stochastic frontier models

49
[Part 3] 1/49 Stochastic FrontierModels Stochastic Frontier Model Stochastic Frontier Models William Greene Stern School of Business New York University 0 Introduction 1 Efficiency Measurement 2 Frontier Functions 3 Stochastic Frontiers 4 Production and Cost 5 Heterogeneity 6 Model Extensions 7 Panel Data 8 Applications

Upload: aurora

Post on 23-Feb-2016

80 views

Category:

Documents


0 download

DESCRIPTION

William Greene Stern School of Business New York University. Stochastic Frontier Models. 0Introduction 1 Efficiency Measurement 2 Frontier Functions 3 Stochastic Frontiers 4 Production and Cost 5 Heterogeneity 6 Model Extensions 7 Panel Data 8 Applications. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Stochastic Frontier Models

[Part 3] 1/49

Stochastic FrontierModels

Stochastic Frontier Model

Stochastic Frontier Models

William GreeneStern School of BusinessNew York University

0 Introduction1 Efficiency Measurement2 Frontier Functions3 Stochastic Frontiers4 Production and Cost5 Heterogeneity6 Model Extensions7 Panel Data8 Applications

Page 2: Stochastic Frontier Models

[Part 3] 2/49

Stochastic FrontierModels

Stochastic Frontier Model

Stochastic Frontier Models Motivation:

Factors not under control of the firm Measurement error Differential rates of adoption of technology

Frontier is randomly placed by the whole collection of stochastic elements which might enter the model outside the control of the firm.

Aigner, Lovell, Schmidt (1977), Meeusen, van den Broeck (1977), Battese, Corra (1977)

Page 3: Stochastic Frontier Models

[Part 3] 3/49

Stochastic FrontierModels

Stochastic Frontier Model

The Stochastic Frontier Model ( )

ln +

= + .

iviii

i i ii

i i

= fy eTE = + v uy

+

x

x

x

ui > 0, but vi may take any value. A symmetric distribution, such as the normal distribution, is usually assumed for vi. Thus, the stochastic frontier is

+’xi+vi

and, as before, ui represents the inefficiency.

Page 4: Stochastic Frontier Models

[Part 3] 4/49

Stochastic FrontierModels

Stochastic Frontier Model

Least Squares EstimationAverage inefficiency is embodied in the third moment of the

disturbance εi = vi - ui.

So long as E[vi - ui] is constant, the OLS estimates of the slope parameters of the frontier function are unbiased and consistent. (The constant term estimates α-E[ui]. The average inefficiency present in the distribution is reflected in the asymmetry of the distribution, which can be estimated using the OLS residuals:

3

1

1 ˆˆ( - [ ])N

N

3 i ii

= Em

Page 5: Stochastic Frontier Models

[Part 3] 5/49

Stochastic FrontierModels

Stochastic Frontier Model

Application to Spanish Dairy Farms

Input Units Mean Std. Dev.

Minimum

Maximum

Milk Milk production (liters)

131,108 92,539 14,110 727,281

Cows # of milking cows 2.12 11.27 4.5 82.3

Labor

# man-equivalent units

1.67 0.55 1.0 4.0

Land Hectares of land devoted to pasture and crops.

12.99 6.17 2.0 45.1

Feed Total amount of feedstuffs fed to dairy cows (tons)

57,941 47,981 3,924.14

376,732

N = 247 farms, T = 6 years (1993-1998)

Page 6: Stochastic Frontier Models

[Part 3] 6/49

Stochastic FrontierModels

Stochastic Frontier Model

Example: Dairy Farms

Page 7: Stochastic Frontier Models

[Part 3] 7/49

Stochastic FrontierModels

Stochastic Frontier Model

The Normal-Half Normal Model

2

2

ln

1Normal component: ~ [0, ]; ( ) , .

Half normal component: | |, ~ [0, ]

1 Underlying normal: ( ) ,

Half

i i i i

i i

ii v i i

v v

i i i u

ii i

u u

y v u

vv N f v v

u U U N

Uf U v

xx

1 1normal ( ) ,0(0)

ii i

u u

uf u v

Page 8: Stochastic Frontier Models

[Part 3] 8/49

Stochastic FrontierModels

Stochastic Frontier Model

Normal-Half Normal Variable

Page 9: Stochastic Frontier Models

[Part 3] 9/49

Stochastic FrontierModels

Stochastic Frontier Model

The Skew Normal Variable

2

2

2 2

| | where ~ [0,1]

2 2[ ] ; [ ]

