Week 5 · Yosi Ganel

Production at Full Employment — Cobb-Douglas, Marginal Products & Returns to Scale

Production at Full Employment — Cobb-Douglas, Marginal Products & Returns to Scale

Part of: Macro-Economics (Intro) | Key concepts: Production Function, Cobb-Douglas Production Function, Marginal Product of Labor, Marginal Product of Capital, Returns to Scale, Output Elasticity, Classical Model


What This Recitation Covers

The big picture

Recitations 1–3 measured output after the fact — how much GDP was produced and how it splits into income and saving. This recitation opens the classical model and asks the prior question: what determines how much output the economy can produce in the first place? The answer is the production function, which maps the economy's stock of capital (KK) and labor (LL) into output (YY). At full employment all factors are fully utilised, so output is pinned down entirely by the quantities of inputs and the technology linking them. The workhorse is the Cobb-Douglas function, and everything else — marginal products, output elasticities, returns to scale — falls out of differentiating it.

The two organising formulas of the whole recitation:

Y=A KαL1−αMPL=∂Y∂L,MPK=∂Y∂K\boxed{Y = A\,K^{\alpha}L^{1-\alpha}} \qquad \boxed{MPL = \frac{\partial Y}{\partial L},\quad MPK = \frac{\partial Y}{\partial K}}

Every problem is a variation on: write down the production function, differentiate to get a marginal product, scale all inputs by a common factor tt to read off returns to scale, or scale a single input by (1+g)(1+g) to find the percentage change in output.


The Production Function & Cobb-Douglas Form

From general function to Cobb-Douglas

The most general statement is that output depends on capital and labor,

Y=F(K,L),Y = F(K, L),

but to compute anything we need a functional form. The course uses the Cobb-Douglas production function:

Y=A KαL1−α,0<α<1\boxed{Y = A\,K^{\alpha}L^{1-\alpha}}, \qquad 0 < \alpha < 1

Two features are doing all the work here:

  • AA is total factor productivity — the general productivity of the economy. It is a pure multiplier: with the same KK and LL, a higher AA (better technology, organisation, institutions) produces more output.
  • The exponents sum to one (α+(1−α)=1\alpha + (1-\alpha) = 1). This is exactly what makes the function constant returns to scale (CRS) — doubling both inputs doubles output (proved below).
Why economists like Cobb-Douglas

It is the simplest form that simultaneously delivers positive but diminishing marginal products, constant factor shares (α\alpha is capital's share of income), and clean constant returns to scale — all the "textbook" properties of the classical production side in one line of algebra.

Plugging in numbers (slide example)

Let A=2A = 2, α=12\alpha = \tfrac12, L=400L = 400, K=100K = 100: Y=A K1/2L1/2=2⋅100 400=2⋅10⋅20=400Y = A\,K^{1/2}L^{1/2} = 2 \cdot \sqrt{100}\,\sqrt{400} = 2 \cdot 10 \cdot 20 = 400 Now raise productivity to A=3A = 3 with the same K,LK, L: Y=3⋅10⋅20=600Y = 3 \cdot 10 \cdot 20 = 600 A 50 % rise in AA raises output by 50 % — because AA enters multiplicatively, output is proportional to it.


The Meaning of α — Output Elasticity of Capital

α\alpha is not just an exponent; it is the output elasticity of capital — the percentage change in output caused by a 1 % change in capital, holding labor fixed:

εY,K=∂Y∂KKY=α\varepsilon_{Y,K} = \frac{\partial Y}{\partial K}\frac{K}{Y} = \alpha
Verifying the elasticity numerically

Keep A=2, α=12, L=400A=2,\ \alpha=\tfrac12,\ L=400 but raise KK by 1 %, from 100 to 101: Y=2⋅101 400≈2⋅10.05⋅20≈402Y = 2\cdot\sqrt{101}\,\sqrt{400} \approx 2\cdot 10.05 \cdot 20 \approx 402 402400=1.005  ⇒  output rose 12%=α×1%.\frac{402}{400} = 1.005 \;\Rightarrow\; \text{output rose } \tfrac12\% = \alpha \times 1\%. A 1 % rise in capital lifts output by α=12\alpha = \tfrac12 %, exactly as the elasticity predicts.

