Comments on International Coercion, Emulation and Policy Diffusion: MarketOriented Infrastructure Re

1
Comments on International Coercion, Emulation
and Policy DiffusionMarket-Oriented
Infrastructure Reforms, 1977-1999

• Alexander Dyck
• Harvard Business School

2
A fact Market oriented reforms increase
significantly in recent years
3
What drives this adoption of market oriented
reforms?

• A conventional (Coasian) approach -
  Privatization and deregulation driven by a
  comparative institutional analysis
• Change approach when net benefits of state
  ownership and regulation less than net expected
  benefits of private ownership and deregulation
• Suggests focus on factors affecting government
  failure (e.g. budget, political constraints) and
  market failure (tech change, economies of scale,
  etc.)
• N.B. - largely focus on efficiency, and domestic
  factors

4
What is approach taken here?

• Choices driven in addition by political and
  sociological factors in the international arena.
• Politically, multilateral lenders are a distinct
  interest group that can influence power balance
  and choice (coercion)
• H1 Rate of adoption of market-oriented reforms
  increases with international coercion
• Sociologically, willing to do something not
  because efficiency suggests better, but because
  others with legitimacy are doing this.
  (emulation)
• H3 Rate of adoption of market-oriented reforms
  increases with emulative pressures

5

• Worthy question both conceptually and from
  policy perspective. Fascinating questions How
  much do lenders matter?(which they address here)
  Do these pressures help or hurt? (which they
  leave to later work)
• Think you are right on lenders
• Lots of rich case studies showing effect of
  lending
• Crude time series variation suggests scope for
  additional explanatory power from international
  factors
• Domestic institutions have limited time series
  variation and/or cyclicality rather than one
  direction variability
• International factors - lending, trade, etc,
  have similar time series variation as reforms.

6
Summary of my comments

• While sympathetic with overall thrust dont
  think youve done enough to convince sceptics
  yet
• Framing and empirics not set up as test of
  conventional and alternative view
• Are independent and dependent variables what we
  want?
• Some data questions
• Remaining opportunities

7
1. Framing concerns

• Approach taken in paper
• Go to great length to define your alternative
• Not clear that 19 pages to get to data
  description and references to other papers does
  justice to your contribution
• There is a considerable body of research on the
  cross-national diffusion of ideas, practices and
  policies, including social security systems
  (Collier and Messick 1975), oil nationalizations
  (Kobrin 1985), decolonization (Strang 1990),
  growth in the welfare state (Strang and Chang
  1993), bureaucratization of national science
  policy (Finnemore 1993), currency crises (Glick
  and Rose 1998), policies to protect the
  environment (Frank, Hironaka, and Schofer 2000a
  Frank, Hironaka, and Schofer 2000b), quality
  certification by firms (Guler, Guillén, and
  Macpherson 2002), deregulation (Eising 2002
  Gilardi 2003), neoliberal macroeconomic policies
  (Yebra 2002a Yebra 2002b Yebra 2003), pension
  privatization (Weyland 2003), current account
  liberalization (Biglaiser and Brown 2003),
  capital account liberalization (Brune and
  Guisinger 2003), Central Bank Independence
  (McNamara and Castro 2003), right to transparency
  laws (Roberts 2003) and privatization (Brune,
  Garrett, and Kogut 2004 Kogut and Macpherson
  2003). While several of these studies emphasize
  the coercive role of multilateral organizations
  or strong states (Strang 1990 Strang and Chang
  1993 Finnemore 1993 Glick and Rose 1999 Frank,
  Hironaka, and Schofer, 2000a, 2000b Eising 2002
  Biglaiser and Brown 2003 McNamara and Castro
  2003 Roberts 2003 Brune, Garrett and Kogut,
  2003 Kogut and Macpherson, 2003) and others
  highlight the role of emulation among peer
  countries (Collier and Messick 1975 Kobrin 1985
  Weyland 2003 Guler et. al. 2002 Yerba 2002a,
  2002b, 2003 Gilardi 2003), only one considers
  both mechanisms simultaneously Brune and
  Guisinger 2003).
• Specify a variety of other factors