[( 2) / ][ ][ ] [( 2) / ]

u

u u

u

v u

u U U N

E u Var u

Var uVar

Page 10: Stochastic Frontier Models

[Part 3] 10/49

Stochastic FrontierModels

Stochastic Frontier Model

Standard Form: The Skew Normal Distribution

Page 11: Stochastic Frontier Models

[Part 3] 11/49

Stochastic FrontierModels

Stochastic Frontier Model

Battese Coelli Parameterization2 2

2 2

2 22 2 2

2 2 2

, ~ [0, ], ~ [0, ]Aigner, Lovell, Schmidt

0, = 0

Coelli, Battese and Coelli

0 1, 0; = 1

v u

uv u

v

uv u

v u

v u v N u N

Page 12: Stochastic Frontier Models

[Part 3] 12/49

Stochastic FrontierModels

Stochastic Frontier Model

Estimation: Least Squares/MoM OLS estimator of β is consistent E[ui] = (2/π)1/2σu, so OLS constant estimates

α+ (2/π)1/2σu

Second and third moments of OLS residuals estimate

Use [a,b,m2,m3] to estimate [,,u, v]

and 0 2 2 3

2 u v 3 u- 2 2 4 = + = 1 - m m

Page 13: Stochastic Frontier Models

[Part 3] 13/49

Stochastic FrontierModels

Stochastic Frontier Model

Log Likelihood Function

Waldman (1982) result on skewness of OLS residuals: If the OLS residuals are positively skewed, rather than negative, then OLS maximizes the log likelihood, and there is no evidence of inefficiency in the data.

Page 14: Stochastic Frontier Models

[Part 3] 14/49

Stochastic FrontierModels

Stochastic Frontier Model

Airlines Data – 256 Observations

Page 15: Stochastic Frontier Models

[Part 3] 15/49

Stochastic FrontierModels

Stochastic Frontier Model

Least Squares Regression

Page 16: Stochastic Frontier Models

[Part 3] 16/49

Stochastic FrontierModels

Stochastic Frontier Model

Page 17: Stochastic Frontier Models

[Part 3] 17/49

Stochastic FrontierModels

Stochastic Frontier Model

Alternative Models:Half Normal and Exponential

Page 18: Stochastic Frontier Models

[Part 3] 18/49

Stochastic FrontierModels

Stochastic Frontier Model

Normal-Exponential Likelihood

2 2n

ui=1

Ln ( ; ) =

(( ) / ( )1-ln ln2

v u

u i i v u i i

v v u

L data

v u v u

Page 19: Stochastic Frontier Models

[Part 3] 19/49

Stochastic FrontierModels

Stochastic Frontier Model

Normal-Truncated Normal2

2

2

2

1

2

~ [0, ]

~ [ , ], | | Nonzero mean for

log log log2 2log2

1 log2

where 1

i v

i u i i

i

u

N i ii

u

v N

U N u UU

NL

Page 20: Stochastic Frontier Models

[Part 3] 20/49

Stochastic FrontierModels

Stochastic Frontier Model

Truncated Normal Model: mu=.5

Page 21: Stochastic Frontier Models

[Part 3] 21/49

Stochastic FrontierModels

Stochastic Frontier Model

Effect of Differing Truncation Points

From Coelli, Frontier4.1 (page 16)

Page 22: Stochastic Frontier Models

[Part 3] 22/49

Stochastic FrontierModels

Stochastic Frontier Model

Other Models Other Parametric Models (we will examine

several later in the course) Semiparametric and nonparametric – the recent

outer reaches of the theoretical literature Other variations including heterogeneity in the

frontier function and in the distribution of inefficiency

Page 23: Stochastic Frontier Models

[Part 3] 23/49

Stochastic FrontierModels

Stochastic Frontier Model

A Possible Problem with theMethod of Moments

Estimator of σu is [m3/-.21801]1/3

Theoretical m3 is < 0

Sample m3 may be > 0. If so, no solution for σu . (Negative to 1/3 power.)

Page 24: Stochastic Frontier Models

[Part 3] 24/49

Stochastic FrontierModels

Stochastic Frontier Model

Now Include LM in the Production Model

Page 25: Stochastic Frontier Models

[Part 3] 25/49

Stochastic FrontierModels

Stochastic Frontier Model

Page 26: Stochastic Frontier Models

[Part 3] 26/49

Stochastic FrontierModels

Stochastic Frontier Model

Test for Inefficiency? Base test on u = 0 <=> = 0 Standard test procedures

Likelihood ratio Wald Lagrange

Nonstandard testing situation: Variance = 0 on the boundary of the parameter

space Standard chi squared distribution does not apply.

Page 27: Stochastic Frontier Models

[Part 3] 27/49

Stochastic FrontierModels

Stochastic Frontier Model

Page 28: Stochastic Frontier Models

[Part 3] 28/49

Stochastic FrontierModels

Stochastic Frontier Model

Estimating ui

No direct estimate of ui

Data permit estimation of yi – β’xi. Can this be used? εi = yi – β’xi = vi – ui Indirect estimate of ui, using E[ui|vi – ui] This is E[ui|yi, xi]

vi – ui is estimable with ei = yi – b’xi.

