Mathematical Notes

1. Because one cannot divide by zero, the ratio meaning. It refers to the dependent variable in the
ΔYΔX cannot be evaluated when ΔX = 0. How- demand function from note 2:
ever, as ΔX approaches zero, the ratio ΔYΔX
QD = D1T, Y, N, Yn , p, pj 2
increases without limit:
It takes on a specific value whenever a spe-
lim ∆Y
= ∞ cific value is assigned to each of the independent
∆X S 0 ∆X
variables. The value of QD changes whenever the
Therefore, we say that the slope of a vertical line value of any independent variable is changed. QD
(when ΔX = 0 for any ΔY) is equal to infinity. could change, for example, as a result of a change
2. Many variables affect the quantity demanded. in any one price, in average income, in the distri-
Using functional notation, the argument of the bution of income, in tastes, or in population. It
next several pages of the text can be anticipated. could also change as a result of the net effect of
Let QD represent the quantity of a commodity changes in all of the independent variables occur-
demanded and ring at once.
Some textbooks reserve the term change
T, Y, N, Yn , p, pj in quantity demanded for a movement along a
demand curve, that is, a change in QD as a result
represent, respectively, tastes, average household only of a change in p. They then use other words
income, population, income distribution, the for a change in QD caused by a change in the
commodity’s own price, and the price of the jth other variables in the demand function. This
other commodity. usage is potentially confusing because it gives the
The demand function is single variable QD more than one name.
Our usage, which corresponds to that in
Q D = D1T, Y, N, Yn , p, pj 2, j = 1, 2, c , n more advanced treatments, avoids this confusion.
We call QD quantity demanded and refer to any
The demand schedule or curve is given by change in QD as a change in quantity demanded.
In this usage it is correct to say that a movement
QD = d 1p 2 2 along a demand curve is a change in quantity
T, Y, N,Yn, pj demanded, but it is incorrect to say that a change
where the notation means that the variables to the in quantity demanded can occur only because of
right of the vertical line are held constant. a movement along a demand curve (because
This function is correctly described as the QD can change for other reasons, for example,
demand function with respect to price, all other a ceteris paribus change in average household
variables being held constant. This function, often income).
written concisely as QD = d(p), shifts in response to
4. Similar to the way we treated quantity demanded
changes in other variables. Consider average income:
in note 2, let QS represent the quantity of a com-
if, as is usually hypothesized, 0Q D>0Y 7 0, then
modity supplied and
increases in average income shift QD = d(p) right-
ward and decreases in average income shift QD = d(p) C, X, p, wi
leftward. Changes in other variables likewise shift this
function in the direction implied by the relationship of represent, respectively, producers’ goals, technol-
that variable to the quantity demanded. ogy, the product’s price, and the price of the ith
input.
3. Quantity demanded is a simple and straightfor- The supply function is
ward but frequently misunderstood concept in
everyday use, but it has a clear mathematical Q S = S1C, X, p, wi 2, i = 1, 2, c , m

