Data Envelopment Analysis

Lesson 1. Conventional productivity

and efficiency concepts

1
 Outline for today

• 1.1. Importance of productivity measures.

• 1.2. Conventional concepts of productivity,
technical efficiency and technical change.
• 1.2.1. Non-frontier methodologies.
• 1.2.2. Frontier methodologies.
• 1.3. Effectiveness, quality and efficiency.

2
 1.1. Importance of productivity
measures.
• Performance measurement
– Productivity measures
• Mainly using partial productivity measures
• Cost, revenue and profit ratios
• Performance of public services and utilities
• Aggregate Level
– Growth in per capita income
– Labour and total factor productivity growth
– Sectoral performance
• Labour productivity
• Share in the total economy
• Industry Level
– Performance of firms
– Market and non-market goods and services
– Efficiency and productivity
– Banks, credit unions, manufacturing firms, agricultural farms, schools
and universities, hospitals, aged care facilities, etc.
• Need to use appropriate methodology to benchmark performance
3
Productivity:
• It measures the level of output per unit of input.
• Partial productivity measures – output per person employed;
output per hour worked; output per hectare etc.
• Total factor productivity measures – Productivity measure which
involves all the factors of production

Efficiency:
(i) How much more can we produce with a given level of inputs?
(ii) How much input reduction is possible to produce a given level of
observed output?
(iii) How much more revenue can be generated with a given level of
inputs? Similarly how much reduction in input costs be achieved?

4
 Partial indicators

• Can be misleading
• Consider two clothing factories (A and B)
• Labour productivity could be higher in firm
A – but what about use of capital and
energy and materials?
• Unit costs could be lower in firm B – but
what if there exists significant differences
in terms of quality?
5
1.2. Frontier and non frontier methods
• The terms productivity and efficiency are
different:
• Productivity = output/input
• Efficiency generally relates to some form
benchmark or target
• An example – where for firm B productivity
rises but efficiency falls:

6
 Production Technology
• We assume that there is a production technology that allows
transformation of a vector of inputs into a vector of
outputs
S = {(x,q): x can produce q}.
• Technology set is assumed to satisfy some basic axioms.
• It can be equivalently represented by
– Output sets
– Input sets
– Output and input distance functions
• A production function provides a relationship between the
maximum feasible output (in the single output case) for a
given set of input
• Single output/single input; single output/multiple inputs;
multi-output/multi-input
7
 Output and Input sets
• Output set P(x) for a given vector of inputs, x, is the set of all
possible output vectors q that can be produced by x.
P(x) = {q: x can produce q} = {q : (x,q)  S}
– P(x) satisfies a number of intuitive properties including: nothing can be
produced from x; set is closed, bounded and convex
– Boundary of P(x) is the production possibility curve
• An Input set L(q) can be similarly defined as set of all input vectors
x that can produce q.
L(q) = {x: x can produce q} = {x: (x , q)  S}
– L(q) satisfies a number of important properties that include:
closed and convex
– Boundary of L(q) is the isoquant curve
• These sets are used in defining the input and output distance
functions

8
Frontier concepts: Output
Distance Function
y2

B

y2A A C
 PPC-P(x)

P(x)

0 y1A y1

Do(x,y)
The value of the distance function is
equal to the ratio =0A/0B.

Technical Efficiency Measure:

TE = 0A/0B = do(x,q)
9
Frontier concepts: Input
Distance Function
x2

x2A A

B L(y)
Isoq-L(y)
C

0 x1A x1

Di(x,y)
The value of the distance function is
equal to the ratio =0A/0B.

Technical Efficiency measure:

= TE = 1/di(x,q) = OB/OA 10
 1.3. Effectiveness and efficiency
• The production technology defines the
technological constraints.
• The objective of the firm could be to maximise
profit
• Or minimise costs when outputs are fixed
• Or maximise revenue when inputs are fixed
• Or … other strategical goals (i.e. to maximize
the market share or to increase the customers’
loyalty)

11
 Profit maximisation

frontier

Profit max

x
12
 Cost minimisation
• The firm must produce output, q0
• Minimum cost is defined as:
c(q, w)  min {wx : (x, q)  S}
x
x1

