Democracy, Autocracy, and Bureaucracy

U. San Andrés Seminar August 5, 2008
Democracy, Autocracy, and Bureaucracy
Avinash Dixit, Princeton University, [email protected]
Background: to place this paper in broader context
Comparison of economic performance in democracies v. autocracies:
Prior favors democracy. [0] Generalized “western” belief.
[1] Rule of law. Decisions by clear rules and due process.
[2] Press freedom; informed citizens (Besley and Burgess).
[3] Political competition (Besley, Persson, and Strum).
[4] Participation, local knowledge (Bardhan, Rodrik).
But others find evidence mixed, or favoring authoritarianism:
Poor fit and U-shaped cross-country relationship (Barro).
Institutions matter less when actual policies controlled for (Glaeser et al).
Economic liberalization first, political later works better (Giavazzi-Tabellini).
Persson stresses effect of type of democracy matters
(presidential v. parliamentary etc.) but Keefer finds little difference.
1
Specific issue: provision of public goods broadly defined –
law and order, political institutions, central banking, regulation,
rules for common resources, health care, education, physical infrastructure.
State failure in this respect hurts incentives for complementary private
investment and effort, leading to more general economic failure.
Almost definition of state failure in political science (Rotberg)
Gradations – weak, failing, failed, collapsed – correlated with ethnic conflict.
Evidence that democracies provide more public goods (Lake-Baum).
But a kleptocratic or predatory state need not be a failing state:
Olson’s “stationary bandit” will optimally invest (provide public goods)
to run economy efficiently and maximize rent extraction.
Olson simply assumes that instrument of extraction is proportional tax;
this builds in distortion that suitable nonlinear tax could avoid.
Must explore better reasons for difference: to explain why
stationary bandit’s optimum more constrained than that of benevolent ruler.
2
Two such possibilities:
[1] If a predatory ruler provides more public goods, esp. education, institutions,
this may only serve to improve the population’s outside opportunities
and make it easier for them to oppose and topple the ruler.
This can be in the model, but main focus here is on a different issue.
[2] Ruler must act through bureaucrats; agency problem.
(a) Solution requires giving away information rent.
Actions distorted downward to reduce this rent-transfer –
suboptimally low provision of public goods.
(Standard Baron-Myerson, Laffont-Tirole mechanism design theory.)
Predatory ruler more reluctant to give up rent; accepts worse second-best
So distorts more, less of public good.
(b) Bureaucrats may care about social welfare (ethics, professionalism)
Possible assortative matching between ruler types and bureaucrat types:
Predatory ruler – selfish bureaucrat; benevolent ruler – caring bureaucrat.
Empirical – Evans-Rauch bureaucracy quality index and Polity III regime indices:
Correlation −0.24 with Autocracy, 0.19 with Democracy.
In right direction but not significant (need 0.29 with 31 DF.)
3
Model structure
Bureaucrat provides public good K, cost 12 γ K 2
Also collects fee F from citizen, remits R to ruler.
Possible dead-weight losses: collect (1 + λC ), (1 + λB ) resp. to deliver 1.
Representative citizen supplies private effort or investment L, cost 12 L2
Many bureaucrats and citizens – choice of scale. Interaction effects mentioned later.
Output Q = min(K, L) – perfect complements
Citizen chooses L = K; utility equal surplus SC = K − 12 K 2 − (1 + λC ) F
Participation constraint SC ≥ SC0 .
Bureaucrat’s surplus SB = F − (1 + λB ) R − 12 γ K 2
Bureaucrat may also be a citizen; SB is rent in excess of that.
Outside opportunity (participation constraint) SB0 .
Ruler’s surplus SR = R
Will consider two types of rulers:
Democrat tightly constrained, maxes total surplus UD = SC + SB + SR
Autocrat maxes own receipt SR = R
Alternative interpretation: ruler innately benevolent or kleptocratic.
Can think of benevolent dictators and democratically elected robbers.
4
Also two types of bureaucrats:
Selfish, maxes SB
Altruistic, maxes UB = SB + α (SC − SC0 )
Ideal first-best solution to maximize net total surplus:
K = L = Q;
max K − 12 (1 + γ) K 2
yields
KF B =
1
1+γ
Departures from ideal:
[1] Deadweight losses in transfers: (zero allowed as a special case)
λC citizen to bureaucrat, λB bureaucrat to ruler
This assumption allows comparison with Olson’s ad hoc assumption
and also helps in comparing predatory and benevolent ruler cases.
Agency issues considered one by one:
[2] Cost types γL < γH bureaucrat’s private information.
[3] Ruler cannot observe bureaucrat’s actions.
[4] Bureaucrat’s private information about his altruism α toward citizen.