M-1

Z01_RAGA3072_15_SE_MN.indd M-1 07/01/16 6:11 PM

 M-2 MATHEMATICAL NOTES

The supply schedule or curve is given by become the accepted one, focused on quantity
demanded and quantity supplied at a given price.
Q S = s 1p 2 2 Thus,
C, X, wi
This is the supply function with respect to price, QD = d 1p 2
all other variables being held constant. This func-
QS = s 1p 2
tion, often written concisely as QS = s(p), shifts
in response to changes in other variables. and the condition of equilibrium is
D S
5. Equilibrium occurs where Q = Q . For specified
d 1p 2 = s 1p 2
values of all other variables, this requires that
d 1p 2 = s 1p 2 [5.1] Walras did not use graphical representation. Had
he done so, he would surely have placed p (his
Equation 5.1 defines an equilibrium value independent variable) on the horizontal axis.
of p; although p is an independent or exogenous Marshall’s work influenced later genera-
variable in each of the supply and demand tions of economists, in particular because he
functions, it is an endogenous variable in the was the great popularizer of graphical analysis
economic model that imposes the equilibrium in economics. Today, we use his graphs, even for
condition expressed in Equation 5.1. Price is Walras’s analysis. The axis reversal is thus one of
endogenous because it is assumed to adjust to those historical accidents that seem odd to people
bring about equality between quantity demanded who did not live through the “perfectly natural”
and quantity supplied. Equilibrium quantity, also sequence of steps that produced it.
an endogenous variable, is determined by substi-
tuting the equilibrium price into either d(p) or s(p). 7. The definition in the text uses finite changes and
Graphically, Equation 5.1 is satisfied only at is called arc elasticity because it is being computed
the point where the demand and supply curves across an arc of the demand curve. The parallel
intersect. Thus, supply and demand curves are definition using derivatives is
said to determine the equilibrium values of the
endogenous variables, price and quantity. A shift h =
dQ
# p
in any of the independent variables held constant dp Q
in the d and s functions will shift the demand or and is called point elasticity.
supply curves and lead to different equilibrium
values for price and quantity. 8. The propositions in the text are proved as follows.
Letting TE stand for total expenditure, we can write
6. The axis reversal arose in the following way.
TE = p # Q
Alfred Marshall (1842–1924) theorized in terms
of “demand price” and “supply price,” which
were the prices that would lead to a given quantity It follows that the change in total expenditure is
being demanded or supplied. Thus,
dTE = Q # dp + p # dQ [8.1]
pD = d 1Q 2 [6.1]
Multiplying and dividing both terms on the right-
p S = s 1Q 2 [6.2]
hand side of Equation 8.1 by p · Q yields
and the condition of equilibrium is
d 1Q 2 = s 1Q 2 dTE = c
dp
+
dQ
d # 1p # Q 2
p Q
When graphing the behavioural relationships
expressed in Equations 6.1 and 6.2, Marshall Because dp and dQ are opposite in sign as we
naturally put the independent variable, Q, on the move along the demand curve, dTE will have the
horizontal axis. same sign as the term in brackets on the right-
Leon Walras (1834–1910), whose formula- hand side that dominates—that is, on which per-
tion of the working of a competitive market has centage change is largest.

Z01_RAGA3072_15_SE_MN.indd M-2 07/01/16 6:11 PM

 MATHEMATICAL NOTES M-3

A second way of arranging Equation 8.1 is to which says that the consumer can get more util-
divide both sides by dp to get ity by increasing consumption of the product;
second,
= Q + p#
dTE dQ
[8.2]
dp dp 0 2U>0X2i 6 0
From the definition of point elasticity in note 7, which says that the utility of additional consump-
however, tion of some product declines as the amount of that

