ECON0702: Mathematical Methods in Economics

Transcription

1 ECON0702: Mathematical Methods in Economics Yulei Luo SEF of HKU January 14, 2009 Luo, Y. (SEF of HKU) MME January 14, / 44

2 Comparative Statics and The Concept of Derivative Comparative Statics is concerned with the comparison of di erent equilibrium states that are associated with di erent sets of values of parameters and exogenous variables. When the value of some parameter or exogenous variable that is associated with an initial equilibrium changes, we can get a new equilibrium. The question posted in the Comparative Statics analysis is: How would the new equilibrium compare with the old one? Note that in the CS analysis, we don t concern with the process of adjustment of the variables; we merely compare the initial equilibrium state with the nal equilibrium. Luo, Y. (SEF of HKU) MME January 14, / 44

3 (Continued.) The problem under consideration is essentially one of nding a rate of change: the rate of change of the equilibrium value of an endogenous variable with respect to the change in a particular parameter or exogenous variable. Hence, the concept of derivative is the key factor in comparative statics analysis. We will study the rate of change of any variable y in response to a change in another variable x: y = f (x). (1) Note that in the CS analysis context, y represents the equilibrium value of an endogenous variable, and x represents some parameter or exogenous variable. The di erence quotient. We use the symbol to denote the change from one point, say x 0, to another point, say x 1. Thus x = x 1 x 0. When x changes from x 0 to x 0 + x, the value of the function y = f (x) changes from f (x 0 ) to f (x 0 + x). The change in y per unit of change in x can be expressed by the di erence quotient: y x = f (x 0 + x) f (x 0 ) x Luo, Y. (SEF of HKU) MME January 14, / 44 (2)

4 Quick Review of Derivative, Di erentiation, and Partial Di erentiation The derivative of the function y = f (x) is the limit of the di erence quotient y x exists as x! 0. The derivative is denoted by dy dx = y 0 = f 0 y (x) = lim x!0 x Note that (1) a derivative is also a function; (2) it is also a measure of some rate of change since it is merely a limit of the di erence quotient; since x! 0, the rate measured by the derivative is an instantaneous rate of change; and (3) the concept of the slope of a curve is merely the geometric counterpart of the concept of derivative. Example: If y = 3x 2 4, dy dx = y 0 = 6x. (3) Luo, Y. (SEF of HKU) MME January 14, / 44

5 (Continued.) The concept of limit. For a given function q = g(v), if, as v! N (it can be any number) from the left side (from values less than N), q approaches a nite number L, we call L the left-side limit of q. Similarly, we call L the right-side limit of q. The left-side limit and right-side limit of q are denoted by lim v!n q and lim v!n + q, respectively. The limit of q at N is said to exist if lim q = lim q = L (4) v!n v!n + and is denoted by lim v!n q = L. Note that L must be a nite number. The concept of continuity. A function q = g(v) is said to be continuous at N if lim v!n q exists and lim v!n g(v) = g(n). Thus continuity involves the following requirements: (1) the point N must be in the domain of the function; (2) lim v!n g(v) exists; and (3) lim v!n g(v) = g(n). Luo, Y. (SEF of HKU) MME January 14, / 44

6 (Continued.) The concept of di erentiability. By the de nition of the derivative of a function y = f (x), at x 0, we know that f 0 (x 0 ) exists if and only if the limit of the di erence quotient y x exists at x = x 0 as x! 0, that is, f 0 y (x 0 ) = lim x!0 x = lim x!0 f (x 0 + x) f (x 0 ) x (Di erentiation condition). Di erentiation is the process of obtaining the derivative dy dx. Note that the function y = f (x) is continuous at x 0 if and only if lim f (x) = f (x 0 ) (Continuity condition). x!x 0 Luo, Y. (SEF of HKU) MME January 14, / 44

7 (Continued.) Continuity and di erentiability are very closely related to each other. Continuity of a function is a necessary condition for its di erentiability, but this condition is not su cient. Consider example: f (x) = jxj, which is clearly continuous at x = 0, but is not di erentiable at x = 0. Continuity rules out the presence of a gap, whereas di erentiability rules out sharpness as well. Therefore, di erentiability calls for smoothness of the function as well as its continuity. Most of the speci c functions used in economics are di erentiable everywhere. Luo, Y. (SEF of HKU) MME January 14, / 44

