Demonstration 1 of 4
Splitting a price rise into substitution and income
How much of the fall in food demand comes from the price change itself, and how much from lost purchasing power?
Compensation shifts the new budget line out until the consumer can afford the old bundle (Slutsky) or reach the old utility (Hicksian). The move from A to S is substitution; the move from S to C is the income effect.
Scroll sideways for the whole equation
x is food, y is a composite good with price one, p is the food price and m = 120 is income. Utility is sqrt(xy). Point A is the old choice, C the new one and S the choice after compensation.
Predict first. As the price rise gets larger, does the gap between the Slutsky and Hicksian substitution effects widen or close?
Choose an example
Scroll sideways for the whole figure
Constructed example: the chapter's hypothetical consumer (income 120, price 2 rising to 3); prices 2.5, 4 and 5 are added for comparison.
Calculated values
- Food before (A)
- 30.00
- Food after (C)
- 20.00
- Compensated income
- 150.00
- Compensated food (S)
- 25.00
- Substitution effect
- -5.00
- Income effect
- -5.00
- Total change
- -10.00
Ordinary demand falls from 120 / (2 x 2) = 30 to 120 / (2 x 3.0) = 20.00. Slutsky income = 120 + (3.0 - 2) x 30 = 150.00, so compensated food is 150.00 / (2 x 3.0) = 25.00. Substitution = 25.00 - 30 = -5.00 and income = 20.00 - 25.00 = -5.00; together -5.00 + (-5.00) = -10.00.
Worked steps
- x before = 120 / (2 x 2) = 30.00
- x after = 120 / (2 x 3.0) = 20.00
- Slutsky income = 120 + (3.0 - 2) x 30 = 150.00
- Compensated x = 150.00 / (2 x 3.0) = 25.00
- Substitution = 25.00 - 30 = -5.00
- Income effect = 20.00 - 25.00 = -5.00
Use the idea
When a price rise hits a household, separate the part of the response a cash transfer would undo from the part that reflects changed relative prices.
Where the conclusion applies
Cobb-Douglas utility sqrt(xy), a fixed income and a single price change. With other preferences the split differs, and the two compensation rules answer different questions.
Check your understanding: If the food price rises from 2 to 4, what are the Slutsky substitution and income effects?
Chapter 2 source: section "Slutsky equation".
Demonstration 2 of 4
Compensating vs equivalent variation
How much money is a fare increase worth to the rider, measured at new prices or at old ones?
The compensating variation is the extra income that restores the old utility at the new price; the equivalent variation is the income that, taken at old prices, leaves the consumer as badly off as the price rise does. Both are vertical gaps between expenditure curves and income.
Scroll sideways for the whole equation
p is the transit price, income is 100 and the composite good has price one. Utility is sqrt(xy), so the expenditure function e(p, u) = 2u sqrt(p) is the least spending that reaches utility u. u0 = 50 before the rise and u1 after it.
Predict first. Is CV always larger than EV for a price rise here?
Choose an example
Scroll sideways for the whole figure
Constructed example: the chapter's hypothetical transit rider (income 100, price 1 rising to 2); prices 1.5, 3 and 4 are added for comparison.
Calculated values
- New utility u1
- 35.36
- e(p1, 50)
- 141.42
- Compensating variation
- 41.42
- e(1, u1)
- 70.71
- Equivalent variation
- 29.29
- Hicksian bundle at p1
- (35.36, 70.71)
At p = 2.0 ordinary demand is 100 / (2 x 2.0) = 25.00 rides, so u1 = sqrt(25.00 x 50) = 35.36. CV = 2 x 50 x sqrt(2.0) - 100 = 141.42 - 100 = 41.42. EV = 100 - 2 x 35.3553 x sqrt(1) = 100 - 70.71 = 29.29. CV is the larger of the two for this price rise.
Worked steps
- x1 = 100 / (2 x 2.0) = 25.00
- u1 = sqrt(25.00 x 50) = 35.3553
- e(p1, 50) = 2 x 50 x sqrt(2.0) = 141.42
- CV = 141.42 - 100 = 41.42
- e(1, u1) = 2 x 35.3553 = 70.71
- EV = 100 - 70.71 = 29.29
Use the idea
When valuing a price change, say which counterfactual you mean: payment to accept the change (CV) or payment to avoid it (EV).
