228 CHAPTER 7 / DERIVATIVES IN USE

NOTE 1 Suppose we consider the Taylor formula on an interval about x = a instead of

x = 0. The first n + 1 terms on the right-hand side of (3) become replaced by those of
(7.5.4), and the new remainder is

1
Rn+1 (x) = f (n+1) (c)(x − a)n+1 (c is between x and a) (7)
(n + 1)!

It is easy to show that (7) follows from (1) and (2) by considering the function g defined by
g(t) = f (a + t) when t is close to 0.

PROBLEMS FOR SECTION 7.6

1. Write Taylor’s formula (3) with n = 2 for f (x) = ln(1 + x).

2. Use the approximation (1 + x)m ≈ 1 + mx + 21 m(m − 1)x 2 to find values of

√
3
√
5
(a) 25 33(b)
√
3 1/3
(Hint: Note that 25 = 3(1 − 2/27) .) Check these approximations by using a calculator.
√ √
3. Show that 3
9 = 2 (1 + 1/8)1/3 . Use formula (3) (with n = 2) to compute 3 9 to the third
decimal.
√
⊂⊃ 4.
SM Let g(x) = 3
1 + x.

(a) Find the Taylor polynomial of g(x) of order 2 about the origin.

(b) For x ≥ 0 show that |R3 (x)| ≤ 5x 3 /81.

√
(c) Find 3 1003 to 7 significant digits.

7.7 Why Economists Use Elasticities

Economists often study how demand for a certain commodity such as coffee reacts to price
changes. We can ask by how many units such as kilograms the quantity demanded will
change per dollar increase in price. In this way, we obtain a concrete number, measured
in units of the commodity per unit of money. There are, however, several unsatisfactory
aspects to this way of measuring the sensitivity of demand to price changes. For instance,
a $1 increase in the price of a kilo of coffee may be considerable, whereas a $1 increase
in the price of a car is insignificant.
This problem arises because the sensitivity of demand to price changes is being measured
in the same arbitrary units as those used to measure both quantity demanded and price. The
difficulties are eliminated if we use relative changes instead. We ask by what percentage
the quantity demanded changes when the price increases by 1%. The number we obtain in

Essential Math. for Econ. Analysis, 4th edn EME4_C07.TEX, 16 May 2012, 14:24 Page 228
 SECTION 7.7 / WHY ECONOMISTS USE ELASTICITIES 229

this way will be independent of the units in which both quantities and prices are measured.
This number is called the price elasticity of demand, measured at a given price.
In 1960, the price elasticity of butter in a certain country was estimated to be −1. This
means that an increase of 1% in the price would lead to a decrease of 1% in the demand,
if all the other factors that influence the demand remained constant. The price elasticity
for potatoes was estimated to be −0.2. What is the interpretation? Why do you think the
absolute value of this elasticity is so much less than that for butter?
Assume now that the demand for a commodity can be described by the function

x = D(p)

of the price p. When the price changes from p to p + %p, the quantity demanded, x, also
changes. The absolute change in x is %x = D(p + %p) − D(p), and the relative (or
proportional) change is
%x D(p + %p) − D(p)
=
x D(p)
The ratio between the relative change in the quantity demanded and the relative change in
the price is
,
%x %p p %x p D(p + %p) − D(p)
= = (∗)
x p x %p D(p) %p
When %p = p/100 so that p increases by 1%, then (∗) becomes (%x/x) · 100, which is
the percentage change in the quantity demanded. We call the proportion in (∗) the average
elasticity of x in the interval [p, p + %p]. Observe that the number defined in (∗) depends
both on the price change %p and on the price p, but is unit-free. Thus, it makes no difference
whether the quantity is measured in tons, kilograms, or pounds, or whether the price is
measured in dollars, pounds, or euros.
We would like to define the elasticity of D at p so that it does not depend on the size of
the increase in p. We can do this if D is a differentiable function of p. For then it is natural
to define the elasticity of D w.r.t. p as the limit of the ratio in (∗) as %p tends to 0. Because
the Newton quotient [D(p + %p) − D(p)]/%p tends to D ′ (p) as %p tends to 0, we obtain:

p dD(p)
The elasticity of D(p) with respect to p is
D(p) dp

Usually, we get a good approximation to the elasticity by letting %p/p = 1/100 = 1% and
computing p %x/(x %p).

