I have the following optimization problem:

$\max_{\{c_t, s_{t+1}\}} \Pi_{t=0}^\infty c_t^{\beta^t}$

$\text{subject to } \space c_t + s_{t+1} = y_t + (1 + r) s_t \text{ and } s_0 = 0$

How do I find the optimal consumption level for each time period $t$?

Here's what I have so far:

Let $p_t = \frac{1}{(1 + r)^t}$

Then, from the budget constraint $p_{t+1} s_{t+1} - p_{t+1} s_t = p_{t+1}(y_t - c_t)$.

If we sum over $T$ periods: $(p_1 s_1 - p_0 0) + (p_2 s_2 - p_1 s_1) + ... (p_{T+1} s_{T+1} - p_T s_T) = \sum_{t=0}^\infty p_{t+1}(y_t - c_t)$

If I use a natural logarithm transformation on the utility function and assume $\lim_{T\to\infty} p_T s_T = 0$ (no ponzi schemes allowed), I can rewrite the optimization problem as:

$\max_{\{c_t\}} \sum_{t=0}^\infty \beta^t\ln(c_t)$

$\text{subject to} \sum_{t=0}^\infty p_{t+1} c_t = \sum_{t=0}^\infty p_{t+1} y_t$

Hence, my lagrangian function should look like (unless I'm missing something):

$\mathcal{L} = \sum \beta^t\ln(c_t) - \lambda(\sum_{t=0}^\infty p_{t+1} c_t - \sum_{t=0}^\infty p_{t+1} y_t) + \mu c_t$

How do I get $c_t^* \space \forall \space t$?


1 Answer 1


Below is the sketch of the solution.

From the Lagrangian $$\max _{\left\{c_{t}, s_{t+1}\right\}} \Pi_{t=0}^{\infty} c_{t}^{\beta^{t}} - \Pi_{t=0}^{\infty}\lambda_t(c_{t}+s_{t+1}-y_{t}-(1+r) s_{t}) \\ \text{s.t.} \ s_0 = 0, c_t > 0 $$, you can get the Euler equation $$\frac{\beta_{t+1}c_{t+1}^{\beta_{t+1}-1}}{\beta_{t}c_{t}^{\beta_{t}-1}}= \frac{\beta_{t+1}\left[y_{t+1}+(1+r)s_{t+1}-s_{t+2}\right]^{\beta_{t+1}-1}}{\beta_{t}\left[y_{t}+(1+r)s_{t}-s_{t+1}\right]^{\beta_{t}-1}} = \frac{1}{1+r}$$, and then you can write $s_{t}^{*} \forall t>1$ as a function of $s_1$ with the initial condition $s_0$. By using the transversality condition as a terminal condition: $$\lim _{t \rightarrow \infty} \lambda_ts_{t+1}= \lim _{t \rightarrow \infty} {\beta_{t}\left[y_{t}+(1+r)s_{t}-s_{t+1}\right]^{\beta_{t}-1}}s_{t+1} =0$$, you can solve for $\left\{s_{t}\right\}_{t=1}^{\infty}$ forwards by guessing an initial value for $s_1$ and yielding the sequence that does not violate the TVC or the nonnegativity constraint on consumption. And once you get $s_{t}^{*} \forall t$ you get $c_{t}^{*} \forall t$.

  • $\begingroup$ Sorry, I do not understand this answer. Are you saying $s_0 = 0$ implies that the agent doesn't save any money in $t=0$? That is, $c_0 + s_1 = c_0 = y_0$? $\endgroup$
    – Arturo Sbr
    Commented Sep 15, 2021 at 13:56
  • $\begingroup$ @ArturoSbr Sorry I made some mistakes in the initial answer. I think I have corrected it. $\endgroup$ Commented Sep 16, 2021 at 1:00

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.