- Indifference curves and budget constraints
- Total and marginal utility
- Optimisation — maximum and minimum problems · coming soon
- Income and substitution effects
- Multivariate optimisation and Lagrangians · coming soon
- Corner solutions and Kuhn-Tucker conditions · coming soon
From searching to solving
In the last lecture, you found Amara's best bundle by sliding along her budget line: 15 cups of coffee and 20 cups of tea a week. That was a search, and it worked because the explorer drew the curves for you.
An exam question gives you no explorer. It gives you a utility function, two prices and an income, and asks for the best bundle. So the picture has to become equations.
The last lecture also promised to deal with cases where the best bundle has none of one good. There, the tangency rule gives a wrong answer, or no answer at all.
So there are two jobs: turn the tangency into algebra, and learn when the tangency is not the answer. The second is the hard one. Spend your time there.
Two conditions, two equations
Amara's problem has two unknowns: C, the cups of coffee, and T, the cups of tea. To find two unknowns, we need two equations. The picture gives exactly two.
- She is on the budget line. Because more is better, she never leaves money unspent. So .
- The curve touches the line. At the tangency, her marginal rate of substitution equals the price ratio: .
The first equation says where she can be. The second says where on that line she wants to be. Neither is enough alone.
To use the second equation, we need her preferences as a formula. The explorer in the last lecture used invented preferences, which it did not show you:
This utility function gives every bundle a number; a higher number means a preferred bundle. As the last lecture showed, only the order of the numbers matters.
Now find the marginal utilities. One more cup of coffee changes U from to . So the extra cup adds exactly T. In the same way, one more cup of tea adds exactly C.
Now solve. Set the MRS equal to the price ratio:
So she spends the same on each good: 3T dollars on tea and 4C dollars on coffee. Put this into the budget: , so and . Then , so .
That is the explorer's bundle, (15, 20), found in four lines. Her utility there is 300. Check two neighbours on the budget line: (12, 24) gives 288, and (18, 16) gives 288. Both are lower. ✓
Why the condition must hold: follow the last dollar
Why must the best bundle have MRS equal to the price ratio? The first lecture in this topic gave the reason in money. Take a bundle that is not best, (6, 32).
There, MUC = 32 and MUT = 6. Coffee costs $4, so a dollar spent on coffee buys about 32 ÷ 4 = 8 utils. Tea costs $3, so a dollar spent on tea buys about 6 ÷ 3 = 2 utils. Move one dollar from tea to coffee. She loses about 2 utils and gains about 8. She is better off, so (6, 32) was not the best.
This keeps working until a dollar buys the same in both goods. Then no move helps. That is the equimarginal principle:
Maximise U(X, Y) subject to .
If the best bundle has some of both goods, then (1) all income is spent, and (2) , which is the same as .
Notice the words "if the best bundle has some of both goods". Most mistakes in this topic come from forgetting them.
The demand curve falls out of the algebra
For U = C × T, Amara spends exactly half her income on each good. This is not luck. The condition 3T = 4C said "spending on tea equals spending on coffee", and it would say the same at any prices. So her coffee spending is always , and
This is her demand for coffee. It came from her preferences and her budget, with no extra assumption. Change the price of coffee below and watch the demand curve being built, one price at a time.
Amara has $120 and tea costs $3. Change the price of coffee. The top graph shows her budget line and best bundle. Each price you visit adds a point to her demand curve below.
Presets:
Amara's best bundle at five prices of coffee, with income $120 and tea at $3 (invented preferences U = C × T).
| Price of coffee ($) | All-coffee end (cups) | Best coffee (cups) | Best tea (cups) | Spent on coffee ($) |
|---|---|---|---|---|
| 2 | 60 | 30 | 20 | 60 |
| 3 | 40 | 20 | 20 | 60 |
| 4 | 30 | 15 | 20 | 60 |
| 6 | 20 | 10 | 20 | 60 |
| 8 | 15 | 7.5 | 20 | 60 |
Every price gave one best bundle, and every best bundle gave one point on a demand curve that slopes down. The law of demand, stated in earlier lectures as an observation, has been built from a consumer's choices.