8 Rules of Di erentiation The central problem of comparative-static analysis, that of nding a rate of change, can be identi ed with the problem of nding the derivative of some function f (x), provided only a small change in x is being considered. Constant function rule: for y = f (x) = c, where c is a constant, then dy dx = f 0 (x) = 0. Power function rule: if y = f (x) = x α where α 2 ( number, then dy dx = αx α 1. Power function rule generalized: if y = f (x) = cx α, then dy dx = cαx α 1., ) is any real Luo, Y. (SEF of HKU) MME January 14, / 44

9 (continued. For two or more functions of the same variable) Sum-di erence rule: d dx [f (x) g (x)] = d dx f (x) d dx g (x) = f 0 (x) + g 0 (x), which can easily extend to more functions " d n # n f dx i (x) = i =1 i =1 Product rule: Quotient rule: d dx f i (x) = n fi 0 (x). i =1 d dx [f (x) g (x)] = f (x) d dx g (x) + g (x) d dx f (x) = f (x) g 0 (x) + g (x) f 0 (x) d dx f (x) g (x) = f 0 (x) g (x) g 0 (x) f (x) g 2. (x) Luo, Y. (SEF of HKU) MME January 14, / 44

10 Marginal-revenue Function and Average-revenue Function Suppose that the average-revenue function (AR) is speci ed by AR = 15 Q,then the total revenue function (TR) is TR = AR Q = 15Q Q 2, (5) which means that the marginal revenue (MR) function is given by MR = d(tr) dq In general, if AR = f (Q), then = 15 2Q. (6) TR = f (Q) Q and MR = f (Q) + f 0 (Q) Q, which means that MR AR = f 0 (Q) Q. Note that since AR = TR Q = PQ Q = P, we can view AR as the inverse demand function for the product of the rm. If the market is perfect competition, that is, the rm takes the price as given, then P = f (Q) =constant, which means that f 0 (Q) = 0 and thus MR = AR. Luo, Y. (SEF of HKU) MME January 14, / 44

11 Marginal-cost Function and Average-cost Function Suppose that a total cost function is C = C (Q), (7) the average cost (AC) function and the marginal cost (MC) function are given by AC = C (Q) Q and MC = C 0 (Q). The rate of change of AC with respect to Q is d 6 C (Q) 7 dq 4 Q 5 = 1 6 Q 4 C 0 C (Q) < > 0 (Q) 7 {z } Q 5 = = 0 : {z } {z } < 0 MC AC AC i 8 < : MC > AC MC = AC MC < AC Luo, Y. (SEF of HKU) MME January 14, / 44

12 Rules of Di erentiation Involving Functions of Di erent Variables Consider cases where there are two or more di erentiable functions, each of which has a distinct independent variables, Chain rule: If we have a function z = f (y), where y is in turn a function of another variable x, say, y = g(x), then the derivative of z with respect to x gives by dz dx = dz dy dy dx = f 0 (y) g 0 (x). (8) Intuition: Given a x, there must result a corresponding y via the function y = g(x), but this y will in turn being about a z via the function z = f (y). Example: Suppose that total revenue TR = f (Q), where output Q is a function of labor input L, Q = g (L). By the chain rule, the marginal revenue of labor is MRL = dtr dl = dtr dq dq dl = f 0 (Q) g 0 (L). (9) Luo, Y. (SEF of HKU) MME January 14, / 44

13 Inverse function rule: If the function y = f (x) represents a one-to-one mapping, i.e., if the function is such that a di erent value of x will always yield a di erent value of y, then function f will have an inverse function x = f 1 (y), (10) note that here the symbol f function f (x). 1 doesn t mean the reciprocal of the For monotonic functions, the corresponding inverse functions exist. Generally speaking, if an inverse function exists, the original and the inverse functions must be both monotonic. For inverse functions, the rule of di erentiation is dx dy = 1. (11) dy dx Examples: Suppose that y = x 5 + x, y 0 = 5x and dx dy = 1 5x Luo, Y. (SEF of HKU) MME January 14, / 44

