Empirical Methods

1
Empirical Methods

• 7.5 credits, Intermediate Level
• Lecturer Mårten Palme
• TA Emma von Essen

2
Goal of the Course

• To give basic knowledge on how to write an
  empirical paper in economics.
• Basic knowledge in statistical tools.
• How to estimate and interpret a regression model.

3
Seminars and Assignments

• Altogether 10 seminars
• First 2 are voluntary classes in handling
  statistical software.
• 7 of the rest 8 classes are compulsory.
• Assignments (available on the course homepage)
  should be completed in groups of at most 3
  students.

4
5 Lectures

• Introduction. Empirical questions in economics.
  Examples of empirical studies.
• Interpretation of the regression model.
  Endogeneity.
• The regression model. Part 2.
• Description how to write a term paper.

5
Literature

• Textbook Introduction to Econometrics. Brief
  Edition by James Stock and Mark Watson.
• Link to the R-guide A Brief Guide to R for
  Beginners in Econometrics.
• Assignments and handouts available on the course
  homepage.

6
Statistical Software

• STATA, SAS, SPSS, Minitab, Eviews.
• We will teach R in this course.
• Why??
• It is free!
• Can be used wherever you are (and when the course
  is finished).
• Available for LINUX, Windows and MacOS.

7
Grading

• The grading is done on the basis of the
  individual term paper.
• Also required
• Presence at 7 of 8 compulsory seminars.
• Passed on the 4 assignments.

8
Data

• We use 2 samples from the Swedish Level of Living
  Survey (LNU).
• Used for research and collected by the Swedish
  Institute for Social Research (SOFI) at Stockholm
  University.
• Random sample of the Swedish population.
• Sample size about 6,000 individuals (about 0.1
  of the population).
• Available for 1968, 1974, 1981, 1991 and 2000.

9
Why LNU 1991?

• Manageable. Register samples much larger (LINDA
  300,000, LOUISE contains the entire population).
• A cross-section, but can be extended to a panel,
  since it contains the same individuals.
• Lots of information on health, work environment,
  family situation, economic conditions.
• Good ducumentation at www.sofi.su.se
• 1991 more recent than 1968, 1974 and 1981
• 2000 not yet available, but can be used for
  undergraduate thesis work.

10
An Empirical Project

• Ask an empirical question.
• Previous literature.
• Data collection.
• Structure of an empirical paper
• Introduction
• Theoretical model
• Empirical strategy
• Description of data
• Results
• Conclusions

11
Ask an Empirical Question

• Which area are you interested in?
• Find appropriate data for the question.
• As you will see, a more precise question makes
  the data collection work (and the entire project)
  much more easy
• Examples.

12

• Regression model
• - dependent variable.
• - independent variable.
• - random disturbance (measures all variables
  omitted in the regression model, all
  measurement errors, all functional form
  errors).
• - parameters to be estimated in the model.

13
Examples on Empirical questions

• What is the background to the gender gap in
  earnings? LNU Yes
• What is the background to the earnings gap
  between natives and immigrants? LNU Yes
• Why is there a higher unemployment rate among
  immigrants in Sweden? LNU - Yes
• Why are earnings higher in big cities? LNU Yes
• What is the economic returns to education?
• What is the effect of class sizes on student
  learning? LNU No
• Are private schools better for student learning?
  LNU No
• What is the effect of labor market programs? LNU
  No
• What is the effect of the replacement level in UI
  on the duration of unemployment spells? LNU - No

14

• How does different properties of an apartment
  affect the price? LNU No
• How does income affect health and mortality? LNU
  Yes
• How does smoking affect earnings? LNU Yes
• Does the size of the police force affect the
  crime rate? LNU No
• Does the length of prison sentences affect the
  crime rate? LNU - No
• Do the marginal tax rate affect labor supply? LNU
  - No
• Do public pension systems affect the savings rate
  in an economy? LNU - No
• Do the pension system affect the retirement
  decision? LNU - Yes
• Do the replacement rate in the sickpay insurance
  system affect the utilization of the insurance
  (moral hazard)? LNU Yes
• Do sporting events affect the probability to call
  sick? LNU Yes
• Does the replacement rate in the disability
  insurance system affect the disability rate? LNU
  - Yes

15

• Does the PPP model hold? LNU No.
• What is the marginal propensity to consume in an
  economy?
• Do people invest rationally? Home bias?
• What affects economic growth?
• Inflation
• Investment
• Human capital
• Public spending and tax rates
• Democracy
• Corruption
• etc. LNU No Penn World Tables Yes.

