148 views

Uploaded by Jennifer Lee

stochastif

- Lecture 4 Day 3 Stochastic Frontier Analysis
- 38th Annual Federal Reserve Bank of St. Louis Fall Conference
- Using GAUSS for Econometrics.pdf
- Econometric Theory%2FOrdinary Least Squares (OLS)
- 3151146
- Modeling and Forecasting Malaria in Tripura, INDIA using NOAA/AVHRR-Based Vegetation Health Indices
- regression
- Useful Stata Commands
- Linear Regression
- FRM_2017_Part_I_GARP_Book_2.pdf
- Chap 14
- Unit Plan May 2016
- BA H DSEC IiApplied Econometrics 5th Sem
- Technical Efficiency of Resource-Conserving Technologies in Rice -Wheat Systems the Case of Bihar and Eastern Uttar Pradesh in India
- Alkali Ref
- Introduction to Econometrics, Tutorial (3
- (9) Who Underreacts to Cash-flow News Evidence From Trading Between Individuals and Institutions
- Production and Cost
- 1-s2.0-S0161893811000743-main
- Regression Analysis

You are on page 1of 45

School of Economics

The University of Queensland, Australia

It is a parametric technique that uses standard

production function methodology.

The approach explicitly recognises that production

function represents technically maximum feasible

output level for a given level of output.

The Stochastic Frontier Analysis (SFA) technique

may be used in modelling functional relationships

where you have theoretical bounds:

Estimation of cost functions and the study of cost efficiency

Estimation of revenue functions and revenue efficiency

This technique is also used in the estimation of multi-output

and multi-input distance functions

Potential for applications in other disciplines

2

Some history.

Much of the work on stochastic frontiers began in

70s. Major contributions from Aigner, Schmidt,

Lovell, Battese and Coelli and Kumbhakar.

Ordinary least squares (OLS) regression

production functions:

fit a function through the centre of the data

assumes all firms are efficient

fit a frontier function over the data

assumes there is no data noise

3

Production Function

It is a relationship between output and a set of

input quantities.

We use this when we have a single output

In case of multiple outputs:

people often use revenue (adjusted for price differences) as

an output measure

It is possible to use multi-output distance functions to study

production technology.

the form:

q f ( x1 , x2 ,..., xN ) v

4

A number of different functional forms are used in the

literature to model production functions:

Quadratic (in inputs)

Normalised quadratic

Translog function

N

1 N N

ln q 0 n ln xn nm ln xn ln xm u

2 n 1 m1

n 1

Translog function is very commonly used it is a generalisation of the

Cobb-Douglas function

It is a flexible functional form providing a second order approximation

Cobb-Douglas and Translog functions are linear in parameters and can

be estimated using least squares methods.

It is possible to impose restrictions on the parameters (homogeneity

5

conditions)

Linear in logs

Advantages:

easy to estimate and interpret

requires estimation of few parameters: K+3

Disadvantages:

simplistic - assumes all firms have same

production elasticities and that subsitution

elasticities equal 1

Quadratic in logs

Advantages:

flexible functional form - less restrictions on

production elasticities and subsitution

elasticities

Disadvantages:

more difficult to interpret

requires estimation of many parameters:

K+3+K(K+1)/2

can suffer from curvature violations

Functional forms

Cobb-Douglas:

Translog:

+ 0.522(lnx2i)2 + 12lnx1ilnx2i + vi - ui

Cobb-Douglas:

Production elasticity for j-th input is: Ej = j

Scale elasticity is: = E1+E2

Translog:

Eji = j+ j1lnx1i+ j2lnx2i

Scale elasticity for i-th firm is: i = E1i+E2i

Note: If we use transformed data where inputs are

measured relative to their means, then Translog

elasticities at means would simply be i.

9

Using sample data file which comes with the

FRONTIER program

H0: 11=22=12=0, H1: H0 false

Chi-square (r) under Ho.

For example, if:

LLF1=-14.43, LLF0=-17.03

LR=-2[-17.03-(-14.43)]=5.20

Since 32 5% table value = 7.81 => do not reject H0

10

In this model all the errors are assumed to be

due to technical inefficiency no account is

taken of noise.

Consider the following simple specification:

xis are in logs and include a constant

ui is a non-negative random variable.

Therefore ln yi xi .