14 Partial Di erentiation In CD analysis, several parameters appear in a model, so that the equilibrium value of each endogenous variable may be a function of more than one parameter. Partial derivatives: Consider a function y = f (x 1, x 2,, x n ),where the variables x i are all independent of one another, so that each can vary by itself without a ecting the others. If the variable x i changes x i while the other variables remain xed, there will be a corresponding change in y, y : y = f (x 1,, x i + x i,, x n ) f (x 1,, x i,, x n ). x i x The partial derivative of y with respect to x i is de ned as y y = f i = lim x i x i!0 x i Techniques of partial di erentiation: PD di ers from the previously discussed di erentiation primarily in that we must hold the other independent variables constant while allowing one variable to vary. Luo, Y. (SEF of HKU) MME 2 January 2 14, / 44

15 Applications to Comparative-Static Analysis How the equilibrium value of an endogenous variable will change when there is a change in any exogenous variables or parameter? The Single Market Model. The model setup is Q d = a bp and Q s = c + dp where a, b, c, and d are all positive. Q d = Q s implies that which means P a P c P = a + c b + d and ad bc Q = b + d, = = 1 b + d 1 b + d > 0, P b > 0, P d = (a + c) (b + d) 2 < 0, = (a + c) (b + d) 2 < 0. Luo, Y. (SEF of HKU) MME January 14, / 44

16 The National Income Model The model setup is Y = C + I 0 + G 0 (12) C = α + β (Y T ) where α > 0, 0 < β < 1. (13) T = γ + δy where γ > 0, 0 < δ < 1, (14) where the rst equation in this system gives the equilibrium condition for national income, while the second and third equations show how C and T are determined in the model. α > 0 means that consumption is positive even if disposable income is 0; β is a positive fraction because it represents the marginal propensity to consume; γ is positive because even if Y is zero the government will still have a positive revenue; δ is an income tax rate that cannot exceed 100%. Combining the three equations gives Y = α βγ + I 0 + G 0. (15) 1 β + βδ Luo, Y. (SEF of HKU) MME January 14, / 44

17 (continued.) From the equilibrium income (15), we can obtain the following comparative-static derivatives that imply the e ects of government policy: Y G 0 = Y γ = Y δ = 1 1 β + βδ > 0 (16) β 1 β + βδ < 0 (17) β (α βγ + I 0 + G 0 ) (1 β + βδ) 2 = βy (1 β + βδ) 2 < 0 (18) Implications: (16) gives us the government-expenditure multiplier, which is positive since β < 1 and βδ > 1. (17) is the nonincome-tax multiplier because it measures the e ect of a change in γ on the equilibrium income. (18) measures the e ect of a change in the income tax rate on equilibrium income. Luo, Y. (SEF of HKU) MME January 14, / 44

18 Note on Jacobian Determinants Partial derivative is useful in CS analysis, but it also tests whether functional (linear or nonlinear) dependence among a set of n functions in n variables. This is related to the notion of Jacobian determinants. Consider the two functions y 1 = 2x 1 + 3x 2 and y 2 = 4x x 1 x 2 + 9x 2 2, all the four partial derivatives are y 1 x 1 = 2, y 1 x 2 = 3, y 2 x 1 = 8x x 2, y 2 x 2 = 12x x 2, which can be arranged into a square matrix which is called a Jacobian matrix J " # y1 y 1 J = x 1 x = (19) 8x x 2 12x x 2 y 2 x 1 y 2 x 2 and its determinant is known as a Jacobian determinant, jjj. Luo, Y. (SEF of HKU) MME January 14, / 44

19 (continued.) More generally, if we have n di erentiable functions in n variables, y 1 = f 1 (x 1, x n ),, y n = f n (x 1, x n ), we can then derive a total of n 2 partial derivatives. The Jacobian determinant is y 1 y jjj = (y 1 y n ) (x 1 x n ) = x 1 1 x n (20) Theorem: jjj will be identically zero for all values of x 1,, x n if and only if the n functions are functionally (linearly or nonlinearly) dependent. E.g., for the above case, J = 2 (12x x 2 ) 3 (8x x 2 ) = 0 for any x 1 and x 2, which then means that the two functions are dependent (y 2 is just y 1 squared: they are functionally nonlinearly dependent). Note that test of linear dependence of a linear equation system is a special application of the Jacobian criterion of functional dependence. Luo, Y. (SEF of HKU) MME January 14, / 44 y n x 1 y n x n