The two elasticities always sum to 1 under CRS

Symmetrically, 1−α1-\alpha is the output elasticity of labor. Because α+(1−α)=1\alpha + (1-\alpha) = 1, a simultaneous 1 % rise in both inputs raises output by 1 % — the CRS property, seen through the elasticity lens.


Marginal Products — Derivation

The marginal product of an input is the extra output from one more unit of that input, i.e. the partial derivative of FF. For Cobb-Douglas both come out in a memorable ratio form.

Marginal product of labor (MPL)

Y=AKαL1−α  ⇒  ∂Y∂L=(1−α)AKαL−αY = A K^{\alpha} L^{1-\alpha} \;\Rightarrow\; \frac{\partial Y}{\partial L} = (1-\alpha)A K^{\alpha} L^{-\alpha}
MPL=(1−α) A(KL)α\boxed{MPL = (1-\alpha)\,A\left(\frac{K}{L}\right)^{\alpha}}

Marginal product of capital (MPK)

Y=AKαL1−α  ⇒  ∂Y∂K=αAKα−1L1−αY = A K^{\alpha} L^{1-\alpha} \;\Rightarrow\; \frac{\partial Y}{\partial K} = \alpha A K^{\alpha-1} L^{1-\alpha}
MPK=α A(LK)1−α\boxed{MPK = \alpha\,A\left(\frac{L}{K}\right)^{1-\alpha}}
Read the ratios, not the raw powers

Both marginal products depend only on the capital-per-worker ratio K/LK/L, never on the scale of the economy. This is the key to the comparative-statics questions below: change LL and you change K/LK/L, which moves both marginal products in opposite directions.

Diminishing marginal product & the cross-effect
  • MPL=(1−α)A(K/L)αMPL = (1-\alpha)A(K/L)^{\alpha} falls as LL rises (more workers on a fixed capital stock ⇒ each adds less). Diminishing marginal product of labor.
  • But raising LL raises MPK=αA(L/K)1−αMPK = \alpha A(L/K)^{1-\alpha} (more workers make each machine more productive).

So in Cobb-Douglas, capital and labor are complements: adding one input raises the marginal product of the other. This single fact answers most of the multiple-choice traps.


PS4 Q1 — Nominal vs Real GDP per Capita (Israel 2000–2024)

The task

Using Bank of Israel / CBS annual series for 2000–2024, compute GDP per capita in current (nominal) and constant (real) prices, find annual growth rates, plot both, and interpret.

The empirical half of the problem set links back to Recitations 1–2. Two conclusions matter for the exam:

  • The base year is 2015. By construction, in the base year nominal and real GDP per capita are equal (the deflator is 1). Before 2015 nominal << real; after 2015 nominal >> real.
  • The widening nominal–real gap reflects inflation. Because prices generally rose, the nominal series drifts above the real series after the base year. The real (constant-price) series is the one that measures genuine improvements in living standards, having stripped out price changes.
Reading the two downturns

The long-run trend of real GDP per capita is upward (growth), but two dips stand out: 2009 (global financial crisis) and a sharper one in 2020 (COVID-19). Note that in 2009 nominal GDP per capita still edged up even as real fell — inflation masked the real contraction. This is precisely why macro tracks real, not nominal, output.

Connects to Rec 1

The nominal-vs-real distinction, the GDP deflator, and per-capita normalisation are all from Rec_01-Nominal & Real GDP, Price Indices & GNP. Q1 is that machinery applied to real Israeli data.

🗣️ In plain English

Real GDP per capita strips out inflation, so it is the true measure of living standards. Nominal and real are equal in the base year 2015, then nominal drifts above real as prices rise, and the 2009 and 2020 dips show real output actually shrinking.


PS4 Q2 — Four Production Functions: MP, Returns to Scale, Complements

For each function: (1) derive every marginal product, (2) classify returns to scale by scaling all inputs by tt, (3) decide whether inputs are substitutes, complements, or neutral by checking the cross-partial — does raising one input change another's marginal product?

The returns-to-scale test

Replace every input by t×t\times input and factor out the power of tt:

F(tK,tL)=tk F(K,L)  ⇒  {k>1increasing RS (IRS)k=1constant RS (CRS)k<1decreasing RS (DRS)F(tK, tL) = t^{k}\,F(K,L) \;\Rightarrow\; \begin{cases} k > 1 & \text{increasing RS (IRS)}\\ k = 1 & \text{constant RS (CRS)}\\ k < 1 & \text{decreasing RS (DRS)}\end{cases}

For a Cobb-Douglas product, kk is simply the sum of the exponents.