Given the frontier nature of the model we can

measure technical efficiency using:

11

We have

yi

exp( xi ui )

TE i

exp( ui )

exp( xi )

exp( xi )

exp(xi) which is non-random.

Estimation of Parameters:

Linear programming approach

Min ui ln yi xi

i

subject to ui 0.

12

Aigner and Chu suggested a Quadratic

prrogramming approach

Min

ui

i

ln yi xi

subject to ui 2 0.

distribution for ui and the use of maximum

likelihood estimation.

Corrected ordinary least squares [COLS]

If we apply OLS, intercept estimate is biased

downwards, all other parameters are unbiased.

13

So COLS suggests that the OLS estimator from

OLS be corrected.

If we do not wish to make use of any probability

distribution for yi then

o (COLS) o (OLS) max imumi ui : i 1,2,...,N

where u i is the OLS residual for i-th firm.

If we assume that ui is distributed as Gamma

then

2

14

Production functions/frontiers

Deterministic

SFA

OLS

15

Production functions/frontiers

OLS:

qi = 0 + 1xi + vi

Deterministic :

qi = 0 + 1xi - ui

SFA:

qi = 0 + 1xi + vi - ui

where

vi = noise error term - symmetric

(eg. normal distribution)

ui = inefficiency error term - non-negative

(eg. half-normal distribution)

16

We start with the general production function as before and add a

new term that represents technical inefficiency.

This means that actual output is less than what is postulated

by the production function specified before.

We achieve this my subtracting u from the production

function

Then we have

q f ( x1 , x2 ,..., xN ) v u

can write the stochastic frontier function for the i-th firm as:

ln qi 0 1 ln xi vi ui

qi exp(0 1 ln xi vi ui )

17

Stochastic frontiers

deterministic frontier

qi = exp(0 + 1 ln xi)

yi

q*B ? exp(0 + 1ln xB + vB)

noise effect

noise effect

inefficiency effect

qB ? exp(0 + 1ln xB + vB uB)

inefficiency

effect

xA

xB

18

qi exp(0 1 ln xi ) exp(vi ) exp(ui )

noise

inefficiency

deterministic

component

In general, we write the stochastic frontier model with several inputs and

a general functional form (which is linear in parameters) as

ln qi xi vi ui

By construction the inefficiency term is always between 0 and 1.

This means that if a firm is inefficient, then it produces less than what is

expected from the inputs used by the firm at the given technology.

We can define technical efficiency as the ratio of observed or realised

output to the stochastic frontier output

qi

exp(xi vi ui )

TEi

exp(ui )

exp(xi vi )

exp(xi vi )

19

Stochastic Specifications

The SF model is specified as:

ln qi xi vi ui

and u.

independence is assumed for elements of vi.

distributed non-negative random variables.

the parameters. Standard distributions used are:

Half-normal (truncated at zero)

ui

iidN (0, u2 ).

Exponential

Gamma distribution

20

var = 1

var= 4

var = 9

-0.5

0.0

0.5

1.0

Distribution of u: ui

1.5

2.0

2.5

3.0

3.5

iidN (0, u2 ).

with mean equal to 0, E(u) is towards zero and therefore

technical efficiency tends to be high just by model

construction.

21

2.5

2

mu = -2

1.5

f(x)

mu = -1

mu = 0

mu = 1

0.5

mu = 2

0

0

This forms the basis for the inefficiency effects model where

u N ( i , u2 )

i 0 k Z ki

k

22

Estimation of SF Models

Parameters to be estimated in a standard SF model

2

2

are: , v and u

Likelihood methods are used in estimating the

unknown parameters. Coelli (1995)s Montecarlo

study shows that in large samples MLE is better than

COLS.

Usually variance parameters are reparametrized in

the following forms.

2 v2 u2

and

2 u2 v2 0.

2

2

2 and u / .

depends upon the parametrization used.

23

In the case of translog model, it is a good idea to

transform the data divide each observation by its

mean

Then the coefficients of ln Xi can be interpreted

as elasticities.

Most standard packages such as SHAZAM and

LIMDEP.

FRONTIER by Coelli is a specialised program for

purposes of estimating SF models.

