CHAPTER

CONSUMERS
Marktes are made of buyers and sellers. In many cases, buyers are consumers and sellers
are firms. In this and in the next chapter we lay the behavioral foundations of consumers
and firms, respectively.

2.1. CONSUMER PREFERENCES AND DEMAND

Creating Demand: Move the Masses to Buy Your Product, Service, or Idea (Updated for the
21st Century). You will find this and many other such titles at your local bookstore (if it
still exists, which is an interesting demand-related question). If there is no demand there is
no business. Therefore, before deciding what to do in business its important to have some
knowledge of what the demand for your product is. The economists answer to this is think
of consumers as having preferences; and from this to derive their demand function: how
much they want to buy of a certain product as a function of a variety of factors, including
price.
Consumer tastes. The traditional theoretical setup for thinking about demand starts
with the tastes or preferences of an individual consumer. To see how this works, suppose
there are two products, A and B. We might express different combinations of purchases in
a diagram with quantities of product A on the vertical axis and quantities of product B
on the horizontal axis. The top left panel in Figure 2.1, for example, shows four possible
combinations, from E1 to E4 . A consumers preferences are reflected by a ranking over all
possible combinations of A and B. For example, a consumer might prefer the bundle E2
best among the four possibilities; be indifferent between E1 and E4 ; and prefer E3 the least.
We can represent our tastes by how various combinations are regarded. We do this
graphically with indifference curves, shown on the top right panel in Figure 2.1. For example,
we might think of 2 units of product A and 1 of product B (point E1 ) as equal to the
reverse (1 unit of product A and 2 units of product B, point E4 ). A line connecting all such
points about which we are indifferent (i.e., like equally well) is called an indifference curve.
Generally, indifference curves are downward-sloping, since we need more of a product to
compensate us for less of the other. (We call this the no satiation principle, which real
means that more is better.) We also assume they are curved away from the origin, since as
Forthcoming in Introduction to Industrial Organization, 2nd Ed. This draft: November 2, 2012.
c Lus Cabral.

Consumers

Figure 2.1
From consumer preferences to consumer demand
qA

qA
......
......
.......
.......
increasing

2
1

E
E2
... ... ... ... ... ... ... ... ... ... 1
..
..
..
..
E
E
... ... ... ... ... ... ... ... ... ..... ... ..3. ... ... ... ... ... ... ... 4
..
..
.
..
..
..
..
...
...

utility

qB

qB
Y
pB

qA

.... E 0 .... E

qB

b0

00

... ... ... ... ... ... ... ... ... ... ... ... ... .. E
..
..
..
..
..
..
..
.

Increase in Y
Initial budget constraint
Increase in pB

qA

qA

......
.......
........
....... ........
.
.
.
.
.
.
......
.......
........
....... ...
.
.......
....... .............
.
.
.
.
.
.
.
.
........ .............
.
.
.
.
.......
........

Y
pA

qB

..
..
..
..
..
..
.

..
..
..
..
..
..
.

00
qB

0
qB

qB

we consider combinations with more and more of one product, we need increasing amounts
of it to keep us equally satisfied.
If we put lots of indifference curves on our graph, we can get a complete description of
our tastes. Since more is better, indifference curves that are up and to the right (farther
from the origin) represent higher levels of satisfaction.
Consumer budgets. The other ingredient in our analysis is what the consumer can
afford: the budget set. With two products, the budget might be expressed by the inequality:
pA qA + pB qB Y,
where pi is the price of product i, qi the quantity purchased of product i (i = A, B), and Y
available income. If we solve for qA we see this is the area below the following downwardsloping line:
Y
pB
qA =

qB
pA
pA
which is line b on the top right panel in Figure 2.1. This is a straight line whose position
and slope depend on income and prices. If you increase income, for example, the budget line
shifts out. This is illustrated by line b0 in the top left panel of Figure 2.1. If you increase pB ,
the line rotates clockwise (around the vertical intercept, Y /pA ). This is illustrated by line b00
in the top left panel of Figure 2.1. And if you increase pA , the line rotates counterclockwise
(around the fixed horizontal intercept, Y /pB ).
Demand. Putting together tastes (represented by indifference curves) and possibilities