20 Comparative Static Analysis of General-functions The study of partial derivatives allows us to handle the simple type of CS problem, in which the equilibrium solution of the model can be explicitly stated in the reduced-form. Note that the de nition of PD requires that the absence of any functional relationship among the independent variables. As applied to CS, this means that parameters or exogenous variables must be mutually independent. However, in some cases, there is no explicit reduced-form solution due to the inclusion of general functions in a model. In such a case, we have to nd the CS derivatives directly from the originally given equations. For example, Y = C + I 0 + G 0 (Equilibrium condition) (21) C = C (Y, T 0 ) (T 0 : Exogenous taxes), (22) which can be reduced to a single equation Y = C (Y, T 0 ) + I 0 + G 0 to be solved for Y. No explicit solution is available due to the general function C.Must nd the CS derivatives directly from this equation. Luo, Y. (SEF of HKU) MME January 14, / 44

21 (continued.) Suppose that an equilibrium Y does exist. Then under certain conditions (will discuss later), we may take Y to be a di erentiable function of the exogenous variables: Y = Y (I 0, G 0, T 0 ) (23) even though we are unable to determine the explicit form of this function. Furthermore, in some neighborhood of the equilibrium value, we have Y = C (Y, T 0 ) + I 0 + G 0, (24) which is called equilibrium condition because Y is replaced by its equilibrium value Y. Since Y is a function of T 0, the two arguments in the C function are not independent. Speci cally, in this case T 0 a ects C not only directly, but also indirectly via Y. Consequently, PD is no longer appropriate for our purposes. We must resort to total di erentiation and total derivatives, which can be used to measure the rate of change in which the arguments are not independent. Luo, Y. (SEF of HKU) MME January 14, / 44

22 The symbol, dy dx, for the derivative of the function y = f (x), has been regarded as a single entity. It can also be interpret as a ratio of two quantities, dy and dx, which are called the di erentials of x and y, respectively. The process of nding the di erential dy from a given function y = f (x) is called di erentiation: dy = f 0 (x) dx. Economic application (Point elasticity of the demand function). Given a demand function Q = f (P), the point elasticity of demand is de ned as ε d = dq/q dp/p = dq/dp marginal function = Q/P average function, (25) 8 < > 1 : the demand is elastic jε d j = 1 : the demand is of unit elastic. : < 1 : the demand is inelastic Example: Given the demand function Q = 100 2P, P ε d = 50 P = 1 when P = when P = 30. Luo, Y. (SEF of HKU) MME January 14, / 44

23 Total Di erentials The concept of di erentials can be extended to a function of two or more independent variables. Consider a saving function S = S (Y, i),where S is savings, Y is national income, and i is the interest rate. The total change in S is then approximated by the di erential ds = S S dy + Y i di = S Y dy + S i di. (26) ds is called the total di erential of the saving function (the process of nding such a total di erential is called total di erentiation) and is the sum of the approximate changes from both sources. S Y and S i play the role of converters that serve to convert the changes dy and di, respectively, into a corresponding change ds. Note that S Y = ds dy, (27) i constant which means that the partial derivative can also be interpreted as the ratio of two di erentials ds and dy given that i is held constant. Luo, Y. (SEF of HKU) MME January 14, / 44

24 For a general function with n independent variables, y = f (x 1, x n ), the total di erential of this function is dy = f dx f n dx n = x 1 x n f i dx i (28) i=1 in which each term on the right side indicates the amount of change in y resulting from an in nitesimal change in one of the independent variable. As in the case of one variable, the n partial derivatives can be written as ε yxi = f x i where i = 1,, n. x i f Luo, Y. (SEF of HKU) MME January 14, / 44

25 Rules of Di erentials Rule 1: for any constant c, dc = 0 Rule 2: Rule 3: Rule 4: Rule 5: d (cu α ) = cαu α 1 du d (u v) = du dv d (uv) = udv + vdu (29) u d = v vdu udv v 2 Luo, Y. (SEF of HKU) MME January 14, / 44

26 Total Derivatives Consider any function y = f (x, w) where x = g (w),unlike partial derivative, a total derivative doesn t require that the argument x remains constant as w varies. w can a ect y via two channels: (1) indirectly, via g and then f, and (2) directly, via f. Whereas the partial derivative f w is adequate for expressing the direct e ect alone, a total derivative is needed to express both e ects jointly. To get the total derivative, we rst get the total di erentials dy = f x dx + f w dw and then divide both sides by dw : dy dw = f dx x dw + f w = y dx x dw + y w. For a more general function y = f (x 1, x 2, w) with x 1 = g (w) and x 2 = h (w), the total derivative of y is dy dw = f dx 1 x 1 dw + f dx 2 x 2 dw + f w. Luo, Y. (SEF of HKU) MME January 14, / 44

