Tip:
Highlight text to annotate it
X
The following content is provided under a Creative
Commons license.
Your support will help MIT OpenCourseWare continue to
offer high quality educational resources for free.
To make a donation or view additional materials from
hundreds of MIT courses, visit mitopencourseware@ocw.mit.edu.
PROFESSOR: So today what we're going to do is continue our
discussion of supply and demand.
This is sort of introduction week, if you will.
We've kind of talked about supply and demand, and you
guys, rightly, immediately were on to where do those
curves come from.
And that's what we'll start next week.
But what I want to do today is talk some more about what
determines the shapes of supply and demand curves and
just think about an overview of how we think about supply
and demand interacting in a market and what determines how
responsive individuals and firms are to prices.
And, once again, remember everyone should have a handout
that you should have picked up in the back on your way in.
So everyone should have a handout.
What we talked about last time was the sort of qualitative
effects, the qualitative version of the supply and
demand model.
We talked about what happens when a supply curve shifts,
what happens when a demand curve shifts.
We talked about how either a supply shock or a demand shock
could lead to the price being increased.
But they could have very different effects on
quantity, et cetera.
What we didn't talk about is how big these effects are.
I made up some numbers.
I threw them on the graphs.
But I didn't talk about where the size of those
effects come from.
And where they come from is the shapes of the supply and
demand curve.
And that's what we'll talk about today is what determines
the shapes of supply and demand curves.
And that will be the focus of today's lecture.
I'll talk both theoretically about what determines these
shapes and empirically about how economists go about
figuring out the shapes of supply and demand curves.
So, to think about this, let's start with Figure 3-1, which
is a standard market diagram we had last time.
With an initial equilibrium at point E1, with an initial
price P1 and a quantity Q1.
That's the equilibrium that's stable.
Because at that price P1, consumers demand Q1 units, and
suppliers are willing to provide Q1 units.
So that's a stable equilibrium.
Now we have some supply shift.
Last time we talked about somehow a
pork-specific drought.
That leads the supply to shift inward.
So the supply curve rises to S2.
At that new price, initially, you would have excess demand.
But quickly the price increases to shut off that
excess demand.
And you end up with a new equilibrium with a higher
price, P2, and a lower quantity Q2, and new
equilibrium point E2.
OK?
And we talked that through last time.
What I want to talk about this time is well, what determines
the size of that shift from Q1 to Q2 and that price increase
from P1 to P2?
What's going to determine it is the elasticity of supply
and demand.
The elasticity of supply and demand is how much do supply
and demand respond?
Do the quantities supplied and the quantities demanded
respond when the price changes?
When we say, how elastic is demand, what we mean is how
sensitive to price is the quantity demanded.
Or, alternatively, what is the slope of that demand curve?
So the slope of the demand curve will be the sensitivity
of quantity demanded to the price consumers face.
And that will determine the market responsiveness.
In economics, it's always true that the best way to think
about things is to go to extremes.
You have to remember that extremes don't exist in the
real world.
But it's a useful teaching device
to think about extremes.
So let's think about one extreme case in Figure 3-2.
Let's think about the case of perfectly inelastic demand.
Perfectly inelastic demand, that's where there's no
elasticity of demand.
What that means is that demand for a good is unchanged
regardless of the price.
So perfectly inelastic demand is a case where demand for the
good is unchanged regardless of the price.
That would lead you to have a vertical demand curve at a
given quantity.
What this says is regardless of the price, people always
demand Q.
Can anyone tell me what would cause demand to
be perfectly inelastic?
In what types of situations would demand be--
it's never perfectly inelastic--
would demand be relatively inelastic?
Yeah?
AUDIENCE: [INAUDIBLE PHRASE].
PROFESSOR: It's all about substitutes.
When there's no substitutes, when there's nowhere to go, it
doesn't matter what the price is.
When there's no substitutes, demand will be perfectly
inelastic, because you have to have Q. It doesn't matter what
the price is.
Because there's no substitute for that good.
So if you wanted amount Q of that good for any reason,
you're always going to want that amount Q
no matter the price.
So a perfectly inelastic good would have no substitutes.
So you'd always want Q no matter what.
Can anyone think of an example?
There's no perfectly inelastic good in the world.
But what sorts of goods?
Yeah?
AUDIENCE: Medicines.
PROFESSOR: Medicines.
Now, not necessarily all medicines.
So give me an example of a medicine which would be more
or less inelastic.
So I don't even need a medical name.
What sort of treatments?
AUDIENCE: Like heart attack maybe?
PROFESSOR: Yeah, something which is sort of lifesaving.
