MathemagiciansLibraryExplorationsDeep Dives
← Economics · Pre-University
EconomicsPre-University

Income and substitution effects

PE3.1 — Consumer Theory › PE3.1.1 — Utility and Indifference CurvesLecture 4 of 4~2760 words

One price, two changes

The price of petrol jumps by 40%. What do you do?

Part of the answer is easy. Petrol is now dear compared with the bus or the train, so you switch some journeys. But something else has happened too. The car journeys you still make cost more, so less money is left for everything else. You are a little poorer, even though your pay has not changed. And poorer people change what they buy.

The last lecture ended on exactly this point. Amara's demand curve for coffee sloped down, and the lecture added a warning: "A price change does two things at once: it makes the good dearer compared with others, and it makes the buyer a little poorer. The next lecture separates these two effects, and shows why, for rare goods, they can push the demand curve the wrong way."

The two effects are the substitution effect and the income effect. Seeing them separately is the hard part, so take your time with the next two sections.

Separating the two: give her the money back

Return to Amara. She has $120 a week. Tea costs $3 a cup. Her preferences are the invented ones from the last lecture, U = C × T. At a coffee price of $4, her best bundle is A: 15 cups of coffee and 20 cups of tea, with utility 300.

Now coffee rises to $6. She still spends half her income on each good, so she buys 60 ÷ 6 = 10 cups of coffee and 20 cups of tea. Call this bundle B. The fall of 5 cups mixes both effects. How can we pull them apart?

Here is the trick. Imagine that, as coffee becomes dearer, a friend gives Amara just enough money to be exactly as happy as before. She stays on her old indifference curve, U = 300. Now the price rise cannot make her poorer. Whatever she still changes is caused by the new price ratio alone.

With the gift, she faces the new prices and chooses the best point on her old curve. The tangency rule still works. It says MRS = T ÷ C = 6 ÷ 3 = 2, so T = 2C. On her old curve, C × 2C = 300. So C² = 150 and C ≈ 12.25 cups, with T ≈ 24.49 cups. Call this bundle A′ ("A prime"). It costs 6 × 12.25 + 3 × 24.49 ≈ $146.97. So the friend's gift is about $26.97.

01020304050010203040Coffee (cups per week)Tea (cups per week)AA′B

- - - old budget line ($4) · —— new budget line · - - - compensated line · —— indifference curves

Coffee rises from $4 to $6. A is Amara's old choice. A′ is on her old indifference curve (U = 300), where a line with the new slope just touches it. B is her new choice on the new budget line, on a lower curve (U = 200). The move from A to A′ is the substitution effect. The move from A′ to B is the income effect.

Now the decomposition is easy to read:

Together, −2.75 + (−2.25) = −5 cups. That is it. That is the whole idea: one movement along the old curve, then one parallel shift of the budget line.

Substitution and income effects (Hicks)

When the price of X changes, draw a compensated budget line: the slope of the new prices, moved until it just touches the old indifference curve at A′.

Substitution effect: the change in X from A to A′ (same utility, new price ratio).

Income effect: the change in X from A′ to B (new prices, real spending power removed).

Total change in X = substitution effect + income effect.

Stop and think

At A, Amara drinks 20 cups of tea. At B, she still drinks 20. Does that mean the price of coffee had no effect on her tea?

Show the answer

No. Two effects cancelled. The substitution effect moved her towards tea, from 20 cups to about 24.49, because tea became relatively cheaper. The income effect took her back to 20, because she was poorer and tea is normal for her. With U = C × T they cancel exactly, because she always spends half her income on tea. "No total change" can hide two large changes.

Why the substitution effect never pushes the wrong way

When a price rises, the substitution effect never raises the quantity of that good. Do not accept this as a rule. Look at the reason.

The move from A to A′ stays on one indifference curve. At A, the curve's slope equals the old price ratio. At A′, it equals the new, steeper one. So where on the curve is it steeper? Indifference curves bow in towards the origin, so they get steeper as you move left, towards less coffee. This is the "preference for balanced bundles": with little coffee left, each cup is worth more tea. So a steeper price line must touch the curve further left. The substitution effect can be zero, but never positive for a price rise. It is zero when the goods are used in a fixed ratio, with L-shaped curves (Worked Example 3).

Move the price of coffee below. Watch where A′ and B go.

Explore — split a price change into two effects

Amara has $120, tea costs $3 and her old price of coffee was $4. Choose a new price of coffee. The diagram shows A, A′ and B; the line below shows the two effects on her coffee.