Available for free downloads from CEPA

website: www.uq.edu.au/economics/cepa

24

Table The FRONTIER Instruction File

1

chap9.txt

chap9_2.out

1

y

344

1

344

10

n

n

n

DATA FILE NAME

OUTPUT FILE NAME

1=PRODUCTION FUNCTION, 2=COST FUNCTION

LOGGED DEPENDENT VARIABLE (Y/N)

NUMBER OF CROSS-SECTIONS

NUMBER OF TIME PERIODS

NUMBER OF OBSERVATIONS IN TOTAL

NUMBER OF REGRESSOR VARIABLES (Xs)

MU (Y/N) [OR DELTA0 (Y/N) IF USING TE EFFECTS MODEL]

ETA (Y/N) [OR NUMBER OF TE EFFECTS REGRESSORS (Zs)]

STARTING VALUES (Y/N)

time-varying inefficiency effects (we will come to this shortly)

The program uses the ratio of variances as the transformation

It allows for the use of single cross-sections as well as panel data

sets

25

FRONTIER output

the final mle estimates are :

coefficient

beta

beta

beta

beta

beta

beta

0

1

2

3

4

5

standard-error

t-ratio

0.27436347E+00

0.15110945E-01

0.53138167E+00

0.23089543E+00

0.20327381E+00

-0.47586195E+00

0.39600416E-01 0.69282978E+01

0.67544802E-02 0.22371736E+01

0.79213877E-01 0.67081892E+01

0.74764329E-01 0.30883101E+01

0.44785423E-01 0.45388387E+01

0.20221150E+00 -0.23532883E+01

beta 6

0.60884085E+00

beta 7

0.61740289E-01

beta 8

-0.56447322E+00

beta 9

-0.13705357E+00

beta10

-0.72189747E-02

sigma-squared 0.22170997E+00

0.16599693E+00 0.36677839E+01

0.13839069E+00 0.44613038E+00

0.26523510E+00 -0.21281996E+01

0.14081595E+00 -0.97328160E+00

0.92425705E-01 -0.78105703E-01

0.24943636E-01 0.88884383E+01

gamma

0.88355629E+00 0.36275231E-01

mu is restricted to be zero

eta is restricted to be zero

log likelihood function = -0.74409920E+02

0.24357013E+02

26

SF Models - continued

Predicting Firm Level Efficiencies:

Once the SF model is estimated using MLE method, we

compute the following:

ui* (ln qi xi ) u2 / 2

and

*2 v2 u2 / 2 .

and compute the best predictor of technical efficiency for

each firm i :

ui*

TE i E exp(ui ) qi *

*2

ui*

exp ui* .

*

2

evaluate technical efficiency.

27

SF Models - continued

Industry efficiency:

Industry efficiency can be computed as the average of

technical efficiencies of the firms in the sample

Industry efficiency can be seen as the expected value of a

randomly selected firm from the industry. Then we have

u2

TE E exp(ui ) 2 u exp .

2

firms and the industry as a whole) can also be computed.

the case of DEA.

28

FRONTIER output

technical efficiency estimates :

firm

eff.-est.

1

0.77532384

2

0.72892751

3

0.77332991

341

0.76900626

342

0.92610064

343

0.81931012

344

0.89042718

mean efficiency = 0.72941885

efficiency.

29

Tests of hypotheses

e.g., Is there significant technical inefficiency?

H0: =0 versus H1: >0

Test options:

t-test

t-ratio = (parameter estimate) / (standard error)

[note that the above hypothesis is one-sided therefore must use Kodde and Palm critical

values (not chi-square) for LR test

LR test safer

30

Steps:

1) Estimate unrestricted model (LLF1)

2) Estimate restricted model (LLF0)

(eg. set =0)

3) Calculate LR=-2(LLF0-LLF1)

4) Reject H0 if LR>R2 table value,

where R = number of restrictions

(Note: Kodde and Palm tables must be

used if test is one-sided)

31

function using sample data file which

comes with the FRONTIER program 344 firms

t-ratio for = 24.36, and N(0,1) critical

value at 5% = 1.645 => reject H0

Or the LR statistic = 28.874, and Kodde

and Palm critical value at 5% = 2.71 =>

reject H0

distribution

32

Distributional assumptions

the truncated normal distribution

N(,2) truncated at zero

More general patterns

Can test hypothesis that =0 using t-test

or LR test

The restriction =0 produces the halfnormal distribution: |N(0,2)|

33

Scale efficiency

For a Translog Production Function (Ray,

1998)

An output-orientated scale efficiency measure

is:

SEi = exp[(1-i)2/2]

where i is the scale elasticity of the i-th firm

and

K K

jk

j 1 k 1

Then SE is in the range 0 to 1.

