the incidence of payroll taxation: evidence from chile 09 级劳动经济学 张素蓉
TRANSCRIPT
The incidence of payroll taxation:Evidence from chile
09级劳动经济学张素蓉
Background
Fact: the increase in payroll taxation burden… OECD countries: 19% to 25% from 1965-88 Sweden: 6% to 40% by the late from 1950 - 1970s
Critique: increase labor cost, lower competitiveness, lead to unemployment?
Question: Is payroll tax really a cost for employers? Or, does payroll tax certainly lead to labor market
efficiency?
Traditional View:(w0, E0) TO (W1, E1)
The incidence :elasticity of DL and SL
Summers(1989):Tax/benefit linkage:
Where:
We get:
Three conditions where full shifting happens:1) Elastic demand2) Inelastic supply3) Full tax-benefit linkage( q = 1 & a = 0)
The study of incidence is important
• If full shifting exists, employer bears no increase in labor cost and there’d be no disemployment( No resulting labor market inefficiency)
• The incidence of payroll tax is subject to empirical studies!
• Reality is complicated: the minimum wage shifting may not be possible
The incidence of payroll taxation• past evidence is mixed… Brittain(1972):full shifting Holmlund(1983): 50% shifted
• problem: omitted variable bias “wage” –complicated other mechanism: the structure is strongly correlated with the structure
of its tax system(Summer, Gruber and Vergara, 1993)
• variance among states within a country Gruber(1991): insurance for workplace injuries – 80% shifted Gruber(1994): childbirth insurance – full shifting Concern: 1) value these two insurance more 2) more able to shift 3) US
specific
Natural experiment: privatization of social security program in chile
• Social security system in Chile: including pension, sickness/maternity, work injury, unemployment, family allowance
• different payroll tax rates for “white-collar” and “blue collar”• covered 60% labor force and 84% in manufacturing sector• privatization: all but work injury• government reimbursement for family allowance taxes after privatization• data: survey of manufacturing plants in Chile over 1979-1986 firm-level data( employees > 10)
Before 1981 After 1981
Pay-as-you-go Fully funded
employer 30% 5%
employee 17% 25%
Financing Employer’s contribution General revenue
Model
Model Basic regression specification:
Cross-sectional regression: difference in difference
Average difference estimator: the difference of the average over the 1979-1980 period and the 1984-1985 period
Model Problem:
1) Recession in the 1980s: wage decreased dramatically, unemployment increased, high inflation(25%)
2) Employers were mandated to give an 18% nominal wage rise due to the “privatization
3) The confounding influence of employee contributions: reduction in tax rate causes an increase in employee’s contribution
w may include employee’s contribution, when (dw/w)/dte would NOT be zero!
Model DDD: difference-in-difference-in-difference
Model Potential bias:spurious variation: bias toward full shifting
measurement error in wage bill - toward full shifting
measurement error in tax payments – toward no shifting
maximum taxable earnings – toward full shifting
missing variable: variation in industry risk
membership in different social security institutions
uncontracted hiring
Model
IV: Within-plant instrumental variables instrumented by the other group’s tax rate for each plant Grouping instrumental variables strategyinstrumented with the average tax rate for the industry or area groups for each plant
Conclusion
• No consequences for labor market efficiency from financing social insurance through payroll taxation in this specific case
• limit applicability because1) High inflation2) Downward rigidity for wages3) what causes the full shifting?
Thanks!