Marginal Products In Multi-Input Functions

Introduction

A production function describes how inputs combine to produce output. When a firm uses more than one input, such as labor and equipment, it is useful to ask how much output would change if it increased just one input while keeping the others unchanged. The answer is that input’s marginal product.

For multiple inputs, this question is expressed using a partial derivative. It measures the local output effect of changing one input, evaluated at specified input quantities. This topic is aligned with the SOA VEE Economics (Validation by Educational Experience) syllabus.

Definition: For a production function $q=f(x_1,x_2,\ldots,x_n)$, the marginal product of input $x_j$ is the partial derivative of output with respect to $x_j$, holding every other input fixed: $$ MP_{x_j}=\frac{\partial q}{\partial x_j}. $$ It measures the local rate of change in output from a small change in $x_j$, at the specified input quantities.

What a Partial Derivative Holds Fixed

Suppose a firm produces output $q=f(L,K)$, where $L$ is labor and $K$ is capital. The marginal product of labor is written $MP_L=\partial q/\partial L$. In calculating it, $K$ is treated as constant. The marginal product of capital, $MP_K=\partial q/\partial K$, is calculated with $L$ held constant.

The phrase holding the other inputs fixed is essential. It describes a comparison in which one input changes and the others do not. If both labor and capital change, the resulting output change is not, by itself, the marginal product of either input.

The derivative may depend on the levels of the other inputs. For example, an interaction term such as $LK$ means that the marginal product of labor can vary with the amount of capital. Therefore, calculate the partial derivative first, then substitute the input quantities specified in the question.

Why Partial Differentiation Gives the Marginal Product

The derivative definition makes precise the idea of changing only one input. For two inputs, consider increasing labor by a small amount $\Delta L$ while keeping capital at $K$. The output change per unit of labor change is the difference quotient below. Its limit as the labor change approaches zero is the partial derivative.

\frac{\partial q}{\partial L} = \lim_{\Delta L\to 0} \frac{f(L+\Delta L,K)-f(L,K)}{\Delta L}.

Capital appears in both output evaluations at the same value $K$, so this comparison changes labor alone. The limit describes the local rate of change, not necessarily the exact output change from a large, finite increase in labor.

For small simultaneous changes in several inputs, the total differential provides a local approximation:

\Delta q \approx \frac{\partial q}{\partial L}\Delta L+ \frac{\partial q}{\partial K}\Delta K.

Each derivative in this approximation is evaluated at the starting input quantities. If only labor changes, then $\Delta K=0$, and the approximation becomes $\Delta q\approx MP_L\Delta L$. It is an approximation for a small change; for a larger change, the marginal product may vary along the way.

Worked Example: Two Inputs with Interaction and Curvature

A workshop’s output, measured in batches per day, is modeled by $q=20+5L+4K+0.2LK-0.5L^2-0.25K^2$. Here $L$ is labor-hours and $K$ is machine-hours. At $L=4$ and $K=6$, find output and both marginal products. Then estimate the output change if labor increases by 0.5 labor-hours while machine-hours stay fixed. Method: take each partial derivative, evaluate at the stated input bundle, and use the local-change approximation. This avoids treating a finite change as though the marginal product must remain constant throughout it.

Output at the stated bundle. Substitute $L=4$ and $K=6$:

$$ q=20+5(4)+4(6)+0.2(4)(6)-0.5(4^2)-0.25(6^2) =20+20+24+4.8-8-9=51.8. $$

The modeled output is 51.8 batches per day.

Marginal product of labor. Hold $K$ fixed and differentiate with respect to $L$:

$$ MP_L=5+0.2K-L. $$

At $K=6$ and $L=4$, $MP_L=5+0.2(6)-4=5+1.2-4=2.2$. At this input bundle, a small increase of one labor-hour is associated locally with about 2.2 additional batches per day.

Marginal product of capital. Hold $L$ fixed and differentiate with respect to $K$:

$$ MP_K=4+0.2L-0.5K. $$

At $L=4$ and $K=6$, $MP_K=4+0.2(4)-0.5(6)=4+0.8-3=1.8$. The local output effect is 1.8 batches per additional machine-hour.

Estimated effect of the labor increase. The local approximation uses the starting marginal product and holds machine-hours fixed:

\Delta q\approx MP_L\Delta L=2.2(0.5)=1.1\text{ batches per day}.

The model’s exact finite change is slightly different. With $K=6$, increasing $L$ by 0.5 changes output by $2.2(0.5)-0.5(0.5^2)=1.1-0.125=0.975$ batches per day. The curvature term makes the marginal product of labor decline as $L$ increases. The derivative-based estimate is local; the exact change accounts for that variation. As a check, substituting $L=4.5,K=6$ gives $q=52.775$, and $52.775-51.8=0.975$.

Three or More Inputs

The same method applies when a production function has three or more inputs. To find one input’s marginal product, differentiate with respect to that input and treat all the others as constants. The answer may depend on several input levels, so evaluate it only after obtaining the derivative.

Worked Example: Marginal Products in a Three-Input Function

A hypothetical production process is represented by $q=2L^2+3KM+M^2$, where output is measured in units per shift and $L$, $K$, and $M$ are quantities of three inputs. Find output and each marginal product at $L=3$, $K=2$, and $M=4$. State what each result means locally. Method: calculate three partial derivatives, holding the other two input quantities fixed in each calculation. This avoids confusing the marginal product of one input with the effect of changing several inputs together.