Cost min

Isoquant (q=q0)
x2 13
 Revenue maximisation
• The firm has input allocation, x0
• Maximum revenue is defined as:
r (p, x)  max {pq : (x, q)  S}
q

y1 Revenue max

PPC (x=x0)
14
y2
 Example of output oriented efficiency analysis

x q1 q2 q1/x q2/x
A 10 40 2 4 0.2
B 10 20 5 2 0.5
C 10 10 20 1 2

15
16
ASSUMING A NON-CONVEX TECHNOLOGY

17
ASSUMING A CONVEX TECHNOLOGY

18
Output prices: p1=1; p2=5.

19
 Returns to Scale
• A production technology exhibits constant returns
to scale (CRS) if a Z% increase in inputs results in
Z% increase in outputs (ε = 1).
• A production technology exhibits increasing
returns to scale (IRS) if a Z% increase in inputs
results in a more than Z% increase in outputs (ε >
1).
• A production technology exhibits decreasing
returns to scale (DRS) if a Z% increase in inputs
results in a less than Z% increase in outputs (ε <
1).
20
 Returns to scale

DRS

CRS
IRS

x
21
 Efficiency Measures
• Using the distance functions defined so far, we can
define:
– Technical efficiency
– Allocative efficiency
– Economic efficiency
• A firm is said to be technically efficient if it operates on
the frontier of the production technology
• A firm is said to be allocatively efficient if it makes
efficient allocation in terms of choosing optimal input
and output combinations.
• A firm is said to be economically efficient if it is both
technically and allocatively efficient.

22
 Example of Technical Efficiency

q
Frontier

B Output orientation: TEO=DA/DB


C  Input orientation: TEI=EC/EA
E A

D x
23
 Productivity?
• productivity = output/input
• What to do if we have more than one input
and/or output?
– partial productivity measures
– aggregation

24
 Example
• Two firms producing t-shirts using labour
and capital (machines).
• The partial productivity ratios conflict.

firm labour capital output q/x1 q/x2

(x1) (x2) (q)
A 2 2 200 100 100
B 4 1 200 50 200
25
Non-frontier total factor productivity (TFP)
• Use an aggregate measure of input:
TFP = y/(a1x1+a2x2)
• What should we use as the weights? – prices?
• Data: Labour wage = $80 per day and
Rental price of the machines = $100 per day
• Calculation:
TFPA = 200/(80×2+100×2) = 200/360 = 0.56
TFPB = 200/(80×4+100×1) = 200/420 = 0.48
=>A is more productive using this measure.
26
Non-frontier total factor productivity (TFP)
• Can decompose TFP difference between 2
firms (at one point in time) into 3 types of
efficiency:
– technical efficiency,
– allocative efficiency and
– scale efficiency.

• Yes, we can! but …

“we need to know the technology”
27
 How do we measure efficiency?
• Depends upon the type of data available for the
measurement purpose.
• Three types:
– Observed input and output data for a given firm over
two periods or data for a few firms at a given point of
time (time-series data);
– Observed input and output data for a large sample of
firms from a given industry (cross-sectional data)
– Panel data on a cross-section of firms over time
• In the first case measurement is limited to
productivity measurement based on restrictive
assumptions.
28
 Overview of Methods:
• Non-frontier index numbers (IN)
– Price and quantity index numbers used in
aggregation.
• Frontier non-parametric methods (convex
or non-convex)
• Frontier parametric methods
(deterministic or stochastic methods (SFA)

29
Relative merits of non-frontier index numbers:

• Advantages:
– only need 2 observations
– transparent and reproducible
– easy to calculate
• Disadvantages:
– need price information
– Lack of theoretical framework
– cannot decompose in substantive terms

30
 Relative merits of frontier methods:

• Advantages of non-parametric methods:

 no need to specify functional form or
distributional forms for errors
 easy to accommodate multiple outputs

• Advantages of parametric methods:

 attempts to account for data noise
 can conduct hypothesis tests with statistical
significance
31
 Example of frontier analysis
(non-convex technology):

BENCHMARK ANALISIS (TWO INPUTS, ONE OUTPUT)

OUTPUT
FIRM (y) x 1/y x 2/y Quality level
A 1 4 1 2 (good)
B 1 2 2 1 (bad)
C 1 1 3 2 (good)
D 1 4 5 2 (good)
E 1 5 3 2 (good)
F 1 2 4 2 (good)

32
33
34
And that’s all folks!

35