# Review: does costless income equalization create an attracting boom–bust cycle? **Short answer:** The arithmetic checks out. The no-redistribution economy settles at total income 6. Full equalization turns the system into a 2-cycle between incomes (1,1) and (3,3). That cycle is locally attracting and averages total income 4, a one-third loss. The construction is deliberately contrived, and the cycle is local rather than reached from every start. ## 1. Precise restatement with hidden assumptions There exists a deterministic two-household economy with the following features: - **Household B** earns 1 every period, regardless of anything. - **Household A** earns, next period, f(y) = 5·exp[−k(y−1)], with k = ln5/2. Here y is B's *current disposable* income, not B's earnings. This is a reduced-form, lagged, decreasing externality. It is not derived from optimizing behaviour. - **Without redistribution**, disposable income equals earnings. The earnings vector converges to the unique steady state (5,1), with total income 6. - **With redistribution**, all earnings are pooled and split equally each period. There is no deadweight cost and no behavioural response other than through f. The production rule f is invariant to policy. Under these assumptions: - (a) The disposable-income dynamics have a 2-cycle {1,3}. - (b) This cycle is locally asymptotically stable. - (c) Total income along the cycle alternates between 2 and 6. Its long-run average is 4, which is ⅓ below 6. "Attracting" means locally attracting. "Average" means the time average along the limit cycle. ## 2. Step-by-step check and independent re-derivation **The two function values.** - f(1) = 5·e⁰ = 5. - f(3) = 5·exp(−2k) = 5·exp(−ln 5) = 1. ✓ **No-redistribution economy.** - B's disposable income is 1 every period, so A's earnings from period 1 onward are f(1) = 5, whatever A started with. - The steady state (5,1) is unique: the map (a, b) ↦ (f(1), 1) is constant after one step. - Total income is 6. "Within two periods" is, if anything, generous; one update suffices. ✓ **Redistribution economy.** - Each household receives half of total earnings f(y_prev) + 1. - So the update is g(y) = (f(y) + 1)/2. - g(1) = 3 and g(3) = 1, so {1,3} is a 2-cycle. ✓ **Stability of the cycle.** - The derivative is g′(y) = f′(y)/2 = −k·f(y)/2. - g′(1) = −5k/2 and g′(3) = −k/2. - Their product is 5k²/4 = 5(ln5)²/16 ≈ 5(2.5903)/16 ≈ 0.8095. - Since 0 < 0.81 < 1, the cycle is locally attracting. ✓ **Income accounting along the cycle.** - In the state (1,1), A's earnings are f(3) = 1, so total is 1 + 1 = 2. - In the state (3,3), A's earnings are f(1) = 5, so total is 5 + 1 = 6. - The average is 4, which is (6 − 4)/6 = ⅓ below the benchmark. ✓ - Inequality of disposable income is zero by construction. ## 3. Attempts to break it **Fixed point of the redistribution map.** - g is strictly decreasing, so it has exactly one fixed point y*. - Solving 5·exp(−k(y−1)) = 2y − 1 numerically gives y* ≈ 1.805. - The slope there is g′(y*) = −k(2y* − 1)/2 ≈ −0.8047·2.61/2 ≈ −1.05. Its magnitude exceeds 1, so y* is unstable. - Starting exactly at y* (or at a pre-image of y*) produces no cycle. - Even then, total income is 2y* ≈ 3.61 < 6. So redistribution lowers income on that path too. **Global behaviour (numerical check).** - Every y > 0 maps into the interval (1/2, g(1/2) ≈ 4.24). - Iterating from y = 2 gives 1.618, 2.02, 1.600, 2.043, … The iterates move away from y* and toward {1,3}. The deviation grows by about 1.10 every two steps, consistent with g′(y*)² ≈ 1.10. - The second iterate g∘g has negative Schwarzian derivative, and its known fixed points are 1, y* and 3. This makes extra 2-cycles implausible, though I have not fully proved it. - The claim only asserts local attraction, which is rigorously established. **Timing and accounting subtleties.** - Over a finite horizon the average depends on the phase; asymptotically it is 4. - Approach to the cycle is asymptotic, not exact. "Permanent" is correct for the limit cycle itself. **No hidden errors found.** The key structural feature is that A's earnings respond to B's disposable income. Redistribution feeds A's own earnings back into that variable. This creates the loop y → f(y) → y with a steep negative gain at y = 1 (slope −5k/2 ≈ −2.01). In the benchmark, B's disposable income is exogenous, so no such loop exists. ## 4. Novelty The mechanism is the generic one in which closing a feedback loop with lagged negative gain produces flip (period-doubling) oscillations. Familiar instances include: - cobweb dynamics; - the Friedman/Phillips argument that lagged stabilization policy can destabilize. The specific two-household construction appears to be a tailored illustration of that known principle, not a new result. ## 5. Fairness of the one-sentence summary The claim says "can," which is an existence statement, and the text supplies a valid example. The supporting text itself disclaims generality. Missing qualifiers in the one-line claim: - The example is reduced-form and engineered: the externality runs from B's *disposable* income to A's future earnings. - "Attracting" is local. - An unstable equal-income fixed point exists. - The one-third figure is a limit-cycle average. None of these makes the summary misleading. ```json {"verdict": "correct", "novelty": "special case or variant of known results", "fair_summary": "yes", "confidence": 4, "one_line": "The arithmetic and stability check are right: equalization turns a constant-map economy into a locally attracting 2-cycle {1,3} with average total income 4 versus 6, an engineered illustration of the known principle that closing a lagged negative feedback loop can create flip oscillations."} ```