Be careful with what this proves. It proves the law of demand for these preferences only. 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.
A family of preferences, and the shares they fix
U = C × T is one member of a family called Cobb-Douglas utility: , with positive powers a and b. The same steps give , and the tangency gives
So the consumer spends a fixed share of income on X, whatever the prices. For U = C × T, a = b = 1, so the share is a half. Remember this prediction. The real case below tests it.
When the tangency is not the answer
Kofi buys rice (invented example). Long-grain rice (X) gives him 3 utils per kilogram, and short-grain rice (Y) gives him 2 utils per kilogram. He sees them as perfect substitutes: U = 3X + 2Y. Long-grain costs $2 a kilogram, short-grain costs $1, and he has $20. Before reading on, commit to an answer.
U = 3X + 2Y, long-grain (X) $2 per kg, short-grain (Y) $1 per kg, income $20. Which bundle is best for Kofi?
Kofi's straight indifference curves are flatter than his budget line everywhere, so his best bundle is the corner with only short-grain rice.
This is a corner solution: the best bundle sits at one end of the budget line, with none of one good. The tangency rule failed because its condition, "some of both goods", is false here.
Look at the slopes to see why. Kofi would give up 1.5 kg of short-grain for one more kilogram of long-grain (his MRS). The market charges 2 kg. So he keeps swapping long-grain for short-grain until he has none left to swap. On the graph, his indifference curves are flatter than the budget line everywhere, so the best point is the corner on the short-grain axis. If his curves were steeper than the line everywhere, the corner would be on the long-grain axis.
The test at a corner uses the same comparison, now as an inequality. If Kofi buys only Y, then a dollar on Y must be worth at least as much as a dollar on X, even at the corner:
Corners are not only for perfect substitutes. For a vegetarian, a dollar on meat is always worth less than a dollar elsewhere. Every good you never buy is a corner in your own choices.
You find a bundle on the budget line where the MRS equals the price ratio. Can you be sure it is the best bundle?
Show the answer
Only if the indifference curves bow in towards the origin (convex). The tangency condition finds the point where the curve and the line have the same slope. That happens at a best point, but it can also happen at a worst one.
With curves that bow outwards, the consumer prefers extremes, so the tangency is the worst point and a corner is best. This is why the last lecture's "preference for balanced bundles" matters. It is what turns the tangency into a maximum.
The same answer with calculus
If you have met differentiation, there is a second method, named after Joseph-Louis Lagrange. Build one function, , from the utility and the budget, with a new letter λ (lambda):
Then differentiate with respect to each letter in turn, treating every other letter as a constant. (This is called a partial derivative, written with ∂.) Set each result to zero:
The third equation is the budget. Dividing the first by the second gives the tangency. So the method gives the same two conditions, plus something new. From the first equation, λ = MUX ÷ PX: λ is the marginal utility of one more dollar of income.
Check it for Amara. At (15, 20), λ = MUC ÷ PC = 20 ÷ 4 = 5. Her best utility at income I is I² ÷ 48: 300 at $120 and about 305.02 at $121. One more dollar adds about 5 utils ✓. Like the tangency rule, this method misses corners unless you check them separately.
Worked examples
U = XY, income $100, PX = $2, PY = $5. Find the best bundle.
Marginal utilities. MUX = Y and MUY = X, so MRS = Y ÷ X.
Tangency. Y ÷ X = 2 ÷ 5, so 5Y = 2X. Spending on Y equals spending on X.
Budget. 2X + 5Y = 100 becomes 2X + 2X = 100, so X = 25. Then 5Y = 50, so Y = 10.
Check. 2 × 25 + 5 × 10 = 50 + 50 = 100 ✓. U = 250. The neighbour (20, 12) gives 240 and (30, 8) gives 240. Both lower ✓.
U = X²Y, income $90, PX = $3, PY = $2.
