Who is "62 Enough"? Identifying Respondents Eligible for Social Security Early Retirement Benefits in the Health and Retirement Study

Social Security Bulletin • Vol. 62 • No. 3 • 1999 51

Who is “62 Enough”? Identifying Respondents

Eligible for Social Security Early Retirement

Benefits in the Health and Retirement Study

by Janice A. Olson, Division of Economic Research, Office of Research,

Evaluation and Statistics, Office of Policy, Social Security Administration

Acknowledgments: The

author wishes to thank

Keith Bender, Ben Bridges,

Barry Bye, Sharmila

Choudhury, Russ Hudson,

John Phillips, and Barbara

Lingg for their helpful

comments on earlier drafts

of this note.

Summary

Workers are not instantly eligible for Social

Security retirement benefits on their 62nd

birthdays, nor can they receive benefits in the

month they turn 62. This note discusses how

well researchers can do using data from the

Health and Retirement Study (HRS) to identify

respondents old enough to receive and report

early Social Security retirement benefits. It

shows that only some workers aged 62 at the

time of an HRS interview will be “62 enough” to

have received a Social Security benefit and

reported it in the survey.

In general, workers become eligible for a

retirement benefit the month after they turn 62,

and they may receive their first payment the

month after that. Until recently, payments were

received very early in the month, but in mid-

1997 and later, the Social Security Administra-

tion (SSA) staggered benefit payments over the

course of a month. Therefore, many beneficia-

ries will not be able to report the receipt of their

first benefit payment until the third month after

their birthday in more recent HRS interviews.

This note describes the best approach for

approximating the pool of HRS respondents

who are old enough to have reported the

receipt of their first retirement benefit. It then

applies the procedure to an analysis by

Burkhauser, Couch, and Phillips, who used the

1994 HRS data to distinguish between those

who took early retirement benefits upon turning

62 and those who postponed the receipt of

benefits. Because these authors did not

provide for respondents who were not “62

enough” to receive a benefit at the time of the

interview, they understated the proportion of

respondents who took retirement benefits at

age 62.

Introduction

Either the normal retirement age (NRA) or

the earliest eligibility age (EEA) for Social

Security retirement benefits would be in-

creased under many proposals for Social

Security reform. As a consequence, research

interest in who retires at early ages and the

potential effects of an increase in the NRA or

EEA has grown. This note discusses how well

researchers can do using data from the Health

and Retirement Study (HRS) in identifying the

pool of respondents who could have received

early Social Security retirement benefits.

Both

an understanding of some policy and opera-

tional features of the Social Security program

and of monthly detail in HRS survey respon-

dent reports are necessary to approximate the

pool of potential early beneficiaries.

It seems simple enough to say that work-

ers can receive early retired worker benefits

at age 62. In fact, only some workers aged

62 at the time of an HRS interview will be “62

enough” to have received a Social Security

benefit and reported it in the survey. To keep

Social Security Bulletin • Vol. 62 • No. 3 • 1999

the focus on age and receipt reports, the discussion is re-

stricted to insured workers, and assumes that the insured group

can be identified (disabled worker and auxiliary beneficiaries

excluded), and that there is no measurement error. To further

crystallize the discussion, the focus is on benefit receipt reports

in the earliest month possible, but the basic ideas are easily

extended to a longer first benefit period. For example, research-

ers may wish to identify those retiring anytime while aged 62 or

anytime prior to the full benefit age in the pool of workers eli-

gible to do so.

Social Security Policies and Procedures

Age and Entitlement

To be entitled to retirement insurance benefits workers must

be at least age 62, fully insured, and have filed an application

for retirement insurance benefits.

Under the age eligibility

requirement, workers must be at least age 62 throughout the

month of first entitlement (Social Security Administration

1997A, paragraph 301).

For example, most workers with 62nd

birthdays in June 1993 do not meet the age requirement until

July 1993. Exceptions are those born on the 1st or 2nd day of

the month. Those born on the 1st

are considered to have at-

tained age 62 on the last day of the preceding month; those

born on the 2nd are considered to have attained age 62 on the

first day of their birthday month.

Both could be entitled to

benefits for the month of their 62nd birthday.

Those retiring early have their benefits reduced by 5/9ths

of 1 percent (or 1/180th ) for each month of entitlement before

age 65.