27 Application to an economic growth model Let the production function is Q = Q (K, L, t) (30) where K is the capital input, L is the labor input, and t is the time which indicates that the production function can shift over time due to technological changes. Since capital and labor can also change over time, we have K = K (t) and L = L(t). Thus the rate of output with respect to time t can be written as dq dt = Q dk K dt + Q dl L dt + Q t. Luo, Y. (SEF of HKU) MME January 14, / 44

28 Derivatives of Implicit Functions A function y = f (x 1, x n ) is called an explicit function because the variable y is explicitly expressed as a function of x 1, x n. But in many cases, the relationship between y and x 1, x n is given by F (y, x 1, x n ) = 0, (31) which may also be de ned as implicit function y = f (x 1, x n ).Note that an explicit function can always be transformed into an equation F (y, x 1, x n ) = y f (x 1, x n ) = 0. The reverse transformation is not always possible. Hence, we have to impose a certain condition under which a given equation F (y, x 1, x n ) = 0 does indeed de ne an implicit function y = f (x 1, x n ). Such a result is called implicit-function theorem. Luo, Y. (SEF of HKU) MME January 14, / 44

29 Theorem (implicit-function theorem) Given F (y, x 1,, x n ) = 0, if (a) the function F has continuous partial derivatives F y, F x1,, F xn, and if (b) at a point (y 0, x 1,0,, x n,0 ) satisfying F (y 0, x 1,0,, x n,0 ) = 0 and F y is nonzero, then there exists an n-dimensional neighborhood of (x 1,0,, x n,0 ), N, in which y is an implicitly de ned function of x 1,, x n, in the form of for all points in N. Moreover, the implicit function f is continuous, and has continuous partial derivatives f 1,, f n. Derivatives of implicit functions. Di erentiating F, we have df = 0 : F y dy + F 1 dx F n dx n = 0. (32) Luo, Y. (SEF of HKU) MME January 14, / 44

30 (continued.) Suppose that only y and x i are allowed to vary, then we have F y dy + F i dx i = 0, (33) which means that dy j other variables constant = y = dx i x i For a simple case F (y, x) = 0, the rule gives F i F y for any i. (34) dy dx = F x F y. (35) Assume that the equation F (Q, K, L) = 0 implicitly de nes a production function Q = Q (K, L). We can then get the marginal product of capital and labor, MPK and MPL, as follows: MPK = Q K = F K F Q and MPL = Q L = F L F Q. Luo, Y. (SEF of HKU) MME January 14, / 44

31 Extension to the Simultaneous-equation Case Consider a set of simultaneous equations, F 1 (y 1,, y n ; x 1,, x m ) = 0 (36) F n (y 1,, y n ; x 1,, x m ) = 0. (37) Suppose that F 1,, F n are di erentiable, we have F 1 dy F 1 F 1 dy n = dx F 1 dx m y 1 y n x 1 x m F n dy F n F n dy n = dx F n dx m, y 1 y n x 1 x m or in matrix form 2 F 1 F y y n 4 F n F y 1 n y n dy 1 dy n 3 5 = F 1 x 1 F 1 x m F n x 1 F n x m dx 1 dx m Luo, Y. (SEF of HKU) MME January 14, /

32 If we want to obtain partial derivatives with respect to x i, let dx j = 0 for any j 6= i and we have the following equation: 2 3 F 1 F y y n = 4 5 dx i =) 2 F n F y 1 n y n 3 2 F 1 F y y n 4 F n F y 1 n y {z n } J dy 1 dy n y 1 x i y n x i = F 1 x i F n x i F 1 x i F n x i (38) Suppose that the Jacobian determinant is nonzero: jjj 6= 0. Then by the Cramer rule, we have y j = jji j j x i jjj where i = 1,, m; j = 1., n. (39) Luo, Y. (SEF of HKU) MME January 14, / 44

33 The National Income Model The model setup is jjj = F 1 Y F 2 Y F 3 Y F 1 = Y (C + I 0 + G 0 ) = 0 (40) F 2 = C [α + β (Y T )] = 0 (41) F 3 = T (γ + δy ) = 0. (42) F 1 F 1 C T F 2 F = C T β 1 β δ 0 1 = 1 β + βδ. F 3 C F 3 T Suppose all exog. variables and parameters are xed except G 0, then: β 1 β = =) Y = G δ β /J Y G 0 C G 0 T G 0 = 1 1 β + βδ. Luo, Y. (SEF of HKU) MME January 14, / 44