The best thing that we often use is insulin for diabetics.
Diabetics without getting that insulin to manage their
diabetes will die.
That seems like that's something where there's not a
whole lot of substitutes.
The substitute is dying.
So basically that's where demand
is relatively inelastic.
Or a heart transplant, when you get a heart transplant or
any kind of transplant, you have medicine you take so you
don't reject the transplanted organ.
That sort of medicine demand should be very inelastic.
Elastic drug, well, our favorite
example is always ***.
It's something where you'd think that you can probably
survive without it.
And people would want less *** if you charged a lot
more for it than if you charged less for it.
So elasticity is going to be about substitutability.
And that's going to determine inelastic demand.
Now, what happens with inelastic demand when there's
a supply shock?
When supply increases, what happens?
Well, in that case, there can never be excess demand,
because demand doesn't change.
So all that happens is price just increases.
If there's inelastic demand, and there's a supply shock,
then all that happens is an increase in price and no
change in quantity.
So with inelastic demand, quantity doesn't change for a
price increase.
Price just goes up.
From a supply shock, prices just goes up.
Now, let's consider the opposite.
Let's look at Figure 3-3 and think about
perfectly elastic demand.
Perfectly elastic demand is demand where consumers,
essentially, don't care about the quantity.
They just care about the price.
That is, there are infinitely good substitutes.
A perfectly elastically demanded good would be one
where there are, essentially, perfect substitutes.
An inelastic good is where there's no substitute.
A perfectly elastic good would be where there's perfect
substitutes.
Technically, if a good is perfectly elastically
demanded, then you are completely indifferent between
that good and a substitute.
Well, if you're completely indifferent, then if the price
changed at all, you would immediately switch.
And so the price can't change.
What's an example?
Once again, there's no good example of a
perfectly elastic good.
Yeah?
AUDIENCE: Candy.
PROFESSOR: What?
AUDIENCE: Candy.
PROFESSOR: Candy.
OK.
So you've got your Wrigley's gum.
I like the sugar-free, minty gum.
You've got Orbit and Eclipse.
And I go to the store, and they're all pretty much the
same price.
If Orbit was more than Eclipse, I just buy Eclipse.
They're the same.
They're minty gum.
It doesn't make a difference.
So basically the price is the same.
If there's a supply shock, I don't know, they're made with
the same ***.
But let's say that Eclipse has some magic ingredient.
And let's say the Eclipse magic ingredient got more
expensive, so the supply curve shifted up.
Well, Eclipse could not respond by raising its price.
Because I just switched to Orbit.
Or we often think of McDonald's and Burger King.
Now, they're less perfect
substitutes, but pretty perfect.
If McDonald's started charging $10 for a hamburger, you
wouldn't go there anymore.
You'd go to Burger King.
So if there's a supply shock to a provider that's facing a
perfectly elastic demand curve, they cannot raise their
price, because people will just switch.
So quantity will fall a lot.
Because if I'm supplying Eclipse gum, and it suddenly
costs a lot more to produce Eclipse gum, but I can't raise
my price, because I will lose all my business to Orbit, I'm
just going to produce a lot less Eclipse.
Because I'm losing money now.
So with perfectly inelastic demand, the
quantity didn't change.
With perfectly elastic demand, we saw a big quantity change.
So, more generally, what determines the quantity change
in response to a price change is the elasticity.
More generally, we're between these two cases of perfectly
elastic and perfectly inelastic.
And what's going to determine the price change is going to
be the price elasticity of demand epsilon which is going
to be the percentage change in quantity for each percentage
change in price or, in calculus terms, dQ/dP.
So it's, basically, the percentage change in quantity
for the percentage change in price.
So, for example, if for every 1% increase in price quantity
falls 2%, that is a price elasticity of
demand of minus 2.
The price elasticity of demand is the percentage change in
quantity for the percentage change in price.
So inelastic demand is an epsilon of 0.
There is no change in quantity when price changes.
Perfectly elastic demand is an epsilon of negative infinity.
Any epsilon change in price leads to a negative infinite
change in quantity.
Immediately, the quantity goes to 0 if you try
to raise your price.
So the price elasticity of demand will typically be
between 0 and negative infinity.
And the larger it is the more quantity will change when
prices change.
Questions about that?
Yeah?
AUDIENCE: So that formula, shouldn't it be dQ/dP times
P/Q because dQ/dP just refers to the change of the quantity
with respect to price, not
necessarily the percent change.
PROFESSOR: Yeah, you're right.
I was trying to get too fancy with my calculus.
You're right.