8
1. Framing concerns

• My preferred approach
• My null non-international factors (such as
  identified by Coase) explain adoption.
• Your alternative coercive and emulative factors
  explain adoption.
• Contribution of paper provide empirical test to
  see if conventional wisdom wrong to ignore these
  international factors

9
Such an approach would reduce and clarify
concerns about omitted variable bias

• Surely a measure of government failure is the
  ability of government to fund new investment in
  sector? (e.g. fiscal imbalance (debt burden,
  budget imbalance), interest rate on government
  debt).
• Time series as well as cross sectional
  variability, so looks promising.
• Table 7 and 8 attempt to address as robustness
  check now negative effect of coercion on
  reform (sign changes, column 7), coefficient
  declines by 1/3 for telecom and sizable drop in
  significance
• Likely that covaries with multilateral lending,
  so probably not right test - Two stage procedure
  (Table 4) helps, but direct measure of debt
  burden should be included in second stage.
• Bottom line Need to control for more other
  variables in core tests to convince skeptics

10
Is foreign investment/GDP a part of alternative
that emphasizes coercion/emulation?

• Test of coercion and emulation also includes
  foreign investment as measure Why?
• Not in theory section up front
• Not sure how to interpret results
• Might be easier if disentangled fdi in sector,
  other fdi
• Do we expect fdi interested in competition?
• concerned that other results may not be there if
  dont include.

11
2. Are independent and dependent variables what
we want?

• Key independent variables are emulation and
  coercion
• Dont have measures of emulation and coercion so
  use proxies
• OK with emulation measure from trade-weighted
  measure of reforms.
• Not very comfortable with multilateral
  lending/GDP as measure of conditionality.

12
2.

• Could very well be true that increases in
  multilateral lending/GDP bring more coercive
  pressures and conditionality.But, no validation
  of this assumption, either cross-sectionally or
  over time.
• Measure closer to theory would be much more
  convincing.
• Likely to have a lot of cross-sectional variation
  for a given level of lending/gdp, quite
  different level of conditionality,
• dont know what time series variation looks
  like increases all at once (step function) or
  gradually?
• Can exploit this variation to identify effect.
• Dont know if this is feasible. Weaknesses
  should at least be recognized, hypotheses stated
  in terms of variables of interest rather than
  imperfect proxies

13
2. Independent variable

• Do you really have 180 observations from each
  country?
• 9 different measures, seem very related, clearly
  within categories (regulation, privatization) and
  also across categories.
• Visual glance of data suggests many of these
  happen at same time
• Does clustering s.e. by country really correct
  s.e. correctly?

14
3. Data questions

• How can you have negative FDI/GDP, Multilateral
  lending/GDP? (Table 2)
• Ratio of multilateral lending/GDP is 12 percent.
  One standard deviation is 15 percent.
• Results relevant to what income ranges? (e.g. for
  middle income aidof gnp just 2.1) What happens
  if exclude low-income?
• 205 (196) countries (American samoa? WB only
  133), pre and post communist, etc.
• Are results true for subsamples that understand
  better (e.g. excluding tiny places, those
  affected by socialism)?
• What actually is subsample for key regressions?
• thought would have 37000 obs 205209, which do
  have. Sample drops to 8217 observations when run
  tests who drops out? Unbalanced sample? Since
  this relevant sample, why not just include this.

15
4. Additional opportunities

• Could do more than refute conventional wisdom.
  Have collected data (type of lending, type of
  reforms) that allow for more subtle tests.
• What are treated as robustness checks could be
  emphasized more
• Right to conclude that multilateral lending a
  disaster for competition? (Table 5)
• Regional development bank lending is disruptive
  of market-oriented reforms? (Table 6)
• Performance shortfalls dont affect privatization
• Multilateral pressures for priv in telecoms and
  for regulatory reforms in electricity, etc.

16
Summary of my comments

• While sympathetic with overall thrust dont
  think youve done enough to convince sceptics
  yet
• Framing and empirics not set up as test of
  conventional and alternative view
• Are independent and dependent variables what we
  want?
• Some data questions
• Remaining opportunities