Political economy of reform

Document Sample
Political economy of reform Powered By Docstoc
					Political economy of reform

   Economics of Transition
       The importance of political
• Political constraints appear to have played a
  major role in the actual transition process in all
  transition countries, especially in blocking parts
  of the restructuring process.
• Some examples:
   – Plans to cut subsidies to SOE‟s were canceled;
   – Fiscal subsidy cuts to firms transformed in in hidden
     subsidies (bank credit and interenterprise arrears);
   – Heavy worker resistance to closing inefficient state
Introduction to Political economy of
           reform (PER)
• Political economy of transition belongs to a more
  general field that tries to integrate political
  science & economics.
• Also there is a general literature on political
  economy of reform summarized by Dani Rodrik
  JEL „94) with the main conclusion: not
  necessarily so that authoritarianism allows for
  more sustained economic reforms, it is rather the
  existence of a deep crisis that forces government
  to really reform).
  Introduction to Political economy of
          reform (PER), cont.
• Political economy arguments are at the center of discussion
  about speed and sequence of reform (big bang vs.
       • big bang  “window of opportunity”
       • gradualism  “build on reforms”
• There is a trade off between a big bang strategy where
  reforms are proposed simultaneously and gradualism where
  they are realized in “sequence” and the expected outcomes
  as well as the realization of reforms will also depend on the
  sequencing of reforms.
• In what follows we always assume the existence of
  complementarities between reforms.
• Which type of approach can be seen as desirable depends
  also on the type of uncertainty that characterizes the process
  (individual uncertainty vs. aggregate uncertainty).
    Theory of political economy of
          transition reform
• Normative PER  decision making problems
  of reformers subject to political constraints.
  – Based on “Agenda-setting hypothesis” 
    government has monopoly over design and
    sequencing of reform packages (parliament
    votes on a take-it or leave-it base or population
    in a referendum), normative PER then
    develops general principles of reform package.
• Positive branch of PER  analysis of clash
  of interest groups.
     Ex-ante and ex-post political
• Two types:
  – Ex-ante    political   constraints       feasibility
    constraints. Block decision-making and prevent
    acceptance of reforms.
  – Ex-post political constraints  backlash and policy
• These two types of constraints must be dealt
  with differently.
  – Ex-ante constraints  compromises, credible
    compensation for losers, decisions delayed.
  – Ex-post constraints  attempts to create
Political constraints without
 (reforms with sure efficiency gains)
 How to relax political (ex-ante)

a) compensating transfer to losers
b) partial reform
c) “packaging” reforms to get majority support
d) creation of institutions making
   compensating transfers credible in a
   sustained way
e) waiting for deterioration of the status quo
   (cf. Dani Rodrik JEL paper)
  Compensating transfers to losers
Compensating transfer is an obvious way to buy off losers
   of reform. In the real world this is not easily done.
i) Redistributive transfer can have high distortionary
   costs, especially in transition countries, where fiscal
   administration is young, large budget deficits are common
   and enforcing tax collection is inefficient.
social safety net is an example, where we have large
   distortions because of a lack of proper targeting
ii) Arguments of information assymmetry between policy
   maker and economic agents who are potential losers.
 Compensating transfers to losers

