Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd Logo
Upload

Welcome to Scribd!

  • 50 views

    Ekonometrika Terapan Lecture 1 2022

    Uploaded by

    erick
    This document provides an outline and introduction to the topics that will be covered in the econometrics course. It discusses what econometrics is, the difference between econometrics and statistics, examples of data types used in econometrics like experimental and observational data, and methods used to establish causal relationships like regression and difference-in-differences. It also covers potential biases in observational data like omitted variable bias and how econometrics approaches can help address these biases to recover causal effects. Finally, it gives examples of how experimental and quasi-experimental data have been used to inform policymaking.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Ekonometrika Terapan Lecture 1 2022

    Uploaded by

    erick
    50 views34 pages
    This document provides an outline and introduction to the topics that will be covered in the econometrics course. It discusses what econometrics is, the difference between econometrics and statistics, examples of data types used in econometrics like experimental and observational data, and methods used to establish causal relationships like regression and difference-in-differences. It also covers potential biases in observational data like omitted variable bias and how econometrics approaches can help address these biases to recover causal effects. Finally, it gives examples of how experimental and quasi-experimental data have been used to inform policymaking.

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document provides an outline and introduction to the topics that will be covered in the econometrics course. It discusses what econometrics is, the difference between econometrics and statistics, examples of data types used in econometrics like experimental and observational data, and methods used to establish causal relationships like regression and difference-in-differences. It also covers potential biases in observational data like omitted variable bias and how econometrics approaches can help address these biases to recover causal effects. Finally, it gives examples of how experimental and quasi-experimental data have been used to inform policymaking.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    50 views34 pages

    Ekonometrika Terapan Lecture 1 2022

    Uploaded by

    erick
    This document provides an outline and introduction to the topics that will be covered in the econometrics course. It discusses what econometrics is, the difference between econometrics and statistics, examples of data types used in econometrics like experimental and observational data, and methods used to establish causal relationships like regression and difference-in-differences. It also covers potential biases in observational data like omitted variable bias and how econometrics approaches can help address these biases to recover causal effects. Finally, it gives examples of how experimental and quasi-experimental data have been used to inform policymaking.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 34

    Ekonometrika

    Terapan
    Week 1
    29 Agustus 2022
    Outline

    01 02 03
    What is econometrics What is regression Data type in econometrics

    04
    Application of econometrics in
    policy

    2
    What is
    econometrics?

    Econometrics may be defined as the


    quantitative analysis of actual
    economic phenomena based on the
    concurrent development of theory
    and observation, related by
    appropriate methods or inference
    (Paul Samuelson)
    What is econometrics

    Theory confirmation?
    Policy/impact evaluation?
    Forecasting?
    Give me an example!
    Econometrics vs
    Statistics

    Statistics:
    Statistical Inference
    Population vs Sample

    Econometrics:
    Statistical Inference
    Causal Inference
    Counterfactuals
    Econometrics vs
    Data Science

    Data Science:
    Concerns with Prediction
    Curve/data fitting

    Econometrics:
    Causal Inference/
    Relationship
    Regression
    Regression is the most common way of estimating the relationship
    between two variables.

    OLS estimates the line that has


    best fitting by minimizing
    squared residuals.

    𝜷 is the coefficient that


    estimates the relationship
    between X and Y.

    Are we done?

    What of the relationship?


    Is it Causal?
    Causal relationship vs
    correlation

    There is a positive Is Leaving the lights


    relationship between on at night, will
    drowning and ice cream. cause children to
    suffer from short
    Does ice cream cause sightedness?
    drowning?

    What is not taken into What is not taken


    account? into account?

    Correlation ≠Causality

    Econometrics is concerned with methods that


    ensures causality
    Any question?
    Ceteris Paribus
    Theoretical descriptions of causal effects contained within economic model
    (comparative statics) always based on the idea of ceteris paribus—or “all else
    constant.
    ” When we are trying to describe the causal effect of some intervention, for instance,
    we are always assuming that the other relevant variables in the model are not
    changing. Otherwise, we cannot separate which one is the effect of X and which one is
    from other than X
    If they were changing, then they would be correlated with the variable of interest and
    it would confound our estimation.
    In the previous example, is the ceteris paribus assumption fulfilled? In some cases
    What variable/s are changing and not controlled in the model? ceteris paribus is
    These variables which are not in the models are called confounders. interpreted as
    balance
    What should we do to get estimate relationship? Is finding the best fit between
    between 2 variables with OLS regression enough? covariates.
    Demand elasticity example
    Q = dependent variable; P = variable of interest
    𝝐 = error term

    What we want
    Reality we have
    Source: Wright (1928) in Cunningham (2018)
    Demand elasticity example

    In the example above, the price elasticity of demand is the solution to the following equation:

    But in this example, we assume the change in P is exogenous. For instance, it holds supply
    fixed, the prices of other goods fixed, income fixed, preferences fixed, input costs fixed,
    and so on.
    Changes in P are completely and utterly independent of the otherwise normal determinants
    of supply and the other determinants of demand.
    Otherwise we get shifts in either supply or demand, which creates new pairs of data for which
    any correlation between P and Q will not be a measure of the elasticity of demand as
    depicted in the right panel.

