Applied Econometrics: Topics in Development Economics and Economic History

1 / 26
About This Presentation
Title:

Applied Econometrics: Topics in Development Economics and Economic History

Description:

Applied Econometrics: Topics in Development Economics and Economic ... Self-selection problem (vegetarians) Survivor bias. Retrospective studies exit of firms ... –

Number of Views:352
Avg rating:3.0/5.0
Slides: 27
Provided by: Compu497
Category:

less

Transcript and Presenter's Notes

Title: Applied Econometrics: Topics in Development Economics and Economic History


1
Applied EconometricsTopics in Development
Economics and Economic History
  • Proseminar 2003/2004
  • M.A. Aravinda Meera and Dipl. Volkswirt Gerhard
    Kling

2
What is the scope of this course?
Introduction to development economics
Applied econometrics -mostly based on empirical
work -useful tools to analyze data
  • Economic history
  • long-term studies
  • experiences from industrialization

3
What is your contribution?
  • Term paper
  • Deadline for submission 3rd March 2004
  • Maximum 18 pages (inclusive figures etc.)
  • We recommend to work with data!
  • You can make your own proposals!
  • Presentation
  • About 20 minutes
  • Highlight your main findings
  • Paper and presentation in English or German

4
Time schedule
  • 13th Oct. How to get started? (Kling)
  • 20th Oct. Introduction to regression analysis
    (Kling)
  • 27th Oct. What is economic development? (Meera)
  • 3rd Nov. Poverty, undernutrition, and health
    (Meera)
  • 10th Nov. Population studies (Meera)
  • 24th Nov. Violations of CLR assumptions (Kling)
  • 8th Dec. Panel Data Analysis (Kling)
  • 15th Dec. Gender studies (Meera)

5
A short introduction to applied econometricsPart
A How to get started?
  • presented by
  • Dipl. Volkswirt Gerhard Kling

6
The structure of an empirical analysis
The problem ? derived hypotheses
Is the Harrod-Domar model valid?
Developing countries over time
Method of sampling
Empirical model
Theoretical model is modified
Empirical findings
Direction of impact correct
Conclusion
7
Example Harrod-Domar model
8
Specific model assumptions
9
Deriving your hypothesis
10
How to draw a sample?
  • Data source Worldbank
  • All countries worldwide
  • Time period 1960 1999
  • Several macroeconomic figures
  • Download as Excel file easy to implement in
    statistical software packages
  • Which GDP/capita measure should be used?
  • Market prices, PPP, etc.
  • Not so important because growth rates!
  • SPSS and STATA to work with the data

11
What is a perfect sample?
  • Sample size The larger the better?
  • Multicollinearity increase size
  • Square root n rule accuracy of estimation
  • Tests if sample size is sufficient MC, Chow etc.
  • Note Statistical inference tries to draw
    conclusions regarding the population by using a
    sample!
  • Population determines sufficient size Indian
    heights versus behavior of German companies

12
What is a perfect sample?
  • Biased sample and selection bias
  • Deviation of the sample from population
  • Special groups are neglected
  • Self-selection problem (vegetarians)
  • Survivor bias
  • Retrospective studies exit of firms
  • Market index problems

13
What is a perfect sample?
  • Is a random draw a way out?
  • Not always possible self collected data
  • Pseudo randomness
  • Random selection criteria name of the company
  • Alternative Monte Carlo Simulations or re-sampling

14
How does the sample look like?
First we focus on the year 1999
Missing values!
This is a typical panel structure
15
Developing an empirical model
  • From theory to empirical specification
  • We use the theoretical relation
  • Add an error term and constant stochastic model
  • Also theoretical magnitude and direction of
    impact
  • Pure empirical model
  • More or less based on theoretical considerations
  • Danger of data fishing
  • Specification problems (over-specification and
    OV)
  • Later discussed in detail!

16
Empirical model in the case of HD
17
How to distinguish savings s and ??
18
Predetermine the capital-output ratio
19
Descriptive statistics
Histogram of GDP/capita growth 1999
Close to a normal distribution
More positive values!
20
Why are histograms useful?
  • Detect non-linearity logarithm
  • Sometimes transformation is sensible
  • Express non-linearity difficult to test for it!
  • Decrease in marginal effects plausible!
  • Multi-modal distributions
  • Distinguish into subgroups forms of takeovers
  • Similar procedure Kernel density

21
Scatter plots
Visualize relation between two variables
Outliners detectable!
22
Example for outliners
population growth of about 16 in Rwanda is far
from being normal. It stems from the stream of
refugees caused by the civil war in central
Africa.
Mistake!
Two extreme values!
23
Why are scatter plots useful?
  • Detect non-linearity inverted U
  • Critical values and thresholds
  • Identify outliners
  • Alternative methods are available but simple
    tool thus, keep your gunpowder dry!

24
How to handle outliners?
  • Dont skip outliners too fast exception error
    in data
  • Try to explain outliners dummy for war period
  • But problem of over-specification
  • Run your regressions with and without outliners

25
Conclusion
  • Deriving the theoretical hypothesis
  • Constructing an empirical model
  • Data collection
  • Some limitations regarding availability
  • Savings easier to get
  • Descriptive statistics
  • Detect some severe biases
  • Useful tools to visualize your data

26
Original datasets (downloadable)
  • For capital output ratio calculation
  • Nehru Vikram and Ashok Dhareshwan (1993)A
    New Database on Physical Capital Stock Sources,
    Methodology and Results
  • ? Gross national investment, saving rate and
    growth rate of GDP per capita
  • William Easterly and Hairong Yu (Worldbank)
  • Global Development Network Growth Database
Write a Comment
User Comments (0)
About PowerShow.com