Example: lay off of all workers of a SOE.  How
  should we compensate workers? Without
  asymmetric information compensation structure
  should be according to “exit costs”  when there
  is asymmetric information: all workers have to
  be paid the highest compensation (those who
  find easily a job elsewhere will not reveal that =>
  have to be paid highest compensation).
 Compensating transfers to losers
Likely solution: Partial layoffs where workers with
  highest exit costs keep their job. From an
  efficiency point of view, this would be the least
  desirable outcome as those with lowest outside
  options remain in the firm. There are, however,
  powerful arguments against this whole set-up,
  see e.g. Gerard Pfann‟s work on downsizing,
  where firm keeps the “best” workers in partial
Compensating transfers to losers
ііі) Weak commitment power of decision makers
   with regard to future compensating transfer.
 Difference between CE and CIS: former
   established rapidly an UB system and other
   elements of social safety net, unlikely that this
   would be dismantled with a later government,
   while latter did develop this system only
               Partial reform
Obvious disadvantages:
 Lower efficiency gains than in a complete reform
  extreme case: partial layoff of essentially best
 Losses of complementarities of bundled reforms
 In an extreme scenario, partial reform generates
  little learning and might become pure noise 
  outcomes worse than status quo absolutely and
  zero learning about results of complete reform (for
  some time in 90‟s maybe Ukraine comes closest to
              Partial reform cont.
i) Asymmetric information and heterogeneity of “exit costs” and
    high efficiency costs of raising funds to compensate losses.
If I only need to compensate “low exit cost worker” then in
    connection with high costs of raising funds, partial reform
    might be more efficient than full reform.
ii) Building constituencies for further reforms.
Strategy: make status quo less attractive
 Divide & rule by example: 2/3 of workers have to be laid off
    in a sector, a majority of workers need to agree to this. By
    having small layoffs, where laid-off are not fully
    compensated, one might get a majority of workers to agree if
    it is clear that compensation packages will be worse down
    the line (then even those immediately laid off might agree).
  Packaging reforms to gain majority
• In some cases (in case without uncertainty), a big bang
  reform packaging several reforms together may obtain
  support by the majority while individual reforms would be
  blocked, even if it was optimal to implement reforms
• This will be the case, for example, when a reform damaging
  a majority is proposed as part of a package benefiting a
• Ex. Fiscal and trade reform (Martinelli and Tommasi, 1997).
  3 groups: one always gains, one always loses, one gains
  only from one (trade).
   – Optimal solution: first fiscal reform → trade reform [together
     excessive unemployment rather than smooth adjustment]. BUT:
     taken separately, no majority.
   – Putting the two reforms together, if the total outcome for 3 is still
     positive, also group 3 supports the package, so there is a majority
     and the package is approved.
   Creation of institutions making
 compensating transfers credible in a
           sustained way
Policy maker can create institutions with a
  commitment to transfers
Extending voting, i.e. enfranchising the poor, is
  such an institution, median income < mean wage,
  one-man-one-vote-system might guarantee that
  redistribution is here to stay.
See also the introduction of massive income
  support system in CE for non-employment
   Waiting for deterioration in the
             status quo

Dani Rodrik’s paper in JEL gives some examples of that
  when the status quo is deteriorated, then it is
  relatively easy to push through a sustained reform
  effort. So waiting for the deterioration of the status quo
  might be a strategy (hard to believe though that such a
  deterioration will not strongly affect future reform path)
On the other hand, one reason Balcerowicz et al. could
  push through a big bang program was that the status
  quo was greatly deteriorated.
Political constraints with
         Ex-ante and ex-post political
              constraints, cont.
• Ex-ante and ex-post constraints will be the same if no uncertainty
  and no reversal costs.
• However,Fernandez and Rodrik (1991) show that, ex-ante and ex-
  post constraints may differ substantially in the presence of
   – Aggregate uncertainty: uncertainty about economy-wide effect of
   – Individual uncertainty: uncertainty who will be winners and losers
• Aggregate uncertainty + reversal costs  harder to implement
• For reforms to take place they have to satisfy both the ex-ante
  (majority must be benefiting from it) and the ex-post political
  constraints (not risking to be reversed).
• For this reason they speak of status quo bias against reforms.
• It is interesting to note that in this model majorities shift over time.
  This aspect is very important in determining the optimal
  sequencing of reforms.
              Individual uncertainty
•   When uncertainty is purely individual or idiosyncratic, it is
    known in advance whether there will be reform reversal or
    not if reforms are adopted, given the (known) ex-ante
    probability of gaining or losing from reform.
•   If there is reversal, no reform will be adopted provided
    reversal costs are big enough.
•   In this framework, big bang is never dominated by
    gradualism and in some cases strictly dominates
    gradualism (thanks to bundling):
    1. No interim reversal under gradualism and no reversal after two
       reforms (payoff identical to big bang if there is no discounting effect
       – otherwise big bang dominates because of complementarities);
    2. Interim reversal under gradualism and reversal after two reforms
       (no reform tried)
    3. Interim reversal under gradualism but no reversal after two reforms
       – interim status quo bias (gradualism dominated by big bang if it
       has a positive outcome).
           Aggregate uncertainty
•   With aggregate uncertainty, it is not clear anymore what
    will be the final outcome of reform. In this case we have
    several possible outcomes both for big bang and
•   This is a reflection of the ignorance about the transition
    process as well as the ignorance of agents involved. It is
    also a reflection of the possible existence of multiple
•   If we assume the existence of complementarities, the net
    present value of the “partial reform” is always considered
    to be lower to the present value of the “full reform”, as we
    miss the benefits of complementarities (interim suffering).
•   However, it is also true that observing the outcome of the
    partial reform may give information about the potential
    outcome of a full reform and allow us to reverse the reform
    before having implemented the full reform (at a lower
     Aggregate uncertainty, cont.
•   Big bang can still be optimal when:
    – There is very high interim suffering;
    – There is no learning from partial reform;
    – The expected outcome of the big bang package is
      positive and large.

