Sunteți pe pagina 1din 24

Consumer Theory

ECON40001/90063

March 3-5, 2014

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

1 / 24

Goals

Existence of utility function Existence and uniqueness of Marshallian (=empirically observable) demand functions. Constrained Optimization (Lagranges method).

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

2 / 24

Consumption Set (Choice Set)


n goods are indexed i = 1, .., n; Quantity of good i is denoted xi , xi R+ ; A typical vector of goods the individual can consume, a consumption bundle, is denoted x = (x1 , ..., xi , .., xn ) Rn + ( R+ ... R+ );
=n

times

n X Rn + is the consumption set. Note that we assume X R+ because we think of a consumption set as all bundles we are able to think about (e.g. me being F1 driver), rather than something that is achievable.

R+ is the set of all non-negative real numbers; In x, bold implies that it is a vector.
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 3 / 24

Primitives: Preferences (Economics)


Faced with the choice between x1 and x2 , an individual is able to express her preferences, e.g. x1 x2 (x1 is weakly preferred to x2 ; weakly means that an individual may be indierent between x1 and x2 .) We feel that there are certain regularities in how people make their choices. We will try to capture these regularities in our mathematical theory.

Source: wikipedia

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

4 / 24

Primitives: Preferences (Mathematics)


Consumption set X An abstract binary relation on set X x2 . Indierence: x1 x2 and x2 x1 . Denoted as x1 x2 . Strict preference: x1 x2 , but not x2 x1 . Denoted as x1

Note: Speaking Economics, you should think of x1 x2 as a YES answer to the question Do you weakly prefer x1 to x2 ? and you should think of x1 x2 as a YES answer to the question Do you weakly prefer x1 to x2 ? AND a YES answer to the question Do you weakly prefer x2 to x1 ?. You should not think of x1 x2 as a YES answer to the question Are you indierent between x1 and x2 . If you do, it implies that I introduced two wiggles, and . If so, I need to have axioms both on and on . Axioms are the objects that ll these wiggles with meaning; before I impose axioms, these wiggles are completely meaningless objects. I do not want many axioms. I would rather describe (via axioms) a single wiggle, and then tell you what and means using a single well-dened wiggle.
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 5 / 24

Primitives: Axioms on

In English, there are some reqularities in how we answer to the question Do you weakly prefer x1 to x2 ? Before we impose any axioms on , in Mathematics there are no regularities; in fact, before we impose axioms we can write f (x1 , x2 ) and, mathematically, it would make no dierence. Axiom 1, Completeness: For all x1 , x2 , either x1 both. x2 or x2 x1 or x2

Axiom 2, Transitivity: For any three bundles x1 , x2 , and x3 , if x1 and x2 x3 , then x1 x3 . For an abstract binary relation , if completeness and transitivity is satised, we will call the binary relation preferences.
Note: what we done here is that we came a step closer matching our mathematical notation, , to what me mean in English when we say preferences.
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014

6 / 24

Axioms 1 and 2: discussion

Completeness: what do you like better, a live chicken or a boiled egg? Failure of transitivity: (redwine , cheese ) (whitewine , cheese ); (beer , cheese ) (redwine , cheese ), (whitewine , cheese ) (beer , cheese ).
Note that (redwine , meat ) (whitewine , meat ); (beer , crackers ) (redwine , crackers ), (whitewine , cheese ) not violate transitivity. (beer , cheese ) does

However, there is a major problem with an individual who violates transitivity in a consistent way (not as a mistake)

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

7 / 24

Violation of transitivity leads to a money pump

x1 x2 , x2 x3 , x3 x1 Start with x3 ; a penny for x2 ; a penny for x1 ; a penny for x3 . The same bundle, but three pennies less. Repeat innitely.
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 8 / 24

A note on Sneetches
Preferences of individual P (initially a plain-belly sneetch): same belly as individual 2. Preferences of individual S (initially a star-belly sneetch): belly that is dierent from individual 1. Formal model: X = {0, 1} {0, 1} (The rst number refers to whether a plain-belly-type of
sneetch has a star (1) or not (0) and the second number refers to whether a star-belly sneetch has a star (1) or not (0). The consumption bundle (1, 1) means that both sneetches have stars.) (1, 1) P (0, 1) and (0, 0) P (1, 0) These are preferences of a plain-belly sneetch. This is indicated by superscript P . Note that these preferences are

not completely specied; other combinations of 0 and 1 are possible. (1, 0) S (1, 1) and (0, 1) S (1, 1) There is no violation of transitivity. Note that we do not compare plain-belly
and star-belly sneetches. They are two dierent individuals and we check the violation of transitivity separately for each of them. Further details on the story: Dr. Seuss, Sneetches
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014

9 / 24

Axioms 3 and 4, I

Axioms 3 and 4 are more technical axioms. If they are violated, some results can still be recovered. Hence, we do not discuss them. Axiom 3, Continuity: there are no sudden changes in preferences, such that 1 orange is better than 0.99999 apple, but 1 apple is better than 1 orange. Mathematically: consider a sequence of pieces of apples, 0.9; 0.99; 0.999; 0.9999. Suppose that each element of this sequence is weakly worse than an orange. This sequence converges to a whole apple. Continuity would then imply that an apple is weakly worse than an orange.

