Giffen data?

[From Rick Marken (2002.11.17.2200)]

Bill Williams and other economists --

Is there any actual quantitative data that shows the Giffen effect? Like a plot
of per capita consumption of bread increasing as the price of bread increases?
I was talking to an economist friend of mine about the Giffen effect this
weekend. He said that there was some question about whether the effect even
occurs. I can understand why he would be happier if the Giffen were not real.
But it made me wonder what data leads economists to think that there is a
Giffen effect at all (Stiglitz clearly things there is such an effect)? All I
know of as far as data on the Giffen effect is the anecdotal observation
(apparently made by Giffen) of an increase in the consumption of potatoes when
the price of potatoes soared during the Irish potato blight. But what was
actually observed? Was data collected on price and consumption of potatoes?
There is some hint of a Giffen effect in the animal data. But I think it would
be very useful to see the phenomenon exhibited in some aggregate economic data.
I think that would also help listeners understand what we are trying to explain
with the models that are being developed.

Best regards

Rick

···

--
Richard S. Marken
MindReadings.com
marken@mindreadings.com
310 474-0313

[From Rick Marken (2002.11.18.0900)]

Bill Powers (2002.11.18.0736 MST)--

Rick Marken (2002.11.17.2200) --

>Is there any actual quantitative data that shows the Giffen effect? Like a
>plot of per capita consumption of bread increasing as the price of bread
>increases?

I don't have any, of course, but I've thought of another dimension in which the
Giffen effect can occur. Maybe this will lead to still others.

Instead of money, consider time. We have only H hours of productive labor we can
produce per day. Suppose we budget this time two ways: we work part of the day
at job A, and the other part at Job B. Job A pays more than job B. What happens
now if the wage for Job A is reduced, although it remains higher than the wage
for job B? The answer, I think, is that we have to work more at Job A, the one
for which the wage was reduced. This assumes that the total time is limited to H
hours, so it's not possible to add time to one job without reducing it for the
other. In that case we have to add time to the job that that pays more when its
wage is reduced, in order to maintain sufficient income.

In this example, money takes the place of calories and time takes the place of
money relative to Bill W.'s example,

If we can't add more time to the job that pays more, then reducing the pay for
the lesser job will lead to increasing the time spent on that job, which also
increases the total time spent working. And if we can't physically work any more
total hours, we have to accept the lower income.

I believe there are even single-good cases that might work: Bill W. and Rick,
what do you think?

I think this is a great start. What I/m trying to think of are cases where there
might be relatively accessible and pertinent aggregate data in the economic data
bases like the Statistical index and Federal Reserve data. Maybe there is
something in the data on double income families that might speak to your first
example. Since the amount of time people work on most jobs is fixed -- most people
can't vary their incomes by working more hours -- maybe we can look at the
relative proportion time that both spouses in two income families are working as a
function of the income or changes in the income of each spouse?

I think you make a good case for something like the Giffen effect being something
expected by control theory -- even in the case of one good. I just think it would
be nice to know what data the economists take as evidence for Giffen effect.

Best

Rick

···

--
Richard S. Marken, Ph.D.
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

[From Bill Powers (2002.11.18.0736 MST)]
Rick Marken (2002.11.17.2200) –

Is there any actual quantitative data that shows the Giffen effect?
Like a >plot of per capita consumption of bread increasing as the
price of bread >increases?
I don’t have any, of course, but I’ve thought of another dimension in
which the Giffen effect can occur. Maybe this will lead to still
others.
Instead of money, consider time. We have only H hours of productive labor
we can produce per day. Suppose we budget this time two ways: we work
part of the day at job A, and the other part at Job B. Job A pays more
than job B. What happens now if the wage for Job A is reduced, although
it remains higher than the wage for job B? The answer, I think, is that
we have to work more at Job A, the one for which the wage was reduced.
This assumes that the total time is limited to H hours, so it’s not
possible to add time to one job without reducing it for the other. In
that case we have to add time to the job that that pays more when its
wage is reduced, in order to maintain sufficient income.
In this example, money takes the place of calories and time takes the
place of money relative to Bill W.'s example,
If we can’t add more time to the job that pays more, then reducing the
pay for the lesser job will lead to increasing the time spent on that
job, which also increases the total time spent working. And if we can’t
physically work any more total hours, we have to accept the lower
income.
I believe there are even single-good cases that might work: Bill W. and
Rick, what do you think? Suppose you have a reference level for good A,
and must pay money (i.e., work) to obtain that amount of A. What happens
when the price of A is raised? There are only two long-term possibilities
that will not lead to chronic error: reduce your reference level for A,
or put out the extra money needed to maintain it at the former reference
level. If A is an optional good, the first solution is feasible. You get
a normal downturning price-demand curve. But if it’s a necessity, like
food, the only option is to spend (work) more on A and continue to eat
enough to stay alive. So the higher the price, the more you pay. Maybe
this comes out the same as the original Giffen effect when you include
the reference level for the “necessity.”
Note that this doesn’t increase the amount of A that you get; in fact it
reduces it a little unless the customer is a perfect integrating control
system. There is no demand for more A when the price is raised.
But neither is there a demand for less, and if you measure demand in
terms of money flow, the demand increases because the increased price is
paid.
I hope we’re not losing track of my other suggestion, the one for
explaining a down-turning supply-demand curve. Suppose there are 100
people who want to see movies. Their reference levels vary from 2 movies
per day to 1 movie per 14 days, with a mean value of 1 movie every 5 days
(0.2 movies per day). Let’s say that nobody wants to see less than one
movie every 14 days, or more than 2 movies per day.
If only one movie is made every 30 days, then nobody will be seeing as
many movies as desired; errors be maximum for just about everybody. The
price can be raised quite a lot before anybody refuses to pay it.
Now increase the rate of movie making from 1 every 30 days to 1 every 25
days. The errors will still all be very large, and the price can be kept
nearly at the same level without losing customers.
When the rate of movie-making becomes something like 1 every 18 days, the
people who want to see 1 every 14 days will experience less than maximum
error, and will refuse to pay so much to have movies made any faster.Now
the price has to be lowered to induce them to keep seeing movies. As the
rate of movie-making rises even more, it will surpass the desires of more
and more of the hundred people, necessitating even more price-drops to
maintain attendance. Finally, at the highest rate of making movies,
everybody will be satisfied and the price will have to go to rock-bottom
to keep attendance up.
In other words, I propose that the observed relationship between supply
and demand for a population is a smeared-out version of the curve
for an individual. If only one person wants one good, then the curve will
show the effort (price willingly paid) decreasing steeply as an
increasing amount of good obtained approaches the amount wanted, going to
zero when the amount equals the reference level. When the amount obtained
equals the reference level, the marginal propensity to consume goes to
zero.

When there are two people with different reference levels for the same
good, the rising price will result in first one person, then the other,
reducing the amount willingly paid for the amount recieved. When there
are 10,000,000 people with widely varyied reference levels and loop gains
relative to a given good or service, there will be a broad decline of the
number willing to pay as the price increases over a wide range. If we
take PCT as a set of first principles, that is a derivation of the
population supply-demand relationship from first principles.

Best,

Bill P.