1st Lecture, STV4346B: Comparative Political Economy

Download Report

Transcript 1st Lecture, STV4346B: Comparative Political Economy

1st and 2nd Lectures, STV4346B:
“Introduction to Comparative
Political Economy: Approaches,
theory and methodology”
(some of the slides will probably be skipped in class
and used in seminars 1 and 2)
Carl Henrik Knutsen, Department of Political Science, UiO
10/11 and 13/11-2008
1
Politics and the economy
Political institutions and
processes
Economic structures
and processes
”Z-variables”:
Geography, culture etc
2
Topics for the course
Methodology and theory




Different traditions of political economy
Basic rational choice/microeconomics + other theoretical approaches
Statistical techniques/econometrics
Substance, general




Allocation of resources; states and markets. Positive and normative
issues. The interaction among: Politicians, bureaucrats, voters, consumers
and producers.
The economic factors that influence regime change and regime stability
(mainly democracy)
The economic effects of political institutions
Substance, specific (with geographic focus)




3
Varieties of capitalism, fiscal policy, economic voting
Industrial policy, land reform, the effects of inequality
Neo-patrimonialism, corruption, natural resources
Practical advice
Read through everything once, quickly
..Then selective reading
Read all abstracts at least four times
Know thy lecture notes
Systematize the readings into a broader framework
Don’t get bogged down in details. Focus on concepts,
general arguments (theoretical) and general empirical
findings.






4
The exam
Approximately:




5
25%: Short questions: Concepts, sketch out theory quickly,
empirical findings
25%: Short essay: Related to 2-3 concrete contributions from
reading list.
50%: Long essay (choice between two alternatives)
Political economy
Empirical interrelations between economy and politics
We need theoretical and methodological frameworks that
are able to deal with such interrelations: Political
economy.
Political economy: different meanings for different groups
of academics and different traditions.
My view: They are not as incompatible as they might look.
Don’t pick your favorite! Choice of framework depends
on research question and focus.



6
Thematical and methodological definitions
of political economy
Methodological definitions:



Study of politics with the tools of an economist (political
economics)
Study of economics with the tools of a political scientist, focus
on power etc, qualitative methodology.
Thematical definition (favored in this course)


Political economy is the study of the interrelations
between political institutions and processes and economic
structures and processes.


7
No restrictions on methodology or theoretical approach; in this
course we will be eclectic.
The definition is very general! Specifications will be made at the topic
level
Definitions continued
COMPARATIVE political economy: comparative studies of
political-economic interrelations.




Differences in political institutions Differences in economic effects.
Differences in economic structures Different effects on politics.
Focus on the national level.
Even though we are here operating with a broad
definition of political economy, we need to know the
various definitions and political economic theories in the
literature to not be confused when reading.

8
State-centered political economy (C&L: 8)
Key point of departure: Other political economy approaches have
underappreciated the importance of the state.
State as independent actor or (at least) state structures matter for
political and economic outcomes.
States are not only vehicles for social interests.
Sharp separation state and private sphere/society (including
economic actors). Politics conducted mainly within state-structures.
States as actors: “State’s interests”. Some go even further: states
assumed to follow “national interests” (whatever this means).
Metaphysical description or analytical simplification? My view: States
are not actors metaphysically, but we can model the state as an
actor in some instances for analytical simplification.






9
State autonomy



State autonomy is the key concept in much of this literature.
“State autonomy… refers to the ability of the state to define and pursue an agenda
not defined for it solely by private interests” (C&L:181)
Ability to pursue own agenda and not be trumped by societal pressures, private
interests when formulating public policy.


Specifications: If we do not buy that the state is an independent actor, how do we
think of state autonomy?





Contrast with C&L’s Utilitarian and Marxist approaches.
State officials and bureaucrats as autonomous? But why do these not act in private
interest?
Rules, regulations and norms matter. These are created through the years and are results
of long processes. Norms and rules can be intentionally created by private actors, but not
always. Often: gradual modification of rules and norms, and strong inertia in institutional
structures.
For analytical purposes therefore, we can often (but certainly not always!) take the rules
and norms as given. Contrast with endogenous institutions and regulation.
Degrees of state autonomy rather than either-or.
Determinants of state autonomy: power concentration, checks and balances,
structure of bureaucracy, nature of social interest groups.
10
The State, a special organization




Monopoly of force within defined boundaries (Weber)
Ability to make binding decisions on issues which private
actors cannot (laws and regulation)
Ability to tax
Implementor of policy
11
Some approaches in C&L (read quickly
through 184-196)