•   Gradualism is more likely to be optimal when:
    – First reform is informative (probability of early
      reversal <1 and >0);
    – Learning has to be fast enough;
    – Early reversal is less costly than full reversal.
     Aggregate uncertainty, cont.
•   Gradualism in this context has the option of learning and
    early reversal at lower cost that is absent in big bang.
•   This can prove a significant advantage, especially when
    costs associated to reversal in big bang are high in the
    case of a bad aggregate outcome.
•   It is interesting to note that the option value of early
    reversal raises a wedge between the ex-ante expected
    payoff of a gradualist program and that of a big bang
•   In this case gradualism might be ex-ante attractive even if
    big bang is not.
•    a gradualist strategy can be a useful way to overcome
    the status quo bias of reform!
•   On the other side, the fact that big bang has larger
    reversal costs makes reversal less likely and this might be
    considered a positive characteristics.
    Aggregate uncertainty, cont.
• This model can have interesting implications
  in terms of timing of transition programs. If
  the default payoff declines with time, and
  the expected payoff of gradualism is larger
  than that of Big Bang
  1. A gradualist program will start earlier;
  2. Big bang program will be observed on countries
     where crisis perceptions are greater (after
     missed opportunities of gradualism –Hungary
     and China vs. Poland and Russia).
      Aggregate uncertainty, cont.
•   This model can have interesting implications in terms of
    timing of transition programs. If the default payoff declines
    with time, and the expected payoff of gradualism is larger
    than that of Big Bang
    1. A gradualist program will start earlier;
    2. Big bang program will be observed on countries where crisis
       perceptions are greater (after missed opportunities of gradualism –
       Hungary and China vs. Poland and Russia).

