10 Integration 320 Indeed, models basically play the same role in economics as in fashion. They provide an articulated frame on which to show off your material 10 advantage, ... a useful role, but fraught with the dangers that the designer may get carried away by his personal inclination for the model, while the customer may forget that the model is more streamlined than reality. J.-H. Drize (1984) The geometric problem of finding the steepness of a curve at @ point leads to the concept of the derivative of a function. The derivative tums out to have important interpretations apart from the geometric one. Particularly important in economics is the fact that the derivative represents the rate of change of a function. The main concept to be discussed in this chapter can also be introduced ge- ometrically. In fact, we begin with the problem of measuring the areas of certain plane regions that are bounded not only by straight lines. Solving this problem will involve the concept of the definite integral of a function over an interval. This con- cept also has a number of important interpretations in addition to the geometric one. As early as about 360 B.C., the Greek mathematician Eudoxos developed a general method for determining the areas of plane regions, known as the method of exhaustion. The idea was to inscribe and circumscribe the region (say, a circular disk) by simpler geometric regions such as rectangles, triangles, or general polys- onal regions—whose area we already know how to measure. Now, if the area of the inscribed region and the area of the circumscribed region tend to the same limit‘Sec. 10.1 / Areas under Curves 321 as more and more refined polygons are chosen, this limit is defined as the area of the region. ‘The method of exhaustion was used by Eudoxos and Archimedes to determine the areas of a number of specific plane regions. Similar methods were developed to determine the lengths of curves and the volumes of solids. However, the method of exhaustion tumed out to work only in a limited number of cases, partly because of the algebraic problems encountered. Nearly 1900 years passed after Archimedes before anyone else made significant progress in measuring areas of plane regions. In the seventeenth century, a new method of finding areas was devised, called integration, that is closely related to differential calculus. Demonstrating the precise relationship between differentiation and integration is one of the main achievements of mathematical analysis. It has even been argued that this discovery is the single ‘most important in aif of science. Barrow, who was Newton’s teacher, and Newton and Leibniz in particular, are the mathematicians associated with this discovery. ‘After these introductory comments, we begin by solving the geometric prob- lem of finding the areas of certain specific plane regions. We then develop the theory of integration based on this foundation. 10.1 Areas under Curves The problem to be considered in this section is illustrated in Fig. 10.1. It can be formulated as follows: How do we compute the area A under the graph of f from @ 10 b, assuming that f(x) is positive and continuous? To answer this question, we first introduce the function A(x) that measures the area under the curve y = f(x) over the interval (a, x], as shown in Fig. 10.2. Clearly, A(a) = 0, because there is no area from a to a, and the area in Fig. 10.1 is A = A(b). Itis obvious from Fig. 10.2 that because f is always positive, A(x) increases as x increases. Suppose we increase x by a positive amount Ax. Then A(x + Ax) FIGURE 10.1 FIGURE 10.2 y y vasa yee i \_F | Aw322 Chapter 10 / integration y y= fe) x xtAx x xtAr FIGURE 10.3 FIGURE 10.4 is the area under the curve y = f(x) over the interval [a,x + Ax]. Hence, A(x + Ax) — A(x) is the area AA under the curve over the interval [x,x + Ax], as shown in Fig. 10.3. In Fig. 10.4, area AA is magnified. It cannot be larger than the area of the rectangle with edges Ax and f(x+ Ax), nor smaller than the area of the rectangle with edges Ax and f(x). Hence, for all Ax > 0, f(x) Ax 5 A(x + Ax) — AQ) < f(x + Ax) Ax [+] But then f0) sAGFAD TAM < Fry ay tes] (If Ax < 0, the inequalities in [+] are reversed, whereas the inequalities in [+*] are preserved. The following argument is equally valid when Ax < 0.) Let us consider what happens to [x] as Ax — 0. The interval [x, x-+ Ax] shrinks to the single point x, and by continuity of f, the value f(x-+Ax) approaches f(x). The Newton quotient (A(x + Ax) — A(x)]/Ax, squeezed between f(x) and a quantity that approaches f(x), must therefore approach f(x) as Ax + 0.' So we arrive at the remarkable conclusion that the function A(x). which measures the area under the graph of f over the interval (a, x), is differentiable, with derivative given by A(a)= f(x) (forall x €(a,b)) This proves that the derivative of the area function A(x) is the curve’s “height” function f (x). 