To figure the number of reduction months, count all

months, beginning with the first month of entitlement to re-

duced benefits, up to the month in which the person reaches

NRA.

As noted in Svahn (1981), the major effect of the

“throughout the month” provision is a delay in benefit entitle-

ment for most of the earliest retirees. That delay, in turn, means

a reduction for early-retirement benefits for most workers based

on their retirement occurring 35 months before age 65 (rather

than 36 months).

Application Processing Times

As noted, to be entitled to retirement benefits, workers must

also apply for them, and an additional lag between a 62nd birth-

day and early benefit receipt may result from the time needed by

SSA to process the application. Appropriate documents and

information must accompany the application (for example, a

birth certificate or the previous year’s W-2). The Social Security

Administration's SSA Accountability Report for Fiscal Year

1997(1997C) showed that in the mid-1990s, when the oldest

HRS cohorts were first reaching age 62, claims processing time

for retirement and survivors insurance claims averaged 15 work-

days. To avoid delays due to processing time, SSA encourages

workers to contact an SSA office 2-3 months before reaching

age 62 (SSA 1997A, paragraph 1503), and workers may file

applications before the first month in which they are entitled.

(Before retirement, SSA also encourages workers to check

periodically on their own Social Security earnings records to

ensure that all earnings have been credited to them.)

Workers who are aware of these SSA policies and who

contact SSA early with the necessary documents are probably

most likely to receive benefits at the earliest possible date.

Delays may result for intended early retirements among the less

knowledgeable (for example, those who believe they must wait

until their 62nd birthday to apply) or those with missing birth

certificates or other documents.

Monthly Payment Procedures: Prior to May 1997

In the 1992 HRS start year and other years prior to May

1997, Social Security benefits for any given month were paid on

the 3rd day of the following month for all beneficiaries.

For

example, a worker born in June who is first eligible in July would

receive the first possible payment—the benefit for July—on

August 3rd. If HRS interviews are spread pretty evenly over the

days of the month, a benefit payment date of the 3rd of the

month means that most HRS interviews in a given month will

have occurred after the benefit was received. Thus, respon-

dents could generally report a first benefit receipt to the HRS

interviewer in the month that the benefit payment was received.

Monthly Payment Procedures: May 1997 and Later

For new filers for Social Security benefits in May 1997 or

later, benefits are paid not on the 3rd of the month but on either

the second, third, or fourth Wednesday of the month, depend-

ing on their day of birth (SSA 1997A, page 77; SSA 1997B,

paragraph 145). Once started, a given person will always be

paid on the same Wednesday. Persons getting both SSI and

Social Security are an exception: their Social Security benefit

will be received (or deposited) on the 3rd of the month. For

other exceptions, see the Social Security Handbook.

Because the change is keyed to the filing date, which is not

available in the HRS, identification of respondents retiring in

the late spring or summer of 1997 who are under the new system

is not possible. In general, however, this new system makes it

more likely that a first Social Security benefit receipt in any

given month will occur after the HRS interview in that month

(again, assuming the interviews are spread evenly over the

days of the month). Insofar as that happens, obtaining a picture

of who took benefits as soon as possible will not be possible

until the third month following the worker’s birthday month in

the 1998 and later HRS interviews. For example, workers born in

June who receive their first benefit (the July benefit) on the

third or fourth Wednesday in August would probably not be

able to report that receipt until a September interview (or the

next biennial HRS interview, since most HRS interviews occur in

the spring and summer).

Identifying EEA Eligibles in the HRS

The HRS is a longitudinal survey of a nationally representa-

tive cohort of persons born in 1931-41. Interviewing began in

1992 when cohort members were aged 51-61 and will continue

biennially as they move through their retirement years. (See

Juster and Suzman 1995, or on-line at: <www.umich.edu/

~hrswww> for more information about the HRS.)

Social Security Bulletin • Vol. 62 • No. 3 • 1999 53

HRS survey data contain interview dates (month, day, and

year (MMDDYY)), date of birth (MMYY), whether the respon-

dent is receiving Social Security benefits and, if so, the date

that the respondent began receiving those benefits (MMYY),

as well as the type of benefit (that is, retirement, disability,

survivor’s benefits, dependent of disabled worker, dependent

of retired worker).

Thus, available HRS information allows age

and first retirement benefit receipt to be measured by month as

well as by year.