(a) Y=5K0.2L0.4Y = 5K^{0.2}L^{0.4}

MPK=5(0.2)K−0.8L0.4=L0.4K0.8,MPL=5(0.4)K0.2L−0.6=2K0.2L0.6MPK = 5(0.2)K^{-0.8}L^{0.4} = \frac{L^{0.4}}{K^{0.8}}, \qquad MPL = 5(0.4)K^{0.2}L^{-0.6} = \frac{2K^{0.2}}{L^{0.6}}
  • Complements: MPKMPK rises with LL and MPLMPL rises with KK.
  • Returns to scale: F(tK,tL)=5t0.2K0.2t0.4L0.4=t0.6F(K,L)F(tK,tL) = 5t^{0.2}K^{0.2}t^{0.4}L^{0.4} = t^{0.6}F(K,L). Since 0.6<10.6 < 1 ⟹ DRS.
🗣️ In plain English

Capital and labor boost each other's productivity, so they are complements. The exponents add up to 0.6, so doubling everything gives less than double output (DRS).

(b) Y=4(K0.5+L0.5)Y = 4\left(K^{0.5} + L^{0.5}\right)

MPK=4⋅12K−0.5=2K0.5,MPL=4⋅12L−0.5=2L0.5MPK = 4\cdot\tfrac12 K^{-0.5} = \frac{2}{K^{0.5}}, \qquad MPL = 4\cdot\tfrac12 L^{-0.5} = \frac{2}{L^{0.5}}
  • Neutral: MPKMPK depends only on KK, MPLMPL only on LL — changing one input leaves the other's marginal product untouched. (This is an additively separable function, not a product.)
  • Returns to scale: F(tK,tL)=4(t0.5K0.5+t0.5L0.5)=t0.5F(K,L)F(tK,tL) = 4\left(t^{0.5}K^{0.5} + t^{0.5}L^{0.5}\right) = t^{0.5}F(K,L) ⟹ DRS.
🗣️ In plain English

Because the two inputs are just added together, changing one does nothing to the other's marginal product, so they are neutral. The exponents add to 0.5, so doubling everything gives less than double output (DRS).

(c) Y=2K0.3L0.6H0.3Y = 2K^{0.3}L^{0.6}H^{0.3} (three inputs: HH = human capital / land)

MPK=0.6 L0.6H0.3K0.7,MPL=1.2 K0.3H0.3L0.4,MPH=0.6 K0.3L0.6H0.7MPK = \frac{0.6\,L^{0.6}H^{0.3}}{K^{0.7}}, \quad MPL = \frac{1.2\,K^{0.3}H^{0.3}}{L^{0.4}}, \quad MPH = \frac{0.6\,K^{0.3}L^{0.6}}{H^{0.7}}
  • Complements: raising any one input raises the marginal product of the other two.
  • Returns to scale: exponents sum to 0.3+0.6+0.3=1.20.3+0.6+0.3 = 1.2, so F(tK,tL,tH)=t1.2FF(tK,tL,tH) = t^{1.2}F ⟹ IRS.
🗣️ In plain English

Each input makes the other two more productive, so they are all complements. The exponents add to 1.2, so doubling everything gives more than double output (IRS).

(d) Y=2K0.5(L+H)Y = 2K^{0.5}(L + H)

MPK=L+HK0.5,MPL=2K0.5,MPH=2K0.5MPK = \frac{L+H}{K^{0.5}}, \qquad MPL = 2K^{0.5}, \qquad MPH = 2K^{0.5}
  • LL and HH are each complementary to capital (MPKMPK rises with LL or HH; MPL,MPHMPL, MPH rise with KK).
  • LL and HH are neutral to each other — MPLMPL doesn't depend on HH and vice-versa (they enter as a simple sum).
  • Returns to scale: F(tK,tL,tH)=2t0.5K0.5⋅t(L+H)=t1.5FF(tK,tL,tH) = 2t^{0.5}K^{0.5}\cdot t(L+H) = t^{1.5}F ⟹ IRS.
🗣️ In plain English

Capital works well with both labor and the other factor, but those two do not affect each other since they are added, so they are neutral to each other. The exponents add to 1.5, so doubling everything gives more than double output (IRS).