Q#h = p#
dQ product consumed increases.
[8.3]
dp 11. Because the slope of the indifference curve is
negative, it is the absolute value of the slope that
which we can substitute into Equation 8.1 to
declines as one moves downward to the right
obtain
along the curve. The algebraic value, of course,
= Q + Q # h = Q # 11 + h2 [8.4]
dTE increases. The phrase diminishing marginal rate
dp of substitution thus refers to the absolute, not the
algebraic, value of the slope.
Because h is a negative number, the sign of the
right-hand side of Equation 8.4 is negative if 12. The relationship between the slope of the budget
the absolute value of h exceeds unity (elas- line and relative prices can be seen as follows.
tic demand) and positive if it is less than unity In the two-good example with prices held con-
(inelastic demand). stant, a change in expenditure (ΔE) is given by
Total expenditure is maximized when the equation
dTEdp is equal to zero. As can be seen from
Equation 8.4, this occurs when elasticity is equal ∆E = pC # ∆C + pF # ∆F [12.1]
to –1.
Expenditure is constant for all combinations
9. The distinction between an incremental change of F and C that lie on the same budget line. Thus,
and a marginal change is the distinction for along such a line we have ΔE = 0. This implies
the function Y = Y(X) between ΔYΔX and the
derivative dYdX. The latter is the limit of the pC # ∆C + pF # ∆F = 0 [12.2]
former as ΔX approaches zero. We shall meet
this distinction repeatedly—in this chapter in and thus
reference to marginal and incremental utility and
∆C> ∆F = - pF >pC [12.3]
in later chapters with respect to such concepts
as marginal and incremental product, cost, and The ratio ΔCΔF is the slope of the budget
revenue. Where Y is a function of more than one line. It is negative because, with a fixed budget,
variable—for example, Y = f(X,Z)—the mar- one must consume less C in order to consume
ginal relationship between Y and X is the partial more F. In other words, Equation 12.3 says that
derivative ∂Y∂X rather than the total derivative the negative of the slope of the budget line is
dYdX. the ratio of the absolute prices (i.e., the relative
10. The hypothesis of diminishing marginal utility price). Although prices do not show directly
requires that we can measure utility of consump- in Figure 6A-3, they are implicit in the budget
tion by a function line: Its slope depends solely on the relative
price, while its position, given a fixed money
U = U 1X1,X2, c , Xn 2 income, depends on the absolute prices of the
two goods.
where X1, …, Xn are quantities of the n products
consumed. It really embodies two utility hypoth- 13. Marginal product, as defined in the text, is really
eses: first, incremental product. More advanced treatments
distinguish between this notion and marginal prod-
0U>0X i 7 0 uct as the limit of the ratio as ΔL approaches zero.

Z01_RAGA3072_15_SE_MN.indd M-3 07/01/16 6:11 PM

 M-4 MATHEMATICAL NOTES

Marginal product thus measures the rate at d 1Q>L2 L # 1dQ>dL2 - Q

which total product is changing as one factor is = [15.4]
dL L2
varied and is the partial derivative of the total
product with respect to the variable factor. In The right-hand side may be rewritten
1 # dQ
d = # 1MP - AP2
symbols, Q 1
c - [15.5]
L dL L L
0TP
MP = Clearly, when MP is greater than AP, the expres-
0L
sion in Equation 15.5 is positive and thus AP is
14. We have referred specifically both to diminish- rising. When MP is less than AP, AP is falling.
ing marginal product and to diminishing average When they are equal, AP is neither rising nor
product. In most cases, eventually diminishing falling.
marginal product implies eventually diminish- 16. The text defines incremental cost. Strictly, mar-
ing average product. This is not true in all cases, ginal cost is the rate of change of total cost with
however, as the accompanying figure shows. respect to output, Q. Thus,
In the figure below, marginal product
diminishes after v units of the variable factor dTC
MC =
are employed. Because marginal product falls dQ
toward, but never quite reaches, a value of m,
From the definitions, TC = TFC + TVC. Fixed
average product rises continually toward, but
costs are not a function of output. Thus, we may
never quite reaches, the same value.
write TC = Z + f(Q), where f(Q) is total variable
costs and Z is a constant. From this we see that
MC = df(Q)dQ. MC is thus independent of the
size of the fixed costs.
Output Per Unit

17. This point is easily seen if a little algebra is used:

MP
TVC
m AVC =
AP Q
but note that TVC = L · w and Q = AP · L,
where L is the quantity of the variable factor
0 v used and w is its cost per unit. Therefore,
L#w
Units of the Variable Factor
w
AP # L
AVC = =
AP
15. Let Q be the quantity of output and L the quan-
tity of the variable factor. In the short run, Because w is a constant, it follows that AVC and
AP vary inversely with each other, and when AP
TP = Q = f1L2 [15.1] is at its maximum value, AVC must be at its mini-
mum value.
We now define
18. A little elementary calculus will prove the point:
d 1L # w 2
Q f1L2
AP = = [15.2] dTC dTVC
L L MC = = =
dQ dQ dQ
dQ If w does not vary with output,
MP = [15.3]
dL
dL #
We are concerned with the relationship between MC = w
AP and MP. Where average product is rising, at a dQ
maximum, or falling is determined by its deriva- However, referring to note 15 (Equation 15.3),
tive with respect to L: we see that

