A Century of Work and A Century of Work and LeisureLeisure

by by

Valerie A. RameyValerie A. Ramey

andand

Neville FrancisNeville Francis

Has Leisure Increased Over the Last Century?

Keynes (1930)Economic Possibilities for Our Grandchildren

Lebergott (1993), Greenwood & Vandebroucke (2005): leisure has increased dramatically over the last century

Prescott (1986) and DGE modelsNo secular trends in leisure

Annual Hours Worked Per Worker in the U.S.(Maddison’s data)

1500

1900

2300

2700

1900 1920 1940 1960 1980 2000year

Annual Hours Worked in Business(Divided by Civilian Non-Institutional Population Ages 16+)

1000

1200

1400

1600

1900 1920 1940 1960 1980 2000year

New Measures of Leisure Per Capita

New measure of “per capita”Entire population

Comprehensive measure of non-leisure time- Work-for-pay hours- School hours- Home production hours

How Should We Measure “Per Capita?”

Standard Measure of “Per Capita”

• Civilian non-institutional population

= total non-institutional population ages 16 and over – armed forces.

• Justification?

Notion of “available workforce”

Why Not Use Total Population?

00 ),(t

tttt NlcUE

Theoretical Basis:

Standard model with explicit population growth

Choose consumption ct and leisure lt to maximize:

Empirical Basis:

The consumption of children is counted in c. Why don’t we count their leisure in l?

where Nt is total population

Importance of Accounting for Children

Consider a model with perfect substitutability of consumption and leisure of adults and children in household utility.

Let c1 = per capita consumption by children, c2 = per capita consumption by adults

h1 = per capita hours worked by children, h2 = per capita hours worked by adults

θ = fraction of population that is children

The representative household maximizes:

subject to:

and h1, h2, c1, c2 0.

If w1 < w2, then it is optimal to set h1 = 0 and

Increase in the fraction of children leads to an increase in h2, hours per capita of adults.

Thus, adult time use is affected by the presence of dependents with lower productivity.

])1(ln[])1(ln[ 2121 hhTccU

212211 )1()1( cchwhw

)1)(1(2

Th

Population Ages 0-15 as a Fraction of Total Population.2

.25

.3.3

5.4

1900 1920 1940 1960 1980 2000year

Population Ages 65+ as a Fraction of Total Population.0

4.0

6.0

8.1

.12

1900 1920 1940 1960 1980 2000year

Comprehensive Measures ofNon-Leisure Time

Work for pay (including government) Commuting time

School Hours Home Production

htstctmtttt HHHHTLwhereLCUU ),(

What is Leisure?

Hawrylshyn (1971) distinguishes leisure from household work by defining household work activities as “those economic services produced in the household and outside the market, but which could be produced by a third person hired on the market without changing the utility to members of the household.”

Ratings of Activity Enjoyment – 1985 (From Robinson and Godbey Appendix O)

9.3 Sex 6.99.2 Play sports 6.89.1 Fishing 6.7 Second job

9 Art, music 6.6 Cook, work at home, shop8.9 Bars, lounges 6.58.8 Play with kids, hug and kiss 6.4 Child care, help adults8.7 6.3 Work commute8.6 Talk/read to kids 6.28.5 Sleep, church, attend movies 6.1 Dress8.4 6 Pet care, classes8.3 Read, walk 5.9 Errands8.2 Work break, meals out, visit 5.8 Housework8.1 5.7

8 Talk with family 5.67.9 Lunch break 5.5 Home repair, grocery shopping7.8 Meal at home, TV, read paper 5.47.7 Knit, sew 5.3 Homework7.6 5.2 Pay bills, iron7.5 Recreational trip 5.17.4 5 Yardwork7.3 Hobbies 4.9 Clean house, dishes7.2 Baby care, exercise, meetings 4.8 Laundry7.1 Gardening 4.7 Child health, doctor, dentist

7 Work, homework help, bathe 4.6 Car repair shop

Accounting for Hours Worked for Pay

The standard RBC measure excludes hours worked in government (civilian and military).

Is that important for trends?