Consumers

(represented by the budget line) we can find out what our hypothetical consumer should do.
The consumers best choice corresponds to the point where the highest indifference curve
has a common point with the budget line intersects. This is illustrated in the bottom left
panel in Figure 2.1, where E is the best combination of A and B that the consumer can
afford with a budget represent by the budget line b.
and q . These are the quantities
The optimal point, E, corresponds to quantities qA
B
demanded by the consumer. Implicitly, these demands depend on tastes (which are built
into the indifference curves). They also depend on income (since a change in income shifts
the budget line and therefore leads to a change in demand) and prices (for the same reason).
We can summarize and abstract from the underlying indifference curves and budget sets
by writing down a demand function, qi (pi , z), denoting the quantity demanded, qi , for a
given price of the good, pi , and for given values of other variables, z, which might include
income, the prices of all other goods, and any other relevant factors that affect the demand
of good i. For example, if i is tickets to the Mets-Braves game on September 24, z might
include variables such as the weather forecast and the identity of the starting pitcher. (Can
you think of other ones? Box 2.1 includes a few.)
The demand curve gives the quantity demanded of a given good as a function of its price
and of other factors; it can be derived from consumer preferences.
Instead of qi (pi , z), we can equivalently write an inverse demand function, pi (qi , z),
which denotes what the price must be if the quantity demanded is to be equal to qi .
To summarize, this analysis suggests that the quantity of a product demanded by an
individual consumer depends on:
The individuals tastes, expressed by indifference curves.
The price of the product. Generally, the lower the price the higher the demand.
Depending on the curvature of the indifference curves, a change in price might have
a small or a large impact on the number of units demanded.
The price of other products. Decisions are not made in isolation. If we spend less on
one product, that necessarily leaves more to spend on others.
Income. At higher levels of income, we can buy more of everything (and generally
do).
Frequently, we graph qi as a function of pi only.1 We refer to the curve that shows qi
as a function of pi as the demand curve. Basically, it corresponds to the demand function
under the assumption that all variables other than pi are constant (that is, z is constant).
The bottom left panel of Figure 2.1 illustrates the process of deriving the demand
curve. Keeping Y and pA fixed, we change the value of pB . For example, we increase the
value of pB from p0B to p00B . This implies a clockwise rotation of the budget line around
the qA axis intercept. The consumers new optimal point is point E 0 . In other words, by
0
increasing the value of pB from p0B to p00B , the quantity of qB demand decreases from qB

Consumers

Figure 2.2
Consumer surplus: (a) demand for pizza slices and (b) general case.
p

$
3.0

A
price = 1.0
0.2

... ... ... ... ... ... ... ... ... ..

..
..
..
..
..
..
... ... ... ... ... ... ... ... ... ..... ... ... ... ... ... ... ... ... ...
.
.

# slices
3

..
..
..
..
..
..
..
..
..
..

D
q

00 . Repeating this exercise for all possible values of p , we get the demand curve for
to qB
B
product B.
It is important to distinguish between movements along the demand curve and shifts
in the demand curve itself. If pi changes, than the quantity demanded changes as we move
along the demand curve. If however some other variable such as pj (the price of another
good) changes, than qi changes because of a shift of the demand curve.

A change in price leads to a shift along the demand curve; a change in other factors leads
to a shift in the demand curve itself.

Consumer surplus. Before going to watch a movie, you stop at a fast-food pizza store
and place your order. Pizza comes at a dollar a slice. Imagine what would be the maximum
price you would pay for one pizza slice. Perhaps three dollars, especially if you are very
hungry and there is no alternative restaurant in the neighborhood. Consumers dont usually
think about this; all they need to know is that they are willing to pay at least one dollar
for that pizza slice. But, for the sake of argument, let us suppose that the maximum you
would be willing to pay is three dollars.
How about a second slice of pizza? While one slice is the minimum necessary to survive
through a movie, a second slice is an option. It makes sense to assume you would be willing
to pay less for a second slice than for the first slice; say, one dollar and 50 cents. How about
a third slice? For most consumers, a third slice would be superfluous. If you are going to
watch a movie, you might not have the time to eat it, anyway. If you were to buy a third
slice, you would probably only eat the toppings and little else. You wouldnt be willing to
pay more than, say, 20 cents.
Putting all of this information together, we have your demand curve for pizza. The left
panel in Figure 2.2 illustrates this. On the horizontal axis, we have the number of pizzas
you buy. On the vertical axis, we measure the willingness to pay, that is, the maximum
price (in dollars) at which you would still want to buy.
There are two things we can do with a demand curve. First, knowing what the price
is (one dollar per slice), we can predict the number of slices bought. This is the number