Z01_RAGA3072_15_SE_MN.indd M-4 07/01/16 6:11 PM

 MATHEMATICAL NOTES M-5

dL 1 pK # K + pL # L = C
=
dQ MP
To do this, form the Lagrangean,

£ = Q 1K,L2 - l 1pK # K + pL # L - C 2
Thus,

w
MC = where l is called the Lagrange multiplier.
MP The first-order conditions for this maximiza-
Because w is fixed, MC varies negatively with tion problem are
MP. When MP is at a maximum, MC is at a mini- QK = l # p K [20.1]
mum.
QL = l # p L [20.2]
pK # K + pL # L = C
19. Strictly speaking, the marginal rate of substitu-
tion refers to the slope of the tangent to the iso- [20.3]
quant at a particular point, whereas the calcula- Dividing Equation 20.1 by Equation 20.2 yields
tions in Table 8A-1 refer to the average rate of
substitution between two distinct points on the
isoquant. Assume a production function QK pK
=
QL pL
Q = Q 1K,L2 [19.1]
That is, the ratio of the marginal products, which
Isoquants are given by the function
is –1 times the MRS, is equal to the ratio of the
K = I 1L, Q 2 [19.2] factor prices, which is –1 times the slope of the
isocost line.
derived from Equation 19.1 by expressing K as
an explicit function of L and Q. A single iso- 21. Marginal revenue is mathematically the derivative
quant relates to a particular level of output, Q. of total revenue with respect to output, dTRdQ.
Define QK and QL as an alternative, more com- Incremental revenue is ΔTRΔQ. However, the
pact notation for ∂Q∂K and ∂Q∂L, the mar- term marginal revenue is used loosely to refer to
ginal products of capital and labour. Also, let both concepts.
QKK and QLL stand for ∂2Q∂K2 and ∂2Q∂L2,
22. For notes 22 through 24, it is helpful first to
respectively. To obtain the slope of the isoquant,
define some terms. Let
totally differentiate Equation 19.1 to obtain

dQ = QK . dK + QL # dL
pn = TRn - TCn
where pn is the profit when Qn units are sold.
Then, because we are moving along a single iso-
If the firm is maximizing its profits by pro-
quant, set dQ = 0 to obtain
ducing Qn units, its profits must be at least as
dK QL large as the profits at output zero. That is,
= - = MRS pn Ú p0 [22.1]
dL QK
This condition says that profits from producing
Diminishing marginal productivity implies QLL
must be greater than profits from not producing.
< 0 and QKK < 0, and hence, as we move down
Condition 22.1 can be rewritten as
the isoquant of Figure 8A-1, QK is rising and QL
is falling, so the absolute value of MRS is dimin- TRn - TVCn - TFCn
ishing. This is called the hypothesis of a dimin- Ú TR0 - TVC0 - TFC0 [22.2]
ishing marginal rate of substitution.
However, note that by definition
20. Formally, the problem is to choose K and L in TR0 = 0 [22.3]
order to maximize
TVC0 = 0 [22.4]
Q = Q 1K, L2
TFCn = TFC0 = Z [22.5]
subject to the constraint

Z01_RAGA3072_15_SE_MN.indd M-5 07/01/16 6:11 PM

 M-6 MATHEMATICAL NOTES

where Z is a constant. By substituting 25. The equation for a downward-sloping straight-

Equations 22.3, 22.4, and 22.5 into Condition line demand curve with price on the vertical axis is
p = a - b#Q
22.2, we get

TRn - TVCn Ú 0 where a is the vertical intercept (when Q = 0)

and –b is the slope of the demand curve. Total
from which we obtain
revenue is price times quantity:
TR = p # Q = a # Q - b # Q2
TRn Ú TVCn
This proves Rule 1.
Marginal revenue is
On a per-unit basis, it becomes