Presets:

- - - old budget line ($4) · —— new budget line · - - - compensated line · —— indifference curves

● A, the old choice · ○ A′, on the old curve at the new prices · ● B, the new choice

Amara's coffee at A′ and B for five new prices of coffee (old price $4, so A is 15 cups; income $120, tea $3, invented preferences U = C × T). The diagram above shows the case of $6. A′, B and the two effects are in cups of coffee.

Coffee price ($)A′BSubstitutionIncome
221.2130+6.21+8.79
317.3220+2.32+2.68
513.4212−1.58−1.42
612.2510−2.75−2.25
810.617.5−4.39−3.11

The income she would need to stay on her old curve is ≈ $84.85, $103.92, $134.16, $146.97 and $169.71 at the five prices. Above $4 she needs more than $120, so the price rise made her poorer; below $4 she needs less, so the price fall made her richer.

The substitution effect always moved coffee against its price. For Amara, the income effect pointed the same way, because coffee is a normal good for her. That is not always true.

The income effect can point either way

You met these words in the lecture on what shifts demand. People buy more of a normal good when their income rises. They buy less of an inferior good, because they can now afford something they prefer. "Inferior" does not mean bad. It only describes the direction of the income response.

A price rise makes the buyer poorer. So the income effect of a price rise is:

How big is the income effect? It depends on how much poorer the price rise makes the buyer: roughly, the quantity she was buying times the rise in price. Amara lost about 15 × $2 = $30 of spending power. Her exact gift was a little less, $26.97, because she can switch. So a good that takes a large share of the budget gives a large income effect. Dearer salt barely makes anyone poorer. Dearer rent, or a dearer main food, can.

Normal goodA 20A′ 16B 13Inferior, not GiffenA 20A′ 16B 18GiffenA 20A′ 18B 23

⟶ substitution effect (A to A′) · ⟶ income effect (A′ to B)

The same price rise for three kinds of good (invented quantities). The substitution effect always points to less. For a normal good the income effect also points to less. For an inferior good it points back, part of the way. For Giffen behaviour it points back further than the substitution effect went.

Giffen behaviour: when the income effect wins

Look at the last row of the picture. If the income effect of a price rise is positive and bigger than the substitution effect, the buyer ends up buying more of the good. The demand curve slopes up. This is Giffen behaviour, named after Robert Giffen, to whom Alfred Marshall credited the idea in 1895.

Three conditions follow from the decomposition. The good must be inferior, or the income effect points the wrong way. It must take a large share of the budget, or the income effect is small. It must have few close substitutes, so the substitution effect is small enough to beat.

Picture a very poor family that lives mostly on rice, with a little meat. Rice becomes dearer, so the family is poorer. It can no longer afford its meat. But it still needs enough food, and the cheapest way to fill the gap is more rice. So it buys more rice, even though rice is dearer.

Now sort each case below.

Sort — normal, inferior or Giffen?

Each card describes a price change for one buyer, with nothing else changing. Put each card in the right box. Read the reason after each choice.

Nine cases, sorted. The sign of the income effect, and its size compared with the substitution effect, decide the box.

CaseBox
Price of X rises. Substitution effect −3 units, income effect −2 units.Normal good
Price of X rises. Substitution effect −3 units, income effect +1 unit.Inferior, not Giffen
Price of X rises. Substitution effect −1 unit, income effect +2 units.Giffen behaviour
Price of X falls. Substitution effect +2 units, income effect +1 unit.Normal good
Price of X falls. Substitution effect +4 units, income effect −1 unit.Inferior, not Giffen
Price of X falls, and the buyer buys less X. Income, tastes and other prices have not changed.Giffen behaviour
Price of X rises. The quantity falls, but by less than the substitution effect alone.Inferior, not Giffen
Price of X rises. The quantity falls by more than the substitution effect alone.Normal good
Rice vouchers cut the price of rice for poor households in Hunan, and many of them bought less rice.Giffen behaviour

Two ways to give the money back

The method so far keeps the buyer's happiness fixed. It is named after John Hicks, who set it out with Roy Allen in 1934. An older method, from Eugen Slutsky in 1915, keeps the old bundle affordable instead. The gift is just enough to buy A again at the new prices.

For Amara, A costs 6 × 15 + 3 × 20 = $150 at the new prices, so the gift is $30. With $150 she buys 150 ÷ 12 = 12.5 cups of coffee and 25 of tea. The substitution effect is 15 → 12.5, or −2.5 cups, and the income effect is 12.5 → 10, or −2.5 cups.

Slutsky's gift is bigger than Hicks's ($26.97) because with it she can buy A, or switch to something better. The two methods give the same signs, and almost the same sizes for small price changes. Say which one you used.

Worked examples

Worked Example 1 — A full decomposition

U = F × C (food and clothing, invented). Income $80, food $2, clothing $4. The price of food rises to $4. Find the substitution and income effects on food (Hicks).

