Keynesian Fiscal Multiplier: Spending Rounds, Leakages and Crowding Out

A cascade of Blender-modelled basins: each is one spending round, and the four leakage buckets must always add up to the original package. Drag to orbit.

Switch to 2D
Controls

Presets

Parameters

Quality

Camera

Model checks

The science

Setup. The government injects a package of G0 = $100: a share 1 - s is purchases of goods and services (income immediately), a share s is a lump-sum tax cut (households first save and import part of it). Round-0 domestic income is y0 = G0[(1 - s) + s c (1 - m)].

One round. Income y is split into tax t y, saving (1 - c)(1 - t) y, imports c(1 - t) m y and a crowding-out term φ y (a larger deficit raises interest rates and lowers private investment; in an IS-LM economy φ = d k/h with investment sensitivity d, money demand parameters k, h). What remains, q y with q = c(1 - t)(1 - m) - φ, is spent at home and becomes the next round's income.

Multiplier. The rounds form a geometric series: ΔY = y0/(1 - q), so the multiplier on the package is k = ΔY/G0. In the textbook closed economy (t = m = φ = 0, spending only) this is 1/(1 - c) = 1/MPS. A tax cut has the smaller multiplier c(1 - m) times the spending multiplier, because part of the cut is saved first. Taxes recouped, t ΔY, reduce the net deficit cost of the package.

Checks. The cascade is iterated round by round. It is compared with the closed form, with the accounting identity (taxes + saving + imports + crowding out must equal the package), with a numerical derivative of the structural model solved by bisection, with a Monte Carlo of random dollars, with the 90 % convergence length ln 0.1/ln q, and with the spending-versus-tax-cut ratio.

Simplifications. This is the Keynesian-cross model with a lag of one round per step. It assumes slack in the economy and fixed prices; at full employment real multipliers are much smaller. We ignore forward-looking households (Ricardian equivalence), the central bank's reaction, wealth effects, supply-side responses and the financing of the deficit; crowding-out is a single constant fraction and is capped at c(1 - t)(1 - m) so that q ≥ 0. Empirical multiplier estimates range from about 0 to 2 depending on the state of the economy and the monetary regime (Ramey 2019, Blanchard and Leigh 2013), so the large values at the slider extremes are illustrations of the algebra, not forecasts.

Loading model...
Loading Blender model...
-
-
Made with 🟠 Blender · Three.js