Let's just stick with the non-calculus formula.
I never should deviate from my notes.
So let's just stick with the non-calculus formula.
OK, other questions about this?
OK.
So, basically, that's the elasticity.
That's going to be the elasticity.
Now, an interesting point about elasticity is now, we're
not going to get into producer theory
for a couple of lectures.
But as a little peek ahead about producer theory, let's
think about how elasticity determines the money that
producers make from selling their goods.
Well, if a producer sells Q goods at a price P, they make
revenues R. Revenues are the price times the quantity.
The amount of money a producer makes when it sells goods, its
revenues, this isn't its profits.
We're not having profits.
It's just the amount of money it makes, not the amount of
money it takes home at the end of the day.
I'm ignoring the cost of making the goods.
The amount of total revenues it makes
is price times quantity.
Well, we can then say that the change in revenues with
respect to price is what?
It's Q plus dQ, plus delta Q-- let me put it this way to make
my math clearer--
plus P times delta Q over delta P. That's how revenues
change with respect to price.
Or, in other words, plugging in from the elasticity
formula, delta R over delta P equals Q times 1 plus epsilon.
So, in other words, what this says is that if you're a
producer, and you're trying to decide whether to raise your
price, whether that will increase revenues, it all
depends on the elasticity.
If the elasticity is between 0 and minus 1, then raising
prices will raise revenues.
If the elasticity is greater than minus 1, then raising
prices will lower revenues.
We're often faced with the issue of why did they charge
this much for this good, or should they raise their prices
or not raise their price.
Well, that's all about the elasticity of demand.
The elasticity of demand will determine whether they're
going to make more money by raising their price or lose
money by raising their price.
For Eclipse gum, their elasticity of demand is well
above minus 1 in absolute value, so they're going to
lose money by raising their price.
If they take the current level of Eclipse, for every penny
they raise, they'll lose money.
For insulin, for every penny they raise,
they'll make money.
And then you might say well, then how come the price of
insulin isn't infinity and the price of
Eclipse gum isn't zero?
Well, that's what we'll talk about in a few weeks.
Because it also depends on the costs of producing it.
But at the end of the day, that's what's going to
determine the money that's made by producers when they
change their prices.
Questions about that?
OK.
So now, that's how we think about the shape
of supply and demand.
The shape of supply and demand is determined by these
elasticities.
So now we have to get into OK, well, where do we get these
elasticities from?
And that is the main topic of empirical economics which is
estimating these kinds of elasticities, estimating these
types of elasticities.
So one of the first distinctions I drew in the
lectures is between theoretical economics and
empirical economics.
Theoretical economics can tell us this is what a graph looks
like and supply and demand.
Theoretical economics can't really tell us how big, for
example, an elasticity is going to be.
It can tell us, there's more substitutes or less
substitutes so we can rank them.
We know the elasticity for Eclipse gum has got to be
higher than the elasticity for insulin.
But from the theoretical model, we can't say what the
elasticity actually is.
To say what an elasticity actually is, we need to go to
an empirical model.
We actually need to bring data to bear on the question.
And this is very difficult.
Because here we face the fundamental conundrum facing
the empirical economist which is distinguishing causation
from correlation.
And the whole guts of empirical economics is all
about this question, distinguishing causation from
correlation.
The classic story that illustrates this, it's due to
my colleague, Frank Fisher, from a textbook many years
ago, was the story of in ancient Russia there was a
cholera outbreak, and many people were dying.
So the government decided to send doctors out to try to
solve the problem.
And where there were more people sick,
they sent more doctors.
Well, the peasants said, wait a second.
We observe that where there's more doctors, more people are
dying from cholera.
So the doctors must be causing the cholera.
So they rose up and killed the doctors.
The peasants confused causation with correlation.
They thought that the fact that you saw more people dying
where there's more doctors meant that doctors were
causing the disease.
Clearly that's wrong.
That's why they were peasants.
But it's not just peasants that make this mistake.
For example, in 1988, Harvard University, our illustrious
neighbor to the south, I guess, west,
east, I don't know.
Which way is Harvard?
I don't know directions, down the street.
A Harvard University dean conducted an interview with a
set of freshmen.
And they found that those that had taken
SAT preparation courses--
now, you all took SAT preparation courses.
But in 1988, not everyone did.
Those who'd taken SAT preparation courses scored an
average of 63 points lower--
this was back when the SAT was 1600 points--
63 points lower on their SATs then those that had not taken
preparation courses.
The dean concluded that preparation courses were
unhelpful, and that the testing industry was preying
on the insecurities of students to
provide a useless service.
