Two ratios from the same budget
Take one household across two years. Its disposable income rises from $40,000 to $50,000, and its spending on goods and services rises from $35,200 to $42,000. Every figure you need for both propensities is in those four numbers.
The average propensity to consume divides total consumption by total income: 42,000 ÷ 50,000 = 0.84 in the second year, and 35,200 ÷ 40,000 = 0.88 in the first. The marginal propensity to consume divides the change in consumption by the change in income: 6,800 ÷ 10,000 = 0.68. Same household, same years — 0.84 on one measure and 0.68 on the other.
Worked example
Year 1: C = 35,200, Y = 40,000 → APC = 0.88
Year 2: C = 42,000, Y = 50,000 → APC = 0.84
ΔC = 6,800, ΔY = 10,000
MPC = 6,800 ÷ 10,000 = 0.68
Why the average sits above the margin
Part of what a household spends does not depend on its income at all: rent, food and the other basics that get paid for in a lean year too, out of savings or on credit. Economists call that part autonomous consumption, a, and write spending as C = a + bY, where b is the MPC. The household above fits that line exactly with a = $8,000 and b = 0.68.
Divide the line by income and the average propensity becomes APC = a/Y + b. The MPC is the constant b; the APC is b plus a fixed amount spread over income. So as long as autonomous consumption is positive, the APC is larger than the MPC, and the gap shrinks as income grows: 0.20 at $40,000, 0.16 at $50,000, 0.08 at $100,000.
Worked example
a = 8,000, b = 0.68
Y = 40,000: APC = 0.20 + 0.68 = 0.88
Y = 50,000: APC = 0.16 + 0.68 = 0.84
Y = 100,000: APC = 0.08 + 0.68 = 0.76
The saving side: APS and MPS
Every dollar of disposable income is either spent or saved, so each consumption ratio has a saving twin that completes it to one. The average propensity to save is saving over income, APS = 1 − APC; the marginal propensity to save is the change in saving over the change in income, MPS = 1 − MPC.
For the household, saving rose from $4,800 to $8,000. Its APS is 0.16 in the second year and its MPS is 3,200 ÷ 10,000 = 0.32. The two saving ratios differ for the same reason the two spending ratios do, in the opposite direction: the fixed spending floor drags the average saving rate down, so the APS sits below the MPS.
Worked example
APS = 8,000 ÷ 50,000 = 0.16
MPS = 3,200 ÷ 10,000 = 0.32
(0.84 + 0.16 = 1 and 0.68 + 0.32 = 1)
Only the marginal propensity belongs in a multiplier
A stimulus cheque, a tax cut and a pay rise are all changes in income, so the question they raise is marginal: how much of the extra dollar gets spent? The spending multiplier in the simplest Keynesian model is 1 ÷ (1 − MPC). With the household's MPC of 0.68 that is 1 ÷ 0.32 = 3.125.
Plugging the average propensity into the same formula gives 1 ÷ (1 − 0.84) = 6.25 — double the correct answer, and a common error in exam answers and in back-of-envelope policy arithmetic alike. The average tells you how a household lives on its income; only the margin tells you how it responds to a change in it. The marginal propensity to consume calculator reports both multipliers from any two observations.
Worked example
Right: 1 ÷ (1 − 0.68) = 3.125
Wrong: 1 ÷ (1 − 0.84) = 6.25
Tax multiplier: −0.68 ÷ 0.32 = −2.125
What real propensities look like
The US Bureau of Economic Analysis publishes an economy-wide APS every month as the personal saving rate. It was 3.0% in July 2026 and a record 31.8% in April 2020, when lockdowns cut spending far faster than income fell — so the economy's APC swung from about 0.68 to about 0.97 in a few years without any change in how people respond to an extra dollar.
Marginal propensities are measured from changes, and they depend on what kind of change it is. Using the randomised timing of the 2008 US stimulus payments, Parker, Souleles, Johnson and McClelland estimated that households spent about 12–30% of the payment on non-durable goods within three months, and 50–90% once cars and other durables are counted.
A raise and a windfall have different MPCs
Milton Friedman's permanent-income hypothesis explains why the same household shows different MPCs at different times. People spend out of the income they expect to last. A promotion changes that expectation and is largely spent; a one-off bonus barely changes it and is largely saved.
So when you estimate an MPC from two years of data, check what moved income between them. An MPC computed across a windfall year describes the response to windfalls, and applying it to a permanent tax change will understate how much of that change is spent.
Common mistakes
Using the APC where a marginal answer is needed.Multipliers, the effect of a tax cut and the response to a raise all run through the MPC; the APC overstates the multiplier whenever some spending is autonomous.Computing the MPC from levels.C ÷ Y is the average. The marginal propensity needs two observations and their differences.Measuring against gross income.Taxes cannot be spent, so both propensities are measured against disposable income.Mixing periods.Monthly spending against annual income gives ratios twelve times too small; both figures must cover the same period.Treating an MPC above 1 as a finding about behaviour.In household data it nearly always means something besides income changed — a car bought, a loan taken — between the two observations.