Why Tangency Identifies an Input Choice
A firm choosing inputs in the long run can often produce a target quantity using many combinations of labor and capital. The isoquant for that target displays those combinations; an isocost line displays combinations with the same total expenditure. The cost-minimizing choice is the least-cost combination that can produce the target. When this choice is a smooth interior point, the isoquant and isocost line are tangent there.
Tangency is a geometric condition, not simply a visual impression that two curves meet. With labor \(L\) on the horizontal axis and capital \(K\) on the vertical axis, both curves typically slope downward: replacing some capital with labor while holding output fixed, or holding expenditure fixed, requires the other input to change in the opposite direction. At an interior tangency, the two slopes are equal, including their negative signs.
Reading the Slopes
Let \(w>0\) be the price per unit of labor and \(r>0\) the price per unit of capital. The expenditure on a bundle is \(C=wL+rK\). Holding expenditure at \(C\), solve for capital as a function of labor. This gives the slope of the isocost line.
The ratio \(w/r\) is positive, but the graph’s isocost slope is negative. For a differentiable production function \(Q=F(L,K)\), with positive marginal products at the bundle, a fixed-output isoquant has slope \(-MP_L/MP_K\). Here \(MP_L\) and \(MP_K\) are the additional output associated with a small increase in labor and capital, respectively, holding the other input fixed. Thus, at a smooth interior tangency, the isoquant slope equals the isocost slope; equivalently, the positive ratio \(MP_L/MP_K\) equals \(w/r\). The geometric equality is the focus here; it is not by itself a guarantee that a bundle is a minimum.
Why Equal Slopes Are Necessary
The tangency condition follows from asking what happens to cost under a small input change that stays on the required-output isoquant. The argument assumes positive input prices, a differentiable isoquant, and a smooth interior minimum, so a small movement along the isoquant is possible in either direction. If cost were changing to first order along that isoquant, moving in one direction would lower expenditure, contradicting minimization.
Along the isoquant, \(dK=(dK/dL)_{\!Q}\,dL\). At an interior minimum, the first-order change in cost along that isoquant must be zero. For a nonzero small change in labor, this gives the slope condition:
This proves why equal slopes are necessary in the stated setting. To conclude that a tangency is a global minimum, additional shape conditions matter. For example, if the set of input bundles capable of producing at least the target output is convex, the cost-minimizing isocost can support that set at the tangency. Without suitable curvature, equal slopes may instead describe a maximum or another stationary point.
Worked Examples
Worked Example: Finding the Least-Cost Bundle From an Isoquant
A hypothetical firm produces \(Q=\sqrt{LK}\) and must produce \(12\) units. Labor costs \(w=\$4\) per unit and capital costs \(r=\$1\) per unit. Find the smooth interior tangency bundle and its expenditure, and check the result.
Method. First write the target isoquant, then equate its slope to the isocost slope. This method avoids treating the slope ratio as positive on one curve but negative on the other.
Step 1: Write the isoquant. The output requirement is \(\sqrt{LK}=12\). Squaring both sides gives \(LK=144\), or \(K=144/L\) for \(L>0\). Its slope is \(-144/L^2\).
Step 2: Apply tangency. The isocost slope is \(-w/r=-4\). Set the slopes equal:
Since \(L\) is positive, \(L=6\), not the negative square root. Substitution into the isoquant gives \(K=144/6=24\).
Step 3: Calculate and check expenditure and output. Expenditure is \(C=4(6)+1(24)=\$48\). Output is \(\sqrt{6\cdot24}=\sqrt{144}=12\), as required. As an additional minimum check along this isoquant, \(C(L)=4L+144/L\); its derivative is \(4-144/L^2\), which is zero at \(L=6\), and its second derivative \(288/L^3\) is positive for \(L>0\). Thus this stationary bundle is the minimum along the isoquant.
Worked Example: Tangency With a Curved Isoquant
A hypothetical workshop uses production technology \(Q=L\sqrt{K}\). It must produce \(10\) units. The labor price is \(w=\$25\) per unit and the capital price is \(r=\$1\) per unit. Find the smooth interior tangency bundle and expenditure, then verify both output and the local minimum condition.
Method. Express capital as a function of labor on the target isoquant and compare the resulting slope with the isocost slope. The common mistake here is to confuse the square root in the production function with a square root when solving for labor.
Step 1: Find the target isoquant. From \(L\sqrt K=10\), divide by \(L>0\) and square to obtain \(K=100/L^2\). Its slope is \(-200/L^3\). The isocost slope is \(-25/1=-25\).
Step 2: Solve the slope equation.
The positive cube root is \(2\). The required capital input is \(K=100/2^2=25\).
Step 3: Verify the bundle. Expenditure is \(25(2)+1(25)=\$75\). Output is \(2\sqrt{25}=2(5)=10\), exactly the target. Along the isoquant, \(C(L)=25L+100/L^2\), so \(C'(L)=25-200/L^3\) and \(C''(L)=600/L^4>0\) for \(L>0\). The stationary choice is therefore a minimum along this isoquant, not a maximum.