Why was the dean confusing causation with correlation?
What did the dean get wrong in drawing that conclusion?
Yeah?
AUDIENCE: What had probably happened is the students who
got worse scores realized that they wanted to try and improve
their scores by taking an SAT prep class.
So that's why there is a lower average score for the people
who had taken the class.
PROFESSOR: Generally, the people who needed the help the
most took the most courses.
And so they had an underlying lower score.
So, in fact, you can't tell anything from the fact that
the people who took the prep course scored worse.
It's just another excellent example of confusing causation
with correlation.
And that's another example.
Another example I like quite a lot is studies of
breastfeeding.
There are numbers of studies of breastfeeding, especially
in developing countries, where they found that the longer
children were breastfed the sicker they were.
So they concluded that
breastfeeding was bad for kids.
Well, that's not the truth.
The truth is the sicker kids need to be breastfed more,
because breastfeeding is actually good for kids.
And they just confused the causation with the
correlation.
Now, these are all fun examples.
But the truth is this is a common mistake made by
citizens, policy makers, everyone in the real world.
It's taking two things that move together and assuming one
causes the other.
And this is the fundamental conundrum facing empirical
economics in trying to address these kinds of things like
measuring elasticities.
So to understand that, let's think about the issue of
trying to estimate the elasticity of demand for pork.
Let's say you have the exciting job of estimating the
elasticity of demand for pork.
That's your assignment.
Well, you say, wait a second.
What we learned in class, as shown in Figure 3-4, is that
the price of pork can rise for very different reasons.
Figure 3-4, we start at an initial equilibrium like E1
with a quantity like Q1 and a price P1.
Now, imagine that there was a shift in demand, because the
price of beef rose, remember?
The price of beef rose.
That shifted demand from D1 to D2.
What did that lead to?
A higher price and a higher quantity.
So if you took that diagram--
forget the supply shift for a minute, just imagine that's
the change--
and you said, aha.
I can measure the elasticity.
I see here there's a change in price.
I can then look at how quantity changed.
And I'll get the elasticity right after all.
It's delta Q over Q or delta P over P.
So I just look, and I take Q2 prime minus Q1 over Q1.
That's the percentage change in Q. I take P2
over P1 over P1.
That's the percentage change in price.
And what do I get?
A wrong signed elasticity is what I get.
I get a positive elasticity, because Q is going up
and P is going up.
Why?
Because I'm confusing causation with correlation.
It's not the price change that caused quantity to change.
In fact, it's the opposite.
It's a taste shift, which caused quantity to increase
which drove up the price.
It was a demand increase which caused the quantity demanded
to increase which drove up the price.
So it's the quantity driving the price, not the price
driving the quantity.
So if you looked at that simple example, as many people
in the real world do, they'd say, hey, look.
Higher prices cause higher quantities.
You're getting the wrong answer.
Because you're confusing correlation which is the
higher price is correlated with the higher quantity.
Because there was a common factor causing both of them
which is the demand shift and not causation.
The higher price did not cause the higher quantity.
What do we need to do?
We need to distinguish why the price increased.
We need to distinguish why the price
increased to measure this.
If, instead, we looked at a shift in supply such as the
case that's shifting from S1 to S2 and moving the
equilibrium from E1 to E2, then you would
get the right answer.
Because then you'd say, look.
Something independent to consumers
shifted up the price.
Some shock to the supply of pork shifted up the price.
And we saw that their quantity fell as a result.
What's the key?
The key is that to measure an elasticity of demand, you're
measuring the slope of the demand curve.
So you need to shift along a demand curve, not shift the
demand curve itself.
So if you look at this figure, what's the concept we want?
We want the slope of the demand curve.
Well, you get that by shifting from E1 to E2, because you
shift along the demand curve.
So by looking at what happens to quantity as price rises
from E1 to E2, you get the slope of the demand curve.
You get that delta Q over delta P you want.
But from E1 to E2 prime, you're not shifting along the
demand curve.
You're actually measuring the elasticity of supply.
You're measuring the elasticity of supply.
You're shifting along a supply curve.
So you're actually answering a different question, a relevant
question, but a different one.
That question is, what's the elasticity of supply?
How willing are pork producers to supply pork as
the price goes up?
So it's the same delta Q over delta P. But here we did the
elasticity of demand.
There's a corresponding elasticity of supply which is
measured the same way.
It's delta Q over delta P, but it's for a
different kind of shock.
It's what you get from moving along the supply curve.
So if we went from E1 to E2 prime, we can use that to
measure the elasticity of supply or the slope of the
supply curve.