'The function f in the figures is increasing in the interval [x. x + Ax]. It is easy to see that the ‘same conclusion is obtained whatever the behavior of f in the interval [x. x + Ax). On the left-hand side of [=]. just replace f(x) by f(C). where c is the minimum point of the continuous function f in the imterva: and on the right-hand side, replace f(x + Ax) by f(d), where dis the maximum point of f in [x.x+ Az).| | Sec. 10.1 / Areas under Curves 323 Suppose that F (x) is another continuous function with f (x) as its derivative, so that F'(x) = A'(x) = f(x) for all x € (a,b). Because (d/dx)[A(x) — F(x)] = Al(x) ~ F'(x) = 0, it must be true that A(x) = F(x) +C for some constant C (see Theorem 7.7 of Section 7.3). Recall that A(a) = 0. Hence, 0 = A(a) = F(a)+C, so C = —F(a). Therefore, A(x) = F(x)— F(a) when F'(x) = f(x) 10.1) This leads to the following. Method for finding the area below the curve y = f(c) and above the z-axis from 2 =a tox = b: 2 1. Find an arbitrary function F that is continuous on [a,b] such that 0.2) F'(x) = f(x) for all x € (a, b). 2. The required area is then F(b) — F(a) A function F with the property that F’(x) = f(x) for all x in some open interval, is often called an antiderivative of f. Note that there are always many such antiderivatives because (d/dx)(F (x) + C] = F’(x) = f(x) whenever C is any real constant. Example 10.1 Calculate the area under the parabola f(x) =x? over the interval [0, 1]. Solution The area in question is the shaded region A in Fig. 10.5. Ac- cording to step 1 of [10.2], we must find a function having x? as its deriva- tive. We look for a power function. Indeed (d/dx)ax" = anx"~! = x? when n= 3 and a = 1/3. So we put F(x) = 4x9 and then F’(x) = x*. Thus, the FIGURE 10.5 >324 Chapter 10 / Integration required area is A=F(1)- FQ) = Figure 10.5 suggests that this answer is reasonable, because the shaded region appears to have roughly 1/3 the area of a square whose side is of length 1 Note: If you wied seriously to use the method of exhaustion for determining the area in Fig. 10.5, you would appreciate the extreme simplicity of the method based on (10.2). Example 10.2 Find the area A under the straight line f(x) = cx-+d over the interval (a, 6] (We assume that the constants c and d are chosen so that f(x) > 0 in (a, b].) Solution The area is shown in Fig. 10.6. If we put F(x) = }ex? +dx, then F’(x) = cx +d, and so A= F(b) - F(a) = (3cb? + db) — 4e(b? — a?) + d(b- a) ca? + da) Compute the same area in another way and check that you get the same answer. The argument leading to [10.2] was based on rather intuitive considerations. However, the concept of area that emerges agrees with the usual concept for regions. bounded by straight lines. Example 10.2 is a case in point. Formally, we choose to define the area under the graph of a continuous and nonnegative function f over the interval (a, b] as the number F(b) ~ F(a). where F’(x) = f(x). Suppose G(x) is any other function with G'(x) = f(x) for x € (a,b). Then G(x) = F(x) + C, for some constant C. Hence, G(b) — G(a) = F(b) + C — [F(a) + C] = F(b) — F(a) FIGURE 10.6 aSec. 10.1 / Areas under Cuves 325 This argument tells us that the area we compute using [10.2] is independent of which antiderivative of f we choose. Moreover, according to Theorem 10.1 of, Section 10.3, any continuous function f in (a,b) has an antiderivative. What Happens if f(x) Has Negative Values in [a,b]? ‘We assumed before that f was continuous and positive-valued. Let us consider the case in which f is a function defined and continuous in (a,b), with f(x) <0 for all x € (a,b). The graph of f, the x-axis, and the lines x = a and x = b still enclose an area. If F'(x) = f(x), we define the area to be —(F(b) — F(a)]. ‘We choose this definition because we want the area of a region always to be positive. Example 103 Compute the area shaded in Fig. 10.7. It is the area between the x-axis and the graph of f(x) = e*/ —3, over the interval [0, 31n3]. Solution We need to find a function F(x) whose derivative is e* — 3. Trial and error leads to the suggestion F(x) = 3e* ~ 3x. (Check that F’(x) = e*/3 — 3.) The area is therefore equal to -[F(3In3) — F@)] = —Ge""? —3-31n3 — 3e°) = -(@—91n3 — 3) =9in3 — 6 © 3.89 Is the answer reasonable? (Yes, because the shaded set in Fig. 10.7 seems to be a little less than 4 units in area.) Suppose f is defined and continuous in (a, 6], positive in some subinter- vals, and negative in others, as is the case in Fig. 10.8 The total area bounded FIGURE 10.7 FIGURE 10.8 y y y{ 326 Chapter 10 / integration by the graph of f, the x-axis, and the lines x = @ and x = b is then calcu- lated by computing the positive areas in each subinterval [a, ci}, [c1. cal, [€2, ¢3)s and [c3, 6] in turn according to the previous definitions, and then adding these areas. Problems | 1. Compute the area under the graph of f(x) = x° over (0. 