Function Form Returns to scale (∑\sum exp.) Input relationship
(a) 5K0.2L0.45K^{0.2}L^{0.4} product DRS (0.60.6) complements
(b) 4(K0.5+L0.5)4(K^{0.5}+L^{0.5}) sum DRS (0.50.5) neutral
(c) 2K0.3L0.6H0.32K^{0.3}L^{0.6}H^{0.3} product IRS (1.21.2) complements
(d) 2K0.5(L+H)2K^{0.5}(L+H) mixed IRS (1.51.5) KK–LL, KK–HH complements; LL–HH neutral
Sum vs product decides "complement vs neutral"

Inputs that enter as a product (or a Cobb-Douglas term) are complements; inputs that enter as an additive sum are neutral to each other. Don't confuse this with returns to scale — a function can be DRS and have complementary inputs (case a), or IRS and have a neutral pair (case d). The two questions are independent.

Slide algebra is sloppy — trust the exponent sum

The lecture's returns-to-scale lines contain stray typos (e.g. "t0.65t^{0.65}", "t1.22t^{1.22}"), but the boxed conclusions are right because kk is just the sum of exponents. When in doubt, add the exponents rather than copying the intermediate line.


PS4 Q3 — Immigration Raises Labor by 5% (Y=6K0.5L0.5Y = 6K^{0.5}L^{0.5})

Which statement is correct after L↑5%L \uparrow 5\%?

a) MPKMPK unchanged · b) MPLMPL rises 5 % · c) output rises < 5 % · d) if KK also ↑5%\uparrow 5\%, GDP rises exactly 5 % · e) none.

Work through each with the ratio formulas. Here α=0.5\alpha = 0.5.

(a) MPK unchanged? MPK=αA(L/K)1−α=3(LK)0.5MPK = \alpha A (L/K)^{1-\alpha} = 3\left(\tfrac{L}{K}\right)^{0.5}. Raising LL raises MPKMPK — false. ✗

(b) MPL rises 5 %? MPL=(1−α)A(K/L)α=3(KL)0.5MPL = (1-\alpha)A(K/L)^{\alpha} = 3\left(\tfrac{K}{L}\right)^{0.5}. Raising LL lowers MPLMPL (diminishing returns) — false. ✗

(c) Output rises < 5 %?

Y1=6K0.5(1.05L)0.5=6K0.5L0.5⋅1.050.5=Y⋅1.050.5≈1.0247 YY_1 = 6K^{0.5}(1.05L)^{0.5} = 6K^{0.5}L^{0.5}\cdot 1.05^{0.5} = Y\cdot 1.05^{0.5} \approx 1.0247\,Y

Output rises ≈ 2.5 %, which is less than 5 %. ✓ (answer c)

(d) KK and LL both ↑5%\uparrow 5\% ⟹ exactly 5 %?

Y1=6(1.05K)0.5(1.05L)0.5=Y⋅1.050.5+0.5=1.05 YY_1 = 6(1.05K)^{0.5}(1.05L)^{0.5} = Y\cdot 1.05^{0.5+0.5} = 1.05\,Y

Output rises exactly 5 % — this is also true (CRS), but the exam keys c as the intended single answer since it is the first correct statement; note both c and d are individually valid statements of the CRS property.

Why 5 % labor → only ~2.5 % output

Under CRS with α=0.5\alpha=0.5, labor's output elasticity is 1−α=0.51-\alpha = 0.5. So %ΔY≈(1−α)×%ΔL=0.5×5%=2.5%\%\Delta Y \approx (1-\alpha)\times\%\Delta L = 0.5\times5\% = 2.5\%. Scaling one input by its share, not the whole economy, is why output rises by less than the input.

🗣️ In plain English

Only labor goes up, and labor's share of the output is 0.5, so a 5% rise in LL lifts output by only about 2.5%. Raising capital too would give the full 5% (CRS), but one input alone gets you less than that.


PS4 Q4 — Natural Disaster Cuts Capital by 10% (Y=5K0.4L0.4Y = 5K^{0.4}L^{0.4})

Note this function is DRS, not CRS

The exponents sum to 0.4+0.4=0.8<10.4+0.4 = 0.8 < 1. This changes the "proportional" answers relative to Q3. Always check the exponent sum before answering the scaling parts.

Which statement is correct after K↓10%K \downarrow 10\%?