And we do that if something shifts demand to move us along
the supply curve.
From E1 to E2, we measure the elasticity of demand as
something shifts supply and moves us
along the demand curve.
So what we need to measure the elasticity of demand is
something which shifts supply but does not,
itself, affect demand.
And the best example of this that we use in economics, a
great example, is government policy which comes along and
changes the supply conditions for a good.
So, for example, let's think about a tax on pork.
So if you go to Figure 3-5, imagine the government came
along and taxed pork.
The government comes along and taxes pork.
Let's think about what a tax on pork does.
The government comes along, and let's say the pork market
is initially in equilibrium at $3.30 with 220 million
kilograms of pork sold.
Now the government comes along and says that it's going to
charge $1.05 in tax for every kilogram of pork.
So it's going to impose a tax of $1.05
per kilogram on producers.
So it's saying to producers of pork, for every kilogram of
pork you sell, you have to send a check to the
government for $1.05.
For every kilogram of pork you sell, you have to send a check
to the government for $1.05.
Now, somebody talk me through how a supplier
thinks through that.
How does a supplier react to that?
What do they think?
They're initially happy at E1 selling 220 million
kilograms at $3.30.
What happens when the government comes in and says
you have to pay $1.05 for every
kilogram of pork you sell?
What happens?
Yeah.
AUDIENCE: The producer decides that the current amount of
money they have will not be able to buy as much inputs to
create their products.
So they can produce less.
PROFESSOR: Exactly.
So, in other words, the cost of producing just rose.
So what do they do?
So, in other words, what they say is look, effectively, if I
was happy before selling 220 million kilograms at $3.30, to
keep me equally happy selling 220 million kilograms, I'm
going to have to raise the price.
We should add this to graph, actually.
If you draw a vertical line up for me, one to the S2 curve.
Draw a little dashed line up from the E1 to the S2 curve
and then over.
That price intersection will be $4.35.
So in other words, if you want me to keep producing 220
million kilograms of pork, I'm going to have
to get $4.35 a kilogram.
And you might say, what gives you the right to get that?
And it's not about rights.
It's about what producers are willing to do.
That same mathematics, that same supply curve that tells
us they're willing to sell 220 million kilograms at $3.30
says, if you want them to keep selling 220 million kilograms
but also pay $1.05 to the government, they're going to
have to get $4.35 a kilogram.
So what happens is that's a supply shift.
And with the same reaction we saw last time with the
drought, the price goes up, consumers demand less, and you
reach a new equilibrium at the price E2.
You reach a new equilibrium where you sell 206 million
kilograms for a price of $4.00.
So someone tell me how I use this example to find
elasticity of demand.
Yeah.
AUDIENCE: I guess you need to know that the change in price
traveled along the demand curve.
So you know that it's not [INAUDIBLE PHRASE].
PROFESSOR: OK, so tell me.
You don't have to do the math in your head.
But how would I compute it?
AUDIENCE: You would take E1 and E2, and then you would do
the price over the quantity change.
PROFESSOR: Right, exactly.
So the quantity change delta Q over Q, is what?
It's minus 14 over 220.
It fell by 14 million kilograms over 220.
The price change, delta P over P, the price rose
from $3.30 to $4.00.
So the price change is $0.70 over $3.30.
And using those, you end up with a price
elasticity of minus 0.3.
Or, in other words, there's a 6.4% change in quantity.
This is minus 6.4% for a 21% change in price.
So quantity falls by 6.4% when price goes up by 21%.
That's a price elasticity of minus 0.3.
Or that's a relatively inelastic demand.
It's not perfectly inelastic, but it's relatively inelastic.
In other words, at that point, pork producers could make
money by raising the price.
Now, you might say well, why didn't they?
That's something we'll discuss in a couple weeks.
But at that point, demand is relatively inelastic.
And you've got a convincing estimate, because you moved
along that demand curve.
You used the supply shift.
Now, we're going to talk about taxation much, much later in
the semester.
Let me just talk for one minute about what we learned
from this graph.
What happens?
Well, the shaded area is the money the government raises
from its tax.
The government has a tax of $1.05 at 206 million
kilograms. So it raises $1.05 times 206 million kilograms
which is that shaded area.
There are two points to note the we'll come back to later
in the semester.
The first point to note is the amount of money the government
raises will depend directly on the elasticity of demand.
Can anyone tell me how much money the government would
raise if you had a perfectly inelastic demand?
Yeah.
AUDIENCE: [INAUDIBLE PHRASE].
PROFESSOR: Right.
If we think about this demand curve being perfectly flat, if
we think about this demand curve being perfectly flat,
then basically the producer can't charge any more for
their good.