Output. Substitute the three input quantities:

$$ q=2(3^2)+3(2)(4)+4^2=18+24+16=58. $$

Marginal product of $L$. The terms involving $L$ give $\partial q/\partial L=4L$. At $L=3$, $MP_L=4(3)=12$.

Marginal product of $K$. While differentiating with respect to $K$, both $L$ and $M$ are fixed. Thus $\partial q/\partial K=3M$. At $M=4$, $MP_K=3(4)=12$.

Marginal product of $M$. Both $3KM$ and $M^2$ depend on $M$, so $\partial q/\partial M=3K+2M$. At $K=2$ and $M=4$, $MP_M=3(2)+2(4)=6+8=14$.

At this input bundle, the local rates are 12, 12, and 14 output units per shift per additional unit of $L$, $K$, and $M$, respectively, with the other two inputs fixed in each case. A check on the differentiation is that the $3KM$ term contributes to both $MP_K$ and $MP_M$: it contributes $3M=12$ to the former and $3K=6$ to the latter.

Worked Example: Input Interactions Change a Marginal Product

A hypothetical food-packaging line has the production function $q=8L+6K+5M+LK+2KM-0.5LM$, with $q$ measured in cases per hour. At $L=2$, $K=3$, and $M=4$, calculate the output and the marginal product of each input. Then explain how an increase in $M$ affects $MP_L$. Method: differentiate with respect to one variable at a time and evaluate the resulting expressions. This avoids assuming that a marginal product is independent of other inputs when the function contains interactions.

Output. Substituting the input levels gives:

$$ q=8(2)+6(3)+5(4)+(2)(3)+2(3)(4)-0.5(2)(4) =16+18+20+6+24-4=80. $$

Marginal product of $L$. Holding $K$ and $M$ fixed gives $MP_L=8+K-0.5M$. At the stated levels, $MP_L=8+3-0.5(4)=8+3-2=9$.

Marginal product of $K$. Holding $L$ and $M$ fixed gives $MP_K=6+L+2M$. Thus $MP_K=6+2+2(4)=16$.

Marginal product of $M$. Holding $L$ and $K$ fixed gives $MP_M=5+2K-0.5L$. Thus $MP_M=5+2(3)-0.5(2)=5+6-1=10$.

Increasing $M$ lowers the expression for $MP_L$, because its $M$-dependent term is $-0.5M$. For example, at fixed $K=3$, raising $M$ from 4 to 5 changes $MP_L$ from $8+3-0.5(4)=9$ to $8+3-0.5(5)=8.5$. This compares the labor marginal product at two different input bundles; it does not mean that $M$ changed while calculating either partial derivative.

Interpreting and Checking Marginal Products

A marginal product has units: output units divided by units of the input being changed. If output is cases per day and labor is labor-hours, $MP_L$ is measured in cases per day per labor-hour. Always state which input is changing and which inputs are held fixed when interpreting the value.

A marginal product can be positive, zero, or negative, depending on the function and the input bundle. A positive value means a small increase in that input is associated locally with more output; a negative value means it is associated locally with less output. These signs describe the modeled local relationship, not a universal claim about every production process.

Use the following sequence to reduce errors:

Name the input.
Identify which marginal product is requested, such as $MP_K$.

Differentiate with the others fixed.
When differentiating with respect to $K$, treat every other input as a constant.

Evaluate at the input bundle.
Substitute all specified input quantities into the derivative, not into the original function alone.

Interpret and check.
State the output-per-input units, identify what is held fixed, and check that every term involving the chosen input was differentiated.

A frequent exam trap is to confuse a marginal product with the exact output change from a finite input increase. For instance, in the workshop example, $MP_L=2.2$ at the starting bundle. Multiplying by a 0.5-hour increase gives the local estimate $2.2(0.5)=1.1$, not the exact change. The exact change is 0.975 because the production function is curved in $L$. Another common error is to change an input that should be held fixed; that calculation no longer isolates the requested partial derivative.

Key takeaway: A marginal product is a partial derivative evaluated at an input bundle. Differentiate with respect to one input while holding all others fixed, then evaluate and interpret the result as a local output rate for that input.

Check Your Understanding

For each problem, identify the relevant partial derivatives before substituting the input quantities.

A production function is $q=30+4L+2K+LK$. At $L=3$ and $K=5$, find output, $MP_L$, and $MP_K$. (Answer: $q=67$, $MP_L=9$, $MP_K=5$.)
A process has $q=100+7A+3B-0.5A^2+AB-B^2$. At $A=4$, $B=2$, find output and both marginal products. (Answer: $q=130$, $MP_A=5$, $MP_B=3$.)
For $q=2L^2+3KM+M^2$, find $MP_L$, $MP_K$, and $MP_M$ at $L=3$, $K=2$, $M=4$. (Answer: $MP_L=12$, $MP_K=12$, $MP_M=14$.)
A packaging function is $q=5X+4Y+XY-0.25X^2$. At $X=2$, $Y=6$, find both marginal products. Use the marginal product of $Y$ at this bundle to estimate the output change from increasing $Y$ by 0.2 while holding $X$ fixed. (Answer: $MP_X=10$, $MP_Y=6$, estimated $\Delta q=1.2$.)
A production function is $q=9A+8B+6C+2AB-BC-AC$. At $A=2$, $B=3$, $C=4$, calculate $MP_A$, $MP_B$, and $MP_C$. (Answer: $MP_A=11$, $MP_B=8$, $MP_C=1$.)

What You'll Learn