Page 29: Stochastic Frontier Models

[Part 3] 29/49

Stochastic FrontierModels

Stochastic Frontier Model

Fundamental Tool - JLMS

2

( )[ | ] , 1 ( )

i ii i i i

i

E u

We can insert our maximum likelihood estimates of all parameters.Note: This estimates E[u|vi – ui], not ui.

2

ˆ ˆˆ ˆˆ ( ) ( )ˆ ˆ ˆˆ[ | ] , ˆ ˆ ˆ( )1i i i

i i i ii

yE u

x

Page 30: Stochastic Frontier Models

[Part 3] 30/49

Stochastic FrontierModels

Stochastic Frontier Model

Other Distributions

2 2

2

2

( / )| = + , = - /

( / )

i u vi

vii it i v i i v u

vi

zE u z z

z

For the Normal- Truncated Normal Model

For the Normal-Exponential Model

Page 31: Stochastic Frontier Models

[Part 3] 31/49

Stochastic FrontierModels

Stochastic Frontier Model

Technical Efficiency

* 2** * *

**

2 2* 2 2 2 u v

i u * 2

[( / ) ][exp( ) | ] exp

[( / )] 2

where = + / and

ii i i

i

i

E u

For the Normal- Truncated Normal Model

For the normal-half normal model, = 0.

Page 32: Stochastic Frontier Models

[Part 3] 32/49

Stochastic FrontierModels

Stochastic Frontier Model

Application: Electricity Generation

Page 33: Stochastic Frontier Models

[Part 3] 33/49

Stochastic FrontierModels

Stochastic Frontier Model

Estimated Translog Production Frontiers

Page 34: Stochastic Frontier Models

[Part 3] 34/49

Stochastic FrontierModels

Stochastic Frontier Model

Inefficiency Estimates

Page 35: Stochastic Frontier Models

[Part 3] 35/49

Stochastic FrontierModels

Stochastic Frontier Model

Inefficiency Estimates

Page 36: Stochastic Frontier Models

[Part 3] 36/49

Stochastic FrontierModels

Stochastic Frontier Model

Estimated Inefficiency Distribution

Page 37: Stochastic Frontier Models

[Part 3] 37/49

Stochastic FrontierModels

Stochastic Frontier Model

Estimated Efficiency

Page 38: Stochastic Frontier Models

[Part 3] 38/49

Stochastic FrontierModels

Stochastic Frontier Model

Confidence Region

Horrace, W. and Schmidt, P., Confidence Intervals for Efficiency Estimates, JPA, 1996.

Page 39: Stochastic Frontier Models

[Part 3] 39/49

Stochastic FrontierModels

Stochastic Frontier Model

Application (Based on Electricity Costs)

Page 40: Stochastic Frontier Models

[Part 3] 40/49

Stochastic FrontierModels

Stochastic Frontier Model

A Semiparametric Approach

Y = g(x,z) + v - u [Normal-Half Normal] (1) Locally linear nonparametric regression

estimates g(x,z) (2) Use residuals from nonparametric regression

to estimate variance parameters using MLE (3) Use estimated variance parameters and

residuals to estimate technical efficiency.

Page 41: Stochastic Frontier Models

[Part 3] 41/49

Stochastic FrontierModels

Stochastic Frontier Model

Airlines Application

Page 42: Stochastic Frontier Models

[Part 3] 42/49

Stochastic FrontierModels

Stochastic Frontier Model

Efficiency Distributions

Page 43: Stochastic Frontier Models

[Part 3] 43/49

Stochastic FrontierModels

Stochastic Frontier Model

Nonparametric Methods - DEA

Page 44: Stochastic Frontier Models

[Part 3] 44/49

Stochastic FrontierModels

Stochastic Frontier Model

DEA is done using linear programming

Page 45: Stochastic Frontier Models

[Part 3] 45/49

Stochastic FrontierModels

Stochastic Frontier Model

Page 46: Stochastic Frontier Models

[Part 3] 46/49

Stochastic FrontierModels

Stochastic Frontier Model

Methodological Problems with DEA Measurement error Outliers Specification errors The overall problem with the

deterministic frontier approach

Page 47: Stochastic Frontier Models

[Part 3] 47/49

Stochastic FrontierModels

Stochastic Frontier Model

DEA and SFA: Same Answer?

Christensen and Greene data N=123 minus 6 tiny firms X = capital, labor, fuel Y = millions of KWH

Cobb-Douglas Production Function vs. DEA

Page 48: Stochastic Frontier Models

[Part 3] 48/49

Stochastic FrontierModels

Stochastic Frontier Model

Page 49: Stochastic Frontier Models

[Part 3] 49/49

Stochastic FrontierModels

Stochastic Frontier Model

Comparing the Two Methods.