Consider a Solow model without technological progress so that the steady state $k^*$ occurs at $sf(k^*) = (n+\delta)k^*$ where $n$ is population growth rate, $\delta$ is capital depreciation rate and $s$ is the proportion of output that is saved and invested. Given that consumption per worker is $c = f(k) - sf(k)$, how does steady state consumption change if $n$ decreases?

Intuitively, output per worker rises so $c = (1-s)f(k)$ should also rise. However, if I try to solve for the change numerically I would do:

$c^* = f(k^*(s, n, \delta)) - (n+\delta)k^*(s, n, \delta)$ at the steady state.

So, $\frac{\partial c^*}{\partial n} = f'(k^*)\frac{\partial k^*}{\partial n} - k^* -(n + \delta)\frac{\partial k^*}{\partial n}$.


$\frac{\partial c^*}{\partial n} = [f'(k^*) - (n+\delta)]\frac{\partial k^*}{\partial n} - k^*$.

We know $\frac{\partial k^*}{\partial n}$ < 0 since a higher slope of the breakeven investment causes it to intersect with $sf(k^*)$ at a lower $k^*$. The sign of the term in the square bracket is ambiguous so $\frac{\partial c^*}{\partial n}$ itself is ambiguous.

There seems to be a contradiction between my intuition and the numerical analysis, so where have I gone wrong?


1 Answer 1


The direction of $\frac{\partial c^{*}}{\partial n}$ is not ambiguous.

An easy way to show this is taking derivative of $c^*=(1-s)f(k^*)$ so that $\frac{\partial c^{*}}{\partial n}=(1-s)f'\frac{\partial k^{*}}{\partial n}$ and because $f'>0$ and we can prove $\frac{\partial k^{*}}{\partial n}<0$ we thus have $\frac{\partial c^{*}}{\partial n}<0$.

Through this you can also easily see that $c^{*}$ will be monotone in $n$ and $\delta$ but it will not be monotone in $s$.

It is somehow harder to prove this from $\frac{\partial c^{*}}{\partial n}=\left[f^{\prime}\left(k^{*}\right)-(n+\delta)\right] \frac{\partial k^{*}}{\partial n}-k^{*}$

  • $\begingroup$ Thank you, this makes perfect sense. For record: The sign in the square term of the bracket is actually not ambiguous. However, I notice that the response function $\frac{\partial c^*}{\partial s} = [f'(k^*) - (n + \delta)]\frac{\partial k^*}{\partial s}$. If we were to reason the same way, wouldn't steady state consumption be ever-increasing in savings, and so no such thing as the golden rule maximum? $\endgroup$ Sep 29, 2021 at 11:36
  • $\begingroup$ @PanhabothK Sorry, you are correct. I made a mistake in my initial answer. I find it is harder than I thought to prove the sign from the equation you are dealing with. But I show that you can easily prove this in an alternative way. And it helps you see the difference between $n$ and $s$. $\endgroup$ Sep 29, 2021 at 13:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.