Given the program features discussed in the

previous section, identifying the pool that would be subject to

the maximum (or near-maximum) reduction for early retirement is

somewhat more complicated than it may initially seem. More-

over, with the available data it is only possible to approximate

that pool,

and the approach described here generally allows

an extra month when the timing is ambiguous. This approach

incorrectly includes among the earliest retirees a few people

who took benefits after their earliest possible month but cor-

rectly captures among the earliest retirees the larger group who

have retired in the earliest possible month but need the extra

month to report it.

Such an approximation can be done as follows. For workers

becoming 62 in a given year prior to 1997, researchers should

identify the pool of respondents who are into at least the sec-

ond month following their 62nd birthday month.

They are “62

enough” to report a first benefit. For all workers reaching age 62

in 1998 or later and many of those reaching age 62 in 1997,

new payment procedures increase uncertainty about benefit

receipt reports in the second month following the birthday

month. Using the third month following their 62nd birthday

month reduces the uncertainty resulting from the new payment

dates.

Regardless of cohort, an additional month’s lag until

the third (or fourth) month following the birthday month would

allow reports from some workers whose applications had longer

claims processing times.

Within the pool, a comparison of the month and year that

respondents report that they started receiving benefits with

the month and year of birth will allow those who took benefits

in the earliest possible month and reported those benefits in

the second or third month following their 62nd birthday month

to be distinguished from those who did not. Note that for re-

spondents with 62nd birthdays late in the year, the earliest

benefit receipt might not occur until early in the following

calendar year.

In 1994 (the second HRS interview year), for example, HRS

interviewing began in May. As a result, all of the 1931 cohort

who had attained age 62 in 1993 could report the month and

year of an early benefit receipt, but for 15 percent of the cohort

who were born in November or December, the earliest receipt

could not occur until January or February 1994, respectively.

That kind of situation—that of respondents who became 62

in a year prior to the interviewing year—is summarized in the

upper part of chart 1. The second kind—that of respondents

interviewed in their 62nd birthday year—is summarized in the

lower part of the chart. Data for some of these respondents

should be set aside because respondents are either still age 61

or not “62 enough” to report an early retirement benefit. For

example, the 1994 HRS interview with the 1932 cohort occurred

in their 62nd birthday year, and they are not all “62 enough” to

be age-eligible for early benefits. In fact, of respondents in the

1932 cohort in the 1994 HRS survey year, only about 43 percent

completed the 1994 HRS interview in or after the first possible

month in which they could report a benefit receipt. The remain-

ing respondents, roughly 57 percent, could not have received a

first benefit in time to report it in their 1994 interview month. For

researchers unwilling to wait for retrospective data from the

1996 HRS interview to establish who in the 1932 cohort retired

in the earliest possible month and who did not, examination

should be restricted to the 43 percent who were “62 enough” to

report a benefit in 1994.

Peak HRS interviewing months may vary across HRS inter-

viewing years. As a result, larger or smaller proportions of each

cohort attaining age 62 in an interviewing year may be “62

enough” to report an early benefit in each interviewing period.

Without the kind of specification described here, the proportion

of those aged 62 in a given interviewing year who take early

benefits may appear to vary from year to year solely as a result

of changing HRS interviewing months.

An Application of the Approach

In developing preliminary estimates of the economic and

health characteristics of early retirees, Burkhauser, Couch, and

Phillips (1996) classified 1994 HRS respondents from the 1931

and 1932 cohorts into “takers” and “postponers” of early Social

Security retirement benefits. In assigning a taker or postponer

status, they used the year of Social Security benefit receipt and

the respondent’s age in 1993 or 1994 (1996, table 2 and foot-

note 7). Respondents aged 62 in 1993 (1994) who reported So-

cial Security receipt in 1993 (1994) were classified as takers. The

remaining respondents, including those not insured for Social

Security, were classified as postponers.

Their analysis would have benefited from a 3-group classifi-

cation: takers of early Social Security retired worker benefits,

postponers of early retired worker benefits, and a group that is

not “62 enough” to determine their taker or postponer status.