A, before. Half the income goes on each good: $40 buys F = 20 and C = 10. Utility U = 200.

B, after. Food now costs $4: $40 buys F = 10, and C = 10 as before. Utility 100.

A′, compensated. New price ratio 4 ÷ 4 = 1, so the tangency gives C = F. On the old curve, F × F = 200, so F = √200 ≈ 14.14, and C ≈ 14.14.

Effects. Substitution: 20 → 14.14, about −5.86. Income: 14.14 → 10, about −4.14. Food is normal, so both are negative.

Check. −5.86 + (−4.14) = −10, and B − A = 10 − 20 = −10 ✓. The gift needed is 4 × 14.14 + 4 × 14.14 ≈ $113.14, about $33.14 more than $80.

Worked Example 2 — Reading the three points

A household buys rice (invented figures, kilograms per month). The price of rice rises. A diagram gives A = 20 kg, A′ = 16 kg and B = 18 kg. What kind of good is rice for this household? What if B were 23 kg instead, with A′ = 18 kg?

First case. Substitution: 20 → 16, −4 kg. Income: 16 → 18, +2 kg. Feeling poorer, the household buys more rice, so rice is inferior. But +2 is smaller than −4, so the total is −2 kg: inferior, not Giffen.

Second case. Substitution −2 kg, income +5 kg. The income effect wins: a higher price and 3 kg more rice. This is Giffen behaviour.

Check. In both cases, substitution + income = B − A: −4 + 2 = −2 ✓ and −2 + 5 = +3 ✓.

Worked Example 3 — When there is nothing to substitute

Kwame always uses one pen refill with one notebook (invented). For him, U = the smaller of X and Y, written min(X, Y). Refills (X) cost $1, notebooks (Y) $1, and he has $60. The price of refills rises to $2.

A. He buys equal numbers: 60 ÷ (1 + 1) = 30 of each. U = 30.

B. Each pair now costs $3: 60 ÷ 3 = 20 pairs. U = 20.

A′. To stay at U = 30, he needs 30 pairs whatever the prices. So A′ = A, and the substitution effect is 0.

Effects. Substitution 0; income 30 → 20, −10 refills. With L-shaped curves, a new price ratio cannot move him along them.

Check. The gift needed is 30 × $3 − $60 = $30, and with $90 he buys exactly A again ✓.

A real case: cheaper rice, less rice

The policy. In 2006 the economists Robert Jensen and Nolan Miller ran a field experiment with 650 poor households in towns in each of two Chinese provinces: Hunan, where rice is the main food, and Gansu, where wheat flour is. Households chosen at random received vouchers for five months, June to October. The vouchers cut the price of the staple by 0.10, 0.20 or 0.30 yuan per jin (half a kilogram). With rice at about 1.2 yuan per jin, these were cuts of roughly 8%, 17% and 25%. Hunan households got about 64% of their calories from rice.

The prediction. The law of demand says cheaper rice means more rice. The decomposition says it depends. The substitution effect pushes towards more rice. But a household that spends much of its budget on rice gains a lot of spending power from the voucher. If rice is inferior for it, that income effect pushes towards less rice and more meat, and it might win.

What happened. In Hunan, households that were poor, but not the very poorest, bought less rice when its price fell. The researchers estimated that for them a 1% rise in the price of rice would raise rice consumption by about 0.45%: a demand curve sloping up. The very poorest households, with more than 80% of their calories from rice, behaved normally; they had almost no meat to switch to. Better-off households also behaved normally. In Gansu the evidence was weaker (American Economic Review, 2008). A follow-up study (2011) found that the vouchers did not improve nutrition.

Verdict. Giffen behaviour is real, but narrow. It belongs to particular buyers at a particular level of poverty, not to rice, which is why the authors call it Giffen behaviour. For policy, a price subsidy works through both effects, and for some families the income effect works against the goal. The older textbook story about potatoes in the Irish famine has not survived historians' checks (Dwyer and Lindsay, 1984; Rosen, 1999).

Trap 1 — The compensated line at the old slope

The compensated line uses the new prices, so it is parallel to the new budget line. If you draw it parallel to the old line, it touches the old curve at A itself, and you will find no substitution effect at all.

Trap 2 — Inferior does not mean Giffen

Most inferior goods still have demand curves that slope down: the income effect works against the substitution effect but is smaller. Giffen behaviour needs it to be bigger.

Trap 3 — "The substitution effect is always negative"

That phrase means it moves the quantity against the price. For a price fall, the substitution effect is an increase.

Try it yourself

Work each one on paper first. Open a hint only when you are stuck.

