class: center, middle, inverse, title-slide # Solow Model (continued) ## Macroeconomics ### Kenji Sato ### 2018-4-19 --- ## Quiz Consider a Solow economy. How fast does the wage grow along the balanced growth path? -- Recall `$$w = A\left[f(k^*) - k^*f'(k^*)\right]$$` --- class: center, middle # Transition and comparative dynamics --- ## Sudden increase of s <img src="solow_model/solow_model.005.jpeg" style="display: block; margin: auto;" /> --- ## Sudden change of `\(C/AL\)` and `\(S/AL\)` <img src="solow_model/solow_model.006.jpeg" style="display: block; margin: auto;" /> --- ## Transition to new steady state <img src="solow_model/solow_model.007.jpeg" style="display: block; margin: auto;" /> --- ## New steady state <img src="solow_model/solow_model.008.jpeg" style="display: block; margin: auto;" /> --- ## Welfare gain in the long run Whether or not the consumption becomes larger than the original steady-state value in the long run depends on the initial saving rate. The steady-state value for consumption per unit of effective labor `$$c^* = \frac{C}{AL}$$` is maximized when `$$f'(k^*) = \delta + g + n$$` --- ## Welfare gain in the long run Let `\(s_G\)` be the saving rate that the corresponding steady state `\(k^*_G\)` satisfies `$$f'(k^*_G) = \delta + g + n$$` * `\(s_G\)` and `\(k^*_G\)` and the Golden-Rule * If an economy in the steady state with `\(s_1 < s_G\)` experiences a hike in the saving rate to `\(s_2 < s_G\)`, then the consumption per unit of effective labor initially drops but eventually gets larger than the original steady state value --- ## Dynamic inefficiency for `\(s > s_G\)` Let's do a similar exercise with `\(s > s_G\)`. Suppose that the economy's saving rate is `\(s_1 > s_G\)` and it's on the balanced growth path. The saving rate suddenly drops to `\(s_2 > s_G\)`. The economy increases the amount of consumption immediately and continue to enjoy larger consumption. This means that if the consumers are allowed to chose their saving rate, they wouldn't choose `\(s > s_G\)` because by reducing `\(s\)` they can consume more at any point in time. --- ## Exercise What happens in response to the following changes of parameter? * Increase/decrease of `\(\delta\)` * Increase/decrease of `\(g\)` * Increase/decrease of `\(n\)` Describe * long-run growth and * transition path to the new steady state --- class: center, middle # some empirical observations --- ## Cobb-Douglas production function Let's suppose `$$\begin{aligned} F(K, AL) = K^\alpha (AL)^{1-\alpha}, \quad 0 < \alpha <1 \end{aligned}$$` Corresponding intensive form: -- `$$\begin{aligned} f(k) = k^\alpha \end{aligned}$$` -- The steady state for the Solow model -- `$$\begin{aligned} k^* = \left( \frac{s}{\delta + g + n} \right)^{\frac{1}{1 - \alpha}} \end{aligned}$$` --- ## What's `\(\alpha\)`? `\(\alpha\)` is the capital share! Exercise: Prove that `$$\begin{aligned} \alpha = \frac{(r + \delta) K}{Y} \end{aligned}$$` --- class: center, middle # Convergence --- ## Growth rate in transition Consider a transition path. The growth rate of output `$$\begin{aligned} \frac{\frac{d}{dt}\left( Y/L \right)}{Y/L} &= \frac{\frac{d}{dt}\left( Ak^\alpha \right)}{Ak^\alpha} = g + \alpha \frac{\dot k}{k} \end{aligned}$$` Since the second term is large and positive when `\(k\)` is small, relatively poor countries tend to grow rapidly. --- ## Testable prediction If the Solow model is correct, the growth rate declines as an economy matures. This trend should be widely observed and we call this situation "convergence." You can estimate the following model to test if the average growth rate is negatively correlated with the initial wealth. `$$\begin{aligned} \ln \left(Y_t/L_t \right) - \ln \left( Y_0/L_0 \right) = \beta_0 + \beta_1 \ln(Y_0 / L_0) \end{aligned}$$` --- class: center, middle # Cross-country income differences --- ## Per-capita income Per-capita income in the steady state is `$$\begin{aligned} \frac{Y}{L} &= A \left( \frac{s}{\delta + g + n} \right)^{\frac{\alpha}{1 - \alpha}}\\ &= A(0)e^{gt} \left( \frac{s}{\delta + g + n} \right)^{\frac{\alpha}{1 - \alpha}} \end{aligned}$$` Taking log, `$$\begin{aligned} \ln \frac{Y}{L} = \text{const.} + gt +\frac{\alpha}{1 - \alpha} \ln s -\frac{\alpha}{1 - \alpha} \ln (\delta + g + n) \end{aligned}$$` --- ## Testable prediction `$$\begin{aligned} \ln \frac{Y}{L} = \text{const.} + gt +\frac{\alpha}{1 - \alpha} \ln s -\frac{\alpha}{1 - \alpha} \ln (\delta + g + n) \end{aligned}$$` The Solow model predicts that, other things being equal, one percent increase in `\(s\)` (e.g. from 30% to 30.3%) results in an increase of `\(Y/L\)` by `\(\frac{\alpha}{1 - \alpha}\times 100\)` percent. Common wisdom suggests `\(\alpha \simeq 1/3\)` and the elasticity of output with respect to the saving rate should thus be `$$\frac{\alpha}{1-\alpha} \simeq \frac{1}{2}$$`