34

Comments

We note the following points with respect to SFA

models

It is important to check the regularity conditions

associated with the estimated functions local and

global properties

This may require the use of Bayesian approach to impose

inequality restrictions required to impose convexity and

concavity conditions.

case of multi-output and multi-input production

functions.

It is possible to estimate scale efficiency in the case of

translog and Cobb-Douglas specifications

35

Investigate technical efficiency change (TEC)

Investigate technical change (TC)

More data = better quality estimates

Less chance of a one-off event (eg. climatic) influencing

results

Can use standard panel data models

no need to make distributional assumption

but must assume TE fixed over time

(time points)

36

Some Special cases:

1.

2.

Time varying effects: Kumbhakar (1990)

uit 1 exp(bt ct ) ui

3.

(1992)

uit exp (t T ui

In FRONTIER Program, this is under Error Components

Model.

37

0

1

BC92 ( = -.01)

10

11

12 13

K90 ( =

14

15

= -.02)

BC92 ( = .1)

over

time. These trends are also independent of any other data on the

firms. There is scope for further work in this area.

38

Technical efficiency is influenced by exogenous factors that

characterise the environment in which production takes place

Government regulation, ownership, education level of the farmer, etc.

In this case firm-level technical efficiency levels predicted will vary

with traditional inputs and environmental variables.

ln qi xi zi vi ui

Inefficiency effects model (Battese, Coelli 1995)

N ( zit , u2 )

program, this is the TEEFFECTS model

39

Current research

There is scope to conduct a lot of research in this area.

Some areas where work is being done and still be done

are:

Modelling movements in inefficiency over time

incorporating exogenous factors driving inefficiency

effects

Possibility of covariance between the random disturbance

and the distribution of the inefficiency term

Endogeniety due to possible effects of inefficiency and

technical change on the choice of input variables

Modelling risk into efficiency estimation and

interpretation

40

Current research

We have seen how technical efficiency can be computed,

but it is difficult to compute standard errors.

Peter Schmidt and his colleagues have been working on a

number of related topics here.

Bootstrap estimators and confidence intervals for

efficiency levels in SF models with pantel data

Testing whether technical inefficiency depends on firm

characteristics

On the distribution of inefficiency effects under

different assumptions

Posterior distribution of technical efficiencies

Estimation of distance functions

41

Facilities

Residential aged care is a multi-billion dollar industry in

Australia

Considered even more important in view of an ageing

population.

Aged care facilities are funded by the Commonwealth

government and are run by local government,

community/religious and private organisations.

Efficiency of residential aged care facilities is considered

quite important in view of reducing costs.

CEPA conducted a study for the Commonwealth

Department of Health and aged Care.

Data was collected by the Department from the

residential aged care facilities.

42

Facilities

Data:

912 Aged Care Facilities

30% response rate response bias

Data validation

Actual observations used: 787

Methods:

DEA and SFA Methods

Peeled DEA

43

Facilities

Preferred model:

Output variables:

High care weighted bed days

Low care weighted bed days

Input variables:

Floor area in square meters

Labour costs

Other costs

44

General findings

Average technical efficiency was calculated to be 0.83. 17%

cost savings could be achieved if all ACFs operated on the

frontier.

Results consistent between SFA and DEA models

Found variation across ACFs in different states (lowest 0.79

in Victoria and 0.87 in NSW/ACT).

Privately run ACFs has higher mean TE of 0.89.

Average scale efficiency of 0.93 was found.

An average size of 30 to 60 beds was found to be near optimal

scale.

A second stage Tobit model was run to see the factors driving

inefficiency.

Potential cost savings was calculated to be $316 m for the

number of firms in the sample greater savings for the

industry!!

Economies of Scope were examined using new methodology

developed but no significant economies were found.

