Consumption: Keynes, Friedman and Modigliani

Keynes’s “fundamental psychological law”: as income rises, consumption rises, but by less than income (Hyderabad 2010). He also expected the rich to save a larger share. The straight line C = a + bY, with a > 0 and 0 < b < 1, captures both. This is the absolute income hypothesis: consumption depends on current disposable income (CUET 2019).

Three consequences, each asked many times: the MPC is constant; the APC is always above the MPC (because a/Y > 0); and as income rises the APC falls towards the MPC. So the share saved rises with income. GATE 2023 asked for the APC formula itself, a/Y + b.

Keynes’s function fitted short-run and cross-section data: richer households, and boom years, had lower APCs. But Simon Kuznets’s long-run estimates for the United States showed the APC roughly constant over many decades, even as income multiplied. If the APC fell with income, saving should have become an ever larger share of income. It didn’t.

A falling APC in the short run with a stable one in the long run is the consumption puzzle. The three theories that follow are three ways of resolving it.

James Duesenberry argued that people care about their income relative to others, and relative to their own past peak.

• The demonstration effect. Households copy the consumption of those around them, so a family’s saving depends on its place in the distribution, not its absolute income. When everyone gets richer together, the APC need not fall. Duesenberry coined the phrase (UGC NET 2024).
• The ratchet effect. Once used to a standard of living, people defend it. Consumption rises easily with income but falls reluctantly when income falls: households find it easier to adjust to rising incomes than to falling ones (Hyderabad 2013). In recessions the APC rises.

Arthur Smithies and James Tobin explained the long-run stability differently, by an upward drift of the short-run consumption function over time; UGC NET matched “drift theory” to them in 2026.

Irving Fisher’s model is the foundation of the modern theories. A consumer lives two periods and can save or borrow at interest rate r. Whatever she doesn’t consume now is worth (1 + r) times as much later, so her choices lie on one budget line:

She discounts future utility at rate ρ, her impatience. At the best choice, the gain from shifting a rupee to the future (1 + r) just balances her impatience (1 + ρ), weighted by how much she dislikes uneven consumption (σ). That condition is the Euler equation.

Two lessons from the lab. First, consumption depends on lifetime wealth, not this year’s income: move income between the periods keeping W fixed and nothing changes. Second, if she can’t borrow, consumption follows current income after all. Borrowing constraints are one reason Keynes’s function fits some households well.

Franco Modigliani, with Richard Brumberg and later Albert Ando, applied Fisher’s idea to a whole life. Income is low in youth, peaks in middle age and stops at retirement; people want smooth consumption. So they borrow when young, save hard in their working years and run their savings down in retirement (Hyderabad 2017). Consumption depends on wealth and expected lifetime labour income, both in present value (UGC NET 2026).

It resolves the puzzle: in a single year, someone whose income is temporarily high (a middle-aged earner) has a low APC; across the economy, the age mix is stable, so the aggregate APC is stable too.

Milton Friedman (1957) split measured income into permanent income, what a household expects to earn on average over the long run, and transitory income, the lucky or unlucky deviation this year. People consume a steady fraction of permanent income. A rise in permanent income raises permanent consumption (Hyderabad 2011); a transitory windfall is mostly saved.

So the marginal propensity to consume out of a permanent change in income is close to one, and out of a temporary change close to zero. That is the cleanest contrast with Keynes, and JNU 2018’s answer: a temporary windfall is consumed under a Keynesian function but saved under the permanent income hypothesis.

Tests of the theory found two anomalies, asked by UGC NET in 2023: excess sensitivity, consumption responding too much to predictable or temporary income changes, and excess smoothness, consumption responding too little to changes in permanent income.