    How to solve it?


    Demand elasticity example

    Given theoretical foundation, we must write out an econometric model as a starting point. One
    possible example of an econometric model would be a linear demand function:

    Where X is a matrix of factors that determine demand like the prices of other goods or
    income.

    We need two things to estimate price elasticity of demand:


    1. We need numerous rows of data on price and quantity.
    2. We need for the variation in price in our imaginary data set to be independent of u. We call
    this kind of independence exogeneity.
    Any question?
    Bias source/confounders

    Some examples of biases that threatened causality:


    1. Selection bias → covariates that determined selection (for
    example for intervention/policy) are not equal.
    2. Omitted variable bias → Important variable/s that affects
    dependent variable is missing from the model (u) and it is
    correlated with variable of interest (x).
    3. Measurement errors → X is measured with errors
    4. etc
    Data type:
    Experimental data vs Observational Data
    Experimental data: this is IDEAL DATA to establish Observational data: data that have been generated
    causality as we generate X and ’isolate’ everything else by something other than a randomized experiment-
    other than X (we will come back into this topic later) typically surveys,censuses, or administrative records.
    making sure ceteris paribus holds.
    Example: SUSENAS, SAKERNAS, IFLS, PODES, tax
    Example: To see whether the sleeping light affect data, DAPODIK data, etc.
    children eyesight, we assign parent-children pair into
    two groups randomly. One group assigned to sleeping We must be very careful in using it since generally
    light and the other asked to turn off the light. data collected in this way cannot be used to observe
    causality readily (ceteris paribus does not hold).
    The idea is like treatment/control group in medical
    experiment where the controls were given placebo
    Quasi Experiment

    There are chances for observational data to be used in making causal inference,
    as long as it close enough to make any variable of interest (X) exogenous.

    We apply modern econometrics approaches (or use opportunity from natural


    experiment- we will come back to this later), that enables as to approach ceteris
    paribus condition, to observational data to estimate causal relationship.

    This is what we will learn in the course and has been the concern of many
    econometricians → cue last year Economics Nobel Prize
    Econometric Methods to recover causal
    relationship (𝛽)
    - Randomized controlled trials (RCT)/experiments → the golden rule
    - Difference in Differences (DiD)
    - Instrumental Variables
    - Fixed effect approach
    - Propensity Score Matching (PSM)
    - Synthetic Control
    - Regression Discontinuity (RD)
    - Good Controls
    - etc

    In many cases these approaches are used in combination


    Any questions?
    Data Structure

    ● Cross Sectional Data


    ● Time Series Data
    ● Pooled Cross Section
    ● Panel Data
    Cross Section Data

    ● A cross-sectional dataset consists of a sample of individuals,


    households, firms, … taken at a given point in time.

    ● Cross-sectional datasets are often obtained from random sampling


    from the underlying population.

    ● If the sample has not been drawn randomly, our methods may
    have to be adjusted. For now, we assume random sampling unless
    I say otherwise.
    Cross Section Data

    o Display here data wage1.dta from Stata


    Time Series Data

    ● A time series data set consists of observations on one


    or several variables over time.
    ● Unlike the arrangement of cross-sectional data, the
    chronological ordering of observations in a time series
    is important.
    ● A key feature of time series data that makes them more
    difficult to analyze than cross-sectional data is that
    observations are unlikely to be independent over time.
    ● Special methodological problems arise when we
    analyze time series data.
    Data on Minimum Wage for Puerto Rico.
    Avgmin is the average minimum wage
    Avgcov is the percentage of workers covered by the minimum wage law.
    Unem is the unemployment rate.
    GNP is the gross national product.
    11
    Pooled Cross Section Data

    ● Some datasets have both cross-sectional and time series


    features.
    ● Example: household surveys from 1985 and 1990 which are
    combined to yield one dataset containing observations from both
    years.
    ● May be a useful basis for analysis of change of policy, for
    example, we often include time (year) as an additional
    explanatory variable in regressions based on pooled cross-
    sections.
    susenas sakernas data series
    are pooled cross section
    12
    IDs
    are different

    13
    Panel (Longitudinal) Data

    A panel dataset consists of a time series for each cross-sectional


    member in the data set.
    Example:

    14
    Any questions?
    Example of experimental data use
    in policy making
    Example of experimental data use
    in policy making
    Example of quasi experimental data use
    in policy making
    Example of quasi experimental data use
    in policy making
    active

    -Januar
    - Marthinus
    - Evan
    - Cintaka
    - Hening
    - Diana
    - Dolly
    - Febrian
    - Deasy

    You might also like