In the Solow model a higher saving rate always raises output per worker. But what people live on is consumption, and saving is what they give up to get capital. So there must be a best saving rate, and there is: the one that makes steady-state consumption as large as it can be. This lesson derives it, shows why for Cobb–Douglas it equals capital’s share, and explains the surprising possibility that an economy can save too much.
Saving buys capital, and capital makes output. But saving is also consumption not enjoyed. If an economy saves nothing, capital wears out and output goes to zero, so consumption is zero. If it saves everything, there is output but nothing to consume. Somewhere in between consumption is highest.
Make this exact. In the steady state, saving equals break-even investment, so consumption is what is left of output after break-even investment is paid for. Writing the break-even rate as n + g + δ (set g = 0 if there is no technical progress):
Both forms are the same number. The first shows why the shape is a hill: the first term rises with k* but flattens (diminishing returns), while the second rises in a straight line, so after some k* the line wins and consumption falls. The second form shows how the saving rate comes in: a higher s lowers the consumed share (1 − s) but raises f(k*).
Pick k* to make c* as large as possible. Differentiate c* = f(k*) − (n + g + δ)k* with respect to k* and set it to zero:
In words, the golden-rule capital stock is the one at which the net marginal product of capital equals the growth rate. The second derivative is f″ < 0, so this is a maximum.
Now find the saving rate that gets the economy there. At any steady state s f(k) = (n + g + δ)k, so s = (n + g + δ)k/f(k). Substitute the golden-rule condition n + g + δ = f′(k_gold):
For Cobb–Douglas, f′(k)k/f(k) = α at every k, so the golden-rule saving rate is simply α, whatever n, g and δ are. Those parameters only change how much capital that saving rate supports:
The top chart plots steady-state consumption against the saving rate. Move the saving rate and watch your position relative to the peak. Change α and the break-even rate and see the peak move. The lower chart answers a practical question: if this economy moved to the golden-rule saving rate tomorrow, what would happen to consumption on the way?
This economy saves less than the golden rule. Saving more would raise steady-state consumption, but consumption falls first, so there is a real trade-off between the present and the future.
Two sides of the hill are very different. If you start above the golden rule, moving down to it raises consumption at once and for ever: there is nothing to pay. If you start below it, moving up lowers consumption first, while capital is being built, and raises it later: there is a price to pay now for the gain later.
An economy is dynamically efficient if no way of changing its saving rate can raise consumption at every date. The golden rule divides the cases:
| Saving rate compared with α | Net return f′(k) − δ | Verdict | Moving to the golden rule |
|---|---|---|---|
| s < α (below) | above n + g | Efficient, but income is lower than it could be | Costs consumption now; gains it later |
| s = α | equal to n + g | The golden rule: maximum steady-state consumption | Nothing to do |
| s > α (above) | below n + g | Dynamically inefficient | Raises consumption at every date, at no cost |
Dynamic inefficiency is surprising: it means people are saving, but the saving earns less than the economy’s growth rate, so it would have been better to eat the capital. In the Solow model it can happen because the saving rate is simply assumed. It cannot happen in a model where households choose their saving, as Lesson 9 shows, except in overlapping-generations economies.
Is it a real problem? There is a simple test: at the golden rule investment equals capital income, so an economy is efficient if capital income exceeds investment. Abel, Mankiw, Summers and Zeckhauser applied this to the United States and other major OECD economies in 1989 and concluded that they appear to be dynamically efficient: they were saving less than the golden rule, not more.
Maximising consumption in the steady state is one objective, and not necessarily the right one. It ignores the generation that must give up consumption to build the capital, and it ignores impatience: people prefer consumption now. When households discount the future, the best long-run capital stock lies below the golden rule. That is the modified golden rule of Lesson 9.
Questions on the golden rule, from CUET-PG and UGC NET. The Delhi School’s golden-rule problems, including the economy with a production function F = L2/3K1/3, are in Lesson 14 when it opens.
Growth accounting, steady-state algebra, statement questions and matching, up to JNU, IIT JAM and UGC NET level.
This topic’s hardest questions, from the DSE entrance, are worked through in Lesson 14: The golden rule.
These are questions from the TestStop bank. A free account answers 10 a day, each with its worked solution; All-Access opens every set with no daily limit.
The golden rule saving rate is the saving rate that makes steady-state consumption per worker as large as possible. Saving too little leaves too little capital to produce with; saving too much uses up output on capital that is not needed. In a Solow economy with Cobb–Douglas production the golden-rule saving rate equals capital's share of income, α.
It is the capital per worker k_gold at which the marginal product of capital net of depreciation equals the growth rate of effective labour: f′(k_gold) − δ = n + g. Equivalently f′(k_gold) = n + g + δ. For Cobb–Douglas f(k) = k^α this gives k_gold = (α/(n + g + δ))^(1/(1 − α)).
At the steady state saving equals break-even investment, s·f(k) = (n + g + δ)k, so s = (n + g + δ)k/f(k). At the golden rule f′(k) = n + g + δ, so s = k f′(k)/f(k), which is the share of output paid to capital. With Cobb–Douglas that share is α. In words: the golden-rule economy saves exactly the income that capital earns.
An economy is dynamically inefficient if it has so much capital that the net return on capital, f′(k) − δ, is below the growth rate n + g. Saving rate above the golden rule is the cause. Everyone could consume more, today and for ever, by saving less and letting capital fall to the golden-rule level, so the extra saving is pure waste.
Not necessarily. The golden rule maximises consumption in the steady state only. It ignores the cost of getting there, since a country that saves less than the golden rule must cut consumption now to reach it, and it ignores impatience. When households discount the future, the optimal long-run capital stock is below the golden-rule level (the modified golden rule, Lesson 9).
They are different ideas. The golden rule (Phelps, 1961) is the saving rate that maximises steady-state consumption per head. The golden age (Joan Robinson) is a path of steady growth with full employment and a constant profit rate, in which the economy grows at the natural rate with capital fully used. Exam questions test the attribution.