1] by using (10.2). 2. For each of the following cases, draw a rough graph of f and indicate (by shading) the area of the set bounded by the x-axis, the lines x = o and x = b, and the graph of f. Also calculate the area in question. a f(x) =3x? in [0,2] b. f(x) =x® in (0.1) ce f(x) =e in [-1.1] 4. f(x) =1/x? in [1, 10] 3. Compute the area A bounded by the graph of f(x) = 1/x3, the x-axis, and the lines x = —2 and x = —1. (Make a drawing.) 4. Compute the area A bounded by the graph of f(x) = }(e*+e™*), the x-axis, and the lines x = —1 and x = 1. 10.2 Indefinite Integrals The problem of computing areas under the graph of a function f leads to the problem of finding an antiderivarive of f—that is, a function F whose derivative is f. Although the name antiderivative is very appropriate, we shall follow the usual practice and call F an indefinite integral of f. As a symbol for an indef- inite imegral of f, we use [ f(x)dx. Two functions having the same derivative throughout an interval must differ by a constam, so we write [teas=Fe+c when F(x) = fix) 103} For instance, [Baattec because (x where (_ )’ denotes differentiation. The symbol f is the integral sign, the func- tion f(x) appearing in [10.3] is the integrand, and C is the constant of inte- gration. The dx part of the integral notation indicates that x is the variable of integration.Sec. 10.2 / indefinite integrals 327 Let a be a fixed number # —1. Because the derivative of x**!/(a + 1) is x” (a#-1) (10.4) This very important integration result states that the indefinite integral of any power of x (except x7!) is obtained by increasing the exponent of x by 1. dividing by the new exponent. and then adding the constant of integration. Here are some examples: A+ When a = —1, the formula in [10.4] is not valid, because the right-hand side involves division by zero and so becomes meaningless. The integrand is then 1/x, and che problem is thus to find a function having 1/x as its derivative. Now lax has this property. but it is only defined for x > 0. Note, however. that In(—x) is defined for x < 0, and according to the chain rule, its derivative is (1/(-x)] (=I) = I/x. Recall that |x| = x when x > 0 and |x| = —x when x <0. Thus, whether we integrate over an interval where x > 0 or x < 0, we have | dr = nisl +C 105) a Consider next the exponential function. The derivative of e* is e*. Thus, Jetdx =e +C. More generally. e+C (a #0) (10.6) / dx because the derivative of (1/a)e% is e**328 © Chapier 10 / Integration For a > Q we can write a* = e*!"*, As an application of [10.6], for Ina #0 (that is, for a # 1), we have [oa=Lexc (a> 0anda #1) (10.7) ina Some General Rules Two rules of differentiation are (a F(x))’ = aF'(x) and (F(x) + G(x) G'(x). They immediately imply the following integration rules: Fixy+ Constant Multiple Property [etooas aa f fayax (a is a real constant) [10.8] The integral of a sum is the sum of the integrals [ t9+200] dx [teoaxe [evar 10.9} Repeated use of these two properties yields the general rule [ison +a fa(x)] dx = » [fonds +f fords [10.10} for the indefinite integral of any linear combination of continuous functions. Example 10.4 Find the integral f (3x4 + 5x? — 2) dx.Sec. 102 / Indefinite inegrais 329 Solution [ox +se-nas [xaces [Pax-2 [ras =3(d84+C1) +5 (fx? +C2) —20 +05) x + 8x3 — 2x +3C, +5C1 - 2C3 $4 $03 2x4 Because C), C2, and C3 are arbitrary constants, 3C; + 5C2 — 2C3 is also an arbitrary constant. So in the last line we have replaced it by C for sim- plicity. It is not necessary to write all the intermediate steps when integrating in this way. More simply, we write [ow + 5x? 2)dx Je oesfe dx — vie a+ ldx By systematically using the proper rules, we can differentiate very complicated functions. On the other hand, finding the indefinite integral of even quite simple functions can be very difficult, or impossible. Note, however, that it is usually quite easy to check whether a proposed indefinite integral is correct. We simply differentiate the proposed function to see if its derivative is equal to the integrand. Example 10.5 Verify that (in an interval where ax +b > 0) x 2 [ae = yplax — 2) Var tb+C Solution We put F(x) = (2/3a*)(ax — 2b)Vax +b = (2/3a7)u - v, where u = ax ~ 2b and v = Jax +b. Now Fix) an i aque + uv) where, after introducing the new variable w = ax +, one has330. Chapter 10 / Integration Hence, 2 a F(x) = x5 |aVax +6 + (ax — 26) =m [a ax (ax ~ 26) =) 2 2a?x +2ab +a?x —2ab L 2Vax+b )* 3a? 2Vax +b = 2, (Awe) ~~ 3a? which shows that the integral formula is correct. Initial-Value Problems ‘As was seen before, there are infinitely many “antiderivatives,” or indefinite integral functions, having @ given function as their common derivative. For instance, the derivative