Econometrics I

Professor William Greene

Stern School of Business
Department of Economics

6-1/41 Part 6: Functional Form

 Econometrics I

Part 6 – Dummy Variables

and Functional Form

6-2/41 Part 6: Functional Form

 Agenda
 Dummy variables
 Interaction
 Categorical variables and transition tables
 Nonlinear functional form
 Differences
 Difference in differences
 Regression discontinuity
 Kinked regression

6-3/41 Part 6: Functional Form

 Monet in Large and Small
Sale prices of 328 signed Monet paintings

Fitted Line Plot

ln (US$) = 2.825 + 1.725 ln (SurfaceArea)
18 S 1.00645
R-Sq 20.0%
17 R-Sq(adj) 19.8%

16

15
ln (US$)

14

13

12

11

6.0 6.2 6.4 6.6 6.8 7.0 7.2 7.4 7.6

ln (SurfaceArea)

Log of $price = a + b log surface area + e

6-4/41 Part 6: Functional Form

 How Much for the Signature?

 The sample also contains 102 unsigned

paintings
Average Sale Price
Signed $3,364,248
Not signed $1,832,712

 Average price of a signed Monet is almost

twice that of an unsigned one.

6-5/41 Part 6: Functional Form

 A Multiple Regression
Scatterplot of ln (US$) vs ln (SurfaceArea)
18 Signed
0
17 1

16

15
d
ln (US$)

14

13

12

11

10
6.0 6.2 6.4 6.6 6.8 7.0 7.2 7.4 7.6
ln (SurfaceArea)

Ln Price = a + b  ln Area + d  (0 if unsigned, 1 if signed) + e

6-6/41 Part 6: Functional Form

 Monet Multiple Regression

Regression Analysis: ln (US$) versus ln (SurfaceArea), Signed

The regression equation is
ln (US$) = 4.12 + 1.35 ln (SurfaceArea) + 1.26 Signed
Predictor Coef SE Coef T P
Constant 4.1222 0.5585 7.38 0.000
ln (SurfaceArea) 1.3458 0.08151 16.51 0.000
Signed 1.2618 0.1249 10.11 0.000
S = 0.992509 R-Sq = 46.2% R-Sq(adj) = 46.0%

Interpretation:
(1) Elasticity of price with respect to surface area is 1.3458 – very large
(2) The signature multiplies the price of a painting by exp(1.2618) (about
3.5), for any given size.

6-7/41 Part 6: Functional Form

 A Conspiracy Theory
for Art Sales at
Auction

Sotheby’s and Christies, 1995 to

about 2000 conspired on
commission rates.

6-8/41 Part 6: Functional Form

 If the Theory is Correct…

Scatterplot of ln (US$) vs ln (SurfaceArea)

18
Sold from 1995 to 17
2000
16

Sold before 1995 15

or after 2000 14
ln (US$)

13

12

11

10

9
3 4 5 6 7 8 9
ln (SurfaceArea)

6-9/41 Part 6: Functional Form

 Evidence

The statistical
evidence seems to
be consistent with
the theory.

Effects on Price Unsigned Signed

Not 1995 - 2000 exp(0.0000) =1.0000 exp(1.2777) = 3.5884
1995 - 2000 exp(0.2009) =1.2225 exp(1.2777 + 0.2009) = 4.3868

6-10/41 Part 6: Functional Form

 Women appear to assess health satisfaction differently from men.

6-11/41 Part 6: Functional Form

 Or do they? Not when other things are held constant

6-12/41 Part 6: Functional Form

 Dummy Variable for One Observation
A dummy variable that isolates a single
observation. What does this do?
Define d to be the dummy variable in question.
Z = all other regressors. X = [Z,d]
Multiple regression of y on X. We know that
X'e = 0 where e = the column vector of
residuals. That means d'e = 0, which says that
ej = 0 for that particular residual. The
observation will be predicted perfectly.
Fairly important result. Important to know.
6-13/41 Part 6: Functional Form
 I have a simple question for you. Yesterday, I was
estimating a regional production function with yearly
dummies. The coefficients of the dummies are usually
interpreted as a measure of technical change with
respect to the base year (excluded dummy variable).
However, I felt that it could be more interesting to
redefine the dummy variables in such a way that the
coefficient could measure technical change from one
year to the next. You could get the same result by
subtracting two coefficients in the original regression but
you would have to compute the standard error of the
difference if you want to do inference.

Is this a well known procedure? YES

6-14/41 Part 6: Functional Form

6-15/41 Part 6: Functional Form
 Example with 4 Periods
The estimated model with time dummies is
y = a +b2*d2+ b3*d3 + b4*d4 + e (possibly some other variables, not needed now).
Estimated least squares coefficients are
b = a, b2, b3, b4
Desired coefficients are
c = a, b2, b3 – b2, b4 – b3
The original model is y = Xb + e.
The new model would be y = (XC)(C-1b) + e = Qc + e
The transformation of the data is Q = XC. c = C-1b
The transformed X is [1,d2+d3+d4, d3+d4.d4]

1 0 0 0 1 0 0 0
0 1 0 0  0 1 0 0 
C1   , C
0 1 1 0 0 1 1 0
   
0 0 1 1 0 1 1 1

6-16/41 Part 6: Functional Form

 A Categorical Variable

6-17/41 Part 6: Functional Form

6-18/41 Part 6: Functional Form
 Nonlinear Specification:
Quadratic Effect of Experience

6-19/41 Part 6: Functional Form

 Model Implication: Effect of Experience and
Male vs. Female

6-20/41 Part 6: Functional Form

 Partial Effect of Experience:
Coefficients do not tell the story

Education: .05654
Experience: .04045 - 2*.00068*Exp
FEM: -.38922

6-21/41 Part 6: Functional Form

 Effect of Experience = .04045 - 2 * 0.00068*Exp
Positive from 1 to 30, negative after.