Statism and national interest (Krasner)


States define (and act upon) national interest
Transformational view of state (Skocpol among others)



12
Differences in state’s organizational structure impacts policy
outcomes even when initiative from private sphere
State affects private agendas and make-up of social groups
State as historical specific organizational structure
Rational choice




A theory of human action
Optimal choice under constraints
Preferences are exogenously given
Predictable irrationality from behavioral economics:





13
Beliefs/understanding of probabilities, some systematic biases
Loss aversion, reference points
Time preferences and hyperbolic discounting
Reciprocity rather than self-interest?
Bounded rationality (Simon)
Rational choice


A “deceptively simple sentence” that summarizes the
theory of rational choice: “When faced with several
courses of action, people usually do what they believe is
likely to have the best overall outcome” (Elster, 1989:22)
1) Thin and 2) instrumental rationality:


14
Ad 1) No initial requirements on what type of goals that should
be pursued.
Ad 2) Actions are chosen because of intended consequences.
Actions are not valued because of themselves (contrast with
Kant)
Przeworski (ch 1)

Decentralized and centralized allocation mechanisms –
final allocation results from decisions by many agents vs
allocation depends on one decision. Examples:





Decentralized: Grab all you can, market exchange
Centralized: Dictator’s allocation, voting, lottery
Does the allocation mechanism result in an equilibrium
outcome (no one has incentive to deviate). Descriptive
question.
Does the allocation result in a Pareto Optimal outcome?
Normative criterion.
Is the allocation equitable? Normative criterion.
15
Przeworski cont’d



State vs market as a too crude dichotomy. More fruitful questions:
What are the properties of an allocation mechanism? How are
states and markets organized?
Always some role for government. Enforcement institutions must
underpin markets.
Political economic game:





1) Political actors reveal preferences for policies through menu of
actions (votes, bribes, threats etc)
2) State maximizes its objectives (whatever this is), given beliefs of
economic actors actions, by setting policy
3)Economic actors max utility (and profit) in markets, given the specific
policy chosen
Political economic equilibrium:
Set of policies by state and political and economic actions by
voters/producers/consumers so that no one would act different
given beliefs and actions of others.
16
A political economic game of voting and
taxation
Voter
Party A
Party B
State ”A”
State ”B”
Tax heavily
Tax lightly
Tax medium
(4, 0)
(2,2)
17
(1,3)
Political economics (Ch6 in C&L)



The Robbins-definition of economics: “The study of
human behavior as a relationship between ends and
scarce means..” Opens up for studying other aspects of
human life than production and consumption of goods
Preferences, beliefs and constraints
Individually rational behavior Efficiency (or “collective
rationality”)?

Not necessarily: Depends on interaction structure


18
Holds in perfect markets (invisible hand)
Does not hold for example in prisoners’ dilemma games or voting in
some instances
Normative political economics




Normative: Analysis of desirable properties of the political
system
Condorcet’s paradox
Voter 1
A>
B>
C
Voter 2
C>
A>
B
Voter 3
B>
C>
A
Vote on alternatives pairwise: A >B>C>A……
Transitive individual preferences but no transitivity in
collective preferences (desired property)
19
Arrow’s impossibility theorem

These are incompatible:






Non-dictatorship
Unrestricted domain
Independence of irrelevant alternatives
Pareto Efficiency
“All voting methods are flawed”. Individual preferences do not
add up to consistent social preference ordering. What is the
“will of the people”?
Ways out of the impossibility theorem:



20
Single-peaked preferences or other restrictions on preferences
Agenda control (institutional structure imposed on decision
procedure)
logrolling
Positive political economics



Two examples in C&L: Mancur Olson’s “Logic of
Collective Action” and Anthony Downs’ “An Economic
Theory of Democracy”.
Down’s (1957) Foreword: “Downs assume that political
parties and voters act rationally in the pursuit of certain
clearly specified goals – it is this assumption in fact, that
gives his theory its explanatory power”
“Starting point” for numerous models on party and voter
behavior
21
Assumptions: A Downsian game









Democracy
Two parties/candidates
Goal for parties: Maximize political support (votes) control
government. Policy only as a mean for politicians.
Goal for voters: Policy (as close as possible to ideal point)
Policy is unidimensional on a scale from 0 (left-wing) to 1 (right
wing)
Voters vote on party that are closest to their ideological preference.
They are distributed uniformly on the interval [0,1]
Party who wins majority forms government
Rational and self-interested voters and parties. No uncertainty in
the baseline model.
Party cannot deviate from proposed policy once elected.
22
Downsian game continued