So it's going to depend on whether the producer is
willing to sell at $1.05 less and how much less they're
willing to sell.
If they're willing to sell a lot less, they're going to
make a lot less money.
It's going to be where that second supply curve intersects
a flat demand curve.
So that quantity is going to be a lot smaller.
We don't have it on the diagram.
But you see where that dashed line at $3.30 intersects S2,
that's way to the left.
Quantity is going to fall a ton in this market.
When quantity falls, the government is going to raise a
lot less money.
Because the government raises $1.05 on every
unit sold at the end.
So if the government taxes very elastically demanded
goods, it's going to raise less money.
If it taxes inelastically demanded goods like insulin,
it's going to raise more money, because the quantity
doesn't change.
Yeah.
AUDIENCE: So cigarettes are relatively inelastic.
PROFESSOR: Yes, exactly.
Cigarettes are relatively inelastic.
The elasticity is around minus 0.5.
So the government will actually raise money by
raising the cigarette tax.
Those of us, as good liberals, think we should tax yachts.
Let's tax yachts.
Only rich guy have yachts.
The problem is yachts are
incredibly elastically demanded.
So you raise a lot less money taxing yachts than you think.
Because guys buy fewer yachts, and you don't raise as much
money as you think you would.
You still raise some, and it still may be worth it.
But you raise less than you think.
So that's one sort of observation about this.
It's basically how much money you'll raise will be a
function of how elastic the demand is.
The other important observation to make is why
it's actually hard for governments to figure out how
much money they're going to raise for a tax.
Because, to figure it out, they need to know these
elasticities.
That is, the naive thing to do would have been to say what?
Well, we're selling 220 million kilograms of pork.
That's $1.05.
We're going to tax each kilogram.
So that's 220 million times $1.05.
And that's how much money we raise.
Well, that's wrong, we know, because that
assumes inelastic demand.
If demand's elastic, they'll raise less than that.
Well, if we want to figure out how much a government is going
to raise from a tax, they've got to know what these
elasticities are.
And those are actually pretty hard things to know.
So that's why there's uncertainty.
That's why when politicians will say, this tax will raise
x and you'll hear the New York Times report, the tax will
raise x, that is a guess.
Those are guesses, because they depend on our best
estimate of the key elasticities that determine
how people respond.
Yeah.
AUDIENCE: But in Washington you have tax
cuts that raise money.
PROFESSOR: Well, some claim you do.
You don't actually.
But some claim you have tax cuts that raise money.
That's because they think the elasticity is very large.
If the elasticity is large enough, a tax
cut can raise money.
So, basically, that's all about that some people think
that elasticities are large enough that tax
cuts can raise money.
Those people are wrong.
But that's what they claim.
Yeah.
AUDIENCE: [INAUDIBLE PHRASE].
PROFESSOR: Yes.
Excellent point.
You'll go through that in section on Friday.
So what I've done is I've done an example of a constant
elasticity curve.
Actually, I've done something here which is logically
inconsistent.
This curve is linear which means it can't be constant
elasticity.
If it's constant elasticity, it would have to curve.
So what I've estimated here is a local elasticity.
I have estimated the elasticity
around that price change.
But the elasticity, if this curve is true, would be
different at different points on this curve.
If the elasticity is going to be constant all over the
curve, and you're going to do a constant elasticity of
demand, that's going to be a curve that bends,
not a linear curve.
So a linear demand curve is not constant
elasticity of demand.
We will typically ignore that issue and focus on local
elasticities.
But that is an important issue.
We'll discuss that in section on Friday, the difference
between constant elasticity of demand curves and linear
demand curves.
But, typically, we're think about local changes.
So if it's local enough, it doesn't really matter.
But, for a broad change, it will matter what the shape of
the curve is.
Good point.
Other questions?
OK.
Let me then turn to another problem we face
in empirical economics.
So this is an example of a problem we're facing in
empirical economics.
Let me turn to an example of another problem we face in
empirical economics estimating elasticities.
It is that individuals often choose the price they face.
Individuals, typically, often don't just face a price that's
given to them.
And then you can say, OK, they're given a price, and we
see how they respond.
They often choose the price they face.
Let me explain what I mean by that.
A classic example of an elasticity that matters a lot
for policies is the elasticity of demand for medical care,
the elasticity of demand for medical care.
That is how much less medical care will you use if you have
to pay for it?
So, for example, most of us have insurance through MIT or
maybe through our parents.
And the way health insurance works is you pay a certain
amount per month or your parents do, and, in return,
that health insurance covers the cost of your medical care,
most of it.