The General Definition of Elasticity

The above definition of elasticity concerned a function determining quantity demanded
as a function of price. Economists, however, also consider income elasticities of demand,
when demand is regarded as a function of income. They also consider elasticities of supply,
elasticities of substitution, and several other kinds of elasticity. It is therefore helpful to see

Essential Math. for Econ. Analysis, 4th edn EME4_C07.TEX, 16 May 2012, 14:24 Page 229
 230 CHAPTER 7 / DERIVATIVES IN USE

how elasticity can be defined for a general differentiable function. If f is differentiable at x

and f (x) ̸ = 0, we define the elasticity of f w.r.t. x as

x
Elx f (x) = f ′ (x) (elasticity of f w.r.t. x) (1)
f (x)

EXAMPLE 1 Find the elasticity of f (x) = Ax b (A and b are constants, with A ̸ = 0).
Solution: In this case, f ′ (x) = Abx b−1 . Hence, Elx Ax b = (x/Ax b )Abx b−1 = b, so

f (x) = Ax b ,⇒ Elx f (x) = b (2)

The elasticity of the power function Ax b w.r.t. x is simply the exponent b. So this function
has constant elasticity. In fact, it is the only type of function which has constant elasticity.
This is shown in Problem 9.9.6.

EXAMPLE 2 Assume that the quantity demanded of a particular commodity is given by

D(p) = 8000p−1.5

Compute the elasticity of D(p) and find the exact percentage change in quantity demanded
when the price increases by 1% from p = 4.
Solution: Using (2) we find that Elp D(p) = −1.5, so that an increase in the price of 1%
causes quantity demanded to decrease by about 1.5%.
In this case we can compute the decrease in demand exactly. When the price is 4, the
quantity demanded is D(4) = 8000 · 4−1.5 = 1000. If the price p = 4 is increased by 1%,
the new price will be 4 + 4/100 = 4.04, so that the change in demand is

D(4.04) − D(4) = 8000 · 4.04−1.5 − 1000 = −14.81

The percentage change in demand from D(4) is −(14.81/1000) · 100 = −1.481%.

EXAMPLE 3 Let D(P ) denote the demand function for a product. By selling D(P ) units at price P ,
the producer earns revenue R(P ) given by R(P ) = P D(P ). The elasticity of R(P ) w.r.t.
P is

P d 1
ElP R(P ) = [P D(P )] = [D(P ) + P D ′ (P )] = 1 + ElP D(P )
P D(P ) dP D(P )

Observe that if ElP D(P ) = −1, then ElP R(P ) = 0. Thus, when the price elasticity of the
demand at a point is equal to −1, a small price change will have (almost) no influence on
the revenue. More generally, the marginal revenue dR/dP generated by a price change is
positive if the price elasticity of demand is greater than −1, and negative if the elasticity is
less than −1.

Essential Math. for Econ. Analysis, 4th edn EME4_C07.TEX, 16 May 2012, 14:24 Page 230
 SECTION 7.7 / WHY ECONOMISTS USE ELASTICITIES 231

NOTE 1
• If |Elx f (x)| > 1, then f is elastic at x.
• If |Elx f (x)| = 1, then f is unit elastic at x.
• If |Elx f (x)| < 1, then f is inelastic at x.
• If |Elx f (x)| = 0, then f is perfectly inelastic at x.
• If |Elx f (x)| = ∞, then f is perfectly elastic at x.