By default, the not “62 enough” were classified with the post-

poners in the Burkhauser, Couch, and Phillips (1996) work. Our

HRS estimates, discussed above, suggest that 15 percent of the

1931 cohort and 57 percent of the 1932 cohort were not “62

enough” in their 62

birthday years (1993 and 1994, respec-

tively) for their status in their first month of possible benefit

receipt to be known. Classifying them all as postponers would

be a misclassification for a large minority who might have re-

tired at the earliest date.

Classifying takers and postponers of early retirement ben-

efits using year alone for the full age-62 group results in a post-

poner group that is too large and a total that is too large be-

cause they both include the not “62 enough” group. Some

implications of these measurement problems are developed

here. For example, the Burkhauser, Couch, and Phillips data

show that 27.9 percent of men take Social Security benefits

while they are aged 62 (1996, table 4). (All of their results are

Social Security Bulletin • Vol. 62 • No. 3 • 1999

given separately by sex.) However, if their estimates had been

restricted to men who were “62 enough” to report even one

month’s benefit in the 1994 interview, then closer to 43 percent

of men took Social Security benefits while they were aged 62.

(For both their estimate and this one, some men have been “62

enough” for only one or a few months when the data were

collected. The true proportion of men in these cohorts who take

benefits at some point in their 62nd year (that is, before their

63rd birthday) will, no doubt, be higher.)

In general, restricting the results in the Burkhauser, Couch,

and Phillips paper to those who are “62 enough” increases

estimates for male takers (and subgroups of male takers) by

about 50 percent. This increase applies to percentages in

Burkhauser, Couch, and Phillips that are based on their sample

of men aged 62, not to percentages based only on their male

takers. The policy relevance of such a change probably varies

with the size of the estimate, and it would be minor for very

small estimates. For example, a corrected estimate would show

an increase of only about 1 percentage point in their findings

that 2.2 percent of men aged 62 took early retirement, were in

poor health, and had no other pension income. Similar calcula-

tions were not done for estimates of women taking early ben-

efits, and no adjustment factor is suggested for estimates sur-

rounding postponers. About half of the men classified as

postponers were not “62 enough” to have reported a benefit.

Concluding Remarks

As HRS interviews continue and the HRS sample matures,

the retirement behavior of all workers while aged 62 can be

identified with a fair degree of accuracy using monthly reports.

However, because workers born on the 1st or 2nd of the month

and individual processing times are unknown, and because

benefit payment days are not available for claimants after May

1997, a single “perfect” pool probably cannot be identified with

HRS data. However, the study of early retirement will be im-

proved by allowing for lags between a 62nd birthday month and

first benefit receipt (including lags that may go into the next

calendar year) and removing those not “62 enough” from data

collected in the 62nd birthday year.

With a more mature HRS sample, it will be possible to iden-

tify all workers who receive benefits reduced for early retirement

rather than just the group with reductions for EEA. The impact

of any misclassifications on a larger group of early retirees

would be much smaller than the impact on only those who are

within the first couple of months of their EEA.

Notes

Information provided here would also be directly useful to re-

searchers using the Panel Study of Income Dynamics and the New

Cohorts of the HRS. (See Willis 1999, for a description of the New

Cohorts.) Persons working with the Survey of Income and Program

Participation or other data sources that collect monthly information on

Social Security benefit receipt and date of birth would also gain from

this discussion.

Under the terminology used in the Social Security retirement

program, “eligibility” refers to meeting the insured status and age re-

quirements regardless of whether or not an application for benefits has

been filed. “Entitlement” refers to the state of meeting all reqirements

for receipt of benefits, including the filing of an application.

The “throughout the month” provision applies only at age 62 to

workers and their spouses and was added with the 1981 amendments

(Svahn 1981). SSA administrative data show that most men who take

retirement benefts at age 62 take them in the earliest possible month.

Under the old law, for example, the majority of the 1915 cohort retiring

at age 62 first took benefits for their birthday month. After the 1981

amendments, the majority of the 1920 cohort retiring at age 62 first

took benefits for the month following their birthday month.

For a discussion of the desirability of changing the throughout-the-

month provision, see the report of the Inspector General (1998),

p. 125, avalable on-line at: <www.ssa.gov/finance>. SSA disagreed

with the Inspector General’s recommendation to change the provision.