  1. U = X × Y, income $100, PY = $4. The price of X rises from $2 to $5. Find A, B and A′, the substitution and income effects on X, and the gift needed (Hicks).

    Hint 1

    With U = X × Y, half the income goes on each good. That gives A and B straight away.

    Hint 2

    For A′, the tangency at the new prices gives 5X = 4Y. Put Y = 1.25X into X × Y = (utility at A).

    Full answer

    A: X = 50 ÷ 2 = 25, Y = 50 ÷ 4 = 12.5, U = 312.5. B: X = 50 ÷ 5 = 10, Y = 12.5. A′: Y = 1.25X and 1.25X² = 312.5, so X² = 250, X ≈ 15.81 and Y ≈ 19.76. Substitution: 25 → 15.81, about −9.19. Income: 15.81 → 10, about −5.81. Gift: 5 × 15.81 + 4 × 19.76 ≈ 158.11, so about $58.11. Check: −9.19 + (−5.81) = −15 = 10 − 25 ✓.

  2. For three goods, a price rise gives these quantities (invented). Good P: A = 30, A′ = 26, B = 22. Good Q: A = 12, A′ = 10, B = 11. Good R: A = 40, A′ = 38, B = 43. Classify each good, with reasons.

    Hint 1

    Work out the two effects for each: A′ − A and B − A′.

    Hint 2

    A positive income effect after a price rise means "inferior". Then compare its size with the substitution effect.

    Full answer

    P: substitution −4, income −4. The income effect adds to the fall, so P is normal. Q: substitution −2, income +1. Inferior, and the total is −1, so not Giffen. R: substitution −2, income +5. Inferior, and the income effect wins: total +3, so this is Giffen behaviour. Check: each pair adds up to B − A: −8, −1, +3 ✓.

  3. Redo question 1 with Slutsky's method. How big is the gift, and how do the two effects compare with Hicks's?

    Hint 1

    Slutsky's gift lets the buyer afford A (25, 12.5) at the new prices.

    Hint 2

    A costs 5 × 25 + 4 × 12.5 at the new prices. Then use "half on each good".

    Full answer

    A costs 125 + 50 = $175 at the new prices, so the gift is $75. With $175, X = 87.5 ÷ 5 = 17.5. Substitution: 25 → 17.5, −7.5. Income: 17.5 → 10, −7.5. Slutsky's gift ($75) is bigger than Hicks's ($58.11) because it lets the buyer do better than A. So his substitution effect is smaller in size (−7.5 against −9.19) and his income effect larger (−7.5 against −5.81). Check: both totals are −15 ✓.

  4. A government adds a tax of $0.50 to a sugary drink that costs $2. A consumer has U = X × Y, where X is drinks and Y is all other spending (price $1), and an income of $100. (a) How many drinks does she buy before and after the tax? (b) Suppose the government gives every consumer back, as cash, exactly the tax she paid. How many drinks does she buy now? (c) Explain the result using the two effects. Is she as well off as before the tax?

    Hint 1

    Before: half of $100 on drinks at $2. After: half on drinks at $2.50. With the refund R, her income is 100 + R, and R = 0.50 × (drinks bought).

    Hint 2

    X = (100 + 0.5X) ÷ (2 × 2.5). Solve for X.

    Full answer

    (a) Before: X = 50 ÷ 2 = 25 drinks. After the tax: X = 50 ÷ 2.5 = 20. (b) 5X = 100 + 0.5X, so 4.5X = 100 and X ≈ 22.2 drinks; the refund is about $11.11. (c) The refund gives back most of the spending power the tax took, so it removes most of the income effect. What is left is mainly the substitution effect: drinks are still dearer than other things, so she still buys fewer, about 22.2 instead of 25. This is why a tax on one good with the money handed back still changes behaviour. But she is not as well off: her utility is 22.2 × 55.6 ≈ 1235, against 25 × 50 = 1250 before. Check: income 100 + 11.11 = 111.11, half of it is 55.56, and 55.56 ÷ 2.5 ≈ 22.2 ✓. (A tax at the same rate on everything she buys would be different: relative prices would not change, so there would be no substitution effect, only an income effect.)

The bigger idea

The decomposition works for any price. A wage is the price of an hour of free time. When the wage rises, free time becomes dearer, so the substitution effect says work more. But the worker is also richer, and free time is a normal good, so the income effect says work less. Which one wins returns in the lecture on labour supply.

Hicks and Allen built this idea in 1934 using only indifference curves, with no need to measure utility. That was the promise of this whole topic: rankings are enough. From rankings alone, you have explained why demand curves slope down, and the rare case where they do not.

Key concepts in this lesson