NOTE 2 If y = f (x) has an inverse function x = g(y), then Theorem 7.3.1 implies that

y ′ f (x) 1 1
Ely (g(y)) = g (y) = = (3)
g(y) x f ′ (x) Elx f (x)

A formula that corresponds to (7.3.5) is

1
Ely x = (4)
Elx y

There are some rules for elasticities of sums, products, quotients, and composite functions
that are occasionally useful. You might like to derive these rules by solving Problem 9.

Elasticities as Logarithmic Derivatives

Suppose that two variables x and y are related by the equation

y = Ax b (x, y, and A are positive) (5)

as in Example 1. Taking the natural logarithm of each side of (5) while applying the rules
for logarithms, we find that (5) is equivalent to the equation

ln y = ln A + b ln x (6)

From (6), we see that ln y is a linear function of ln x, and so we say that (6) is a log-linear
relation between x and y. The transformation from (5) to (6) is often seen in economic
models, sometimes using logarithms to a base other than e.
For the function defined by (5), we know from Example 1 that Elx y = b. So from (6)
we see that Elx y is equal to the (double) logarithmic derivative d ln y/d ln x, which is the
constant slope of this log-linear relationship.
This example illustrates the general rule that elasticities are equal to such logarithmic
derivatives. In fact, whenever x and y are both positive variables, with y a differentiable
function of x, a proof based on repeatedly applying the chain rule shows that

x dy d ln y
Elx y = = (7)
y dx d ln x

Essential Math. for Econ. Analysis, 4th edn EME4_C07.TEX, 16 May 2012, 14:24 Page 231
 232 CHAPTER 7 / DERIVATIVES IN USE

PROBLEMS FOR SECTION 7.7

1. Find the elasticities of the functions given by the following formulas:

√ √
(a) 3x −3 (b) −100x 100 (c) x (d) A/x x (A constant)

2. A study in transport economics uses the relation T = 0.4K 1.06 , where K is expenditure on
building roads, and T is a measure of traffic volume. Find the elasticity of T w.r.t. K. In this
model, if expenditure increases by 1%, by what percentage (approximately) does traffic volume
increase?

3. (a) A study of Norway’s State Railways revealed that, for rides up to 60 km, the price elasticity
of the volume of passenger demand was approximately −0.4. According to this study, what
is the consequence of a 10% increase in fares?
(b) The corresponding elasticity for journeys over 300 km was calculated to be approximately
−0.9. Can you think of a reason why this elasticity is larger in absolute value than the
previous one ?

4. Use definition (1) to find Elx y for the following (a and p are constants):
(a) y = eax (b) y = ln x (c) y = x p eax (d) y = x p ln x

5. Prove that Elx (f (x))p = p Elx f (x) (p is a constant).

6. The demand D for apples in the US as a function of income r for the period 1927 to 1941 was
estimated as D = Ar 1.23 , where A is a constant. Find and interpret the elasticity of D w.r.t. r.
(This elasticity is called the income elasticity of demand, or the Engel elasticity.)

7. Voorhees and colleagues studied the transportation systems in 37 American cities and estimated
the average travel time to work, m (in minutes), as a function of the number of inhabitants, N.
They found that m = e−0.02 N 0.19 . Write the relation in log-linear form. What is the value of m
when N = 480 000?

8. Show that
! "
Elx Af (x) = Elx f (x) (multiplicative constants vanish)
! " f (x) Elx f (x)
Elx A + f (x) = (additive constants remain)
A + f (x)

HARDER PROBLEMS

⊂⊃ 9.
SM Prove that if f and g are positive-valued differentiable functions of x and A is a constant, then
the following rules hold (where we write, for instance, Elx f instead of Elx f (x)).
(a) Elx A = 0 (b) Elx (f g) = Elx f + Elx g
% &
f f Elx f + g Elx g
(c) Elx = Elx f − Elx g (d) Elx (f + g) =
g f +g
f Elx f − g Elx g ! "
(e) Elx (f − g) = (f) Elx f g(x) = Elu f (u)Elx u (u = g(x))
f −g

Essential Math. for Econ. Analysis, 4th edn EME4_C07.TEX, 16 May 2012, 14:24 Page 232