Strategy sets for two candidates/parties, S1=S2 = [0,1]
Choose policy (strategy) within this strategy set; denoted s1 and s2
If s2>s1  all voters to right of (s1+s2)/2 votes for 2
Example






2 chooses policy 0.7, 1 chooses 0.6. All voters to the right of 0.65 votes for 21
wins.
Can this be a proper solution to the game? No! This is not a Nash Equilibrium:
Player 2 is not playing best response to 0.6. Will win majority if plays for example
0.59
But then 1 will not play a best response.. Can for example choose 0.58
(s1,s2)= (0.5 , 0.5) is the only Nash equilibrium: both strategies are best
responses to other. (50% probability of winning when tie). Nobody wants
to unilaterally deviate.
In equilibrium: Both parties propose the same policy. Policy convergence!
23
What does Downs tell us? Remember that a
model is a model.



The model can be extended to incorporate uncertainty about
positions of the voters, and we can assume that parties are
motivated by pushing a specific policy. If so, the convergence
result does not necessarily hold.
If interested: See my homepage for a presentation of “Games
in the Normal Form” for extended models where there is
uncertainty and where politicians care about policy (and not
number of votes)
Simplest possible political economic model of voting and does
not describe real-world politics precisely. BUT: Captures
central political mechanism: Why did Obama talk more about
tax cuts and killing Bin Laden in general election than in
primaries? But: What was McCain thinking when going to the
right in the general election?
24
Preference requirements and utility
functions, rational choice

Preference requirements:







Actors must be able to rank different outcomes. >, < or =.
(Complete preferences)
If a>b and b>c  a>c (Transitivity)
Utility functions: U(x), ordinal level of measurement
Note that preferences are exogenous and given.
Cost functions: C(x).
First-and second order derivatives. U’(x)>0, C’(x)>0,
U’’(x)<0 and C’’(x)> = or < 0 (depends on productiontechnology)
Optimization (minimize or maximize): Differentiate and
set equal to zero.
25
Example: A single market, partial
equilibrium


One good: x. Price for good x is p.
Demand side of the market: Consumers will buy until U’(x)=p
because:


Supply side of the market: Assume C’(x), the marginal cost, is
increasing in x (C’’(x)>0). Producers will supply until C’(x)=p.
Because profits π is given by:



Max U(x)-px  U’(x)-p=0U’(x)=p
Π(x) = px – c(x)  Π’(x) =0 ↔p=c’(x)
Market equilibrium: The price clears the market. Assume one
consumer and one producer, both take prices as given. They
will both adjust their demand and supply so that U’(x)=p and
C’(x)=p U’(x)=C’(x)
In equilibrium: Marginal utility of consumption equals marginal
cost of production
26
The choice between two goods (Knutsen,
2008a)
Max U( x1, x2)
Subject to the constraint: p1 x1 + p2 x2 = m, where m is income


U( x1, x2) = U( x1, m/p2 - p1/p2*x1)

Optimum condition: U’(x1) + U’(x2)*- p1/p2 = 0
 U’(x1) /U’(x2) = p1/p2



Marginal rate of substitution is equal to relative prices in optimum
It can be shown that also marginal rate of transformation (relative cost of
producing two goods on the margin) is equal to relative price ratio: Prices
carry information about opportunity costs


 In equilibrium: Marginal rate of substitution is equal to marginal rate of
transformation.
Prices enable consumers and producers to adjust so that relative
production costs equal relative subjective evaluation of goods.
27
Graphical sketch of general equilibrium
Equilibrium relative prices and equilibrium production and
consumption of X1 and X2
Indifference curves:
combinations of X1 and X2 that
yields a particular level of utility
X2
Production
possibilities set:
viable combinations
of X1 and X2
-P1/P2
X1
28
Normative criteria for efficiency: Pareto
optimality




A movement from one allocation to another that can make at
least one individual better off without making any other
individual worse off is called a Pareto improvement.
An allocation is Pareto optimal when no further Pareto
improvements can be made; that is, when no one can be made
better off without anyone else being made worse off.
First welfare theorem: Market equilibrium is Pareto optimal
(given a set of conditions)
Second welfare theorem: Any Pareto optimal allocation can be
supported by a market equilibrium (given set of conditions
including cost-free redistributions)
29
When does the first welfare theorem break
down?








