Professional Documents
Culture Documents
Lecture 3 - Consumer Choice
Lecture 3 - Consumer Choice
Class Notes
©Syed M. Ahsan
Spring, 2023
Lecture 3: Consumer Choice
2
Outline of Lecture
▪ 3.1 Introduction
▪ 3.2 Commodities
▪ 3.3 The Consumption Set
▪ 3.4 Competitive Budgets
▪ 3.5 Demand Functions & Comparative
Statics
▪ 3.6 WARP & the Law of Demand
▪ 3.7 Conclusion
Note that time or ‘states of nature’ can be built into the concept of a
commodity space by alternatively interpreting l (= 1, 2,.., ), as
consumption at date- l or consumption in state- l.
Definition: ‘States of nature’. Under conditions of uncertainty, it is
often convenient to describe the yet unknown realised value at a future
point in time of, say consumption or wealth, to depend on the state of
nature (e.g., some good , some bad). In the context of a finite number
of possibilism, we may have l –states of nature.
Arrow had defined a ‘states of nature’ as a description so complete that
if/once known, all uncertainty would be resolved.
Examples of ‘states’: portfolio returns in financial markets, farmer's
harvest of rain-fed crops, milk yielded by a cow/buffalo on the quality
of the feed.
Asf
The left panel is of curse convex, but not the ones to its right. The last has
holes on its boundary (all sides), though the interior of the set is convex.
Note that the intersection of all convex sets is always convex, but not their
union. An easy illustration: Take a square and a rhombus of equal base.
Adv Micro-I, Lect 01_Ahsan 3/23/2023
… Math Appendix, Fig 3.4 Convex Hull
11
Convex Hull: For a set B ⊂ RN, has a convex hull, C0B, if that is the
smallest convex set containing B (i.e., containing all points in B), that is
the intersection of all convex sets that contain B.
Obviously the issue arises when B itself is not convex. Note that the non-
convex part (rt-panel) is being made convex by a straight line at the
boundary points.
Q: What would be a convex hull of the square, i.e., the figure in the
right panel of Fig 3.3?
Returning to the idea consumption sets, RL+, is a convex set.
Of the six examples of consumption sets, in figures 3.1 and 3.2,
determine if any of these are not convex?
Most of the discussion on consumption sets to follow require the
assumption of convexity.
Incidentally, Figs 3.3 and 3.4 are taken from lecture notes (Boyd &
Vandenberghe, www.stanford.edu/class/ee364a/lectures/sets.pdf).
You would agree that many figures in the MWG text are poorly
executed.
dfsdf
Df
The set {x ∈ RL+ : p.x w}, namely the upper boundary of the budget
set, is called a budget hyperplane, which of course in 2-good case,
becomes the budget line. See illustrations in Fig 3.5.
Convexity of the Budget set: The Walrasian budget set Bp,w is convex.
Thus if both bundles x and x are elements of Bp,w , so is the bundle
x = {x +(1- )x}, ∈ [0,1].
This can be verified easily. First note that so long as both x and x >0,
so is x ∈ RL+. Next, we show that the cost of x will not exceed w.
Since both p.x w and p. x w. Thus p.x = {(p.x) +(1- )(p. x)} w.
Thus x∈ Bp,w = {x ∈ RL+ : p.x w}.
It happens that the convexity of the budget set is derived form he
convexity of the consumption set, RL+.
While we mostly analyse convex budget sets, an important non-convex
example has been studied in labour economics (supply of work hours).
Adv Micro-I, Lect 01_Ahsan 3/23/2023
Non-convex Budget: A Labour Market Example
17
Constraints are:
(1) C1 = (y1 –S) and C2 = y2 +S(1 +r),
where r-denotes the rate of return on saving. Substituting for S in the
period-2 constraint, we have
C2 =[y2 +(y1 - C1)(1 +r)], or, re-arranging,
(2) C1(1+r) +C2 = [y1 (1+r) +y2]
Now dividing through by (1+r),
(3) [C1 + C2 (1+r)-1]= [y1 +y2 (1+r)-1].
Here the lhs denotes the present value (PV) of life-time consumption,
while the rhs the PV of life-time earnings.
If we allow bequest at the end of life, in eq (1) the second part would
read (C2+B) = y2 +S(1 +r),
Adv Micro-I, Lect 01_Ahsan 3/23/2023
… Lifetime Consumption Choice
22
Exercise next slide: Note that here that the demand functions are such that
(i) p1x1 = (1w), (ii) p2x2 = (2w), and if =1, also (iii) p3x3 = (3w).
Now it is easy to verify that {1+ 2+ 3} = 1. Hence in this case, i.e., if =1,
we have p.x = w, i.e., the Walras law is satisfied.
If < 1, for the third good, we get p3x3 = ( 3w). Thus
p.x = {1+ 2+ 3}w < w since {1+ 2+ 3} <1.
To check for HOD‘0’, it is trivial to establish that there is no issue here.
Later in chapter 3, we shall see that both the above properties hold
under rather general behavioural conditions and hence are not mere
assumptions, but till then these will remain convenient assumptions.
Normalization: Note that in principle, our demand correspondence
x(p,w) have L+1 independent variables, but we can reduce it to L,
either by setting the price of a (numéraire) good- l to unity, i.e., pl =1,
or alternatively by setting w =1. Thus the effective L-arguments in
x(p,w).
Gong forward, we also assume x(p,w)-single valued. Thus we can write
the x-vector in terms of individual commodities, x l (p,w):
x(p,w) = [x1(p,w), x2(p,w)…., xL(p,w)]T,
where T-denotes the transpose o the row vector.
From here on, we also assume that x l (p,w) to be both continuous and
differentiable.
Adv Micro-I, Lect 01_Ahsan 3/23/2023
Comparative Statics
26
For a given (p,w), {xl (p,w)/w} is known as the wealth effect of good-l .
Normalcy/Inferiority: A good x l is normal at (p,w) if {xl (p,w)/w}0,
and is inferior at (p,w) if the derivative is negative.
Adv Micro-I, Lect 01_Ahsan 3/23/2023
… Consumer’s Budget
28
If all goods are normal at all (p,w), one says that demand is normal.
Normality makes sense if goods are defined as aggregates such as food,
shelter, holidays, … , rather than spinach, cucumber, …
In matrix notation, the wealth effects may be expressed as
Dw.x(p,w) = [x1(p,w)/w, x2(p,w)/w, …. , xL(p,w)/w]T ∈ RL
The derivative, xl (p,w)/pk, denotes the price effect of a change in pk on the
demand for good-l.
Own Price Changes: While one would ordinarily expect people to buy
more of a good when its price falls, i.e., xl (p,w)/pk < 0, it is not
unthinkable that the opposite may happen on occasion. Such goods are called
Giffen goods, i.e, xl (p,w)/pk > 0.
In Fig 3.7 above, in the bottom panel, good-2 is inferior throughout. Moreover
it becomes a Giffen good at {𝑝ഥ1, p2, w).
Imagine a very poor person spending a lot of money on a basic good such as
potatoes; now if the potato price falls, income remaining the same, he now
becomes able to afford other goods (say corn) and ends up buying less
potatoes.