Each bar is a separate economy following the discrete Solow growth law for capital per worker k:
k(t+1) = k(t) + s·k(t)^α − (δ+n)·k(t)
Steady state: k* = ( s / (δ+n) )^(1/(1−α))
Because α < 1, capital has diminishing returns: an economy with little capital per worker gets a bigger output boost from each extra unit of capital than a capital-rich one, so it grows faster in percentage terms. Plotted against the whole group, this produces β-convergence — the negative relationship between an economy's distance from its steady state and its growth rate, shown live in the scatter panel on the right.
- Savings rate s and capital share α — set the production/investment side of the Solow model; both raise the steady-state k*.
- Depreciation + population growth (δ+n) — capital lost to wear and dilution across a growing workforce; a higher value lowers k*.
- Heterogeneous savings — splits the economies into two groups with different savings rates. With one shared rate, every bar converges to the same height (absolute convergence). With two rates, each group converges to its own steady state (conditional convergence) — the real-world pattern, since countries differ in savings, institutions and technology.
- The dispersion (CV) readout is the coefficient of variation of k across all economies — it should fall toward zero under absolute convergence, and toward a nonzero plateau under conditional convergence.
- The bottom strip plots average capital per worker over simulated time against the dashed steady-state target(s), so you can watch the whole trajectory, not just a snapshot.
Real-world relevance: this is the mechanism behind why poorer countries with similar policies and institutions tend to catch up faster than rich ones (e.g. postwar Japan/Germany vs. the US), and why countries with very different savings/institutional quality settle at persistently different income levels instead of converging to one another.