Marginal utilities. Using derivatives with the other letter held constant, MUX = 2XY and MUY = X². So MRS = 2XY ÷ X² = 2Y ÷ X.
Tangency. 2Y ÷ X = 3 ÷ 2, so 4Y = 3X, which means 2 × (2Y) = 3X. Spending on X (3X) is twice spending on Y (2Y).
Budget. Two thirds of $90 go on X: $60 buys X = 20. One third, $30, buys Y = 15. This matches the share rule a ÷ (a + b) = 2 ÷ 3.
Check. MRS = 2 × 15 ÷ 20 = 1.5 = 3 ÷ 2 ✓. U = 400 × 15 = 6000. On the budget line, (18, 18) gives 5832 and (22, 12) gives 5808. Both lower ✓.
U = 3X + 2Y, PX = $1, PY = $1, income $20.
Compare value per dollar. MUX ÷ PX = 3, and MUY ÷ PY = 2, at every bundle. X always wins.
Answer. All income on X: X = 20, Y = 0, U = 60. Only Y would give 40.
Change one price. If PX rises to $2, a dollar on X gives only 1.5, and the corner switches to Y = 20. With perfect substitutes, demand jumps from one corner to the other instead of sliding.
Check. At PX = $1, the MRS (1.5) is more than the price ratio (1), so the curves are steeper than the line and the corner is on the X axis ✓.
A real case: does spending stay fixed when a price rises?
The policy. On 1 January 2014, Mexico introduced an excise tax of 1 peso per litre on drinks with added sugar, about a 10% rise in their price. Studies found that the tax was passed on in full: prices rose by about 1 peso per litre through 2014 (Colchero and others, BMJ, 2016).
The prediction. With Cobb-Douglas preferences, spending on these drinks stays a fixed share of income. A 10% price rise with spending unchanged means the quantity falls to 1 ÷ 1.1 of its old level, about 9.1% less. Spending on everything else stays the same too.
What happened. Colchero and her colleagues used the shopping records of more than 6000 households in 53 Mexican cities. They compared 2014 purchases with what the trend before the tax predicted. Purchases of taxed drinks were about 6% lower on average over 2014. The fall grew during the year and reached about 12% by December. Poorer households cut back more: about 9% on average, and about 17% by December. Purchases of untaxed drinks, mainly bottled water, rose by about 4%.
Verdict. Over the whole year, quantity fell by less than 9.1%. So spending on taxed drinks rose, by roughly 1.1 × 0.94 ≈ 1.03, about 3%. By December, the fall was larger than 9.1%. And spending moved towards water, which Cobb-Douglas says should not happen. So the fixed-share prediction failed: the response grew with time, and buyers switched to a close substitute. The method survives, because households did respond to prices. What failed is one convenient formula for preferences. Economists treat a utility function as a hypothesis, and data like these test it.
"Set MRS equal to the price ratio" only works when the best bundle has some of both goods. Always compare MU per dollar at the corners too. If the equation has no solution, or gives a negative quantity, the answer is a corner.
The budget line's slope is −PX ÷ PY, negative, with the price of the good along the bottom on top. The MRS is a positive number, so the condition compares two positive numbers: MRS = PX ÷ PY.
MRS = price ratio gives a relationship such as 5Y = 2X, not a bundle. Many bundles satisfy it. Only the budget equation picks the one the consumer can afford with all her income.
Work each one on paper first. Open a hint only when you are stuck.
-
U = XY, income $60, PX = $3, PY = $2. Find the best bundle and its utility.
Hint 1
MRS = Y ÷ X. Set it equal to 3 ÷ 2.
Hint 2
The tangency says 3X = 2Y: equal spending, $30 on each good.
Full answer
Y ÷ X = 3 ÷ 2 gives 2Y = 3X, so spending is equal. $30 on X buys X = 10. $30 on Y buys Y = 15. U = 150.
Check: 3 × 10 + 2 × 15 = 60 ✓. MRS = 15 ÷ 10 = 1.5 = 3 ÷ 2 ✓.