Consumers

of slices such that willingness to pay is greater or equal to price. Or, to use the demand
curve, the quantity demanded is given by the point where the demand curve crosses the
line p = 1: two slices.
A second important use of the demand curve is to measure the consumers surplus.
You would be willing to pay up to three dollars for one slice of pizza. That is, had the price
been $3, you would have bought one slice of pizza anyway. In fact, you only paid $1 for that
first slice. Since the pizza is the same in both cases, you are $3$1 = two dollars better
off than you would be had you bought the slice under the worst possible circumstances (or
not bought at all). Likewise, you paid 50 cents less for the second slice than the maximum
you would have been willing to pay. Your total surplus as a consumer is thus $2.50.
Consumer surplus is the difference between willingness to pay and price.
More generally, the consumer surplus is given by the area under the demand curve and
above the price paid by the consumer. This is illustrated in the right panel in Figure 2.2.
If price is p , then quantity demanded is q and consumer surplus is given by the shaded
area A.

2.2. DEMAND ELASTICITY

Having established the foundations of the demand curve, we are now interested in characterizing its shape, that is, understanding how it depends on a variety of factors: how
it depends on price of the product in question, but also on the prices of other goods, as
well as on the consumers income. For this purpose, economists use the concept of demand
elasticity.
Price elasticity of demand. We presume (based on lots of evidence) that the demand for
a product increases if we lower its price. This doesnt have to be the case, but it invariably
is. The question is how sensitive demand is to price. If I were to tell you that the world oil
demand decreases by 1.3 million barrels a day when price increases from 50 to 60 dollars
per barrel, would you consider the demand for oil very sensitive or not very sensitive to
price? Its hard to tell, unless you have a very good idea of the size of the oil market. What
if I told you that the demand for sugar in Europe decreases by 1 million tones per day
when average retail price increases from 80 to 90 euros cents per kilo: can you compare the
demand for sugar in Europe to the worldwide demand for oil? Its even more difficult.
The problem is that, by measuring the slope of the demand curve, we are stuck with
units: barrels, dollars, kilos, euros, and so on. The solution is to measure things in relative
terms, that is, in terms of percent changes. Specifically, we measure the sensitivity of
demand by the price elasticity of demand:
=
In words

dq
q
dp
p

dq p
q p

dp q
p q

Consumers

Table 2.1
Price elasticity of demand for selected products
Product

Elasticity

Milk

-0.5

Cigarettes

-0.5

Beer

-0.8

Apples

-1.3

US luxury cars in US

-1.9

Foreign luxury cars in US

-2.8

The price elasticity of demand is the ratio between the percentage change in quantity and
the percentage change in price, for a small change in price.
By small change in price we mean, strictly speaking, an infinitesimally small change
in price, something we represent by d p. In practice, we observe small but not infinitesimal
changes in p, which we denote by p. The estimate of the demand elasticity we obtain
based on such price variations is only approximately equal to the value of , thus we use
the sign in the above equation.
Note that elasticities are generally negative, since demand declines when price increases. The question is how negative. We say that products for which || > 1 are elastic, meaning the quantity demanded is very sensitive to price; the higher || is, the more
sensitive to price. Conversely, if 0 < || < 1 (elasticity is small), we say that demand is
inelastic, meaning that demand is relatively insensitive to price; the closer  is to zero,
the less sensitive demand is to price. Later we will see why the threshold 1 is so important
when classifying demand elasticity.
Table 2.1 presents the values of the demand elasticity for selected products. Would
you expect the demand for milk to be inelastic? Would you expect the demand for foreign
luxury cars in the U.S. to be elastic? Why?
Note that the elasticity is defined at a point: it generally differs from one point to
another along a demand curve. The left panel in Figure 2.3 considers the case of a linear
demand curve. As we go from the extreme when p is equal to zero to the extreme when q
is equal to zero, the value of  varies from 0 to . (You can check this by looking at the
definition of elasticity.) At some intermediate point (the mid point, if the demand curve is
linear), we have || = 1.
Although the value of demand elasticity varies from point to point, there is such
thing as a demand curve with constant elasticity, that is, with the same value of demand
elasticity at every point. The right panel in Figure 2.3 depicts several examples. There are
two extreme cases: a vertical demand curve ( = 0), such that for any price the quantity
demanded is always the same; and a horizontal curve ( = ), the extreme case such
that even a very small change in price leads to an infinite increase in quantity demanded.
These extreme examples are not found in any real-world situation, though some market
demands may be close to it. (Can you think of examples?) For the majority of real-world

