Topic 1 - Empirical and policy aspects of labour supply Professor Christine Greenhalgh P Cahuc and A Zylberberg (2004) Labor Economics, Chapter 1 Labor.

Topic 1 - Empirical and policy aspects of labour supply

Professor Christine GreenhalghP Cahuc and A Zylberberg (2004) Labor Economics,

Chapter 1 Labor Supply, part 2. D Bosworth, P Dawkins and T Stromback (1996)

The Economics of the Labour Market Chapter 5 A Manning (2003) Monopsony in Motion, Chapter 4:

The Elasticity of the Labor Supply Curve to an Individual Firm.

T Boeri and J van Ours (2008) The Economics of Imperfect Labour Markets, Chapter 5: Regulation of Working Hours; Chapter 7: Family Policies.

Estimating aggregate labour supply elasticities – why useful?

• Future planning/projection of employment and unemployment under changing policy

• Responses to fiscal policy affecting net pay– Will a more progressive income tax affect

average hours of work?– Does income support for lone parents affect

decisions to participate in work?

• Policies towards working hours, retirement– Should we relax mandatory retirement?

Trends in Male and Female Participation (US)

Patterns of work in G5 countriesSource: Boeri and van Ours Table 5.2

1955

Annual

2005

Annual

%

Change

Weekly hours

Weeks

(2002)

France 2040 1434 - 12.1 36.2 40.5

Germany 2265 1437 - 16.6 36.5 40.6

Japan 2081 1775 - 6.1

UK 2156 1624 - 10.6 38.2 40.5

US 2030 1790 - 4.8

Weekly working hours of women and men aged 18-64 in Britain

Source: G Paull in Economic Journal Feb 2008

No Kids

Pre- Sch

Sch

Age

Post

Kids

W M W M W M W M

Hours per week

38 43 25 47 28 47 33 47

% FT

(>31h) 85 91 34 96 41 97 58 96

Basic labour supply theorySupply of work responds to the hourly wage but sign

of this effect is ambiguous because– Substitution effect is positive (higher wage

leads to more work)– Income effect is negative (higher income leads

to less work and more leisure) – So may observe ‘backward-bending’ supply

curve if plotting Wage v Hours Worked– Or ‘hump-shape’ if plotting Hours Worked v

Wage (Cahuc and Zylberberg)

Econometrics - Issues with Data and Estimation

Necessary variables: hours of work, hthe individual’s hourly wage, wincome other than the wage, Rvector of personal characteristics, θ(e.g. married, children)

Estimate:

ln h = αw ln w + αR ln R + x.θ + ε

Tricky bits in relation to the wage elasticity of hours

• Hours and wages are not observed for those choosing zero hours of work

• Observed ε is a random error but without a fully observed distribution (observe all positive elements but not larger negative ones)

• If estimate by OLS this gives biased estimate as violates basic assumption of statistical model

• Can use estimation techniques that deal with these ‘truncation biases’

• Can estimate jointly a model of decision to participate and hours worked

• Then have to estimate a potential wage for all those who chose not to work (based on their θ)

Tricky bits in relation to the ‘other income’ elasticity

• Other income R is f(wealth) so depends on the age/ stage of life cycle of person, past job history and savings, not all exogenous

• Again OLS model relies on lack of correlation between RHS variables and the error term ε, so more biases– Use a more complex inter-temporal model– Replace R with an estimate of MU of wealth– Take first differences of equation (panel data)

Tricky bits in relation to the tax structure

• Both the net wage received and the level of income other than the wage are affected by taxes and benefits

• Replace w with w(1- t) and R with R+S• Those with high w.h will have a high t and a

low S and vice versa• Problem for estimation is that each person

has chosen where on these schedules to put themselves by working more or less hours

Priors about relative wage elasticity for men v. women?

• Suppose labour supply is backward bending

(hump–shaped in hours v wages)

• Know that men tend to earn more than women per hour due to more continuous work experience

• Expect men to be on the flat part of curve (i.e. zero wage elasticity?)

• Expect women to be on the upward sloping part (i.e. positive wage elasticity?)