= a - 2#b#Q
TRn TVCn dTR
Ú [22.6] MR =
Qn Qn dQ

where Qn is the number of units produced. Thus, the MR curve and the demand curve are
Because TRn = Qn × pn, where pn is the both straight lines, they have the same vertical
price when n units are sold, Condition 22.6 may intercept (a), and the (absolute value of the) slope
be rewritten as of the MR curve (2b) is twice that of the demand
curve (b).
pn Ú AVCn
26. The marginal revenue produced by the factor
23. Using elementary calculus, we may prove Rule 2. involves two elements: first, the additional out-
put that an extra unit of the factor produces and,
pn = TRn - TCn second, the change in price of the product that the
each of which is a function of output Q. To max- extra output causes. Let Q be output, R revenue,
imize π, it is necessary that and L the number of units of the variable factor
hired. The contribution to revenue of additional
dp labour is ∂R∂L. This, in turn, depends on the
= 0 [23.1]
dQ contribution of the extra labour to output ∂Q∂L
(the marginal product of the factor) and the contri-
From the definitions,
bution of the extra output to revenue ∂R∂Q (the
firm’s marginal revenue). Thus,
dp dTR dTC
= - = MR - MC [23.2]
0Q 0R
dQ dQ dQ 0R
= #
0L 0L 0Q
From Equations 23.1 and 23.2, a necessary con-
dition for attaining maximum π is MR – MC = 0, We define the left-hand side as marginal revenue
or MR = MC, as is required by Rule 2. product, MRP. Thus,
24. To prove that for a negatively sloped demand MRP = MP # MR
curve, marginal revenue is less than price, let
p = p(Q). Then 27. The proposition that the marginal labour cost
is above the average labour cost when the aver-
TR = p # Q = p 1Q 2 # Q age is rising is essentially the same mathematical
proposition proved in note 15. Nevertheless, let
us do it again, using elementary calculus.
= Q#
dTR dp
MR = + p The quantity of labour supplied depends on
dQ dQ
the wage rate: Ls = f(w). Total labour cost along
For a negatively sloped demand curve, dpdQ is the supply curve is w · Ls. The average cost of
negative, and thus MR is less than price for posi- labour is (w · Ls)Ls = w. The marginal cost of
tive values of Q. labour is

Z01_RAGA3072_15_SE_MN.indd M-6 07/01/16 6:11 PM

 MATHEMATICAL NOTES M-7

d 1w # Ls 2
= w + Ls #
dw If z is less than 1, the series in parentheses con-
dLs dLs verges to 1(1 – z) as n approaches infinity. The
total change in expenditure is thus ΔA(1 – z). In
Rewrite this as the example in the box, z = 0.80; therefore, the
change in total expenditure is
MC = AC + Ls #
dw
= 5 # ∆A
dLs ∆A ∆A
=
1 - z 0.2
As long as the supply curve slopes upward,
dwdLs > 0; therefore, MC > AC. 31. The “rule of 72” says that any sum growing at
28. In the text, we define MPC as an incremental ra- the rate of X percent per year will double in ap-
tio. For mathematical treatment, it is sometimes proximately 72X years. For two sums growing
convenient to define all marginal concepts as de- at the rates of X percent and Y percent per year,
rivatives: MPC = dCdYD, MPS = dSdYD, and the difference between the two sums will double
so on. in approximately 72(X – Y) years. The rule of
72 is only an approximation, but at low annual
29. The basic relationship is rates of growth it is extremely accurate.
YD = C + S 32. A simple example of a production function is
GDP = z(LK)½. This equation says that to find
Dividing through by YD yields
the amount of GDP produced, multiply the
amount of labour by the amount of capital, take
YD C S
= + the square root, and multiply the result by the
YD YD YD constant z which is a technology parameter. This
and thus production function has positive but diminishing
marginal returns to either factor. This can be seen
1 = APC + APS by evaluating the first and second partial deriva-
tives and showing the first derivatives to be posi-
Next, take the first difference of the basic rela- tive and the second derivatives to be negative.
tionship to get For example,
∆YD = ∆C + ∆S 0GDP z # L1>2
= 7 0
Dividing through by ΔYD gives 0K 2 # K1>2
and
z # L1>2
∆YD ∆C ∆S
= + 02GDP
= - 6 0
4 # K3>2
∆YD ∆YD ∆YD
0K2
and thus
33. The production function GDP = z(LK)12 dis-
1 = MPC + MPS plays constant returns to scale. To see this, multi-
ply both L and K by the same constant, u, and see
30. The total expenditure over all rounds is the that this multiplies the whole value of GDP by u:
z1uL # uK 2 1>2 = z1u 2 # LK 2 1>2 = uz1LK 2 1>2 = u # GDP
sum of an infinite series. If we let A stand for
autonomous expenditure and z for the marginal
propensity to spend, the change in autonomous
expenditure is ΔA in the first round, z · ΔA in the 34. This is easily proved. The banking system wants
second, z2 · ΔA in the third, and so on. This can sufficient deposits (D) to establish the target ratio
be written as (v) of deposits to reserves (R). This gives RD = v.
Any change in D of size ΔD has to be accompa-
∆A # 11 + z + z2 + c + zn 2 nied by a change in R of ΔR of sufficient size to
restore v. Thus, ΔRΔD = v, so ΔD = ΔRv and