Consumers

Figure 2.3
Demand elasticity
p

p
=0
|| > 1
|| = 1

 =

|| > 1

|| < 1
 =

=0

|| < 1

markets, demand elasticity lies somewhere between the two extremes. Table 2.1 provides a
few examples.
At this point, it should be clear that elasticity is not the same thing as slope, although
slope is an input to it (recall that the slope is given by dq/dp). The difference between slope
and elasticity is illustrated by the left and right panels in Figure 2.3. On the left panel, we
have a curve with constant slope but varying elasticity; on the right panel, we have a series
of curves with constant elasticity but varying slopes.
Technical point: logarithms. A useful variant of the elasticity formula uses logarithms.
You might recall that2
dq
q
dp
p

d log q
d log p

If we do our usual approximation, replacing ds with s, we have

=

dq
q
dp
p

d log q
log q

d log p
log p

The approximation is exact if the elasticity is constant along the demand curve. In this
case, the demand function is
log q = a + b log p
and the elasticity is  = b. Given two sets of price and demand combinations for a given
product, (p1 , q1 ) and (p2 , q2 ), we can recover the elasticity from
log qq21
log q
d log q
=

=
d log p
log p
log pp21
since log x log y = log xy .
Numerical example. To see how the concepts of demand elasticity can be used in
practice, consider the following example. Suppose you have access to historical sales data
from the Yellow Note jazz label. When price was set at $10, $11 and $13, total units sold
(thousands) were 6.31, 5.63, and 5.07. (As you can see, Yellow Note has not bee a great

Consumers

commercial success.) Suppose that these historical data correspond to different points of
the demand curve (that is, suppose that the demand curve has remained constant over
these periods). This is a big and possibly unrealistic assumption, to which we will
return later; but it will make our lives easier for the time being. Given this big if, how do
we estimate the value of the demand elasticity at the price of 10?
We can approximate the value of the demand elasticity by the percent change formula
applied to prices 10 and 11:



q p
10
5.63 6.31

= 1.08.
=
p q
11 10
6.31
This is an approximation, since were using discrete changes. We could get a better approximation by using a finer price grid. In fact, the numbers in the present example were
generated by the demand curve, log q = log 100 1.2 log p, which has an elasticity of 1.2.
So, while the estimate of 1.08 is greater than one in absolute value as the true value
(that is, we correctly estimate that the demand for Yellow Note CDs is elastic), there is a
considerable estimation error (1.08 as opposed to 1.2).
Suppose I didnt have the value of demand when price was set at 13. How could I use
q/q
the elasticity estimate to predict demand when price is 13? Since 
p/p , we conclude
that
p
q

q
p
Specifically, consider a variation in price from 11 to 13. This corresponds to p/p = 2/11 =
18.18%. We therefore estimate that q/q = 1.08 18.18% = 19.64%, which implies a
value of q given by 5.63 (1 19.64%) = 4.53. This overestimates the decrease in q: as we
know, the true value is 5.07.
Alternatively, we can do the above calculations using the logarithms. First, we estimate
the demand elasticity as:


log 5.63 log 6.31

.1140
=
= 1.20
log 11 log 10
.0953

In fact, given that the demand has constant elasticity, the logarithmic formula yields the
exact value of the demand elasticity. Next, we can estimate the new value of q when p = 13
by solving the equation
log 5.63 log q
= 1.20

log 11 log 13
The result is q = 5.07, the true value of q. Once again, we see that if the demand has
constant elasticity the logarithms approach delivers more precise results.
Elasticity and revenue. From a firms perspective, the elasticity of demand is a critical
piece of information, since it determines the change in revenue that results from a given
change in price. Recall that revenue is simply price times quantity demanded. It is easy
to see that by increasing a products price by 15%, each unit sold will yield more money.
But if overall demand for the product drops as a result of the price increase, the positive
effects of higher per unit prices will be offset by a decline in the number of units demanded.
Which effect is larger?