But, typically, you have to pay some of it.
So how many people have gone to the doctor in
the last six months?
Did you have to pay something?
How much did you pay?
Did you pay a copayment?
No?
None of you?
Yeah.
How much did you pay?
AUDIENCE: I think like $20.
PROFESSOR: $20, $10, $5, that's what's called the
copayment, or $0.
Most insurance these days has what's called copayments.
A copayment is what you pay when you go to the doctor.
Insurance picks up the rest. You don't know.
You didn't know how much the whole doctor visit cost. You
just went, you gave them your card.
They said your copayment is $20.
You gave them $20.
You don't know.
The visit might have cost $100, $200, $500, $1,000.
You don't know.
Your insurer picks up the rest. You pay the copayment.
Copayments are rapidly on the rise in health insurance.
There's a rapid rise in copayments.
Increasingly, insurers are saying, look, health care
costs are out of control.
One way we're going to combat them is by making people bear
more of the cost that they use.
I could go on forever about how I'm a health care
economist. I could go on about health care forever.
But just to fix ideas on why this is an issue, in 1950, the
US economy spent 5% of our gross domestic product, 5% of
our size of the economy went to health care.
Today it's 17%.
By 2075, it's projected to be 40%.
That is of every dollar that's made in America, $0.40 will go
to medical care.
By 100 years later, it's about 100%.
Literally, if we do nothing, the entire economy will be
health care.
Obviously, that can't happen.
We've got to deal with this.
And one way that insurers and some policy makers are saying
we need to deal with this is we need to make consumers bear
more of the costs of their medical care.
We need to make consumers pay more when they go to the
doctor, so that they understand the consequences of
their decision.
Well, if we're going to do that, a key question we need
to know is well, does it affect their behavior?
If we make consumers pay more, and it doesn't at all affect
their demand for medical care-- it's just a tax on
them, essentially--
then that's different than if it causes them to use less
medical care.
It may be good, may be bad.
We'll come back to that.
But the key empirical question is what is the elasticity of
demand for medical care?
If you pay $20 and you pay $0, how much less like are you to
use the doctor when you pay $20 versus when you pay $0.
Well, we can all introspect this and think about it.
But, in fact, to answer this we have to go to the data and
ask, well, what's the difference?
So people, for many years, went to the data.
And they said, look, there's all sorts of differences out
there across people and what they pay for their copayments.
Some people have insurance where they pay nothing, some
where they have $20.
Some people have what they call high-deductible plans.
A deductible plan is where you pay the full cost of your
visits until you reach some limit.
So a $2000 deductible plan will be one where you pay all
of your medical costs until you've spent $2,000.
It's a big copayment.
So we look across those people, and people did.
And they found, look, the people that have plans where
they spend more for health care, where they have a high
copayment, use a lot less health care than where they
don't have to spend anything.
The elasticity of demand looks very, very high.
What is wrong with those studies?
What is wrong with the conclusion those people drew?
They drew it by comparing people who had plans where
they paid a lot to go to the doctor, and therefore use a
lot less care to people who didn't pay anything when they
went to the doctor and used a lot more care.
I pick the $20 person, because I picked on you already.
AUDIENCE: Probably they chose to have a high-deductible
plan, because they don't often go to the doctor already.
PROFESSOR: The rational choice, if you're young and
healthy, for almost everyone in this room, is going to be a
very high-deductible, high copayment plan.
Because it will cost you less money, because the insurer is
shifting the money to you.
But you don't use the doctor anyway.
So who cares?
So the healthier people are going to choose the plans
where they pay more.
So, of course, you're going to find in the plans where people
pay more they use less medical care.
But is it because they're paying more, or is it because
healthy guys choose those plans?
It's causation versus correlation.
We don't know.
Well, how can we figure that out?
Well, if we were doctors, what we'd do-- real doctors, not a
doctor like me, a real doctor, a medical doctor--
what we'd do is we'd run a randomized trial.
So if doctors want to figure out whether a drug works or
not, they don't just look at guys who take the drug versus
guys who don't.
They run a randomized trial.
They randomly assign some people to take the drug and
some people not.
Now, when you run a randomized trial, by definition, you get
a causal effect.
Well, this room isn't quite big enough.
We all know the law of large numbers.
But imagine there were four times as many people in this
room or five times as many people in this room.
OK?
And I had you come up to the front.
I flipped a coin and said half of you are going to take the
drug, and half of you are not, randomly by
the flip of a coin.
Then, by definition, any statistically noticeable
differences I get between the group the takes the drug and
the group that doesn't is caused by the drug.
