Effort under taxation: an experimental inquiry

1

• Effort under taxation an experimental inquiry
• Guido Ortona Stefania Ottone Ferruccio Ponzano
  Francesco Scacciati
•  
• () Università del Piemonte Orientale
• () Econometica, Università di Milano Bicocca
• () Università di Torino

2
Related Literature

• Workers effort decreases in presence of taxation
  
• Experimental Evidence ? Sutter and
  Weck-Hannemann (2002) Lévy-Garboua (2005)
  Masclet et al. (2005) Swenson (1988) Sillamaa
  (1999)
• BUT the tax revenue was lost for tax payers or
  assigned to other non-deserving subjects ? no
  relevance of the use of the tax revenue
  (Leviathan notion of State)
• Our previous experiments (WP 77 of the Dep. POLIS
  of the Università del Piemonte Orientale, Oct.
  2006) if the tax revenue used to produce a
  welfare state, the effort does not decline

3
Our previous experiment

• Relation between effort and taxation in presence
  of a non-Leviathan state universalistic
  insurance and other universalistic public goods
  compared with a State of Nature where there is
  no State.
• Participants were requested to carry out a
  secretarial task
• Two contracts were submitted, one corresponding
  to the State of Nature (SN) and the other to the
  Welfare State (WS)
• State of Nature ? there is the risk of loosing
  the whole wage or half of it, and there are no
  taxes, insurance and public goods
• Welfare State ? there is the same risk of loosing
  the whole wage or half of it, a proportional
  income tax, a (high but not full) insurance and a
  public good
• Note we avoided terms like "welfare state",
  "cooperation", etc.

4
Possible biases

• Our result is coherent with the theoretical model
  of Bird (J. of Public Economics, 2001) the
  welfare state may enhance productivity due to
  risk reduction
• However, there are possibly two biases in our
  results
• a) Cultural bias - Are the results due to the
  cultural and/or political orientation of the
  subjects?
• b) Lab bias - are the results generalizable to
  real world?
• To control for a) a subset of the European
  Value Survey added at the end of the experiment
• To control for b) the experiment was replicated
  in a questionnaire ("What would you do for real
  if.."), with different subjects

5
Experimental Design

• Experiment ? WS and SN as in our previous
  experiment questions from the European Value
  Survey
• Questionnaire ? virtual experiment questions
  from the European Value Survey

6
Experimental Procedure

• Overall, 123 subjects. 31 subjects in the
  Experiment and 92 in the Questionnaire
• Each task is paid 1 euro (before taxes, if
  present)
• 50 fine for those who carried out less tasks
  then those freely established
• Control Questions
• Strategy method ? subjects decide the number of
  tasks they want to carry out under each contract,
  and are informed that the assignment to one of
  the two contracts will take place, randomly, only
  after that
• Two-thirds of the subjects work is ruled by the
  WS contract, one third by the SN contract (this
  because the marginal contribution should be as
  small as possible)
• After choosing the number of tasks, but before
  the assignment of the contracts, subjects are
  requested to state their preference for one of
  the two contracts
• No time constraint in the Experiment
• Show-up fee of 8 in the Experiment (but
  participants were told at the end), while the
  Questionnaire was expressly paid 10

7
Results (I)

• 1) Lab bias Not present

Number of Tasks Number of Tasks Number of Tasks
gt SN gt WS SN WS
Experiment (N 31) 7 7 17
Questionnaire (N 92) 31 29 32
Chi-square test (P 0.14)
8
Results (II)
2) Political bias
POSITION EFFORT (experiment) EFFORT (experiment) EFFORT (experiment) EFFORT (experiment)
POSITION gt SN gt WS SN WS Total
Left 2 0 4 6
Center 0 4 5 9
Right 3 2 4 9
Total 5 6 13 24
Fisher-exact test P 0.18
9
Results (III)
POSITION EFFORT (questionnaire) EFFORT (questionnaire) EFFORT (questionnaire) EFFORT (questionnaire)
POSITION gt SN gt WS SN WS Total
Left 2 11 10 23
Center 12 10 13 35
Right 17 8 8 33
Total 31 29 31 91
Chi-square test P 0.022
10
Results(IV)
OLS regression, robust standard error dependent
variable effort Variable Coefficient
(robust std. err.) Trust Institutions -1.4 (
factor 1) (12.8) Welfare state 23.73 (factor
2) (9.46) Int. Institutions -18.45 (factor
3) (11.93) Meritocracy 8.4 (factor
4) (10.8) Gender 25.19 (35.06) Age -0.18
(4.76) Constant -14.92 (95.42) N 92 F
2.27 significance at 5
11
Main Conclusions

• Main conclusions
• a) It is confirmed that a welfare state financed
  by taxation does not reduce workers effort (and
  therefore the overall output)
• b) The coherence between the experiments and the
  questionnaire confirms the validity of both
• c) The political orientation matters in the
  questionnaire but in the experiment its time
  that mostly affects subjects behaviour

12

• Thank You