Consumers

Table 2.2
Automobile demand elasticities
Model

323

Cavalier

Accord

Taurus

Century

BMW

Mazda 323

-6.4

0.6

0.2

0.1

0.0

0.0

Cavalier

0.0

-6.4

0.2

0.1

0.1

0.0

Accord

0.0

0.1

-4.8

0.1

0.0

0.0

Taurus

0.0

0.1

0.2

-4.2

0.0

0.0

Century

0.0

0.1

0.2

0.1

-6.8

0.0

BMW 735i

0.0

0.0

0.0

0.0

0.0

-3.5

Formally, the percentage change in revenue induced by a (small) change in price is

d (p q)
dp dq
dp dq p dp
dp
=
+
=
+
= (1 + )
pq
p
q
p
dp q p
p
That is, the percent change in revenue following a price change is (1 + elasticity)
(% change in price). Since  < 0, the direct effect of the price change (the 1) and the
indirect effect of demand (the ) have opposite signs. If demand is elastic ( < 1), the
demand effect is larger and an increase in price reduces revenue. This is a basic point, but
one that some have missed: that to increase revenue in markets with elastic demand, you
need to lower price, not raise it.
Cross-price elasticity. We have seen that demand for a product depends not only on
its own price, but on prices of other goods. When there are lots of other products its easy
to lose sight of this, but its always there. Some examples are obvious. The demand for ski
boots depends on the demand for skis: if skis get more expensive, we might expect people
to buy fewer skis and fewer boots, too. And as we saw above, the demand for commuter
rail tickets may be influenced by the price of gasoline.
We summarize the sensitivity of demand to the price of another product with the cross
price elasticity:
12 =

d q1
q1
d p2
p2

That is, the ratio of the percent change in demand for product 1 to the percent change in
the price for product 2.
The essential distinction here is between substitutes and complements. If the crossprice elasticity is positive, we say that the products are substitutes. Hence Coke and Pepsi
are substitutes: If Coke gets more expensive, wed expect some people (but not all) to
switch to Pepsi. (We know that you would never do such thing, but some consumers just
dont have any principles.) Similarly, gas and commuter rail tickets are substitutes, since
an increase in the price of gas would lead some people to switch from car to train travel.
Conversely, if the cross-price elasticity is negative, we say the products are complements.
The language of business is filled with competitive metaphors for which substitutes seem
appropriate (Coke v. Pepsi). But there are lots of examples of complements, in which a price

Consumers

10

reduction for one product increases demand for others. Skis and boots are one example.
Others include: Windows operating system and Intel microprocessors, beer and pretzels,
gas and cars.
Speaking of cars, Table 2.2 presents the values of direct and cross-price elasticities for
a selected car models. For example, the number 0.2 on the third column, second row means
that a 1% increase in the price of the Accord leads to a 0.2 1% increase in Cavalier sales.
Are the different car models complements or substitutes? What is the model closest to the
Cavalier, in terms of demand? Exercise 2.2 extends the analysis of this example.
Income elasticity. Changes in income also affect demand. Higher income generally
means greater demand for all products, but some products benefit more than others. We
define the income elasticity of a product by
y =

dq
q
dy
y

That is, the percent change in demand induced by a one percent change in income (represented by y).
Economists have names for products with different income elasticities. Inferior goods
have negative income elasticities. Although inferior goods arent all that common, its fun to
try to think of examples. Spam comes to mind, on the assumption that anyone with enough
money would buy something else. (You can see the source of the term.) Normal goods have
positive income elasticities. Within normal goods, those with elasticities between zero and
one are referred to as necessities, and those with elasticities greater than one as luxuries.
Can you explain why?
Application: demand for gasoline. Let us consider a specific example where the above
concepts play an important role. Using historical data on gasoline price and consumption
in the U.S. from 19532004 (see Figure 2.4), the following relation was estimated:
log qG = 9.72 0.0544 log pG + 1.580 log y 0.285 log pC
where qG is gasoline consumption, pG is the price of gasoline, y is income, and pC is the
price of new cars.
What is the price elasticity of gasoline demand? From the analysis above,
=