Chart 1.—Summar

of methods for identif

ersons a

ed “62 enou

h” to re

ort recei

t of earl

Social Securit

retirement benefits

HRS

rou

HRS example Method Notes

Cohorts with 62nd 1931 cohort who Identifying those with first benefit receipt Properly measured,

birthda

s in

ears

rio

became 62 in 1993 re

orts for at least the second month followin

all in the cohort are

to the interview year (an HRS noninter- the 62nd birthda

(

Note that for those “62 enou

h” to re

ort

viewing year) born in November or December, allow benefit receipt for at

reports for January or February in the year least 1 month.

following the 62nd birthday year.)

Cohort with its 62nd 1932 cohort who Identify those with 62nd birthdays at least Some in the cohort

birthda

in the interview became 62 in 1994 2 months

rior to the interview month.

Set are set aside as not “62

year interviewing year others aside as not “62 enough” for their early enou

h” in this inter-

benefit receipt status to be known. view year.

After the 1997

ment

rocedure chan

allow an extra month.

After the 1997

ment

rocedure chan

allow 3 months.

Social Security Bulletin • Vol. 62 • No. 3 • 1999 55

In SSA regulations, the rule is that “an individual attains a particu-

lar age on the first moment of the day preceding the anniversary of his/

her birth” (SSA 1999). (See also the Inspector General’s 1998 report,

p. 125, available on-line at: <www.ssa.gov/finance>. The age-attain-

ment rule results from commonlaw definitions.

This discussion of benefit reductions for early retirement assumes

that no benefits are withheld due to the earnings test.

For those born in 1938 or later, the retirement age will increase

gradually from age 65 until it reaches age 67 in 2022. The increase will

affect the amount of reduction for those retiring early with reduced

benefits (SSA, Social Security Handbook 1997A, paragraphs 723-725).

This note assumes a normal retirement age of 65.

Only those born on the 2nd of the month are subject to the full 36-

month reduction. They attain age 62 on the lst of their birthday month,

are eligible throughout their birthday month, and attain age 65 in their

birthday month 3 years later. On the other hand, those born on the 1st

of the month attain age 62 in the previous month, are eligible through-

out their birthday month, but they attain age 65 in the month preceding

their 65th birthday month. The “throughout the month” provision

does not apply at age 65, so they are subject only to a 35-month reduc-

tion. Using June as an example month, patterns by day of birth can be

summarized as follows:

_______________________________________________________

Day of birth 1st 2nd 3rd-30th

62nd birthday 6/1/93 6/2/93 6/3-6/30/93

Attain age 62 5/31/93 6/1/93 6/2-6/29/93

Age 62 throughout month 6/93 6/93 7/93

NRA month 5/96 6/96 6/96

Maximum reduction months

for early retirement 35 36 35

Because day of birth is suppressed for confidentiality reasons in

the HRS (discussed subsequently), respondents born on the lst or 2nd

of the month cannot be distinguished from other respondents.

If the 3rd of the month is a Saturday, Sunday, or federal holiday,

checks are dated and delivered on the first day preceding the 3rd of the

month if that is not a Saturday, Sunday, or federal holiday. For ex-

ample, if the 3rd is a Saturday or Sunday, checks are delivered on the

preceding Friday (SSA, Social Security Handbook 1997A, paragraph

145). (As a result of this procedure, beneficiaries can sometimes re-

ceive two payments in 1 month. In 1999, for example, January 3rd was

a Sunday, January 1st was a holiday, and the payment was made on

December 31st. Thus, beneficiaries could have received two payments

in December 1998—one on the 3rd and one on the 31st.) For ease of

exposition, this payment feature is not discussed in the text.

Benefit data derived from SSA records are available for consenting

HRS respondents on a restricted basis. However, these data do not

allow direct measurement of retirement benefit payments for HRS

respondent cohorts because the benefit data are for the period ending in

December 1991, and the HRS target cohorts do not begin to be eligible

for Social Security retirement benefits until 1993, when the oldest

cohort members become 62.

Covered earnings history data from SSA records are similarly

available on a restricted basis. They would establish which HRS re-

spondents are insured for Social Security retired worker benefits (as of

December 1991). While nearly all men in the HRS cohorts will be

insured, women’s status is more problematic. Without the restricted

data, estimates of the insured will be based on survey questions on

work history, including a special series in the 1996 interview designed

to aid in estimating years of covered earnings.

Day of birth is collected in the HRS, but it is suppressed for

confidentiality reasons.