6-22/41 Part 6: Functional Form

 Specification and Functional Form: Nonlinearity

Population Estimators
y  1  2 x  3 x 2  4 z   yˆ  b1  b2 x  b3 x 2  b4 z
E[ y | x, z ]
x   2  23 x ˆ x  b2  2b3 x
x

6-23/41 Part 6: Functional Form

 Log Income Equation
----------------------------------------------------------------------
Ordinary least squares regression ............
LHS=LOGY Mean = -1.15746 Estimated Cov[b1,b2]
Standard deviation = .49149
Number of observs. = 27322
Model size Parameters = 7
Degrees of freedom = 27315
Residuals Sum of squares = 5462.03686
Standard error of e = .44717
Fit R-squared = .17237
--------+-------------------------------------------------------------
Variable| Coefficient Standard Error b/St.Er. P[|Z|>z] Mean of X
--------+-------------------------------------------------------------
AGE| .06225*** .00213 29.189 .0000 43.5272
AGESQ| -.00074*** .242482D-04 -30.576 .0000 2022.99
Constant| -3.19130*** .04567 -69.884 .0000
MARRIED| .32153*** .00703 45.767 .0000 .75869
HHKIDS| -.11134*** .00655 -17.002 .0000 .40272
FEMALE| -.00491 .00552 -.889 .3739 .47881
EDUC| .05542*** .00120 46.050 .0000 11.3202
--------+-------------------------------------------------------------
Average Age = 43.5272. Estimated Partial effect = .066225 – 2(.00074)43.5272 = .00018.
Estimated Variance 4.54799e-6 + 4(43.5272)2(5.87973e-10) + 4(43.5272)(-5.1285e-8)
= 7.4755086e-08. Estimated standard error = .00027341.

6-24/41 Part 6: Functional Form

 Objective: Impact of Education
on (log) Wage

 Specification: What is the right model to

use to analyze this association?
 Estimation
 Inference
 Analysis

6-25/41 Part 6: Functional Form

 Application: Is there a relationship between (log) Wage and Education?

6-26/41 Part 6: Functional Form

 Group (Conditional) Means (Nonparametric)

6-27/41 Part 6: Functional Form

 Simple Linear Regression (semiparametric)

LWAGE = 5.8388 + 0.0652*ED

6-28/41 Part 6: Functional Form

 Multiple Regression

6-29/41 Part 6: Functional Form

 Interaction Effect
Gender Difference in Partial Effects

6-30/41 Part 6: Functional Form

 Partial Effect of a Year of Education
E[logWage]/ED=ED + ED*FEM *FEM
Note, the effect is positive.
Effect is larger for women.

6-31/41 Part 6: Functional Form

 Gender Effect Varies by Years of Education
-0.67961 is misleading

6-32/41 Part 6: Functional Form

 Difference in Differences
With two periods,

y it = y i2 -y i1 = 0 + (x i2 -x i1 )β + ui
Consider a "treatment, Di ," that takes place between
time 1 and time 2 for some of the individuals
y i = 0 + (x i )β + 1Di + ui
Di = the "treatment dummy"

This is a linear regression model. If there are no regressors,

ˆ
1  y | treatment - y | control
= "difference in differences" estimator.
ˆ
0  Average change in y i for the "treated"

6-33/41 Part 6: Functional Form

 Difference-in-Differences Model
With two periods and strict exogeneity of D and T,
y it = 0  1Dit  2 Tt  3 TtDit  it
Dit = dummy variable for a treatment that takes place
between time 1 and time 2 for some of the individuals,
Tt = a time period dummy variable, 0 in period 1,
1 in period 2.
This is a linear regression model. If there are no regressors,

Using least squares,

b3  (y 2  y1 )D1  (y 2  y1 )D 0

6-34/41 Part 6: Functional Form

 Difference in Differences
y it = 0  1Dit  2 Tt  3Dit Tt  βx it  it , t  1, 2
y it = 2  3Di 2  (βx it )  it
= 2  3Di 2  β(x it )  ui
 y it | D  1   y it | D  0 
 3  β (x it | D  1)  (x it | D  0) 
If the same individual is observed in both states,
the second term is zero. If the effect is estimated by
averaging individuals with D = 1 and different individuals
with D=0, then part of the 'effect' is explained by change
in the covariates, not the treatment.

6-35/41 Part 6: Functional Form

 SAT Tests

Using least squares,

d3  (Score2  Score1 ) TestPrep 1  (Score2  Score1 ) TestPrep=0
Potential x = Income, Parents' Education, GPA

6-36/41 Part 6: Functional Form

 Abrupt Effect on Regression at a Specific Level of x

6-37/41 Part 6: Functional Form

6-38/41 Part 6: Functional Form
 Useful Functional Form: Kinked Regression

6-39/41 Part 6: Functional Form

6-40/41 Part 6: Functional Form
 Kinked Regression and Policy Analysis: Unemployment Insurance

6-41/41 Part 6: Functional Form