•   Going back to political constraints, we can identify an
    ex-ante/ex-post trade off between big bang and
 looking at ex-post political constraints big bang is
  preferable (reduces the probability of reversals – “window
  of opportunity”);
 looking at ex-ante political constraints, instead, gradualism
  has the advantage of reducing resistance to the adoption
  of reforms (as it is more easily reversible).
     Aggregate uncertainty, cont.
•   The previous results were obtained in a representative
    agent framework (all individuals             had identical
    “perceptions” about the probability distributions associated
    with various aggregate outcomes).
•   It is possible, however to imagine an alternative
    framework, where agents have different subjective beliefs
    about the likelihood of given aggregate outcomes
    (“reformers” vs. “conservatives”).
•   In such a framework, gradualism might be the best
    solution to allow “reformers” to start the reform process
    even in presence of a majority of “conservatives” with a
    preference for the status quo.
•   More generally, the flexible approach to reforms provided
    by gradualism may explain why politicians often take a
    gradual approach to large-scale reforms even in presence
    of complementarity of reforms.
    Complementarity and momentum
             of reforms
•   Complementarities can allow gradualism to build a
    momentum for reforms that might not be built in case
    reforms were separable.
•   It may be good to start with a reform having a high
    likelihood of bringing benefits to a majority, in order to
    build support for more difficult reforms;
•   Example of momentum under gradualism:
    –   decollectivization in agriculture (China 1978). From 1% in 1979
        to 90% in 1984. Increased agricultural output;
    –   support of rural constituencies for second phase: dual-track
        price system (market prices on excess production - to achieve
        support for reforms in cities);
    –   New constituencies for further reform: restructuring of SOE‟s
        with layoffs and privatization.
 We go back to the issue: which sequence is best?
             Optimal sequencing
•    What is the optimal sequence of reforms when
     gradualism dominates a big-bang strategy?
•    Three aspects are particularly important in
     defining a “correct” sequencing of reforms:
    1. Unbundling of reform package in given sequences
       should not lead to the loss of “informativeness”;
    2. Sequencing should make the reform process ex-
       ante acceptable;
    3. Sequencing should aim at building constituencies
       and momentum for further reform and satisfy ex-
       post political constraints.

•    Note: Objectives 2 and 3 may or may not be conflicting
     with each other in determining an optimal sequencing
     of reforms.
        Optimal sequencing, cont.
•   Different expected outcome: if we have two reforms
    with positive expected outcomes it is better to start with
    the one showing the highest expected outcome (lower
    probability of first-period reversal and increased
    probability of both being achieved);
•   Differences in riskiness: introducing the risky reform
    first increases the option value of reversibility
    (increased volatility), increasing the expected outcome.
•   Differences in constituencies: the sequencing of
    reforms may affect reform reversibility after
    implementation of a first reform, depending on the
    expected costs and benefits of reform continuation for
    the pivotal voter, by affecting the “identity” of the pivotal
    voter at the interim stage (the implementation of the
    first reform may cause the “pivotal” voter to change).
         Optimal sequencing, cont.
•   Exogenous vs. endogenous “reelection” or
    future support and related literature: the case
    for big bang or gradualism or a particular
    reform sequencing may be influenced by the
    “nature” of the probability of “reelection”.
    –   If the loss of power of the incumbent government is
        caused by exogenous forces, the advantage of a big
        bang approach is that the government can constrain
        a potential successor government by choosing today
        policies more difficult to reverse.
    –   On the other side, if policymakers face an
        endogenous probability of reelection, they should
        work to build constituencies by starting with more
        popular reforms. In this case, political backlash may
        be just the consequence of wrong sequencing.
        Optimal sequencing, cont.