45

- Lecture 4 Day 3 Stochastic Frontier AnalysisUploaded bytrpitono
- 38th Annual Federal Reserve Bank of St. Louis Fall ConferenceUploaded byFederal Reserve Bank of St. Louis
- Using GAUSS for Econometrics.pdfUploaded byGregort Kloy
- Econometric Theory%2FOrdinary Least Squares (OLS)Uploaded bySajib Jahan
- 3151146Uploaded byImmersohn Sohn
- Modeling and Forecasting Malaria in Tripura, INDIA using NOAA/AVHRR-Based Vegetation Health IndicesUploaded bySEP-Publisher
- regressionUploaded byNdivhuho Neosta
- Useful Stata CommandsUploaded bySuman Chattopadhyay
- Linear RegressionUploaded byrose00flower
- FRM_2017_Part_I_GARP_Book_2.pdfUploaded byMohammad Khataybeh
- Chap 14Uploaded bywhist888
- Unit Plan May 2016Uploaded byD Dávíd Fung
- BA H DSEC IiApplied Econometrics 5th SemUploaded bythanks naved
- Technical Efficiency of Resource-Conserving Technologies in Rice -Wheat Systems the Case of Bihar and Eastern Uttar Pradesh in IndiaUploaded bynirajpjoshi
- Alkali RefUploaded byOde Abah
- Introduction to Econometrics, Tutorial (3Uploaded byagonza70
- (9) Who Underreacts to Cash-flow News Evidence From Trading Between Individuals and InstitutionsUploaded byfatiaputri
- Production and CostUploaded bycarendleon
- 1-s2.0-S0161893811000743-mainUploaded byZaheer Khan
- Regression AnalysisUploaded byFreddie Yuan
- Optimization of an Accelerated ..Uploaded byvasek28
- Trade Flows Ans Spatial Effects - The Gravity Model RevisitedUploaded byChristian Pacheco
- Powerpoint 2017 (1)Uploaded byTina Chen
- 517Uploaded byFendy Hananto
- Clad-01Uploaded byMaster Plop
- Proxy VariablesUploaded byWill Matcham
- Journal of Chemical Information and Modeling Volume 51 issue 10 2011 [doi 10.1021%2Fci200220v] Wang, Jui-Chih; Lin, Jung-Hsin; Chen, Chung-Ming; Perryman, Alex -- Robust Scoring Functions for Protein–Ligan.pdfUploaded byVictor Frankenstein
- Belbin Group AnalysisUploaded byMei Yin Ho
- Searching Appropriate Methods for Survey Data AnalysisUploaded byPrashanta Pokhrel
- 1Uploaded byHarsh Jain

- Student Partners Program (SPP) ChecklistUploaded byAdip Sukhija
- 2016 Saln FormUploaded byAlain Odartseam
- DisequilibriumUploaded byMadhu Mahesh Raj
- EMERGENCE OF NIGERIAN NATIONALISMUploaded byAdeshina Adedunmoye Seyi
- Summary of EC102 - Economics BUploaded byKiṣn Mandaliya
- Syllabus J375_Eric Zhao Fall 2017 (2)Uploaded byphilmore
- kmintroUploaded bydatsabbath
- Managerial Economics - TestsUploaded bysl00red
- Catálogo PC200 8M0 PC200LC 8M0 Inglés DigitalUploaded byLUIS NUÑEZ
- Britannia to Restructure Its Top ManagementUploaded byNipul Bafna
- Pl Tax Compliance j14 Exam PaperUploaded bykarlr9
- Tata NYK ShippingUploaded bysameer15847
- Puckane Development Association Submission to North Tipperary County Council on Draft County Development Plan 2010-2016 (Nov 20th 2009 Final Copy)Uploaded byPuckane Pocan
- Chap 006Uploaded byHyunjoo Noh
- Individual Asset Allocation ExerciseUploaded byWaqas Sher Zaman
- Project Charter EUploaded byDunie Navarro
- Women_Workers_in_the_Maquiladoras_and_th.pdfUploaded bySald9398
- Ch3Uploaded byMadyoka Raimbek
- NPC v Cabanatuan DigestsUploaded bypinkblush717
- Eco561 r9 Student Guide Week5Uploaded byJohn Altman
- Gatsby MarxistUploaded byAshik Mahmud
- Dam Dan LuyUploaded byyencoi90
- Register of Members Interests Dáil TD 2000Uploaded bylostexpectation
- CBAUploaded byJelena Brašnović
- the-global-risks-report-2018.pdfUploaded byrazvancalinescu
- Taxation of CorporationsUploaded byGlory Mhay
- Study 20Materials 20 20(Short 20Notes)Uploaded byDickson Neethiraj
- Case Study on Maruthi Sujuki India Ltd Employees StrikeUploaded byDiya Jagtap
- gdUploaded byMahesh Narvankar
- Master Circular on IRAC NormsUploaded bysoham