Uncertainty arises principally because the day of birth is un-

known, individual processing times are unknown, and for those retiring

in 1997, filing dates are unknown. Uncertainty about an exact identifi-

cation of the pool who are “62 enough” to receive and report a benefit

will be greater in data from the 1998 and later interviews because of the

change in SSA payment dates.

In a small number of cases, retired worker benefits may be re-

ceived in the earliest possible month, but not result in permanent ben-

efit reductions. For example, workers may take a reduced retirement

insurance benefit (RIB) pending the outcome of a disability insurance

benefit application. If disabled worker benefits are later awarded retro-

active to the RIB entitlement, then retired worker benefits at age 65 are

not actuarially reduced. (However, retired worker benefits can be per-

manently reduced if a worker files for disability and elects retirement

benefits that start before the retroactive entitlement date to disability

benefits, for example, in the 5-month waiting period.)

In another small number of cases, those entitled for the earliest

possible month but who lose all of their first month’s benefits to the

earnings test are correctly not counted as early retirees.

Greater accuracy would be obtained by treating those interviewed

on the 1st or 2nd, and possibly the 3rd of the month—before the

payment date of the 3rd of the month—separately and allowing an

additional month for their first benefit report. For ease of exposition,

the text simply uses months.

As noted, persons filing for benefits after May 1, 1997, will no

longer be paid on the 3rd of the month, and one cannot determine

exactly which respondents in the 1935 cohort are affected. Early filers

among those born in the first half of 1935 (age 62 in 1997) probably

received their earliest benefit on the old SSA payment date of the 3rd of

the month, but others in the 1935 cohort probably received their first

benefit on one of the new payment dates of the 2nd, 3rd, or 4th

Wednesday of the month.

Note that if no adjustment is made for the additional payment lag,

a small decrease in the proportion of 62-year olds who appear to retire

at EEA may appear in the 1998 HRS data compared with similar pro-

portions in earlier HRS interview years.

This and the following HRS estimates assume that the earliest

possible report could occur in the second month following the 62nd

birthday month, do not correct for day of interview, and use

unweighted data. They also assume that birthdates of workers insured

for Social Security are distributed like those of other respondents.

Respondents with disability or widow’s benefits and those

reporting benefits prior to age 62 were excluded. They noted that

subsequent work would exclude those not insured for Social Security

benefits.

This estimate assumes that the earliest possible report could

occur in the second month following the 62nd birthday month, that

birthdates of workers insured for Social Security are distributed like

those of other respondents, and that “takers” do not differ from others

on these characteristics. The estimates do not correct for day of inter-

view, are from unweighted data, and still suffer from the inclusion of

noninsured persons in the denominator.

Social Security Bulletin • Vol. 62 • No. 3 • 1999

References

Burkhauser, Richard V., Kenneth A. Couch, and John W. Phillips.

1996. “Who Takes Early Social Security Benefits? The Economic

and Health Characteristics of Early Beneficiaries.” The Gerontolo-

gist, Vol. 36, No. 6, pp. 789-799.

Inspector General. 1998. “Report to Congress.” Final section in Social

Security Administration, FY 1998 Accountability Report, available

on-line at: <http://www.ssa.gov/finance/finance_intro.html>.

Juster, F. Thomas and Richard Suzman. 1995. “An Overview of the

Health and Retirement Study.” The Journal of Human Resources,

Vol. 30: Supplement, pp. S7-S56.

Social Security Administration. 1999. Program Operations Manual

Systems, CD-ROM (June).

_______. 1997A. “SSA’s New Payment-Cycling Initiative.” Social

Security Bulletin, Vol. 60, No. 2, p. 77.

_______. 1997B. Social Security Handbook, SSA Publication

No. 65-008 (August).

_______. 1997C. “SSA’s Accountability Report for Fiscal Year 1997,”

available on-line at: <http://www.ssa.gov/finance/

fy97_accountability.html>.

Svahn, John A. 1981. "Omnibus Reconciliation Act of 1981: Legislative

History and Summary of OASDI and Medicare Provisions.” Social

Security Bulletin,Vol. 44, No. 10, pp. 3-24.

Willis, Robert J. 1999. “Theory Confronts Data: How the HRS is

Shaped by the Economics of Aging and How the Economics of

Aging Will Be Shaped by the HRS.” Labour Economics, Vol. 6,

No. 2, pp. 119-145.