Increasing returns to scale
Market power
Externalities
Public goods
Imperfect information
Transaction costs
Missing markets
Property rights are non-existent or poorly enforced
30
Monopoly



Market power Ability to influence prices
When prices are not taken for given 1st welfare theorem breaks down
Why monopoly/oligpoly?




Increasing returns to scale: the role of fixed costs (innovation, investment, advertising
etc),
Regulation
differentiated products
The monopoly’s profit-maximizing equation
 Π(x) = p(x)x – c(x)  Π’(x)= p’(x)*x + p(x) –c’(x)=0
 p’(x)*x + p(x) =c’(x)
 The monopoly sets x lower than the ”social optimum” (given by c’(x)=p).
 Intuition: Monopoly reduces supply to keep prices high
31
Externalities (actions with unintended
consequences on other actors)




Externalities from consumption and production.
Both positive and negative externalities
Actor does not take ext. into account when making
decisions Does not provide the socially optimal
amount of consumption/production in free market
Examples
32
Effect\Type
Consumption
Production
Positive
Education (with
knowledge
spillover)
Innovation (with
knowledge
spillover)
Negative
Smoking (health
effects)
Polluting activities
Public goods


1) Non-rivalry: One actor’s consumption does not reduce the utility of
another from consuming.
2)Non-excludability: Not possible to stop other actors from consuming.

1) inefficiency in market because of positive externality (does not take
into account other actors’ benefits from providing good)

2)Free rider problem. Even if would like to have the public good, hope
that others will provide and free-ride. In free market, no one has incentive
to supply.
Przeworski defines public goods only according to 1). I prefer to call nonrival but excludable goods for club goods. Public goods are defined as nonrival and non-excludable


Note that excludability and rivalry are continuous dimensions Degrees
of public goods.
33
Neo-classical political economy







Role for government and politics when markets fail/market
equilibrium is not Pareto-optimal
How can government correct the workings of the market?
Example 1: Negative externalities Taxation so that the actor bears
the full cost (own cost + externality) will act so that optimum is
realized
Example 2: Assymetric information: Provide regulation, issue
standards etc so that f.ex consumers can identify quality of a
product in the market
Example 3: Public goods: Tax and provide correct amount of public
good.
Example 4: Increasing returns to scale and natural monopoly:
Regulate price
The state’s role of securing property rights to ensure voluntary
exchange.
34
Some criticisms of NCPE



Keynes (1936) and inherent rigidities of the market. Slow
adjustment towards equilibrium and the benefits of
activist macroeconomic policy.
Schumpeter (1942): Dynamic efficiency is not ensured by
a statically efficient market (profit goes towards zero).
Monopoly/oligopoly surpluses and incentives for
innovation. Innovation  Long run economic growth.
The utilitarian underpinnings. Other important normative
concepts than efficiency/PO: distribution and justice.
35
Uncertainty

Uncertainty: General lack of knowledge about outcomes. Unproblematic for
markets if uncertainty is symmetric (and actors are risk neutral).

Instead of maximizing utility: Maximize expected utility!

Von Neumann-Morgenstern utility functions:
 EU(p) = p1u1 + p2u2+…+pnun

We can take into account that actors are not risk neutral.

Risk aversion, two outcomes:
 u(px1+(1-p)x2)>pu(x1)+(1-p)u(x2) [u(50)>0.5*u(0)+0.5*u(100)]
 Means that a risk averse actor would rather take the expected value for certain,
rather than gamble between two extremes.
36
Asymmetric information





Markets in trouble when there is asymmetric information.
Note that governments also would have trouble with
asymmetric information. How to make actors reveal their
private knowledge is a general phenomenon. However,
governments might have other means to deal with such
problems
1) Private information about actions, effort
2) Private information about quality, type
1) can lead to so-called “moral hazard” problems
2) can lead to so-called “adverse selection” problems
37
Some concepts



Moral hazard: Incentives to act otherwise than agreed
upon in contract, due to private information about own
actions.
Adverse selection: The tendency of an agent (or a good)
of a particular type to self-select into a contract in a way
not desired by the other contracting party, due to private
information about own type (quality).
A contractual relation can be modeled as a principalagent relationship. Principal wants a job to be done and
agent is assigned to do the job
38
Principal-agent theory



Principal wants to maximize profits/utility, and so does agent.
Principal moves first, and can post contract as “ultimatum” to agent.
When full information:



When private info:



Principal can design contract so that agent is pushed down to “reservation utility”
(indifferent between accepting and rejecting contract).
Can also choose to accept only “good” types of agents, or design different contracts
to good and bad types.
Agent can shirk (moral hazard) to obtain a higher utility, to the despair of the
principal.
Agents can also pretend to be other types than they actually are to achieve a higher
utility (adverse selection).
Example: voter-politician as principal-agent.
39
Some solutions for the principal

When moral hazard Design contracts that make agents
“stake holder” in the outcome.