d log qG
= 0.0544
d log pG

Is gasoline demand elastic or inelastic? Well, since || < 1, we conclude that it is inelastic.
No major surprises here: with the exception of New Yorkers, people need to drive to work,
and there arent many reasonable alternatives to get there (at least not in the short run).
What is the income elasticity of gasoline demand? The answer is
=

d log qG
= 1.579
d log y

Consumers

11

Figure 2.4
U.S. gasoline consumption (1953-2004)
p index (2000=100)
120
100
80
60
40
20

q index
(2000=100)

0
20

40

60

80

100

Is gasoline a normal good? Yes, since the income elasticity is greater than zero. In fact,
since > 1, gasoline is a luxury good. (Well, with the price you now have to pay for it,
gasoline is indeed a luxury that few can afford.) Notice that the threshold that defines a
luxury is not purely arbitrary. In fact, if the income elasticity is greater than 1 then an
increase in income implies that the fraction of income spent on that good increases. So,
richer people not only spend more on gasoline ( > 0) but spend a greater proportion of
their income on gasoline ( > 1).
Are gasoline and new cars complements or substitutes? In order to answer this question, we compute the cross-price elasticity of gasoline demand with respect to the price of
cars. This is given by
d log qG
= 0.285
GC =
d log pC
Since we get a negative value, we conclude that gasoline and cars are complements. This
is probably not very surprising as gasoline is not particularly useful unless you have a car;
and a car is totally useless unless you have gasoline to fuel the engine. (We hope reality
will soon falsify the latter statement!)
To conclude, let us consider the evolution of price and consumption over the period
under consideration. From 1953 to 2004, gasoline consumption increased by 2.6% a year;
price increased 4.2% per annum, income 2.2%, and the price of cars increased at the rate
of 2.8%. The above regression implies an increase in consumption of
qG
qG

= .0544 4.2% + 1.579 2.2% 0.285 2.8%

= 0.230% + 3.473% 0.798%
= 2.445%

which is pretty close to the actual number, 2.6%. The regression model thus suggests that
the main factor pushing up consumption levels is income. The price of gasoline and the
price of cars had relatively little effect on the growth in consumption from 1953 to 2004.

Consumers

12

Figure 2.5
Supply and demand in two different markets.
pA

pA
D

S2

D2

S1

D1

S0

qA

D0

qA

This example helps explain an apparent paradox, which is illustrated in Figure 2.4:
for some time now, the price of gasoline has been increasing and so has consumption level.
Doesnt this violate the regularity that demand curves are downward sloping, that is, that
demand decreases when price increases? The answer is no, not when there are other factors
varying at the same time. In the next section, we explore this issue in the context of demand
curve estimation.

2.3. ESTIMATING THE DEMAND CURVE

In an ideal world, firms would know the demands for their products. In practice, its not
so easy. One reason is that its hard to get reliable market data: how much was bought
by whom and at what price. Another reason is that its inherently difficult to tease out
the effect of price from the effect of other variables, especially when the latter might be
changing at the same time as price (or, even worse, when they are not known to us).
We have colleagues who provide this service for a reasonable fee. But you should be
aware of the main challenges in estimating the demand curve and the demand elasticity
from actual data. A central problem is what econometricians refer to as the identification
problem. Consider the case of two different markets depicted in Figure 2.5.
In market A, the demand curve remains constant over three periods, whereas the
supply curve shifts around. The idea is that there were changes in some input price or
the number of suppliers, which in turn shifted the supply curve during the time period in
consideration. The historical data that we observe corresponds to the equilibrium values
of p and q for each period, that is, the points where supply crosses demand. Notice that
these are points along the (stable) demand curve. Implicitly, case A is the situation weve
been assuming all along, that is, the case when the historical pairs of q and p trace out the
demand curve.
But now consider the case of market B. Here the supply curve is fixed over the period
in consideration, whereas the demand curve shifts from period to period, perhaps because
of changes in product quality or consumer income or the price of a substitute product. Now
the historical pairs of p and q trace out the supply curve, not the demand curve. In fact, if
the only source of variation consists in shifts in the demand curve, then there is little hope
of estimating the shape of the demand curve.
More generally, both the supply and the demand curves will shift over time (and across