of $x> + C is x* for all choices of the constant C. The graphs of these functions are all translates of each other in the direction of the y-axis. Given any point (xo, yo), there is one and only one of these curves that passes through (xo. Yo) Example 10.6 Find all functions F(x) such that F(x) =-(@- 1)? a) and draw some of the graphs in the xy-plane. whose graph passes through the point (xp, Yo) Solution Equation [1] implies that Find in particular the function aD. FQ)= [-« Pia te io All values of C are possible. Some of the associated graphs are drawn in Fig. 10.9. The curve that passes through (I. 1) is found by solving the equation F(I) = 1, or -j0-34+C=1 This gives C = {. so the required function is F(x) -3@-03 The last part of Example 10.6 can be formulated this way: Find the unique function F(x) such that F'(x) = —(x— 1)? and F({) = 1. This is called an initial- value problem and the requirement that F(1) = | is called an initial condition.Sec. 10.2 / indefinite Integrals 331 FIGURE 10.9 F(x) = $x — 1) +C. Example 10.7 ‘The marginal cost of producing x units of some commodity is 1 + x + 3x? and fixed costs are 150. Find the total cost function. Solution Denoting the total cost function by c(x), we have U1) cz) =14x43x? and (2) (0) = 150 because c(0) is the cost incurred even if nothing is produced. Integrating [1] yields ex) = xt hte 4 (3) Substituting x = 0 in [3] gives c(0) = C, and so C = 150 because of (2]. Hence, the required total cost function is e(x) =x + $x? +23 +150 So far, we have always used x as the variable of integration. In economics, the variables often have other labels. Example 108 Find the following: B @ fasar b) fla t+bg +97) dq332 Chapter 10 / Integration Solution (a) Writing B/r?> as Br~25, we see that formula [10.4] applies, and so Bore ase [wees] dr = Bz (b) f(a +bq +0q?) dq = aq + $bq? + }cq> + Problems 1. Find the following integrals using [10.4]: a. fo ax b. [ae « [aa a [fale 2. Find the following integrals: a. [e+u-nae b. fo-vtax « [o-na+aax 3x+4 a [ora & fer-e + et )dx [oes sae 3. Find the following integrals: 2 2 ais af Fe v. [Apa e faa xySdx (Hint: In part (a), first expand (y — 2) and then divide each term by /. In part (b), do long division, In part (c), what is the derivative of (1 +.x°)!62) 4. a, Show that +b)? dx = ———_(ax +b)" = fo + b)Pdx = partyr+c @#0. p#- b. Find the following: @ fox+vtds Gi) [artes Git) i 5. Show that [wa Fbdx = ee = 2bYax + bP +C 6. Solve the following initial-value problems: a, Find F(x) if F’(x) =} —2x and F(0) = 1/2. b. Find F(x) if F’(x) = x(1 —x°) and F(1) = 5/12. 7. In the manufacture of a product, the marginal cost of producing x units is c'(x) = 3x +4. If fixed costs are 40, find the total cost function c().See. 10.3 / The Detinte integra 333 8. Find the general form of a function f whose second derivative is x°. If we require in addition that f(0) = 1 and f/(0) = —1, what is f(x)? 9. a. Suppose that f(x) = 2 for all x, and (0) = 2, f/(0) = 1. First find f(x) and then f (x) b. Similarly, suppose that f"(x) = I/x?+x° +2 for x > 0, and (1) = f'() = 1/4. Find f(x). 10.3 The Definite Integral Let f be a continuous function defined in the interval |a, bJ. Suppose that the function F is continuous in (a, 6] and has a derivative satisfying F(x) = f(x) for every x € (a,b). Then the difference F(b) — F(a) is called the definite integral of f over (a,b). As observed in Section 10.1, this difference does not depend on which of the infinitely many indefinite integrals of f we choose as F. The definite integral of f over (a, b] is therefore a number that depends only on the function f and the numbers a and b. We denote it by 5 { f(x)dx (10.11) This notation makes explicit the function f(x) we integrate, which is called the integrand. and the interval of integration (a, b]. The numbers a and b are called, respectively, the lower and upper limits of integration. The letter x is a dummy variable in the sense that the integral is independent of its label. For instance, [ serar= [soar [seas In many other mathematical writings, the difference F(b) — F(a) is often denoted > by F(x) |, or by [F(x)]f. But we shall use. Thus: o F(z) is also common, ané this is the notation Definition of the Definite Integral ° . [ F@)dx=) F(x) = F)- Fla) (10.12) where F’(x) = f(x) for all x € (a,b) Definition [10.12] does not necessarily require a < b. However, if a > b and f(x) is positive throughout the interval (b. a]. then f° f(x) dx is a negative number.334 © Chapter 10 / Integration Note that we have defined the definite integral without necessarily giving ita geometric interpretation. In fact, depending on the context. it can have dif- ferent interpretations. For instance. if f(r) is an income density function, then Jf f(”) dr is the proportion of people with income between @ and b, (See the next section.) With the new