•   Observed    “patterns”    in     transition
    economies (creation of support):
    1. Development of small private sector;
    2. Best firms privatized first (profit channel vs.
       job creation channel);
    3. Setting up new institutions (political reform
       and other institutional changes);
    4. Phase of restructuring and closing money-
       losing enterprises is generally delayed.
    Speed of reform and endogenous
•    The design of the “reform package” can affect the final
     outcomes that can be attained.
•    For example, the speed of reform (big bang vs.
     gradualism) can affect the response of foreign
     investment to the reform process itself.
•    In his book, Roland derives two main results:
    1.   When the level of investment DOES NOT affect the reversal
         probability, the investment response under gradualism is
         ALWAYS higher before uncertainty resolution, because of the
         lower option value of waiting (you lose less);
    2.   When the level of investment DOES affect the probability of
         reversal, then both gradualism and big bang may generate
         higher investment response before TOTAL uncertainty
         resolution. (Gradualism better when probability of good states
         is lower). Before ANY uncertainty resolution, however, in this
         case more irreversibility is created – and investment is higher
         – under big bang.
Back to Normative PER…
      Normative PER cont.
Increasing ex-post constraints, lowering the
  possibility of reversal
Normally low ex-ante constraints go hand in hand
  with high ex-post constraints. Has we have seen,
  however, good sequencing and well-designed
  compensation schemes will make reversal more
Sometimes the strategy was used to take advantage
  of a window of opportunity and make a reform
        Normative PER cont.
Ex.: Mass privatization in Russia, insider
  privatization but here irreversibility (very hard
  to give it back to the state) is very strong;
  existing state structures had been deliberately
  weakened => impossible to organize a reversal.
The downside of this irreversibility in the Russian
  case: asset stripping, weak performance, 
  inequality and corruption,  capture of the state
  by oligarchs =>  welfare. => Greater reversibility
  with a better privatization policy might have meant
  a smaller fall in welfare.
And, finally, positive PER…
                     Positive PER
Explains differences in rent seeking and how special interests may
  capture regulatory bodies. In general we look at institutions
  affecting costs and benefits of rent seeking.
• Some questions:
   – Is it easier (or preferable) to pass reforms when the executive branch
     of government has relative autonomy, or is it better to have it closely
     checked by the legislature?
   – Is a presidential regime preferable to a parliamentary regime in terms
     of deciding reforms?
   – Is it better to have coalition governments or majoritarian governments
     to pass reforms?
   – Is a large number of veto players a good or a bad thing (blocking
     good/bad reforms)?
Normally these things are exogenous, but in transition they are more
  or less endogenous.
              Positive PER, cont.
Hellman (1998) – democratic transition countries – more fragmented
  party systems, coalition governments, uninsulated executives 
  most progress with reforms.
Hypothesis: better accountability /closeness of politicians to voters /
  less capture by vested interest groups.
Drawback: mechanisms of checks and balances creating
  accountability can also create attrition/ paralysis of decision.

“Reformers” are usually seen as “special politicians” having in mind
  the success of efficiency-enhancing reforms. In this case, insulating
  them from the pressure of interest groups might be a good thing.
Drawback: possible insulation also from voters and opportunity for well
  organized interest groups or for the “opportunistically oriented
  politician” (to commit abuse).
            Positive PER, cont.
One interesting paper makes point that e.g. in Russia (and this
   can be extended to Ukraine) oligarchs are interested in weak
   property rights => they then can easily convert social and
   state assets into private ones for their own use.
 Oligarchs try to capture state regulatory bodies to prevent
   the establishment of strong property rights (unexpected
   result- maybe a priori one would have thought that the rich are
   interested in secure property rights).
Differences in the strategy of rent seeking between e.g. Poland
   and Russia is not fully explained yet
{ Poland has a collective memory of capitalist society but also
   within socialism powerful social networks – catholic church,
   solidarity trade union, nothing of the kind in Russia and FSU.
   => one might speculate whether this lack of social networks
   made it easier for insiders and nomenklatura to capture the
         Positive PER, cont.
 Understanding properties of various political
 institutions in generating efficiency-enhancing
 reforms is an important orientation of future
        PER and performance
 Wide differences in economic performance can be
 explained as consequences of different reforms
 implemented under different political constraints.

Other possible aspects:
 Geopolitics and economic performance.
CE shift to Western Europe (shedding satellite image)
   and linking in with EU => after initial output fall, fall
   into lawlessness, corruption, and extreme state
   capture could be prevented.
It is not so in Russia, where we have the lost empire
 PER and performance cont.
2. PER and disruption of CMEA trade.
Collapse of CMEA trade was not purely
EBRD and IMF tried a dual-track approach as
  far as trade was concerned (2nd track –
  additional trade with western businesses).
  When CE countries wanted to freely trade with
  West, SU insisted on world prices for all its
  trade => CMEA broke down.
Conscious policy choice by e.g. CR and Poland to
  see trade with West as a priority.