Consumers

different geographical markets). The key to econometric estimation is then to understand

the source of variation: what are the variables that are shifting each curve over time. If
we have enough information about these shifters, then there is hope of identifying both the
demand and the supply curves.
Failure to take into consideration the sources of variation may lead to identification
errors, sometimes gross identification errors. Consider, for example, the case of ticket sales
at the New York Mets, summarized in Box 2.1. Based on the graph at the bottom of the
box, what can you say about the price elasticity of demand? How does the Mets case relate
to the above discussion regarding identification?
Demand for characteristics. To be completed
Alternative paths to estimating demand. Statistical analysis is not the only avenue for
demand estimation. An alternative approach is to obtain data by means of surveys. One
problem with this method is that we are never sure how accurate the responses are going
to be: until your own money is at stake you dont have an incentive to think hard.
Still another approach is to do experiments in markets. Thus catalog companies sometimes send out catalogs to different customers in which some of the prices are different.
These experiments run the risk of alienating customers (what if you find out you got the
high price?), but you can see the value to the firm of knowing the demand for its products.
Finally, if you do not have historical data to estimate demand or the resources to
experiment with price changes, you should at least have an idea of whether demand is more
or less elastic depending on the characteristics of the good in question. In that spirit, here
are some rules of thumb that might help in this process:
Elasticities are higher (in absolute value) on luxuries than necessities. Compare, for
example, food and Armani suits. Consider too the elasticities of the various products
listed in Table 2.1.
Elasticities are higher for specific products (e.g., the iPhone) than for a category as a
whole (smartphones). Why is this so? Because when the price of a specific product
rises, people are willing to buy fewer units. Some of this reduction leads to purchases
of other products in the same category (e.g., Samsung phones), and part to a reduction
in the category as a whole (smartphones). Only the latter shows up in the elasticity
of the category as a whole, so its typically less elastic.
Elasticities are lower (in absolute value) in the short run than in the long run. A
good example is gasoline demand. Can you see why? Suppose, for example, that the
government plans to levy a 100% tax on gasoline for the next 3 years. In the next
day or week consumers would probably still drive their cars, but in the longer term
their demand for gas could change for many reasons. They might buy more fuelefficient cars, carpool, or take the bus or train to work. Perhaps some would work
from home. As a result, the quantity of gasoline demanded at the new price would
gradually decrease.

13

Consumers

14

Box 2.1. Ticket demand at the Mets.

What makes baseball fans go the ballpark? Clearly having a good team helps but its
not the only factor. Based on ticket sales at the New York Mets Shea Stadium during
the 19942002 seasons, a statistical regression can be performed where the dependent
variable is the number of tickets sold. The following table displays the estimated coefficients of such a regression for a specific section of the Mets stadium: Upper Reserved.
Weekend

1078.63

Evening

-905.58

Season opener

8196.82

July

2410.27

August

1425.13

September

1555.44

October

3774.67

Yankees

9169.82

Constant

401.53

Each dependent variable is a dummy, or indicator, variable, taking the value 0 or

1. For example, if the game is played on a weekend, then everything else constant
1,078.63 more tickets were sold on average. Considering that the Upper Reserve
capacity is about 17,000 seats, these are economically significant coefficients: for
example, playing against the New York Yankees leads to an increase in ticket sales
equivalent to more than one half of capacity.
What about ticket prices? Until 2002, prices did not vary within each season. They did
vary from season to season, but so did team quality, which is hard to measure but is
certainly an important demand shifter. For these reasons, it is not possible to estimate
the price elasticity of demand. This is corroborated by the graph below, which plots
average price and average attendance on a yearly basis. Notice the resemblance with
the right panel in Figure 2.5 and the accompanying discussion.
p ($)
10

q (year total)
(000)

4
0

250

500

750

All estimated coefficients are statistically significant at the 2% level; year dummies and a constant are
also included but not reported; N = 651, R2 = 0.44.