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

10 / 24

Axioms 3 and 4, II
Mathematically: consider a sequence of pieces of apples, 0.9; 0.99; 0.999; 0.9999. Suppose that each element of this sequence is weakly worse than an orange. This sequence converges to a whole apple. Continuity would then imply that an apple is weakly worse than an orange.

Formally: Consider a sequence yn X such that x yn and yn y. Then x y. Consider a sequence zn X such that zn x and zn z. Then z y. Alternative denition (as in JR): for any x X , the sets (x) = {y X |x y} (lower contour set) and (x) = {z X |z (upper contour set) are closed. x}

If it is not obvious to you that these two denitions are equivalent, please have a look at the denition of a closed set and convince yourself that this is true. We will use open/closed sets later, so it would be useful to be familiar with the concepts. Notation: {a A|B} means all a from A such that B is true.
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 11 / 24

Axioms 3 and 4, III


Axiom 4, strict monotonicity: For all x0 , x1 X , if x0 x1 , then x0 while if x0 >> x1 , then x0 x1 . Note that Axiom 4 allows an individual not to care about a particular good; she can be indierent between (3, 1) and (3, 5), but she cannot prefer (3, 1) to (3, 5).
A note on , >> x0 and x1 are vectors. How do we compare (3, 1) and (1, 3)? We say that x0 x1 if some components of x0 are larger than components of x1 and none are smaller. We say that x0 >> x1 if all components of x0 are larger than components of x1 . Neither applies to (3, 1) and (1, 3); I have dened a partial order in Rn .

x1 ,

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

12 / 24

Utility function as representation of preferences


It is not very convenient to work with an abstract binary relation (e.g. maximisation). Can we simplify the representation of preferences? Utility function u : X R represent preference relation 0 1 0 x0 , x1 Rn x1 + , u (x ) u (x ) x , if for each

Note. Suppose u (x) = x1 + x2 represents preference relation . Let us 2 + 2x x + x 2 . Utility function v must represent the look at v (x) = x1 1 2 same preference relation because 2 + 2x x + x 2 = (x + x )2 x1 1 2 1 2 For any two bundles x 0 , x 1 , whenever u (x 0 ) > u (x 1 ), we know that 2 2 v (x 0 ) = u (x 0 ) > u (x 1 ) = v (x 1 ) For any increasing function f : R R, u (x) and f (u (x)) represent the same preference relation (more details: Theorem 1.2 in JR).
How to make a person twice as happy does not make sense. How much of good 2 is needed to compensate for the loss of one unit of good 1 makes sense. f : A B : f takes an element from A and maps it to an element in B , f (a) = b .
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 13 / 24

Existence of utility function, proof, step 1


Theorem (Theorem 1.1, JR) If preference relation is continuous and strictly monotonic, there exists a continuous function u : X R, which represents . Proof outline: 1 Pick a vector e, e.g. e = (1, 1). 2 For any bundle x, nd a number, such that a person is indierent between x and a corresponding fraction of e. For example, how (2, 3) and (1, 1) compare? There may be a number, e.g. 2.5, such that a person is indierent between (2, 3) and (2.5, 2.5). Why do we need this? A utility function needs to map into R, so we are looking for a number. 3 Once we have found a number for each x, we call it a utility function u . We need to check that u indeed represent preferences ; that is, x1 x2 if and only if u (x1 ) u (x2 ), where u (x1 ) and u (x2 ) are the numbers we have constructed.
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 14 / 24

With Axioms 1-4, continuous utility function exists


Theorem (Theorem 1.1, JR) If preference relation is continuous and strictly monotonic, there exists a continuous function u : X R, which represents .
Idea of the existence proof: pick a bundle e (for example (1, 1, . . . , 1)) and measure any other bundle x in relation to this bundle (how much of bundle e is needed to make an individual indierent between it and x?)

Why is it not obvious? Consider lexicographic preferences: for any two 0 bundles x0 , x1 R2 x1 if +, x 0 > x 1 or x1 1 0 = x 1 , then x 0 > x 1 if x1 1 2 2 How would you measure bundle (1, 3) in terms of the bundle (1, 1)? We know that (1, 3) (1, 1), but if we improve (1, 1) a bit, e.g. (1.001, 1.001), suddenly (1.001, 1.001) (1, 3) (example hints at the importance of continuity).
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 15 / 24

Existence of utility function, proof, step 1

Theorem (Theorem 1.1, JR) If preference relation is continuous and strictly monotonic, there exists a continuous function u : X R, which represents . Proof: For each x, dene u (x) R (note, this is just a number) as a number such that u (x)e x. Is u (x) well dened? That is, does it exist for any x and is it unique (if not, we would need to specify a selection rule to pick a particular number)

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

16 / 24

Number u (x) exists, I

Consider two sets, A = {t |t e x} and B = {t |x t e}. Note that both are the sets in R. We would intuitively think that one such set is [, ) and another is [0, ] and would be exactly u (x). We need to show it using axioms. Observation 1: if we can show that there is t such that t A B , then t would dene u (x). We show that such t exists in three steps:

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

17 / 24

Number u (x) exists, II


1

Note that strict monotonicity implies that if t e x, and > t , it must be that e x. Also, if x t e and < t , it must be that x e. From step 1, it follows that set A must be of the form (t , ) or [t , ), where t is an arbitrary number. From continuity, the set A must be closed, so it is [t , ) (I am sloppy here, because the sets A and (x) are not the same and we assumed that (x) is closed, not A). Set B is [0, t ]. Suppose t < t ; that is, there is t such that it is not in A or in B . From completeness, it must be the case that either (i) t e x or (ii) x t e. Yet, if (i) holds, t must belong to A and if (ii) holds, it must belong to B . Thus, we conclude that t t t t As t t t t , t (as well as t ) belongs to both A and B . The number t denes u (x).

In summary: for every x we have found a way to associate a number. Note, that we still have to show that these numbers represent preferences . Before we turn to that, we need to show that if we pick t , we would not get a dierent number.
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 18 / 24

Number u (x) is unique

Suppose that there are two numbers, t and t such that t e x t e. By transitivity, t e t e. By strict monotonicity, t = t (Note how it would not
be true if, at some point, an individual would not have wanted any more goods)

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

19 / 24

u (x) represent preferences

Consider two bundles, x1 and x2 . We have devised a way to construct a number for each of these bundles, u (x1 ) and u (x2 ). Can we be sure that x1 x2 if and only if (denoted by ) u (x1 ) > u (x2 ). By construction: u (x1 )e x1 x2 u (x2 )e. By transitivity, u (x1 )e u (x2 )e. By strict monotonicity, u (x1 ) u (x2 ). Hence, if x1 x2 , u (x1 ) u (x2 ). To show the reverse direction, we repeat the steps in reverse order. We skip continuity argument.

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

20 / 24

Lexicographic preferences, Axioms 1, 2 and 4


We conjectured that lexicographic preferences on R2 cannot be represented by a continuous utility function. This conjecture is correct, but requires a proof. We will check that lexicographic preferences do not satisfy the conditions of Theorem 1.1 and nd the place where the proof fails. Note that this does not prove that lexicographic preferences cannot be represented, because Theorem 1.1 is not if and only if statement.
1

is a preference relation.
1

Completeness: for any two bundles, we have specied a procedure how to compare them Transitivity: Suppose there are three bundles, x1 , x2 , x3 such that x1 x2 and x2 x3 . We need to show that x1 x3 . 1 2 1 2 1 2 If x1 x2 , it means that either x1 x1 and, if x1 = x1 , x2 x2 . 2 3 2 3 2 3 2 3 If x x , it means that either x1 x1 and, if x1 = x1 , x2 x2 . Combining these two observations, we see that it must be the case that 1 3 1 3 1 3 x1 x1 and, if x1 = x1 , then x2 x2 . That means that x1 x3 .

Strict monotonicity is also satised (trivially).


ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 21 / 24

Lexicographic preferences, Axiom 3


Lexicographic preferences are not continuous. Consider set (1, 3) = {(x1 , x2 ) R2 |x1 1 or x1 = 1, x2 3} A compliment to this set is C = {(x1 , x2 ) R2 |x1 > 1 or x1 = 1, x2 > 3}. This is not an open set; that is, we cannot nd a small ball around any point that is contained within the set. Consider a point (1.01, 2). Take a ball of radius 0.001 around this point. Note that the worst point for an agent within this ball is (1.009, 1.999). Note that this point belongs to C . Hence, all other points (which are better) also belong to C . Consider a point (1, 4). Take a ball of radius around this point. Note that the worst point for an agent within this ball is (1 , 4 ). Note that, for any , this point does not belong to C . Hence, the set C is not open and the set (1, 3) is not closed.
ECON40001/90063 L1-2, Consumer Theory March 3-5, 2014 22 / 24

Lexicographic preferences, failure of the proof

In step 2 on page 16 we use continuity to show that set A is [t , ). For point (1, 4), this set is, actually, (1, ) because (1, 4) e, but for any > 1, e (1, 4). Note that if we do not require that u maps into R, but, instead, require it to map into R2 , then the function is already given, which is these numbers themselves. That is, x = (x1 , x2 ) X maps into u (x) = (x1 , x2 ) R2 .

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

23 / 24

Working with proofs


Check where assumptions are used. In this proof, I have put an emphasis every time I have used a particular assumption. We can ask a question: what would happen if we do not make this assumption? Why the proof would fail? Consider strict monotonicity in step 1. What do I need it for? Having more of everything is better. Note that this uses only the second part of strict monotonicity. We do not need, in this particular place, the assumption that giving more of a single good cannot make you worse. Working out the proof with a particular example is very useful (taking some weired case, such as lexicographic preferences, that is expected to fail somewhere, is even more useful).

ECON40001/90063

L1-2, Consumer Theory

March 3-5, 2014

24 / 24

S-ar putea să vă placă și