Title: Positive Policy Analysis Is Hard
1Positive Policy Analysis Is Hard
- Why is it so hard to predict what will happen
when government changes a tax or spending policy?
- Textbook Example Will lowering marginal income
tax rates increase work effort and labor supply? - Unlike the physical and natural sciences,
difficult to perform carefully controlled
experiments on the economy. - Oftentimes, there is no control group or
comparison group for a policy.
2Alternatives to an Experiment
- Use of statistical tools to study impact of
public policies. - Will use the debate over the effect of taxes on
labor supply to illustrate how positive analysis
is done in public finance.
3Role of Economic Theory
- Consider analyzing Table 2.1
- As marginal tax rates have increased, average
weekly hours have decreased from 1955 to 2001. - Can we conclusively say that taxes have depressed
labor supply? No. - Nonlabor income rose (dividends, interest)
- Attitudes may have changed
4Table 2.1
5Role of Economic Theory
- In reality, an unlimited number of factors change
over time, and could affect labor supply. - Economic theory helps isolate a small set of
variables that are important influences on
behavior.
6Role of Economic Theory
- Theory would suggest that person maximizes
utility -- and would include factors like the
persons own wage rate. - Theory is often too simple may ignore important
considerations. But whole point of model
building is to reduce a problem to its essentials.
7Role of Economic Theory
- In this labor supply case, the after-tax wage
changes with the policy. - Theory predicts that the effect on hours is
ambiguous. - The substitution effect predicts that as the wage
(price of leisure) falls, consumers substitute
toward leisure. - The income effect says that if leisure is a
normal good, consumers consume less of it as
income falls. - Only empirical work analysis based on
observation as opposed to theory can answer
this question.
8Methods of Empirical Analysis
- There are three main methods
- Interviews
- Experiments
- Econometric studies
9Methods of Empirical Analysis
- Interviews
- Most straightforward way to find out whether a
policy will affect behavior is to ask. - Reporters often do this.
- Pitfalls of interviews
- The fact that people say something about their
behavior does not make it true. - People may be embarrassed about the subject or
have other agendas. - Can only do limited number of interviews.
10Methods of Empirical Analysis
- Social Experiments Although difficult, it is
not impossible to run policy experiments. - Requires random assignment.
- Pitfalls of experiments
- Selection issues, even if initial random
assignment. - Subjects know they are in experiment.
- Cost.
11Methods of Empirical Analysis
- Laboratory Experiments some economic theories
are tested in laboratory settings, often with
college students. Similar approach as used by
psychologists. - Usually offer different rewards.
- Setting is artificial, however.
- College students not really representative of
population as a whole.
12Methods of Empirical Analysis
- Econometrics statistical analysis of data.
- Effects of policies are inferred from the
analysis of observed behavior.
13Methods of Empirical Analysis
- Choose specific algebraic form to summarize the
relationship. For instance
- Where L is hours worked, wn is the net wage, and
A, X1, and X2 are other factors that affect work.
?0- ?4 are the parameters, and ? is a random
error.
14Figure 2.1
15Methods of Empirical Analysis
- Ignoring all of the other factors except for the
wage rate, the goal is to fit a line through
these data points. - No straight line can fit through them, but the
purpose of multiple regression analysis is to
find the parameters that has the best fit. - The slope of such a regression line gives the
regression coefficient on the wage rate.
16Methods of Empirical Analysis
- When ?10, the net wage has no impact on hours
worked. - When ? 1gt0, the net wage increases work.
Substitution effect dominates. - When ? 1lt0, the net wage decreases work. Income
effect dominates. - Presence of random error reflects influences on
labor supply that are unobservable to the
investigator.
17Methods of Empirical Analysis
- In practice, method does not always lead to
conclusive results. - After ? 1 is estimated, its reliability must be
considered. Is it close to the truth? The
standard error indicates how much the estimated
parameter can vary from its true value, and when
the standard error is small in relation to the
estimated parameter, the coefficient is
statistically significant.
18Methods of Empirical Analysis
- Pitfalls of econometric analysis
- Heterogeneity across groups
- Changes in parameters over time
- Omitted variables bias
- Some variables, such as motivation, are
inherently unmeasurable. - Reverse causality (simultaneity)
- Observed variables dont always correspond to
theory - Hours of work is not the same as work effort.
19Recap of Tools of Positive Analysis
- Role of Theory
- Methods of empirical analysis