And how do I know that?
Because I know the groups are otherwise identical by the law
of large numbers.
By the law of large numbers, I know that as long as I have
enough people, they're identical.
So if the only difference between them is that one's
taking the drug and one's not, that's a randomized trial.
That would be how I could solve the causation versus
correlation problem.
In medicine, thousands of randomized trials every day
are being run.
In fact, the FDA, before it will approve a drug, will
typically require a randomized trial.
Well, in the social sciences, it's harder to
run randomized trials.
Because we're actually trying to understand things like
people's demand for medical care, not whether a
drug works or not.
But, in fact, one of the most famous social randomized
trials in history was called the RAND Health Insurance
Experiment run in the 1970s.
This is where some innovative health economists who
understood this problem that we laid out about the fact
that you can't just compare more or less generous health
insurance policies, actually randomized health insurance
policies across people.
They recruited volunteers, and they literally said, we're
going to randomize.
Some people are going to have policies where the health care
is free, and some people are going to have policies where
they have to pay, essentially, all the costs of health care.
So they, essentially, randomized across these
different groups.
And, therefore, they can assess what the price
elasticity was.
Because they knew the price difference between groups.
For one, the price was zero.
For one, the price was one.
They actually had a range of prices they varied it across.
They could look at the quantity response, and they
knew that was a quantity response to the price, because
people weren't choosing their prices.
The prices were being assigned to them.
What did they find?
Well, they found that medical demand is elastic, although
not as elastic as the previous study.
It's somewhat elastic.
It's not as elastic as the previous studies found.
They found that the elasticity of demand for medical care is
around minus 0.2.
So when the price goes up, people use less medical care
but not that much less.
Now, let's be clear.
Remember what elasticity is.
That delta Q over Q. The same study showed that if you take
someone who paid nothing and make them pay almost
everything, their utilization of medical care falls by 45%.
That's consistent with that small elasticity.
Because that's a huge delta P, percent delta P. So,
basically, that's comes to the question about local versus
global elasticities.
So it's not saying that prices don't matter.
But it's not a very, very elastically demanded good.
So that's how they measure that price of
elasticity of demand.
That experiment, which was run over 35 years ago now, that
result drives much of what we do in health policy.
So a lot of the estimates that we saw for the recently passed
health reform bill derived from how
do we get that estimate.
We'll have to figure out how people are going to respond
with their medical care when we give them health insurance.
The recently passed health care bill just gave 32 million
people health insurance.
Well, how are they going to respond to
having health insurance?
We go back to the RAND estimates and say, well, we
have this elasticity of demand.
We know what we're doing to the price.
We figure out how much medical care is going to go up.
But here's the other thing.
Here's the question in the lecture that
that we'll close with.
Is that a good thing or a bad thing that medical care fell
when the price went up?
And how would we tell whether it's a good
thing or a bad thing?
So we know when we raise the price, people use
less medical care.
How can we tell if that's a good thing or a bad thing?
In the same experiment, how could we tell?
What could we do?
Yeah.
AUDIENCE: Maybe you'll get death rates or like--
PROFESSOR: You look at their health.
You say, look, the same trial can
answer a different question.
We know that when you charge someone for health
care, they use less.
Well, are they sicker?
The answer, not at all.
People use less health and were no sicker.
Why?
Because we waste a huge amount of health care in the US.
A huge amount of health care is wasted.
So, in fact, we could cut back quite a lot on health care,
and we'd be no sicker.
And that's what the RAND experiment showed, that we can
charge people to use medical providers.
And they'll use less medical care, and they won't be sicker
as a result.
Which suggests that, actually, as we try to think about
getting our health care costs under control in America,
making people pay something to go to the doctor is not a
crazy thing to be thinking about.
How much?
Well that depends on efficiency versus equity.
We can't make someone who has no income pay $1,000 to go to
the doctor.
That, clearly, is a mistake.
But we can take a rich guy like me and make me pay $50 to
go to the doctor.
There's no reason not to do that.
So, basically, that's a lesson of how you can use elasticity
of demand to help inform the kind of
policies we need to make.
OK, let me stop there.
By the way, if you at all find this stuff interesting, and
you haven't yet read Freakonomics--
how many of you have read Freakonomics?
That's amazing.
OK.
If you haven't read Freakonomics, you should.
It's a great book.
If you're lazy, the movie is coming out.
And Freakonomics the movie is premiering on
Friday the 30th at LSC.
So if you're interested in learning more about empirical
tools in economics, you can watch Freakonomics the movie
on Friday the 30th.