Government Hours as a Fraction of Total Work Hours0

.05

.1.1

5.2

.25

1900 1920 1940 1960 1980 2000year

Measuring Total Hours

Includes private hours (establishment, self-employed, unpaid family workers) plus government hours

Use Kendrick data for early period

Use BLS private hours index upweighted by BEA full-time

equivalent employment numbers

New Estimates of Annual Market Hours Per Worker

1800

2100

2400

2700

1900 1920 1940 1960 1980 2000year

Commuting Time• Time diary estimates from 1965 – 2003 suggest commute times are

a relatively constant 10 % of hours worked. • Scant evidence early in the century

- Average commute distances for shorterurban workers, farmers

- But modes of urban transportation were slower- Hours per worker, days per week

• Rodrigue (2004) argues time spent commuting for urban workers

was relatively constant over 20th Century • We assume commute time is 10% of hours worked for entire

century

Accounting for Hours Spent in School

School Enrollment Rates

025

5075

100

1900 1920 1940 1960 1980 2000year

High School Enrollment Rate for Ages 14-17

100

130

160

1900 1920 1940 1960 1980 2000year

Average Days Attended per Enrolled Student, K-12

020

4060

1900 1920 1940 1960 1980 2000year

Higher Education Enrollment Rate for Ages 18-24

Estimating School Hours

Annual school hours

= (enrollment in grades K – 8 ) ∙ (avg. days attended by enrollee) ∙ 5.5 hours

+ (enrollment in grades 9 - 12 ) ∙ (avg. days attended by enrollee) ∙ 7 hours

+ {(enrollment in college) ∙ [(fraction full-time)+ 0.3 ∙ (fraction part-time)] ∙ 165 days ∙ 8 hours}

Annual Per Capita Hours Spent on School for Ages 5-22

300

500

700

900

1900 1920 1940 1960 1980 2000year

Hours Spent in School as a Fraction of Total Market Work.1

.15

.2.2

5.3

.35

1900 1920 1940 1960 1980 2000year

Accounting for Hours Spent in Home Production

Conventional Wisdom

“The diffusion of household utilities and appliances dramatically reduced the hours spent in household chores.”

Estimating Home Production Hours

• We use data from time diaries when possible, since they are considered the most reliable measure of both market work and home production hours

• Strategy:

(i) gather time diary estimates by sex-age-employment status cells

(ii) interpolate between years for each cell(iii) weight cell by fraction of population in that cell.

Estimates of Hours of Housework per Week by Nonemployed Women Ages 18-64

3040

5060

70ho

urs

per w

eek

1900 1920 1940 1960 1980 2000 2020year

housewives nonemployed

Are the Early Studies Representative?

Samples were not nationally representative Urban samples tended to have above average

income But most samples were rural, which had less access

to electricity, market goods, etc. Evidence suggests that poor urban households did

not do more housework – “being poor meant being dirty”, relied on “bakery bread.”

Bryant (1996) adjusts for non-representativeness. Our estimates are consistent with his.

Why Didn’t the Diffusion of AppliancesReduce Housework?

• Appliances replaced low-wage immigrant labor

• Decline in “maiden aunts”% of nonemployed women living in other’s house with no children of own: 18% in 1900, 7.6% in 1960.

• Cross-section and time series studies on appliances: more appliances lead to more household production output

• Betty Friedan (1963) The Feminine Mystique • Mokyr (2000): Revolution in sanitation, germ theory of disease and

nutrition theory increased demand for cleanliness just as appliances were diffusing

Estimates of Housework by Employment & Sex Category (ages 18-64)0

2040

60

1900 1920 1940 1960 1980 2000year

nonemployed women employed womennonemployed men employed men

Children’s Home Production

Estimates from the 1920s are similar to those from the 1980s:

Ages 5-14: 3 hours a week

Ages 15-17: 5 hours a week

Average Weekly Hours of Housework of Adults

010

2030

4050

hour

s pe

r wee

k

1900 1920 1940 1960 1980 2000year

females males

Ages 18-64, by Sex

010

2030

hour

s pe

r wee

k

1900 1920 1940 1960 1980 2000year

ages 18-64 ages 65+ages 0-17

Both Sexes, by Age

Hours Per Capita in Various Activities

700

800

900

1000

1100

annu

al h

ours

1900 1920 1940 1960 1980 2000year

Market Hours

010

020

030

040

0an

nual

hou

rs

1900 1920 1940 1960 1980 2000year

Commute Hours (Market and School)0

100

200

300

400

annu

al h

ours

1900 1920 1940 1960 1980 2000year

School Hours

700

800

900

1000

1100

annu

al h

ours

1900 1920 1940 1960 1980 2000year

Home Production Hours

Per Capita Market-Oriented Hours

600

800

1000

1200

1400

1900 1920 1940 1960 1980 2000y ear

market market plus school

Measuring Leisure

Time endowment is 24 hours per day, 365 days per year

Most personal care time ranks high on enjoyment index (sleeping, eating), so we do not subtract it from leisure

Personal care time is relatively constant at 75 hours per week

Annual Leisure Hours Per Capita

6400

6500

6600

6700

6800

1900 1920 1940 1960 1980 2000year

Conclusions

New measures suggest leisure per capita now is about equal to leisure per capita in 1900

Our results are different from the standard ones because we track the leisure of the entire population and we don’t count schooling as leisure.

Keynes prediction has not come true yet for the US