-
U = XY², income $90, PX = $1, PY = $3. Find the best bundle, and check it with the MRS.
Hint 1
Here a = 1 and b = 2. What share of income goes on X?
Hint 2
MUX = Y² and MUY = 2XY, so MRS = Y ÷ (2X).
Full answer
The share on X is 1 ÷ 3, so $30 buys X = 30. The other $60 buys Y = 20.
Check: MRS = Y ÷ (2X) = 20 ÷ 60 = 1/3, and PX ÷ PY = 1/3 ✓. Budget: 30 + 3 × 20 = 90 ✓.
-
Lin's utility is U = 2X + 5Y. PX = $4, PY = $8, and she has $40. (a) Find her best bundle. (b) Above what price of Y would she switch to buying only X?
Hint 1
Compare utils per dollar: 2 ÷ 4 and 5 ÷ 8.
Hint 2
She switches when 5 ÷ PY falls below 2 ÷ 4 = 0.5.
Full answer
(a) A dollar on X gives 0.5 utils and a dollar on Y gives 0.625. So she spends everything on Y: Y = 40 ÷ 8 = 5, X = 0, U = 25. Only X would give X = 10 and U = 20.
(b) 5 ÷ PY = 0.5 when PY = 10. Above $10, X gives more per dollar, and she buys only X. At exactly $10 she is indifferent between every bundle on the line.
Check: at PY = $10, only Y gives 5 × 4 = 20 utils, the same as only X ✓.
-
A household spends 5% of its income on soft drinks and has Cobb-Douglas preferences. (a) A tax raises the price of soft drinks by 10%. By what percentage does the quantity bought fall? What happens to spending on soft drinks? (b) In Mexico in 2014, the quantity fell by about 6% on average. What does this say about spending? (c) Give one reason why the fall was larger by December than over the year as a whole.
Hint 1
With Cobb-Douglas preferences, the spending share stays at 5%, so spending does not change.
Hint 2
Same spending at 1.1 times the price buys 1 ÷ 1.1 of the old quantity.
Full answer
(a) Spending stays at 5% of income. The quantity becomes 1 ÷ 1.1 ≈ 0.909 of what it was, a fall of about 9.1%.
(b) Spending becomes 1.1 × 0.94 ≈ 1.034 times its old level: about 3% higher. So the households spent a bigger share of income on soft drinks than before, which Cobb-Douglas does not allow. Their demand responded less than the model predicts.
(c) Habits take time to change, and people need time to find substitutes such as bottled water. A model of one choice at one moment has no time in it, so it cannot show a response that grows.
Check: by December the fall was about 12%, and 1.1 × 0.88 ≈ 0.97, so by then spending had fallen by about 3%. The data crossed the model's prediction during the year ✓.
Choosing the best option within a limit describes a firm choosing workers and machines, a student dividing hours between subjects, and a government dividing a budget. The answer always has two parts: use the whole limit, and make the last unit equally valuable in every use. When that cannot happen, the choice is a corner.
The Cobb-Douglas formula has its own history. It is named after the economist Paul Douglas and the mathematician Charles Cobb, who used it in 1928 to describe how American factories combined labour and capital. Economists later borrowed it for consumers because it is easy to solve. The Mexican data show the cost of that convenience. A formula chosen because it is easy must still be tested against what people do.
- An interior best bundle is found from two equations: the budget line, and the tangency condition MRS = PX ÷ PY.
- The tangency condition is the same as equal marginal utility per dollar in both goods, because otherwise moving a dollar would raise utility.
- Solving the consumer's problem at different prices gives the demand curve, so demand can be built from preferences and a budget.
- With Cobb-Douglas utility U = XaYb, the consumer spends a fixed share a ÷ (a + b) of income on X.
- A corner solution, with none of one good, happens when one good gives more utility per dollar even at the corner, and the tangency rule then gives a wrong answer.
- A tangency is the best bundle only when indifference curves bow in towards the origin, and the Lagrange multiplier λ measures the extra utility from one more dollar.