example of tpe

download report

Transcript example of tpe

XXI Encuentro de Economía Publica
Girona 30 - 31 Enero 2014
The social cost of non-compliance and the optimal
punishment of corruption.
Can European policy measures be helpful?
Bruno Bosco (DEMS) and Margherita Savona (DSG)
University of Milan-Bicocca
Milan, Italy
1
Motivations 1
•An estimated 120 billion Euros per year, 1% EU GDP, is
lost because of corruption. Similar phenomena exist
everywhere and since long time. Empirical works show the
damages to GDP, GDP growth, investment and overall
budget policy.
•Economic theory initially saw corruption as “oil in the
wheels” of economic systems and as a form of self-defence
of mkts against burdensome public regulation.
•Myrdal (1968) and Rose-Ackerman (1975, 1978) showed
that corruption is “sand and not oil in the wheels” of
economic systems
2
Existing economic theory
• It generally analyses corruption/bribing decisions as a two-party utility
maximization choice under uncertainty
• Government sets the rules of conduct for privates, the fines for non
complying subjects and somehow affects the prob. of discovering a
corrupt transaction
• The value of the pay-off from the corrupted deal (e.g. the evaded tax
yield, the exact fee requested at full compliance, etc.) is supposed to be
known to the two parties but it can be unknown to the Government
• Private compliance costs are not generally considered.
• The main theoretical results (concavity of the penalty function,
differences in fines between corrupted parties, properties of wages
structure of public officers, etc.) are frequently case specific
3
Possible Frameworks
(Relationship among parties involved)
Current approach
Private Subject vs Public Officer
Privates bargain to max a Pay-off
given E[penalty] = Prob(discovery)
× fine
All information is equally shared by
both parties
Results:
NE about the bribe and the
“optimal” level of non compliance
Government deduces the best
measures to curb corruption
No incentives to comply; only
repression
Alternative approach
Private Subject and Public Officer
vs Government
Pay-off ex-ante uncertain conditional
to government regulation
There is asymmetric information
between Gov. and Agents about
compliance costs and/or behaviour
Privates and Officers form NE to be
used by Gov. in P-A model with 1 P
and 2 potentially colluded As to
determine fines and regulation in a
welfare max framework
Gov. should simultaneously search
for efficient regulation, incentives
4
and fines
Purposes of this paper
i) Present a simple but general welfare maximization model of optimal
regulatory activity generating compliance costs to private subjects
who, under the hypothesis of asymmetric information with G, may
resort to bribing public officers.
ii) Derive conditions for efficient costly regulation as well as the
properties of the fine structure to be imposed to both privates and
officers.
iii) Discuss the main EU policy guidelines against corruption in the
lights of the above theoretical results
5
Model Assumptions
H1 Government:
U G  V ( x(r ))   X  x(r )  D  r  w
where V(.) = public benefit from compliance, r = expenditure for instruments to implement regulation, X =
requested compliance, x(r) = actual compliance, D = damage from incompliance, V’x > 0 [Example: X =
number of filters; x = filters actually deployed; r = expenditure for chemical analyses of air pollution; D =
health damages from pollution]
H2 Private individuals:
U P  U  E[C]x(r )
where C is the compliance cost
UB  w
H3 Officers:
where w is the wage
H4 C is a r.v.
C
F[C] with f (C) / [1  F (C)] nondecreasing in C [C, C ]
6
G is tolerant and benevolent. He wants to determine the optimal r for any X but he
does not want to impose “excess” full compliance costs to privates.
G wants full compliance (state 1) if V(X) – r  CX, i.e. when the public advantage
is greater than or equal to the full compliance cost. G tolerates some non
compliance (state 2) when V(X) – r < CX.
Evaluate F [C ] at C = [V(X) - r] /X and define
F [C ] = probability of state 1
1 – F [C ] = probability of state 2
We analyse two situations
FB: costs are ex-post observable
SB: G does not observe realized costs and employs perfectly informed public
7
officers