Where the conclusion applies
Utility sqrt(xy), income fixed at 100 and no other price changes. These are partial-equilibrium measures for one household, not a verdict on social welfare.
Check your understanding: If the price rises from 1 to 4, what are CV and EV?
Chapter 2 source: section "Hicksian-Marshallian demand duality".
Demonstration 3 of 4
When is an inferior staple Giffen?
How large must the budget share and how negative the income elasticity be for demand to rise with price?
Substitution always pushes demand down when price rises. For an inferior good the loss of purchasing power pushes it up, by the budget share times the income elasticity. The good is Giffen only when that push is larger.
Scroll sideways for the whole equation
eps_M is the ordinary own-price elasticity, -0.20 the compensated elasticity, s the staple's budget share and eta its income elasticity. The Slutsky equation in elasticities is eps_M = eps_H - s eta.
Predict first. At budget share 0.5, how negative must income elasticity be for Giffen behavior?
Choose an example
Scroll sideways for the whole figure
Constructed example: the chapter's hypothetical calibration (-0.20, share 0.70, eta -0.60 and -0.10); shares 0.3 and 0.5 and eta of -0.3 and -0.9 are added for comparison.
Calculated values
- Substitution term
- -0.20
- Income term -s x eta
- 0.42
- Ordinary elasticity
- 0.22
- Change in demand for a 10% price rise
- 2.2%
- Giffen?
- yes
- Threshold eta for Giffen
- below -0.29
eps_M = -0.20 - (0.70)(-0.60) = -0.20 + 0.42 = 0.22. The income term outweighs substitution, so demand rises with price: Giffen behavior. At this budget share, Giffen behavior needs income elasticity below -0.20 / 0.70 = -0.29.
Worked steps
- Income term = -(0.70) x (-0.60) = 0.42
- eps_M = -0.20 + 0.42 = 0.22
- A 10% price rise changes demand by 10 x 0.22 = 2.2%
- Threshold: -0.20 / 0.70 = -0.29
Use the idea
Before reading a positive price response as Giffen behavior, check that the good takes a large share of the budget and is strongly inferior.
Where the conclusion applies
A local, hypothetical calibration with a fixed compensated elasticity of -0.20. It is not an estimate from any study; Giffen behavior is local to a price and income range.
Check your understanding: With share 0.5 and eta = -0.3, is demand Giffen?
Chapter 2 source: section "Giffen paradox".
Demonstration 4 of 4
Add trips until marginal benefit falls below marginal cost
How many evening trips should run, and does a sunk depot cost change the answer?
Each trip is judged on its own forward benefit and cost. Trips run while the net increment is positive. A cost already paid that no choice can recover changes no increment, so it cannot change the count.
Scroll sideways for the whole equation
Marginal benefits of the five successive trips are 900, 700, 520, 360 and 220; marginal costs are 300, 380, 460, 540 and 620, less any cost cut per trip. The vectors are the net increments.
Predict first. With a cost cut of 100, how many trips run?
Choose an example
Scroll sideways for the whole figure
Constructed example: the chapter's hypothetical transit planner (cuts of 0 and 200 and the sunk depot); cuts of 100 and 300 are added for comparison.
Calculated values
- Marginal costs
- 300, 380, 460, 540, 620
- Net increments
- 600, 320, 60, -180, -400
- Trips run
- 3
- Net gain from trips run
- 980
- Depot cost
- left out (sunk)
Net increments are MB - MC = 600, 320, 60, -180, -400. Trip 3 passes: 520 - 460 = 60. Trip 4 fails: 360 - 540 = -180. So 3 trips run.
Worked steps
- MC after the cut of 0: 300, 380, 460, 540, 620
- Net increments: 600, 320, 60, -180, -400
- Last trip that passes: trip 3, net 60
- Trip 4 fails with net -180
Use the idea
When someone argues for one more unit to use an earlier investment, ask whether that unit's own benefit exceeds its own cost.
Where the conclusion applies
Benefits and costs per trip are known and ordered, and the depot cost cannot be recovered either way. A trip that preserves a licence or lowers future costs would add a forward benefit.
Check your understanding: With a cost cut of 300, how many trips run?
Chapter 2 source: section "Marginal analysis".