A) MPKMPK decreases · B) output falls > 10 % · C) function is DRS · D) if LL also ↓10%\downarrow 10\%, output falls exactly 10 % · E) output unchanged.

A) MPK decreases? MPK=5(0.4)L0.4K0.6=2L0.4K0.6MPK = 5(0.4)\dfrac{L^{0.4}}{K^{0.6}} = \dfrac{2L^{0.4}}{K^{0.6}}. As KK falls, K0.6K^{0.6} falls, so MPKMPK rises — false. ✗

B) Output falls > 10 %?

Y1=5(0.9K)0.4L0.4=Y⋅0.90.4≈0.959 YY_1 = 5(0.9K)^{0.4}L^{0.4} = Y\cdot 0.9^{0.4} \approx 0.959\,Y

Output falls ≈ 4.1 % — less than 10 %, not more — false. ✗

C) Function is DRS?

F(tK,tL)=5(tK)0.4(tL)0.4=t0.8F(K,L),0.8<1  ⇒  DRS.✓F(tK,tL) = 5(tK)^{0.4}(tL)^{0.4} = t^{0.8}F(K,L), \quad 0.8 < 1 \;\Rightarrow\; \textbf{DRS}. ✓

This is the correct answer. (answer C)

D) KK and LL both ↓10%\downarrow 10\% ⟹ exactly 10 %?

Y1=5(0.9K)0.4(0.9L)0.4=Y⋅0.90.8≈0.919 YY_1 = 5(0.9K)^{0.4}(0.9L)^{0.4} = Y\cdot 0.9^{0.8} \approx 0.919\,Y

Output falls ≈ 8 %, not 10 % — false (exactly what DRS predicts: scaling everything down by 10 % cuts output by less than 10 %). ✗

E) Output unchanged? From (B), output falls ≈ 4.1 % — false. ✗

DRS vs CRS on proportional shocks

With CRS, cutting all inputs by 10 % cuts output by exactly 10 %. With DRS (k=0.8k=0.8), the same cut reduces output by only ≈10%×k\approx 10\%\times k-ish → about 8 %. The whole-economy scaling factor is 0.9 ∑exp0.9^{\,\sum\text{exp}}; single-input scaling is 0.9 that input’s exp0.9^{\,\text{that input's exp}}.

🗣️ In plain English

Cutting only capital by 10% costs about 4.1% of output, and cutting both inputs by 10% costs about 8%, so with DRS the damage is always smaller than 10%. The function is DRS (exponents add to 0.8), and MPK actually rises because capital got scarcer, so the answer is C.


The Standard Exam Recipe

Production-function questions in four moves

1 — Marginal products. Differentiate. For Cobb-Douglas memorise MPL=(1−α)A(K/L)αMPL=(1-\alpha)A(K/L)^{\alpha} and MPK=αA(L/K)1−αMPK=\alpha A(L/K)^{1-\alpha}; each rises in the other input and falls in its own. 2 — Returns to scale. Scale all inputs by tt and read the power of tt = sum of exponents (>1>1 IRS, =1=1 CRS, <1<1 DRS). 3 — Single-input shock. Multiply that one input by (1+g)(1+g); output changes by the factor (1+g)that input’s exponent(1+g)^{\text{that input's exponent}}. 4 — Complement vs neutral. Product/Cobb-Douglas terms ⟹ complements; additive terms ⟹ neutral. Independent of returns to scale.


Summary — Recitation 5 in Five Ideas

  1. Output comes from a production function. At full employment, Y=A KαL1−αY = A\,K^{\alpha}L^{1-\alpha} — technology AA times a Cobb-Douglas combination of capital and labor.
  2. α is capital's output elasticity. A 1 % rise in KK raises output by α\alpha %; 1−α1-\alpha plays the same role for labor, and the two sum to one under CRS.
  3. Marginal products depend on K/LK/L and diminish. MPLMPL falls as labor grows but rises as capital grows — capital and labor are complements in Cobb-Douglas.
  4. Returns to scale = sum of exponents. Scale every input by tt; the exponent on tt tells you IRS/CRS/DRS. Under DRS a proportional 10 % cut lowers output by less than 10 %.
  5. Single-input scaling ≠ whole-economy scaling. Raising one input by gg changes output by (1+g)(1+g) raised to that input's exponent, so a 5 % labor rise with α=0.5\alpha=0.5 lifts output only ≈ 2.5 %.