5
1. Ruler has full information
Can dictate Ki , Fi , Ri for each case γi , i = L or H;
subject only to participation constraints.
Optimum level of K is
KF I =
1
1 + γi (1 + λC )
for i = L, H
whether ruler democratic/benevolent or autocratic/predatory.
This confirms intuition that Olson’s stationary bandit should run economy
efficiently unless available instruments have unavoidable dead-weight losses.
The only difference is in transfers: autocrat ruler sets R
as high as possible, leaving citizen and each type bureaucrat on their PCs
Democrat ruler sets R = 0, bureaucrat on PC; gives all surplus to citizens
6
2. Cost type is bureaucrat’s private information
Revelation principle (Myerson): outcome of any mechanism chosen by the ruler
is as if it comes from a direct revelation mechanism constrained by the requirement
that truthful type revelation is optimal for each type of bureaucrat.
Cost-type i bureaucrat’s incentive-compatibility constraint ICi:
Fi − (1 + λB ) Ri − 12 γi (Ki )2 ≥ Fj − (1 + λB ) Rj − 12 γi (Kj )2
and participation constraint P Ci requires LHS of this to be ≥ SB0 .
Results: P CH and ICL are binding; ICH and P CL are slack. Standard intuition:
Low-cost type has temptation to pretend to be high-cost and get bigger F
So this type must get rent 12 (γH − γL ) (KH )2 for truthful revelation
KL = K F I ; KH is distorted downward to reduce rent loss.
Democrat Ruler’s KH
Autocrat Ruler’s KH
1
1
θL
1 + γH (1 + λC ) +
(γH − γL ) (1 + λC )
θH
θL
1 + γH (1 + λC ) +
(γH − γL ) λC
θH
7
Democrat/benevolent ruler: no distortion if λC = 0.
Ruler regards taking rent from citizen to give to bureaucrat
as pure transfer; only reason to save rent is any DWL in transfer.
Autocrat/predatory ruler: distortion even if λC = 0:
ruler dislikes losing his own rent.
Numerical example (θL = θH = 12 , γH = 1, γL = 0.5, λC = 0.25, λB = 0):
Situation
KH
KL
Ideal, ignoring λC : KiF B
Full info with λC > 0, both ruler types’ KiF I
Democrat ruler; λC > 0 and agency
Autocrat ruler; λC > 0 and agency
0.500
0.444
0.421
0.348
0.667
0.615
0.615
0.615
Therefore difference between democrat and autocrat for reason of
agency (bureaucracy) should be expected to be quantitatively significant.
8
3. Bureaucrat’s choice of K unobservable
Ruler can force a choice of higher K by stipulating higher F
if he can observe violation of the citizen’s participation constraint.
Extractive ruler can achieve same outcome as when both actions observable.
But this keeps citizen on participation constraint, so benevolent ruler
cannot achieve same outcome; and can do better only if he can find
a sufficiently altruistic bureaucrat with parameter α, who will then
choose K = α/(α + γ) to max − 12 γ K 2 + α [ K − 12 K 2 ].
9
4. Bureaucrat’s altruism type unobservable
Different cases arise depending on type of ruler, nature of information,
and extent of bureaucrat’s altruism depending on α (1 + λC ).
Normal case: < 1; bureaucrat wants to take money from citizen.
Super-altruism: > 1; bureaucrat (NGO?) prefers to leave money with citizen.
(a) Autocrat ruler, normal altruism:
Citizen’s participation constraint is binding.
Bureaucrat’s altruism parameter irrelevant (within this range).
Revelation not needed; K = K F I with full or private info.
(b) Autocrat ruler, super-altruism:
Ruler does better to extract from NGO than from citizen.
NGO has to pay fee R that makes its participation constraint binding.
Then FA = 0, KA = α/(α + γ) so moral hazard not an issue.
Citizen is incidental beneficiary of ruler’s greed!
10
(c) Democrat ruler, normal altruism:
With full info, K = K F I regardless of α.
But F smaller when α higher so long as net total surplus > 0.
Then bureaucrat gains by pretending to be selfish.
Can achieve separation by (i) raise RS , FS , lower FA ; (ii) distort KS .
(i) may not be practical; can also hit limit FA = 0.
Then use (ii) and distort KS down as much as compatible with PC.
If selfish type too likely (ωS too large), too costly to distort.
Then ruler tolerates pooling; lets altruist type keep all surplus.
(d) Democrat ruler, super-altruism:
With full info, ruler wants altruist type to implement first-best K F B
If α > 1, the bureaucrat does not want to pretend to be selfish.
If 1/(1 + λC ) < α < 1, incentive to pretend at low end of this:
pretense would force larger F but K closer to preferred.
Interval likely to be too small to justify pedantic analysis.