notation, the results in Examples 10.1 and 10.2 can be writ- ten as 1 1 \ [ drs f lo » (fex? +ax) f (cx+d)dx= yeh? + db) — ($ca? + da) }e(? — 2) + dba) Although the notations for definite and for indefinite integrals are similar, they are entirely different concepts. In fact, J f(x) dx denotes a single number, whereas J f(0)ax represents any one of the infinite set of functions all having f(x) as their derivative. The relationship between the two is that f f(x)dx = F(x) + CC over an interval J, if and only if [? f(x)dx = F(b) — F(a) for all a and bind. Properties of the Definite Integral From the definition of the definite integral in [10.12], a number of properties can be derived. If f is a continuous function in an interval that contains a, b, and c. then [10.13] (10.14) (@ is an arbitrary number) [10.15] < » f soeee f fosds f S(xdx (10.16) Alll these rules follow easily from [10.12]. For example, [10.16] can be proved as follows. Let F be continuous in [a. 6], and suppose that F'(x) = f(x) for all xSec. 10.3 / The Definite integral 335 FIGURE 10.10 f? f(x) dx = f% fox) dx + [2 f(x) dx. in the interior of an interval long enough to include a, b, and c. Then i fadax+ f f)dx = [Flo) - Fa] + [FO - Fo] =F) -Fla)= f flayax When the definite integral is interpreted as an area, [10.16] is the additivity prop- erty of areas, as illustrated in Fig. 10.10. Of course, [10.16] easily generalizes to the case in which we partition the interval [a,b] into an arbitrary finite number of subintervals. The constant multiple property (10.8] and the summation property (10.9] are also valid for definite integrals. In fact, if f and g are continuous in (a, 5), and if, @ and 8 are real numbers, then ’ ° [ [a F(x) + Bg(x)] dx = fQdx+ af g(x)dx (10.17) ‘The proof is simple. Let F'(x) = f(x) and G'(x) = g(x) for all x € (a,b). Then (@F (x) + BG@)Y = @F (x) + BG'(x) = af (x) + g(x). Hence, [are +pe00] dx =| [aF(x) + 6G(x)) = [aF(b) + BG(b), — (a F(a) + BG (a) = a[F(b) ~ F(a)] + B[G(b) - G(a)] ’ if fovax~o[ B(x) dx The rule in (10.17) can obviously be extended to more than two functions.336 © Chapter 10 / integration Some Important Observations It follows directly from the definition of the indefinite integral that the derivative of the integral is equal to the integrand: & | fea = f(x) (10.18) Also / F(x)dx = F(x) +C (10.19) Moreover, [ f@)dx=| FQ) =F) - Fla) So, differentiating w.r.. 1 with a fixed, it follows that ar al fQxddx =F) = fi) [10.20] In other words: The derivative of the definite integral w.r.t. the upper limit of inte- gration is equal to the integrand as a function evaluated at that limit. Correspondingly, [toa F(x) = F(a) — F(t) so that a a +f fQ@dx =-F'() =-f() (10.21) In other words: The derivative of the definite integral w.r.t. the lower limit of inte- gration is equal to minus the integrand as a function evaluated at that limit. The results in (10.20] and (10.21) can be generalized. In fact, if a(r) and (0) are differentiable and (x) is continuous, then q pro aa F(x)dz = fOM)'O - fama (10.22) ott To prove this formula, let F(x) = f(x). Then f f(x)dx = F(v) — F(w), so in particular, bi) f(x) dx = F(b(t)) — Flatt) 20)Sec. 10.3 / The Definite integral 337 Using the chain rule to differentiate the right-hand side of this equation w.r.. 1, we obtain F’(b(0))b(t) — F'(a(t))a'(t). But F'(b(0)) = f(b(t)) and F'(a(t)) = f (a(t), so [10.22] results. (Formula [10.22) is a special case of Leibniz’s formula discussed in Section 16.2.) Continuous Functions Are Integrable Suppose f(x) is a continuous function in (a, 5). Then we defined f? f(x) dx as the number F(b) — F(a), provided that F(x) is some function whose derivative is f(x). In some cases, we are able to find an explicit expression for F(x). For instance, we can evaluate fj x° dx as 1/6 because (1/6)x° has x° as its derivative. ‘On the other hand, for the integral f ede 0 (closely related to the “normal distribution” in statistics), there is no standard func- tion whose derivative is e~*”.? Still, the integrand function is continuous in (0, 2] and there should be an area under the graph from 0 to 2. In fact, one can prove that any continuous function has an antiderivative: Theorem 10.1 If f is a continuous function in {a. 6], then there exists a continuous function F (x) in (a, b] such that F(x) = f(x), forall x ¢ (a, b). A sketch of a proof: Let x € (a,b). Subdivide the interval a, x] into m equal parts so that the points of subdivision are a+(x—a) /n, a+2(x—a)/n, ...,a+(n—1)(x—a)/n. For each natural number n, define the new function F, as an approximation to F using the formula x xa Fax) ===" [pa + sf (a+ + “+ F(ata-0 (ry to illustrate this definition of F,(x).) Define F (2) = lim, ao Fa (x). Itis possible (but not easy) to show that this limit exists for each x < [a,b]. that F is continuous in (a. 5), and finally that F(x) has f(x) as its derivative in (a. 6). The Riemann Integral The kind of integral discussed so far, which is based on the antiderivative, is called the Newton-Leibniz (N-L) integral. Several other kinds of integral are considered by math- ematicians. For continuous functions, they all give the same result as the N-L integral. 