Z01_RAGA3072_15_SE_MN.indd M-7 07/01/16 6:11 PM

 M-8 MATHEMATICAL NOTES

v #
ΔDΔR = 1v. This can be shown also in terms ∆R = X
of the deposits created by the sequence in Table v + c
26-7. Let v be the reserve ratio and e = 1 – v be c #
the excess reserves per dollar of new deposits. If ∆C = X
v + c
X dollars are initially deposited in the system, the
successive rounds of new deposits will be X, eX, For example, when v = 0.20 and c = 0.05, an
e2X, e3X, . . . . The series injection of $100 will lead to an increase in
X + eX + e2X + e3X + c reserves of $80, an increase in cash in the hands
of the public of $20, and an increase in deposits
= X # 3 1 + e + e2 + e3 + c 4 of $400.
36. Let d be the government’s debt-to-GDP ratio and
has a limit of X # 1 let Δd be the annual change in d. The percent-
1 - e age change in d over the year is therefore Δdd,
which for small percentage changes is very closely
approximated by
= X#
1 X
=
1 - 11 - v2 v
∆d>d = ∆D>D - ∆GDP>GDP
This is the total new deposits created by an injec-
tion of $X of new reserves into the banking where ΔD is the budget deficit and is equal to
system. For example, when v = 0.20, an injection G – T + iD. The second term is the percentage
of $100 into the system will lead to an overall change in nominal GDP, which is approximately
increase in deposits of $500. equal to g + π, where g is the growth rate of real
GDP and π is the rate of inflation. We therefore
35. Suppose the public wants to hold a fraction, rewrite the expression as
c, of deposits in cash, C. Now suppose that X
dollars are injected into the system. Ultimately, ∆d>d = 1G - T + iD2 >D - 1g + p2
this money will be held either as reserves by the
banking system or as cash by the public. Thus, Now, multiply both sides by d to get:
we have
∆d = 1G - T + iD2 >GDP - 1g + p2d
∆C + ∆R = X
From the banking system’s reserve behaviour, we We can now let x be the primary budget deficit
have ΔR = v · ΔD, and from the public’s cash (G – T) as a share of GDP. The equation then
behaviour, we have ΔC = c · ΔD. Substituting becomes
into the above equation, we get the result that
∆d = x + 1i - p - g2d
X
∆D = Finally, note that the real interest rate on govern-
v + c
ment bonds is r = i 2 π, and so our final equation
From this we can also relate the change in becomes
reserves and the change in cash holdings to the
initial injection: ∆d = x + 1r - g2d

Z01_RAGA3072_15_SE_MN.indd M-8 07/01/16 6:11 PM