So democratic ruler, altruistic bureaucrat may be a happy combination.
11
Extensions etc. for future work
1. If public good makes it easier for citizens, bureaucrats to revolt.
Replace citizen’s PC by SC ≥ SC0 + δC K; bureaucrat’s by SB ≥ SB0 + δB K.
This multiplies autocrat ruler’s optimal Ks by (1 − δB − δC ) in all cases.
Democrat ruler’s only by (1 − δB ) because he keeps citizen’s PC slack.
So this explanation of autocracy’s failure is easy to treat in the model.
Robinson has more complex model focusing on this aspect.
2. Combination of private information about γ and α:
Interesting to examine correlation: altruist may be less (or more) efficient
than selfish bureaucrat. Then which who wants to reveal / conceal type?
3. Ex ante competition among applicants for bureaucrat’s job.
May make revelation easier with less rent transfer, e.g. auction job.
Does this work better for one type of ruler than for the other?
4. Bureaucrats can be incentivized by career concerns instead of
immediate monetary rewards. Does this work better for one type of ruler?
12
5. Multidimensional cases:
(a) Many public goods with different degrees of observability
and possible complementarity in production.
(b) Different bureaucrats choose different public goods
(i) Yardstick competition; (ii) Externalities
(If perfect substitutes, like competitive contracting for private supply.)
(c) Different bureaucrats choose K and F
Problem ensuring participation constraint
6. Endogenize information structure
Of course ruler must use still other bureaucrats to find information
Raises issues of hierarchical agency, collusion
7. Ruler is actually a coalition; common agency problem.
Can generate feedback cycle – common agency weakens incentives, lower K,
if this affects groups unequally, distributive conflict worsens, etc.
Eventually ethnic conflicts etc, leading to failed or collapsed state.
Rotberg et al. suggest such dynamic process with two-way causation.
13
References
Bardhan, Pranab. 2005. Scarcity, Conflicts, and Cooperation: Essays in Political
and Institutional Economics of Development. Cambridge, MA: MIT Press.
Baron, David P. and Roger B. Myerson. 1982. “Regulating a Monopolist with Unknown Costs.” Econometrica 50(4), July, 911–930.
Barro, Robert J. 1999. Determinants of Economic Growth: A Cross-Country Empirical Study. Cambridge, MA: MIT Press.
Besley, Timothy, and Robin Burgess. 2002. “The Political Economy of Government
Responsiveness: Theory and Evidence from India.” Quarterly Journal of Economics 117 (4): 1415-1451.
Besley, Timothy, Torsten Persson, and Daniel Sturm. 2005. “Political Competition
and Economic Performance: Theory and Evidence from the United States.”
London School of Economics, working paper.
Evans, Peter and James E. Rauch. 1999. “Bureaucracy and Growth: A CrossNational Analysis of the Effects of ‘Weberian’ State Structures on Economic
Growth.” American Sociological Review, 64(4), October, 748–765.
Giavazzi, Francesco, and Guido Tabellini. 2004. “Economic and Political Liberalizations.” Bocconi University, Milan: IGIER Working Paper No. 264.
14
Glaeser, Edward, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer.
2004. “Do Institutions Cause Growth?” Journal of Economic Growth 9 (2):
271-303.
Keefer, Philip. 2004 a. “What Does Political Economy Tell Us About Economic
Development and Vice Versa?” Annual Review of Political Science 7: 247-272.
Laffont, Jean-Jacques and Jean Tirole. 1993. A Theory of Incentives in Procurement
and Regulation. Cambridge, MA: MIT Press.
Lake, David A. and Matthew A. Baum. 2001. “The Invisible Hand of Democracy: Political Control and the Provision of Public Goods.” Comparative Political Studies
34: 587–621.
Myerson, Roger B. 1982. “Optimal Coordination Mechanisms in Generalized PrincipalAgent Problems.” Journal of Mathematical Economics 10, 67–81.
Olson, Mancur. 1993. “Dictatorship, Democracy, and Development.” American Political Science Review 87(3), September, 567–576.
Persson, Torsten. 2005. “Forms of Democracy, Policy, and Economic Development.”
Cambridge, MA: NBER Working Paper No. 11171.
15
Robinson, James A. 2001. “When is a State Predatory?” Working paper, Harvard
University.
Rodrik, Dani. 2000. “Institutions for High-Quality Growth: What They Are and
How to Acquire Them.” Studies in Comparative International Development 35
(3): 3-31.
Rotberg, Robert I. 2004. “The Failure and Collapse of Nation-States: Breakdown,
Prevention, and Repair.” In Robert I. Rotberg (ed.) When States Fail: Causes
and Consequences. Princeton, NJ: Princeton University Press.
16