2ee (11.4) in Section 11.2 for other examples of “unsolvable integrals.”338 Chapter 10 / Integration We briefly skerch the so-called Riemann integral. The idea behind the definition is closely related to the exhaustion method that was described in the introduction to this chapter. Let f be a bounded function in the interval 2, 6]. and let n be a natural number. Subdivide (a, b] into n parts by choosing points a = x9 < x) < x2 < +++ 0) is 3000000 x F(&) = 4000 — x - a. Find the output that maximizes profit. Draw the graph of f. b. The actual output varies between 1000 and 3000 units. Compute the average profit 1 7000 io 5305 Jen f(x)dxSec, 10.3 / The Definite integral 339 4, Evaluate the integrals: 3 3x ps ipa / 10 * 1 e. [act ar #0) 5 3. By using [10.22] or otherwise, evaluate the following: afi, wk fsa 10 a. $f Inxdx e +f xSdx ada & Jin 6. Compute f22x2(2 - x)?dx. Give a rough check of the answer by drawing the graph of f(x) = 2x°(2 ~ x)? over [0.2]. 7. Find the area between the two parabolas defined by the equations y +1 = (x — 1)? and 3x = y*. (The points of intersection have integer coordi- nates.) 8. Compute the following: 1 er xt+b ad a [rer max w [4 +2) as lo | OS xee x f ztyvxFl e f ta A x+1 Harder Problems 9. A theory of investment has used a function W defined for all T > 0 by Faye wh=> i 8 de (K and @ are positive constants) Evaluate the integral, and prove that W(T) takes values in the interval (0, X’) and is stricly decreasing. 10. a. Show that if f is continuous in (a, b], then there exists a number x” [@. b} such that f(r") = x! F(edx ‘This is called the mean-value theorem for integrals, and f(x*) is called the mean value of f in (a,b). (Hint: Put F(x) = f% f(t)dt, and use Theorem 7.5 of Section 7.3.) b. Find the mean value of f(x) = / in [0,4], and illustrate.340 chapter 10 / integration 10.4 Economic Applications of Integration ‘We motivated the definite integral as a tool for computing the area under a curve. However, the integral has many other important interpretations. For instance, we are led to a definite integral when we want to find the volume of a solid of revolution or the length of a curve. Several of the most important concepts in statistics are also expressed by integrals of continuous probability distributions. This section presents some examples showing more directly the importance of integrals in economics. Extraction from an Oil Well Assume that at time 1 = 0 we start extracting oil from a well that contains K barrels of oil. Let us define x(t) = amount of oil in barrels that is left at time 1 In particular, x(0) = K. If we assume that we cannot pump oil back into the well, then x(t) is a decreasing function of t. The amount of oil that is extracted in 2 time interval (, 1+ Ar] (where Ar > 0) is x(r) — x(t + At). Extraction per unit of time is, therefore, x(t) (Han x@tAn-x0 ar ar te] If we assume that x(t) is differentiable, then the limit as Ar approaches zero of the fraction [+] is equal to —i(r). Letting u(r) denote the rate of extraction at time 1, we have X)=-uQ) with = xQ=K (10.23) The solution to the initial-value problem [10.23] is x)= K i u(c)dt [10.24] 0 Indeed, we check [10.24] as follows. First. setting t = 0 gives x(0) = K. More- over, differentiating (10.24] w.r.t.r according to rule (10.20] in Section 10.3 yields H(t) = —u(t). The result [10.24] may be interpreted as follows: The amount of oil left at time 1 is equal to the initial amount K, minus the total amount that has been extracted during the time span [0,1], namely fj u(t) dt If the rate of extraction is constant, with u(r) = i, then (10.24) yields snk [ idt=K-| areK-it 0 0 In particular, we see that the well will be empty when K — ir = 0, or when‘Sec. 10.4 / Economic Applications of integration 341 1 = Ki. (Of course, this particular answer could have been found more directly, without recourse to integration.) The example illustrates two concepts that are important to distinguish in many economic arguments. The quantity x(t) is a stock, measured in barrels. On the other hand. u(r) is a flow. measured in barrels per unit of time. A Country's Foreign Exchange Reserves Let F(t) denote a country’s foreign exchange reserves at time 1. Assuming that F is differentiable, the rate of change in the foreign exchange reserves per unit of time will be f= F(a) (10.25) If f(2) > 0, this means that there is a net flow of foreign exchange into the country at time 1, whereas f(t) <0 means that foreign exchange is flowing out. From the definition of the definite