Worked Example: Diagnosing a Non-Tangent Bundle
A hypothetical producer has \(Q=\sqrt{LK}\), must produce \(8\) units, and faces \(w=\$4\) and \(r=\$1\). A proposed bundle uses \(L=2\) and \(K=32\). Determine whether it is a tangency and show how a small movement along the isoquant affects cost. Then find the tangency bundle and compare expenditures.
Method. Check feasibility first, compare the two slopes at the proposed bundle, and use the direction of the cost change to diagnose whether the bundle can be cost-minimizing. This avoids assuming that every feasible input combination is efficient.
Step 1: Check feasibility and slopes. The target isoquant is \(LK=64\), or \(K=64/L\). The proposed bundle is feasible because \(2(32)=64\). At \(L=2\), the isoquant slope is \(-64/2^2=-16\), while the isocost slope is \(-4\). The slopes differ, so this is not a tangency.
Step 2: Determine the cost-reducing direction. Along the isoquant, the cost as a function of labor is \(C(L)=4L+64/L\). Its derivative at \(L=2\) is \(4-64/4=-12\). Therefore, a small increase in labor while remaining on the isoquant reduces expenditure near the proposed bundle.
Step 3: Find and check the tangency. Set the isoquant slope \(-64/L^2\) equal to \(-4\). Then \(L^2=16\), so \(L=4\), and \(K=64/4=16\). Its expenditure is \(4(4)+16=\$32\), compared with \(4(2)+32=\$40\) at the proposed bundle. The tangency bundle produces \(\sqrt{4\cdot16}=8\) units. Also \(C'(L)=4-64/L^2\) changes from negative to positive at \(L=4\), confirming the minimum along this isoquant.
What Tangency Does—and Does Not—Tell You
At a smooth interior cost minimum, equal slopes express a balance between the firm’s technological trade-off and the market’s price trade-off. The isoquant slope describes how much capital must change to offset a small change in labor while output stays fixed. The isocost slope describes how much capital can be exchanged for labor without changing expenditure. If the slopes differ, the two trade-offs are not aligned, and a local movement along the isoquant can reduce cost.
The tangency condition is a local condition. In a well-behaved convex input requirement set, it can identify the least-cost bundle. But slope equality alone does not prove global minimization; checking curvature or comparing feasible costs may be needed. The condition also presumes a smooth interior choice. If the optimum is at a boundary, or the isoquant has a kink, the smooth tangency argument does not apply directly.
A crossing deserves careful interpretation. A smooth crossing with different slopes allows a local cost-reducing movement along the isoquant. However, a crossing by itself does not prove that the slopes differ: curves can cross at a higher-order intersection while having the same slope. In that case the first-order slope test is inconclusive, and the cost change beyond first order must be examined.
A reliable solution sequence is to state the target isoquant and input prices, compute both slopes using the same axis convention, solve for a positive bundle, and then verify output and expenditure. When the question asks whether the bundle is a minimum rather than merely a tangency, examine the direction of cost change or the curvature of cost along the isoquant. This geometric reasoning is aligned with the SOA VEE Economics (Validation by Educational Experience) syllabus.
Check Your Understanding
For each problem, use the stated axis convention of labor \(L\) horizontal and capital \(K\) vertical. Check feasibility and expenditure as well as the requested slope or tangency.
- A hypothetical firm has \(Q=\sqrt{LK}\), needs \(15\) units, and faces \(w=\$9\) and \(r=\$1\). Find the smooth interior tangency bundle and its expenditure. (Answer: \(L=5\), \(K=45\), expenditure \(\$90\).)
- A hypothetical producer uses \(Q=L\sqrt K\), must produce \(6\) units, and faces \(w=\$9\) and \(r=\$1\). Find the tangency bundle and verify the target output. (Answer: \(L=2\), \(K=9\); output \(2\sqrt9=6\), expenditure \(\$27\).)
- For \(Q=\sqrt{LK}\), a firm must produce \(10\) units and pays \(w=\$4\), \(r=\$1\). Find its tangency bundle, then compare its expenditure with the feasible bundle \(L=10,K=10\). (Answer: Tangency at \(L=5,K=20\), costing \(\$40\); the comparison bundle costs \(\$50\).)
- An isoquant has slope \(-3\) at a feasible bundle \(L=6,K=9\). If \(w=\$12\) and \(r=\$4\), calculate the isocost slope and expenditure, and state whether the bundle satisfies the smooth tangency condition. (Answer: Isocost slope \(-3\), expenditure \(\$108\), and the slopes match, so it satisfies the tangency condition.)
- Near \(L=1\), an isoquant is \(K=3-L+(L-1)^3\), and an isocost line is \(K=3-L\), with \(w=r=1\). At their intersection, compare slopes and test cost at \(L=0.9\) on the isoquant. What does this show? (Answer: They intersect at \(L=1,K=2\) with equal slopes \(-1\), yet at \(L=0.9\), \(K=2.099\) and cost is \(2.999<3\); equal slopes and an intersection alone do not establish a minimum.)