Consumers

15

The price elasticity of demand tends to be higher (in absolute value) for luxuries; for narrow
product categories; and in the long run.

Summary
The demand curve gives the quantity demanded of a given good as a function of its
price and of other factors; it can be derived from consumer preferences.
A change in price leads to a shift along the demand curve; a change in other factors
leads to a shift in the demand curve itself.
Consumer surplus is the difference between willingness to pay and price.
The price elasticity of demand is the ratio between the percentage change in
quantity and the percentage change in price, for a small change in price.
The price elasticity of demand tends to be higher (in absolute value) for luxuries; for
narrow product categories; and in the long run.

Key concepts
preferences indifference curve budget set demand function inverse
demand function demand curve willingness to pay consumer surplus
price elasticity of demand cross price elasticity substitutes.
complements income elasticity inferior goods normal goods
necessities luxuries identification problem

Review and practice exercises

2.1. Village microbrew. Village microbrew raised its price from $10 to $12 a case (wholesale). As a result, sales dropped from 10,500 to 8,100 (in units).
(a) What is the elasticity of demand?
(b) Based on your estimate of the demand elasticity, what percent change
would you predict if price were cut from $10 to $9? What demand
level would this correspond to?
2.2. Cars. Table 2.2 gives the own and cross-price elasticities for selected automobile
models.3 Specifically, each cell corresponds to the demand elasticity of the car model in the

Consumers

16

row with respect to changes in the price of the car model in column.
(a) Why are the own elasticities so high?
(b) Are the Accord and Taurus complements or substitutes?
(c) What are the Tauruss closest competitors?
(d) If GM lowers the price of its Chevy Cavalier, does it cannibalize its
Buick Century sales?
(e) Why is the direct elasticity for the Mazda not lower than the elasticity
for more expensive models (as the rule of thumb would suggest)?

Consumers

17
(f) Suppose Honda sold 300,000 Accords in 2001. In 2002, the price of the
Accord decreased by 2%, whereas the price of the Taurus decreased
by 3%. What is the likely change in Accord sales?

2.3. Demand elasticity. Based on the values in Table 2.1, provide an estimate of the impact
on sales revenues of a 10% increase in each products price.

Challenging exercises
2.4. Netflix and Blockbuster. Suppose the demand for Netflix is given by
qN = a bN pN + bB pB ,
where qN is the number of Netflix subscriptions, pN the price of a Netflix plan, and pB the
price of a Blockbuster plan.
(a) What is the price elasticity of Netflix subscriptions?
(b) Suppose a = 500, bN = 10, bB = 5, and pB = pN = 50. What
are N s elasticity and cross-price elasticity? Are products N and B
substitutes or complements?
(c) How much do consumers get in surplus at these prices?
2.5. Lamborghini. The current U.S. demand for the Lamborghini Gallardo SE is elastic;
specifically, it is estimated that demand elasticity is given by  = 3. The current price is
p = $120, 000. Annual sales at this price amount to q = 160 (number of cars).
(a) What do you estimate would be the impact of an increase in price to
$140,000?
The cross price elasticity of Lamborghinis with respect to the price of the Maserati MC12
is LM = .05; and with respect to the price of gasoline, LG = .1.
(b) What are the definitions of a substitute and of a complement? Are
Maserati MC12 and gasoline substitutes or complements with respect
to Lamborghinis? Can you think of other substitutes and complements to the Lamborghini Gallardo SE?
(c) Suppose that, in addition to the price increase considered in (a), there
is also an increase in the price of the Maserati MC12 (from $110,000
to $115,000); and an increase in the price of gasoline (from $2 to $2.8
per gallon). What do you estimate will be the new demand for the
Lamborghini Gallardo SE?

Consumers

Endotes
1. Although we think of qi as a function of pi , we normally measure price on the vertical axis. British
economist Alfred Marshall started doing it this way back in the 19th century and things havent changed
ever since.
2. This is basic calculus; the equation is valid for any base logarithm, including natural log, sometimes
written ln.
3. Source: BLP, 1990 data.

18