integral, it follows that F (th) ~ F(to) =| fat (10.26) hs ‘We see that this expression measures the change in the foreign exchange reserves over the time interval [f, )]. An example is illustrated in Fig. 10.11. Here there is anet flow of foreign exchange into the country from f to 1, then a net flow out of the country from 1’ to 1”, and, finally, there is a net flow into the country from 1” to ty. (Note that ff" f (1) dr does not denote the total area bounded by the graph, the x-axis, and the lines 1 = fp and t = 1 in this case. See the end of Section 10.1.) Income Distribution In many countries, anonymous data from income tax authorities can be used to reveal some properties of the income distribution within a given year, as well as how the distribution changes from year to year. FIGURE 10.11 The rate of change of foreign exchange.342 Chapter 10 / Integration ‘We measure income in dollars and let F(r) denote the proportion of indi viduals who receive no more than r dollars. Thus, if there are n individuals in the population, n#(r) is the number of individuals with income no greater than 1. If ro is the lowest and r; is the highest (registered) income in the group, we are interested in the function F in the interval [ro, ri]. By definition, F is not continuous and therefore also not differentiable in [r9. r1] because r has to be a multiple of $0.01 and F(r) has to be a multiple of 1/n. However, if the popula- tion consists of a large number of individuals, then it is usually possible to find a “smooth” function that gives a good approximation to the true income distribution. Assume, therefore, that F is a function with a continuous derivative denoted by f, so that F()=F(r) (forall r € (r.71)) According to the definition of the derivative, we have $(r) Sr & F(r + Ar) - Fir) for all small Ar. Thus, f(r) Ar is approximately equal to the proportion of individuals who earn between r and r+ Ar. The function f is called an in- come density function, and F is the associated cumulative distribution func- tion Suppose that f is a continuous income distribution for a certain population with incomes in the interval {ro. ri]. If ro < @ 0. In this income interval, there are n{F (r+ Ar) — F(r)] individuals each of whom ears at most. r+ Ar and at least r. Thus, nr [Fr + Ar) — FO) S M+ ON — MO) en(rt an[F + Or)- FO) 0 If Ar > 0. division by Ar yields For + Sr) — Fir). M(r + Br) — Mir) F(r + Sr) ~ F() ni i a: a Snr + Sr): in fea) (If Ar < 0. then the inequalities in [1] are left unchanged. whereas those in [2] are reversed.) Letting Ar — 0 gives nrF'(r) < M'(r) 1, because f° f(r) dr > 00 as r + 0 (See Section 11.3).344 © Chapter 10 / Integration Example 10.9 Ina population with incomes between @ and b, suppose the income distribu- tion is given by f(r) =Br?5 (Ba positive constant) O) Determine the mean income in this group. Solution Here [sorare [art ar Also So the mean income of the group is 2B(a~°5 — 5-05) (2/3) B(a~ (2) Suppose that b is very large. Then b~°> and b!5 are both close to 0. and so [2] implies that m * 3a. The mean income of those who earn at least a is therefore approximately 3a. The Intiuence of Income Distribution on Demand Assume that the individuals in a population are offered a commodity for which demand depends only on the price p and the income r of each individual. Let D(p.r) be a continuous function that denotes the number of commodity units demanded by an individual with income r when the price per unit is p. If the incomes in the group vary between a and , and the income distribution is f(r), what is the total demand for the commodity when the price is p? Let the price p be fixed, and denote by T(r) the total demand for the com- modity by all individuals who eam less than or equal to r. Consider the in- come interval {r,r + Ar]. There ae approximately nf(r) Ar individuals with incomes in this interval. Because each of them demands approximately D(p.r) units of the commodity, the total demand of these individuals will be approxi- mately nD(p,r) f(r) Ar. However, the actual total demand of individuals with incomes in the interval [r,r + Ar] is given by T(r + Ar) — T(r). So we must‘Sec. 10.4 / Economic Applications of Integration 345 have T(r + Ar) = T(r) © nD(p.r) f(r) Ar, and thus T(r+4r)-T(r) ee te nD(p.r) f(r) The approximation improves (in general) as Ar decreases, and by taking the limit as Ar —> 0, we obtain T'(r) = nD(p. r) f(r). By definition of the definite integral. T(b) — T(a) =n J? D(p.r) f(r) dr. But T(b) — T(a) is che desired measure of total demand for the commodity by all the individuals in the group. This will naturally depend on the price p. So we denote it by x(p), and thus have » x= [ nD(p.r)f(r)dr (total demand) (10.31) Example 10.10 Let the income distribution function be that of Example 10.9. and let D(p.r) = Ap7'5r208_ (This function