For example: let agent have some of the surplus (part of profit
instead of pure wage contract),
or threaten to punish if moral hazard is detected/strongly
suspected
When adverse selection: Design different contracts that
make agents self-select into different contracts

40
Impose costs on bad agent or reward good agent for revealing
information about type
Some comments from Evans (Ch2)



“Getting the prices right” as an insufficient approach to
development
Neoclassical economics and focus static efficiency: What about
transformation and dynamic efficiency?
Gerschenkron and late developers:


Hirschman and late developers:



State’s role in amassing capital.
State’s role in inducing entrepeneurship
The role state institutions for economic development.
Structure of state and state-society relations matter.
But: rejection of “naïve statism”.
41
Some comments on markets..

Market/economic sphere is embedded in society (Polanyi).



Norms and exchange. Granovetter and the non-existence of
informalized market exchanges. Trust as crucial to non-contractual
exchange.
Repeated exchanges between actors that know each other, rather
than anonymous one-shot exchanges
Transaction costs (Williamson/Coase): In some instances,
hierarchies reduce transaction costs related to exchange when
compared to markets.

42
Internalization of exchanges within firms and other organizations.
Evans cont’d



The neo-utilitarian depiction of “state failure”
 Politicians and bureaucrats as self-interested agents and not as altruistic
and omniscient welfare maximizers. Can the people (principal) ensure
that the agents (buer and pol) act in their best interest?
 Capture of state by interest groups
 Rent seeking: Wasteful use of resources to redistribute gains/affect policy
to own benefit
 Collective action problems
Implications: The nightwatchman state? Does the conclusion follow from
the premise of state failures? (I don’t think so..)
Weber: Bureaucracy and capitalism together: How does the state operate
and how does it regulate and interact with the economy. Not how much
state!
43
Evans cont’d: why need other
approaches than “neo-utilitarianism”

Differences in state structures and state society relations
matter for the selection and the effectiveness of policies.



Context contingencies and comparative approach necessary. Need to
draw on concrete historical, case-based knowledge
Methodological individualism at the root of failure to
understand states.
Weber and the structure and norms of the idealized
bureaucracy Bureaucrats do not follow private agendas
(more on this next lecture)

44
Differences bureaucratic structures Diff development
Evans cont’d


Institutionalist revisions of neo-utilitarian model:
 Institutions as “rules of the game”. Creates incentives for economic
actors and are thus crucial for economic outcomes.
 Douglass North (property rights, institutions and transaction costs,
endogenous institutions and functionalism)
Comparative institutional variation
 Migdal and strong vs weak state (state vs society in zero-sum game).
 Social power of local elites? Local, traditional elites related to rural
interests: often opposing industrial transformation.
 The transformative role of the stat: Strong states transform economy
and are not objects for rent seeking
45
Pierson

“The notion of path dependence is generally used to
support a few key claims:





46
Specific patterns of timing and sequence matter;
starting from similar conditions, a wide range of social
outcomes may be possible;
large consequences may result from relatively "small" or
contingent events;
particular courses of action, once introduced, can be virtually
impossible to reverse;
and consequently, political development is often punctuated by
critical moments or junctures that shape the basic contours of
social life”
Pierson




Path dependence related to increasing returns: Self-reinforcing or positive
feedback processes
 Other broader def: “History matters”. Event at t=0 affects events at t=1.
“Switching costs” or “reversal costs” increase as time passes
Particular importance: Formal institutional arrangements that are costly to
alter once initiated and historically entrenched. Inflexibility.
Potential path inefficiency
 A funny example: QWERTY keyboards (David).
 Not so funny: Dysfunctional institutional arrangements like neopatrimonial political arrangements that are hard to break up, but clearly
economically inefficient.
47
Technology and increasing returns

Four features that generate increasing return-structures





Large fixed costs
Learning effects
Coordination effects
Adaptive expectations
Direct analogies to politics, institutions: Institutions are




