.5 percent in

1975 and in 1999, and the contribution rate for all plans varied between 5and 6 percent for most of the period When scaled by private and by all wageand salary earnings, the contribution rates are also stable, although they aregreater than the rates scaled by personal disposable income The retirementplan contribution rate for all plans, including those in the federal and stateand local government sector, is near 8 percent for most of the period, orabout 2 percentage points higher than the rate for the private sector alone.The relative stability in the retirement plan contribution rates was bro-ken only by the large increase in the plan contribution rate when the IRAprogram was initiated, and the decrease when the program was curtailed in

1986 For example, relative to earnings, both the private and the all planrates are about 2 percentage points higher during the IRA period—over 8and 10 percent, respectively

1.2.1 Time Series Changes in the Retirement Plan Contribution RateThe relative stability of the retirement plan contribution rate concealsfluctuations in some of the factors that affect this rate Contributions toprivate DC type plans increased sharply over the 1975–99 period, while DB

26 James M Poterba, Steven F Venti, and David A Wise

contributions varied widely At the end of this period, DB plan tions were only slightly higher than at the beginning.

contribu-Retirement plan contributions are the product of the number of pants and the average contribution per participant Figure 1.7 shows thesum of the number of active participants in all DB and DC plans.3It illus-

A

B

Fig 1.6 A, Ratio of private and total pension contributions to disposable income;

B, ratio of private and total pension contributions to wage and salary earnings

3 These data, from Form 5500 filings and IRS tabulations of tax returns, show the number

of persons participating in each type of retirement saving plan Many persons participate in

trates in particular the rapid growth of 401(k) plans The number of ticipants in these plans, which first became available in 1982, grew to al-most 38 million by 1997 While 401(k) plan participation grew in the 1980sand 1990s, participation in DB plans declined from about 30 million in

par-1984 to about 23 million by 1997 Participation in non-401(k) DC plansincreased until about 1986 and then declined, ending the period about 30percent higher than at the beginning There is a clear “IRA effect” on planparticipants, as well as plan contributions, in the early 1980s In total, thenumber of plan participants increased from about 39 million in 1975 toover 80 million in 1997

Panel A of figure 1.8 shows contributions per participant in DB, DC,and 401(k) plans Panel B of figure 1.8 shows IRA and 401(k) contribu-tions, while panel C of figure 1.8 shows contributions to Keogh plans DBcontributions per participant fluctuated substantially during the last twodecades, and they were about 40 percent higher at the end of the periodthan at the beginning Non-401(k) DC contributions per participant in-creased about twofold over the period, and on average were higher than

DB contributions Over the past fifteen years contributions per participant

to 401(k) plans were, on average, twice as large as contributions per ticipant to DB plans Contributions to 401(k)s increased almost 50 percent

par-28 James M Poterba, Steven F Venti, and David A Wise

more than one plan, so the total number of participants overstates the number of persons who participate in at least one plan For 401(k) plans, participants include all persons eligible to contribute, regardless of actual contributions.

Fig 1.7 Active participants in private pension plans (with double counting)

between 1982 and 1996 alone.4During the “unrestricted” IRA period,1982–86, IRA contributions on average were greater than 401(k) contri-butions.

4 401(k) contributions are calculated by dividing total contributions to 401(k) plans by the

total number of employees eligible to contribute, not the number that actually make tions There is much less change during this period in the participation rate of 401(k) eligibles,

contribu-conditional on eligibility, than in the eligibility rate Most of the change in the number of tributors is therefore due to changes in eligibility.