PRJ 5078 Capstone Research Project 1
Lecture 5: Econometrics: Multivariate Analysis
Shandre M Thangavelu
THE LINEAR REGRESSION MODEL (LPM)
ØThe general form of the LPM model is:
• Yi = B1 + B2X2i + B3X3i + … + BkXki + ui
ØMultivariate: more than one independent variables
ØTake note of unit of analysis: i=individual, firms,
organization,
ØOr, as written in short form:
• Yi = BX + ui
Ø[Reduced-form representation
ØY is the regressand (dependent), X is a vector of regressors
(independent), and u is an error term (iid).
Øu is identically and independently distributed and random
– not affected by individuals – Example: individual 1 does
not affect individual 2
POPULATION (TRUE) MODEL
Yi = B1 + B2X2i + B3X3i + … + BkXki + ui
ØThis equation is known as the population or true
model.
ØIt consists of two components:
Ø(1) A deterministic component, BX (the conditional mean of
Y, or E(Y|X)).
Ø(2) A nonsystematic, or random component, ui.
REGRESSION COEFFICIENTS
ØB1 is the intercept.
ØB2 to Bk are the slope coefficients.
ØCollectively, they are the regression coefficients or
regression parameters.
ØEach slope coefficient measures the (partial) rate
of change in the mean value of Y for a unit change
in the value of a regressor, ceteris paribus.
SAMPLE REGRESSION FUNCTION
ØThe sample counterpart is:
• Yi = b1 + b2X2i + b3X3i + … + bkXki + ei
ØOr, as written in short form:
• Yi = bX + ei
where e is a residual.
ØThe deterministic component is written as:
Ù
Yi = b1 + b2 X 2i + b3 X 3i + ... + bk X ki = bX
THE NATURE OF THE Y VARIABLE
ØRatio Scale:
ØRatio of two variables, distance between two variables, and ordering
of variables are meaningful. [output per worker, consumption per
household, etc]
ØInterval Scale:
ØDistance and ordering between two variables meaningful, but not
ratio. [Wage gap between skilled and unskilled, income distribution
using Gini coefficients, etc]
ØOrdinal Scale:
ØOrdering of two variables meaningful (not ratio or distance). [height,
weight, number of goods, number of workers, etc.]
ØNominal Scale:
ØCategorical or dummy variables, qualitative in nature. [scale of
satisfaction: 1, 2, 3,..; scale of Happiness: 1, 2, 3, ..
THE NATURE OF DATA
ØTime Series Data
ØA set of observations that a variable takes at different times, such
as daily (e.g., stock prices), weekly (e.g., money supply),
monthly (e.g., the unemployment rate), quarterly (e.g., GDP),
annually (e.g., government budgets), quinquenially or every
five years (e.g., the census of manufactures), or decennially or
every ten years (e.g., the census of population).
Time-Series Data
Cross sectional data
THE NATURE OF DATA
ØCross-Section Data
ØData on one or more variables collected at the same point in
time.
ØExamples are the census of population conducted by the Census
Bureau every 10 years, opinion polls conducted by various
polling organizations, and temperature at a given time in several
places.
THE NATURE OF DATA
ØPanel, Longitudinal or Micro-panel Data
ØCombines features of both cross-section and time series data.
ØSame cross-sectional units are followed over time.
ØPanel data represents a special type of pooled data (simply time
series, cross-sectional, where the same cross-sectional units are
not necessarily followed over time).
ØPrimary data: Census, administrative data, survey
ØSecondary data: Aggregated data, national sources, Penn
World Tables
METHOD OF ORDINARY LEAST SQUARES
ØMethod of Ordinary Least Squares (OLS) does not
minimize the sum of the error term, but minimizes
error sum of squares (ESS):
å i å i 1 2 2 i 3 3i
u 2
= (Y - B - B X - B X - .... - Bk X ki ) 2
ØTo obtain values of the regression coefficients, derivatives
are taken with respect to the regression coefficients and set
equal to zero.
SRF: Sample Regression Function - Linear
Regression (identify outliers)
Conditional Distribution: variance is the same,
homogenous
Variance is not homogenous:
heteroscedasticity
No correlation in errors
Model Misspecification
CLASSICAL LINEAR REGRESSION MODEL
ØAssumptions of the Classical Linear Regression
Model (CLRM):
ØA-1: Model is linear in the parameters.
ØA-2: Regressors are fixed or nonstochastic.
ØA-3: Given X, the expected value of the error term is
zero, or E(ui |X) = 0.
CLASSICAL LINEAR REGRESSION MODEL
ØAssumptions of the Classical Linear Regression Model
(CLRM):
ØA-4: Homoscedastic, or constant, variance of ui, or
var(ui|X) = σ2.
ØA-5: No autocorrelation, or cov(ui,uj|X) = 0, i ≠ j.
ØA-6: No multicollinearity, or no perfect linear
relationships among the X variables.
ØA-7: No specification bias.
ØR-square: Goodness of fit, R-square close to 1
GAUSS-MARKOV THEOREM
ØOn the basis of assumptions A-1 to A-7, the OLS method
gives best linear unbiased estimators (BLUE):
Ø(1) Estimators are linear functions of the dependent
variable Y.
Ø(2) The estimators are unbiased; in repeated applications
of the method, the estimators approach their true values.
Ø(3) In the class of linear estimators, OLS estimators have
minimum variance; i.e., they are efficient, or the “best”
estimators.
Consumption Estimation