34 The Market Model (with general functions) The model setup is Q d = D (P, Y 0 ), Q s = S (P) where D D P < 0, Y 0 > 0, ds dp > 0, Q d is a function not only of P but also of Y 0, and both demand and supply functions have continuous derivatives. The equilibrium condition Q d = Q s implies that F (P, Y 0 ) = D (P, Y 0 ) S (P ) = 0. (43) We assume that there does exist a static equilibrium (for otherwise there would be no point in raising the equation of CS) and expect that P = P (Y 0 ). (44) Note that with the IF theorem, the satisfaction of the conditions of the IF theorem will guarantee that every Y 0 will yield a unique P in the neighborhood around a point satisfying (43) that de nes the initial equilibrium. In that case, we can write the IF P = P (Y 0 ). Luo, Y. (SEF of HKU) MME January 14, / 44

35 (continued.) A straightforward application of the IF theorem gives dp = F / Y 0 dy 0 F / P = D/ Y 0 D/ P > 0, (45) ds/dp in which all derivatives are evaluated at the equilibrium point. Economic (qualitative) implication: an increase (decrease) in the income level will always result in an increase (decrease) in the equilibrium price P. If we know the values of the derivatives at the equilibrium, this formula gives a quantitative conclusion. Furthermore, at the equilibrium, we have Q = S (P ) and apply the chain rule gives dq = ds dp dy 0 {z} dp > 0, (46) dy 0 >0 which also means that the equilibrium quantity is positively related to Y 0. Luo, Y. (SEF of HKU) MME January 14, / 44

36 Simultaneous-equation Approach The above analysis was carried out on the basis of a single equation. We rst derived dp dy 0 and then infer dq dy 0. Now we shall show how both can be determined simultaneously. As there are two endogenous variables, we need accordingly set up a two-equation system. First, let Q = Q d = Q s in the market model and rearranging gives F 1 (P, Q; Y 0 ) = D (P, Y 0 ) Q = 0 (47) F 2 (P, Q; Y 0 ) = S (P) Q = 0, (48) which is in the form of (36),,(37), with n = 2 and m = 1. First, we check the conditions of the implicit-function theorem: (1) since the demand and supply functions are both assumed to have continuous derivatives, so must F 1 and F 2 ; (2) the endogenous-variable Jacobian (the one involving P and Q) indeed turns out to be nonzero regardless of where is evaluated because jjj = F 1 F 1 P Q F 2 F 2 = D P 1 ds P Q dp 1 = ds D > 0. (49) dp P Luo, Y. (SEF of HKU) MME January 14, / 44

37 (continued.) Hence, if an equilibrium solution exists, the implicit-function theorem implies that P = P (Y 0 ) and Q = Q (Y 0 ) even though we cannot solve for P and Q explicitly. These implicit functions are known to have continuous derivatives. Furthermore, (47) and (48) will have the status of a pair of identities in some neighborhood of the equilibrium state, so we may also have F 1 (P, Q ; Y 0 ) = D (P, Y 0 ) Q = 0 (50) F 2 (P, Q ; Y 0 ) = S (P ) Q = 0, (51) from which, dp dy 0 and dq dy 0 can be found simultaneously by using the implicit function rule (38) (Note that here we have two endogenous variable and one exogenous variable): # " # " # dp " F 1 P F 1 Q F 2 P F 2 Q dy 0 dq dy 0 = F 1 Y 0 F 1 Y 0. (52) Luo, Y. (SEF of HKU) MME January 14, / 44

38 (continued.) More speci cally, D P 1 ds dp 1 " dp dy 0 dq dy 0 # = D Y 0 0. (53) By Cramer rule, we have dp dy 0 = D Y 0 jjj and dq dy 0 = ds D dp Y 0, jjj where all the derivatives of the demand and supply functions are to be evaluated at the initial equilibrium. Note that the results here are identical to those from solving the single equation. Instead of directly applying for the implicit function rule, we can also reach the same result by rst di erentiating totally (50) and (51) to get a linear equation system in terms of dp and dq : D P dp dq = D Y 0 dy 0 and ds dp dp dq = 0. Luo, Y. (SEF of HKU) MME January 14, / 44