16
Overview of Previous Literature

• Always a first step in an empirical paper What
  is the state of knowledge in this area?
• International scientific journals
  www.econlit.org, www.nber.org
• Sweden Ekonomisk debatt, Swedish Economic Policy
  Review.
• Google or Google scholar.

17
Data Collection/Data Sources

• Collect your own data? Feasible? Not very common
  in economics compared to other social sciences.
• Combine register data. The Swedish personal id
  number system is very useful for this strategy.
• Use an existing data set.
• Time series. Aggregate macro data. National
  accounts, financial series, labor for survey
  (www.konj.se or www.scb.se)
• Cross-section data. Sample of observations at a
  particular point of time. LNU, AKU (Labor Force
  Survey), HUS, LINDA, LOUISE, LISA, PSID, CPS

18
Data Collection/Data Sources

• Panel data. Same individuals could be observed in
  more than one time period. (LNU, AKU, HUS, LINDA,
  LOUISE, PSID, Country data (OECD, IMF, Penn World
  tables).
• More data.
• Allows us to study how individuals react to
  changes.
• Event data. Data on events and duration of states
  linked to individuals observed over a particular
  time period. HÄNDEL (AMS) on unemployment
  durations.

19
When You have decided on Data

• Read in data in the software. Text files often
  most convenient.
• Cross-section data most often organized as one
  observation per line.
• Time series organized in chronological order.
• Clean the data set.
• Read the documentation. What do the variables
  measure? What is the original source?
• Missing observation code? NA, ., 99999 etc.
• Variable types. Binary, ordinal, continuous.?
• In what unit are the variables measured?

20
Introduction

• Example What is the background to the gender
  earnings gap?
• Theoretical model or background
• The human capital theory. Observable differences
  such as educational investments or on-the-job
  training may be different (Becker, 1967).
• Theories of discrimination.
• Statistical discrimination. Discrimination in
  wage offers towards the group mean productivity.
• Preferential discrimination. Employers may have
  preferences for giving a lower wage to women,
  even if it is inefficient to do so.

21
Empirical strategy

• Regression model
• - dependent variable.
• - independent variable.
• - random disturbance (measures all variables
  omitted in the regression model, all
  measurement errors, all functional form
  errors).
• - parameters to be estimated in the model.

22
Example with Background to the Gender Gap in Wages

• lnW log of hourly wage rate.
• FEMALE binary (0, 1) indicator for being
  female.
• EDUC number of years of schooling.
• EXP number of years of work experience.

23
Estimation strategy

• First step estimate the unconditional gender gap
  by only including the dummy variable for FEMALE
  in the equation.
• Second step include the EDUC variable.
• Third step include also the EXP variables.
• There are, in addition, lots of other potential
  variables that can be included.

24
Presentation of data

• Data should be presented so that the study can be
  replicated by other researchers in the future.
• Data source. Such as LNU 1991.
• Population. What population do you make inference
  to?
• Sample size.
• How many observations were deleted and why?
• Table that defines the variables may be useful.
• Table containing descriptive statistics mean,
  standard deviation, max/min.

25
Variable Description
26
Descriptive Statistics
27
Results

• Estimation results.
• Start with the most simple model in most cases
  the bivariate regression.
• Present then the alternative models.
• All results should be included in the same model.
• Discuss economic and statistical significance.
• Expected sign?
• Are the estimates statistically significant?
• What about the magnitude of the results? Economic
  significance.
• Joint significance tests? F-tests.
• Possible problems and caveats with the analysis?

28
Table. Estimation results from earnings equation.
Dependent variable hourly wage rate. Number of
observations 2,249.