Aggregate supply elasticities for married women and men

Source: Cahuc and Zylberberg Tables 1.1 & 1.2

MW

wage

MW

Income

MM

wage

MM

income

US 0.97 to 0.99

- 0.12 to - 0.33

0.0 to

0.05

0.0 to

- 1.03

UK 0.09 to 2.03

- 0.2 to - 0.4

0.02 - 0.29

EU

(miscel)

0.05 to 1.00

-0.2 to

- 0.3

0.08 to 0.12

- 0.01 to

- 0.04

Uncompensated and compensated wage elasticity

Source: Bosworth, Dawkins & Strombach Tables 5.1, 5.2

Women Women Women Men Men Men

Wage Pure Subn

Income Wage Pure Subn

Income

Ist

Gen

0.2 to

0.9

0.1 to

2.0

- 0.1 to - 0.2

0.0 to

- 0.4

0.0 to

0.4

0.0 to

- 0.2

2nd Gen

0.6 to 1.1

0.7 to 1.2

- 0.1 to

- 0.2

- 0.2 to - 0.5

0.1 to 0.2

- 0.1 to

- 0.4

Interpretation of trends using wage and income elasticities

• Over time men’s and women’s wages have increased with productivity growth

• Women’s wages have increased relative to those of men in advanced countries

• At the same time average non-wage income from assets has tended to increase (more home owners and more saving for retirement)

• Prediction using sum of wage and income elasticities is that men will work less and women will work more – fits the facts

Implications for income taxation • Most countries use progressive rather than

proportional taxation

• Larger wage elasticities for women suggest higher proportional responses in female labour supply (absolute responses similar, as noted in Bosworth et. al)

• Estimates (Sweden) of effect of existing taxes - % decrease in hours worked - cp. with: Proportional No tax Lump Sum

Males -6.2 -13.4 -13.6

Females -9.3 -23.0 -23.3

Implications for income support• Many countries support incomes of the low paid

e.g. supplementary income payments or tax benefits for families with children

• Historical problem of disincentive to work or to increase hours if working

• UK has moved strongly to Welfare to Work approach to avoid trade-offs in participation

• System makes benefits higher if in work rather than non-participant

• Lower marginal rates of withdrawal of benefits as weekly earnings rise

Estimating labour supply to firmsWhat are the issues?

• Competition v Monopsony – which model fits labour supply to firms?

• If firm has no monopsony power then it pays going wage – elasticity of labour supply is infinite to the firm

• General presumption is that men are mobile between firms, but women are less so

• If true, this reverses above estimates, with higher labour supply elasticity to firms for men

• Need to know for estimating effects of minimum wages and union bargaining

Monopsony evidence? • Employer size wage effect – consistent

empirical evidence that larger firms pay higher wages

• Competitive labour market => firms employing more or fewer workers should pay same wage

• Other features can explain size-wage effect- Higher labour quality in large firm- Compensation for disamenity of size- Rent sharing by larger profitable firm

Manning’s estimates of labour supply elasticity to firms

• Regressions of employer size on wages• Controls for personal characteristics,

education, region, industry and occupation• In US 1.0 to 1.6; in UK 1.9 to 2.7

(Manning Table 4.5 col. 3) • More complex estimation using model of

separations into other jobs and non-employment

• In US 0.7 to 1.4; in UK 0.8 (Manning Table 4.10 row 3)

Implications of these supply elasticities

Under monopsony:

Wage = [1/(1+e)] . MPL

e = 1/ αw

where αw is supply elasticity to firm

(Perfectly comp. αw infinite and wage = MPL)

UK most elastic case is αw = 2.7

Wage is only 73% of MRP

Lower values of αw give worse ratios

Even Manning thinks these αw are too low!

Firm Supply Elasticity Differences between Women and Men?

• Heterogeneous preferences for ‘leisure’? In truth much leisure is spent in home work

• Women have greater compar. advantage in home work? Combining home care and work leads to constraints on job mobility

• Travel-to-work times are lower for women than men and lowest for women with kids

• Motivation for work differs – men place higher value on monetary reward; women more influenced by non-pecuniary factors

Policy aspects (1)Regulation of Working Hours

European Working Time Directive: • limit of average 48 hours a week which a worker can be

required to work (can voluntarily work more if want to) • limit of an average of eight hours’ work in 24 which

nightworkers can be required to work • right to at least one day off each week • right to a rest break if working day is longer than six

hours • right to 11 hours of rest per day • right to four weeks of paid leave per yearNB Table 5.1 of Boeri and van Ours is out of date re UK

Policy aspects (2)Work-Life Balance in Families

• Should there be more opportunities for parents to work shorter hours?

• Are women happy to work part-time?

• Are men happy to work part-time?

• What jobs are available to part-timers?

• What can legislation do to change what is on offer?

Part-time Employment in G5Source: Boeri and van Ours Table 5.3

% PT % PT Invol. PT %

M W M W

France 5 23 53 39

Germany 7 39 18 13

Japan 14 42 19 4

UK 10 39 24 10

US 8 18 7 8