39 Use of Total Derivatives In both the single equation and the simultaneous equation approaches illustrated above, we have taken the total di erentials of both sides of an equation equation and then equated the two results to arrive at the implicit function rule. However, it is possible to take and equate the total derivatives of the two sides of the equilibrium equation with respect to a particular exogenous variable or parameter. For example, in the single equation approach, D (P, Y 0 ) S (P ) = 0 where P = P (Y 0 ). Taking total derivatives of this equation w.r.t. Y 0 gives D dp P + D dy 0 Y 0 S dp P = 0 =) dp = dy 0 dy 0 S P D Y 0. (54) D P Luo, Y. (SEF of HKU) MME January 14, / 44

40 Application to the IS-LM Model Equilibrium in the IS-LM model is characterized by an income level and interest rates that simultaneously determine equilibrium in both the goods market and the money market. A goods market is described by the following set of equations: Y = C + I + G; (55) C = C (Y T ) ; I = I (r); (56) T = T (Y ); G = G 0 (57) where Y is the level of GDP (gross domestic product) or total income. C, I, G, and T are consumption, investment, government spending, and taxes, respectively. Note that (1) consumption is a strictly increasing function of disposable income Y d = Y T, that is, the marginal propensity to consume is dc dy d = C 0 Y d 2 (0, 1). Luo, Y. (SEF of HKU) MME January 14, / 44

41 (continued.) (2) Investment is a strictly decreasing function of the interest rate di dr = I 0 (r) < 0. (58) (3) The government sector is described by two variables: government spending (G) and taxes (T ). G is assumed to be set exogenously, whereas T is assumed to be an increasing function of income: the marginal tax rate dt dy = T 0 (Y ) 2 (0, 1). Slope of the IS curve. The IS equation can be written as Y C Y d I (r) G 0 = 0, (59) then take the total di erential with respect to Y and r gives dy C 0 Y d 1 T 0 (Y ) dy I 0 (r) dr = 0 =) slope of the IS curve dr dy = 1 C 0 Y d (1 T 0 (Y )) I 0 (r) < 0. where we use the fact that dy d dy = 1 T 0 (Y ). Luo, Y. (SEF of HKU) MME January 14, / 44

42 (continued.) The money market can be described by the following three equations Money demand: M d = L (Y, r) where L Y > 0, L r < (60) 0, Money supply: M s = M s 0 (M s 0 is exogenously), (61) Equilibrium condition: M s = M d, (62) which implies the LM equation (curve): L (Y, r) = M s 0. (63) Slope of the LM curve. Take the total di erential with respect to Y and r gives L Y dy + L r dr = 0 =) slope of the LM: dr dy = L Y L r. Luo, Y. (SEF of HKU) MME January 14, / 44

43 Comparative static analysis (the e ects of two exog. variables on endo. variables) The simultaneous macro equilibrium state of the goods and money markets can be characterized by the IS-LM equations: Y = C Y d + I (r) + G 0 (64) L (Y, r) = M s 0. (65) Taking the total deferential of the system gives dy C 0 Y d 1 T 0 (Y ) dy I 0 (r) dr = dg 0 which can be written in matrix form 1 C 0 Y d (1 T 0 (Y )) I 0 (r) L Y {z } J L Y dy + L r dr = dm s 0, L r dy dr dg0 = dm s 0. Luo, Y. (SEF of HKU) MME January 14, / 44

44 (continued.) The Jacobian determinant jjj is h jjj = 1 C 0 Y d 1 T 0 (Y ) i L r + I 0 (r) L Y < 0. (66) Note that since jjj 6= 0, this system satis es the conditions of the implicit function theorem and the implicit functions Y = Y (G 0, M0 s ) and r = r (G 0, M0 s ) can be written even though we are unable to solve for them explicitly. To do comparative static analysis, rst we need to set dm0 s = 0, and divide both sides by dg 0 : 1 C 0 Y d (1 T 0 (Y )) I 0 " # dy (r) dg 0 1 L Y L dr =. (67) r 0 dg 0 Using Cramer rule gives dy = dg 0 1 I 0 (r) 0 L r /jjj = L r jjj > 0 and dr dg 0 = L Y jjj > 0. Luo, Y. (SEF of HKU) MME January 14, / 44