48
a) costly to generate,
b) require learning as you go along (learning by doing) from different actors,
c) interaction between actors within rule-based framework depend on coordination
d) actions depend on beliefs about what others do
Pierson cont’d



Read Pierson’s interpretation of North
Institutional complementarity (see also Hall and Soskice):
complementary organizational forms and institutions A
source of increasing returns
North: “interdependent web of an institutional matrix”
49
Politics and increasing returns (A)

Four aspects of political life that make it conducive to
increasing returns:

1) Central role of collective action


2) high density of institutions


(use initial position to entrench position and increase power even more)
4) intrinsic complexity and opacity

50
(legal constraints on behavior)
3) political authority and asymmetries of power


(coordination and adaptive expectations plus difficulty of changing politics
individually)
(where are the prices? Multiplicity of political issues and goals of actors, long lags
and complex causal chains)
Politics and increasing returns (B)

Three aspects of politics that make it more susceptible to
increasing returns than economic processes

1) weakness of competition and learning enhancing
mechanisms


2) short time horizons of political actors


(switching costs up front!)
3) status quo bias built into political institutions

51
(self correction possible, but difficult due to opacity of politics)
(constitutions as example, credible commitment devices, political
consensus and uncertainty when change)
Pierson page 263: Effects of increasing
returns on politics




1. Multiple equilibria. Under a set of initial conditions conducive to
increasing returns, a number of outcomes-perhaps a wide range-are
generally possible.
2. Contingency. Relatively small events, if they occur at the right moment,
can have large and enduring consequences.
3. A critical role for timing and sequencing. In increasing returns processes,
when an event occurs may be crucial. Because earlier parts of a sequence
matter much more than later parts, an event that happens "too late" may
have no effect, although it might have been of great consequence if the
timing had been different.
4. Inertia. Once an increasing returns process is established, positive
feedback may lead to a single equilibrium. This equilibrium will in turn be
resistant to change.
52
Some implications




“Historical” causes  Implications for analysis
The role of time
Problems with functionalist explanations (institutions are
not necessarily efficient)
Problem with the approach: How to test hypotheses from
path dependent arguments?
53
Statistics/econometrics



OLS
Pooled Cross Section - Time Series (OLS with Panel
Corrected Standard Errors as example)
Panel Data




Fixed Effects
(Random Effects)
Instrumental Variable Analysis/2SLS
Matching
54
Repetition of multivariate
regression/OLS


Remember interpretation of coefficient: Average increase
in Y when X increases one unit and all other variables are
held constant.
OLS as best linear unbiased estimator (BLUE) when
certain conditions are satisfied







55
The underlying relationship is linear
No autocorrelation
No heteroskedasticity
No omitted variables (important: Omitted variable bias!)
No endogeneity (y affects x)
No measurement error in x
See Knutsen 2008b for full list
What to do when we have a panel data
structure?




Panel units assumed to be countries here, but can be
regions, individuals, firms etc.
If no country specific effects, and we want to use both
intra-national variation (over time) and variation between
countries: OLS with PCSE.
Interpretation: Just as OLS!
But: takes into account heteroskedasticity between panels,
contemporraneous correlation (between panels), and
autocorrelation (within panels). These adjustments are
crucial when using panel data.
56
Panel data: Fixed Effects




What if there are country specific effects?
Control for these effects by using dummies for all
countries Fixed Effects
Only intra-national variation is used
Strict technique. Reduces omitted variable bias, but also
reduces the amount of information used for inference
57
2SLS



Main application: endogenous independent variables
Intuition: Use only exogenous part of variation in X.
Must find at least one instrument:




Instrument should be correlated with X, but not directly linked with Y. Instrument
should be as highly correlated with X as possible.
Method:
 First run OLS regression with X as dependent variable and instrument(s) and all control variables as independent
 Then take the predicted value of X from the first regression rather than
the actual X and use in the second (original) regression.
2SLS is consistent, and does not run into the endogeneity bias problems of
OLS
The major problem with 2SLS is that it yields large standard errors
(particularly when low correlation between instrument and X) difficult
to find significant results.
58
Matching






Non-parametric: Does not need to assume anything
about the functional form of relationship.
Treatment and control variables
Find (one or more) most similar units to use as matches
for each unit.
Similarity calculated on the basis of control variables.
BUT: Unit and match must differ on the treatment
variable (dichotomous)
Find treatment effect for each comparison (difference in Y
between unit and match)
Calculate average treatment effect
59