Under FB G solves different variants of the following model
Max r E[W State1] = F éëC%ù
û[V ( x(r )) - ( X - x(r ) )D] +
U-
C%
x(r ) òC Cf
(C )dC - r - w(r )
s.t. F [C%]U P x(r ) = X ³ 0
Different definitions of D; different wage structures, different properties of V(.)
Results for different definitions of the damage and w (full ex-post observability)
• x = X conditional of costs below cut-off value. Then no corruption when
costs below cut-off
• No bribes and no fines
But
• Always under investment in r w.r.t cost certainty (G wants to avoid to impose
excess costs to privates)
• Extent of underinvestment may depend on the definition of D
• Officers’ wage have a fix part plus an incentive depending on F[C%] and the
gain generated by r
9
COSTS ARE NOT OBSERVED BY G who wants to max E[W]
Privates and Officers can bargain. P does not fully comply and B accepts a bribe b for
underreporting. Define
Penalty for P  Residualcompliance  C ( X  x (r )) f P
Penalty for B  (1   )b
E[U P ]  (1  )[U  Cx(r )  b]  [U  Cx(r )  b  C( X  x(r ))(1  f P )]
E[U B ]  (1  )[ w  b]  [ w   b]
As in previous literature, NE about b generates
b (C, x(r ), )
*
There exists C* > 0 such that b = 0 and C* used to define E[W|C ≥ C*]
10
COSTS ARE NOT OBSERVED BY G
G:
maxr
E[W|C ≥ C*]] = [V(x(r)) – (X – x(r ))D – r – w(r)] Prob[C ≥ C*]
= [1 – F(C*] [V(x(r)) – (X – x(r ))D – r – w(r)]
= [V ( x( r )) -
s.t.
PC
IC
( X - x(r ) D - r - w(r ) ]
C
òC*
f (C )dC
E[UP] ≥ 0, E[UB] ≥ 0
dE[U*P] / dC* = – x(r) – Θ(X – x(r ))(1+ fP) < 0
Solving the P-A model we find r*, fP*.
Properties of the fine and wage structures
11
COSTS ARE NOT OBSERVED BY G
•
No underinvestment in r
Penalty:
1 - Q h(C ) é '
êVx + D fP =
+
Q
Q êë
æ1 + wr' öù
÷
çç
ú
÷
çè xr' ÷
øú
û
•
fP = (1 – Θ)/Θ with cost certainty: fP (Θ/ (1 – Θ)) = (UP’(Bad-State) / UP’(Good-State))
•
decreasing in Θ: the higher the prob. of discovering corruption the smaller the fine can be
(general result); non-decreasing in C and increasing in the hazard rate (price regulation theory)
for
Vx'
æ1 + wr' ÷
ö
ç
+ D> ç ' ÷
çè xr ÷
ø
•
Increasing in the measure of the gain generated by regulation if > 0 (general result made more
explicit)
•
But since (V’x + D) interacts with dx/dr the net effect depends on the sign of dx/dr, i.e. on
privates’ “reaction” to regulation (new result to be tested)
•
fP > α: Privates proportionally punished more than officers (result already obtained by some
other authors)
12
EUROPE
• Public procurement The current EU legal framework on public procurement does not include
specific provisions for incentive to “compliance”. European Procurement Passport
• Privatization and selling of public property Our results show that corruption in
privatization may result from the way in which the privatized markets in which firms will operate
are regulated.
• Privates may win the “competition for the field” by using for corruption purposes the resources that
they expect from the “competition within the field” that follows the privatization procedure. This
will reduce the advantage (if any) of privatization.
• Accounting standards The procedures on statutory audit were harmonised, introducing a
requirement for external quality assurance, provisions on public supervision, duties and
independence of statutory auditors and the application of international standards. These measures
increased the credibility, quality and transparency of financial reporting, reducing the risks of
corruption
13
Pooled OLS and Fixed Effects Estimates. T = 10; N = 31 Dependent variable CPI
OLS
Regressors
Coefficient
FE
Prob.
Coefficient
Prob.
1) C
4.465317 (12.91669)
0.0000
2.321507 (6.289525)
0.0000
2) Law
1.601130 (8.776411)
0.0000
1.055597 (5.771911)
0.0000
3) Regulation
-0.273752 (-1.111285)
0.2673
0.644212(8.645114)
0.0000
4) TPE (Total Public Expenditure)
% GDP
-0.013866 (-1.910077)
0.0571
– 0.002385 (– 0.271791)
0.7860
0.008284 (2.722872)
0.0069
1.674330 (4.412740)
0.0000
0.023336 (5.98129)
0.0000
5) (TPE)*(GPE)
6) Education level
0.035686 (2.16397)
0.0001
7) Rel1 (Catholic/Orthodox)
-0.642405 (-2.611241)
0.0095
8) Rel2 (Protestants)
-0.211286 (-0.771637)
0.4409
8) Local Expenditure % of TPE
0.035678 (2.89234)
0.0000
9) Gov. Policy Effectives (GPE)
1.395484 (7.649022)
0.0000
14