describes the demand for milk in Norway during the period 1925-1935. See Exampie 15.2.) Compute the total demand. Solution Using [10.31] gives . x(p) = f nape Hence, * 8B r*5 dy snappy f Caer » Loss _ nAB a a15 (058 _ 4058 058 oer FD x(p) = nABp7'> Present Discounted Value of a Continuous Future Income Stream Section 6.6 discussed the present value of a series of future payments made at specific discrete moments in time. It is often more natural to consider revenue as accruing continuously, such as the proceeds from a large growing forest. ‘Suppose that income is to be received continuously from time t = 0 to time 1 = T at the rate of f(r) dollars per year at time 1. We assume that interest is compounded continuously at rate r. Let P(r) denote the present discounted value of ail payments made over the time interval (0. r]. This means that P(r) represents the amount of money you would have to deposit at time t = 0 in order to match what results from (continuously) depositing the income stream f(t) over the time interval (0. 7]. If dr is any number, the present value of the income received in the interval [t,1-+dr] is P(t +dz)— P(t). If dt is a small number, the income received in this interval is approximately f(r)dr, and the present discounted value (PDV) of this amount is approximately f(t)e~”' dt. Thus, P(t+dt)— P(r) © f(n)e~" dr and so Pirtd)- Pe), a © fe346 chapter 10 / Integration This approximation gets better the smaller is dr, and in the limit as dr — 0, we have P= fine" By the definition of the definite integral, P(T) — P(0) = [7 f(t)e~"' dt. Because P(0) = 0, we have the following: The present discounted value (at time 0) of a continuous income stream at the rate of f(t) dollars per year over the time interval 0, 7], with continu- ously compounded interest at rate r, is given by 1 PDV if Fe dt (10.32) Equation [10.32] gives the value at time 0 of income stream f(t) received during time interval (0, T]. The value of this amount at time T,, with continuously compounded interest at rate r, is e’7 [J f(t)e~"' dt. Because the number e’7 is a constant, we can rewrite the integral as fy) f(t)e"7—" dr. This is called the future discounted value (FDV) of the income stream: The future discounted value (at time T) of a continuous income stream at the rate of f(t) dollars per year over the time interval (0, 7], with continuously compounded interest at rate r, is given by 7 rv= [ fe dt 0 fact, the DV at time s of income, f(r) received in the small time interval {1 is f()e~“'- de. So we have the following: The discounted value at time s of a continuous income stream at the rate of f(t) dollars per year over the time interval [s. 7]. with continuously compounded interest at rate r, is given by 7 Dv=/ fae ar (10.33) ‘An easy modification of [10.32] will give us the discounted value (DV) at time s € [0, T] of an income stream f(t) received during time interval [s, T). In .t+dt) (10.34)r Sec. 10.4 / Economic Applications of Integration 347 Example 10.11 Find the PDV and the FDV of a constant income stream of $1000 per year over the next 10 years, assuming an interest rate of r = 8% = 0.08 annually, compounded continuously. Solution 10 0 0.081 1000 pov = {1000 dr =| 10 S) = — 8) & 6883: f 1 =|, 1000 (—Saa-) = gag tl — e708) = 6883.39 FDV = 2 ppv ~ 68 . 6883.39 = 15,319.27 Problems 1, Assume that the rate of extraction u(r) from an oil well decreases exponen- tially over time, with u(t) = de®", where a is a positive constant. Given the initial stock x(0) = xo, find an expression x(t) for the remaining amount of oil at time 1. Under what condition will the well never be exhausted? 2. a. Follow the patter in Example 10.9 and find the mean income m over the interval (b,2b] when f(r) = Br-?. ‘b. Assume that the individual’s demand function is D(p.r) = Ap’r®, A > 0, y < 0.6 > 0,5 # 1. Compute the total demand x(p) by using formula [10.31], assuming that there are n individuals in the population. 3. Let K(t) denote the capital stock of an economy at time r. Then net in- vestment at time t, denoted by /(r), is given by the rate of increase K (1) of K(t). a. If 1(¢) = 32? +2 +5 (¢ > 0), what is the total increase in the capital stock during the interval from t = 0 to ¢ = 5? b. If K(to) = Ko, find an expression for the total increase in the capital stock from time s = 1p to t = T when the investment function I(t) is as in part (@). 4, Find the present and future values of a constant income stream of $500 per year over the next 15 years, assuming an interest rate of r = 6% = 0.06 annually, compounded continuously. 5. a, Find the present discounted value (PDV) of a constant income stream of a dollars per year over the next T years, assuming an interest rate of r annually, compounded continuously. b. What is the limit of the PDV as T > 00? Compare this result with (6.22) in Section 6.6.