Source
Schram, A. & Ule, A. (2024), Public Choice 198: 403–425, LIT-1340, read in NOTE-1053.
What was actually shown
A 2 × 2 between-subjects design at CREED, Amsterdam: 351 subjects in 59 groups of six (one operator, two policymakers, three voters), nine periods, each group playing once. Capture: policymakers’ payoffs equal the operator’s, or the consumers’. Independence: the policymaker elected in period 1 fixes the price for all nine periods, or the elected policymaker sets it each period. Prices in {0, …, 5}; the efficient no-loss price is 1 after investment.
Mean price over periods 1–6 (Table 4):
| flexible | independent | |
|---|---|---|
| consumer-aligned | 0.16 | 0.47 |
| operator-aligned | 1.78 | 3.19 |
- Consumer-aligned: independence raised prices toward 1 (permutation test, N = 41, p = 0.015) and cut the mean deviation from the efficient price from 0.98 to 0.53 (p < 0.001). Its effect on investment by period 5 (63% to 84%) was not significant (p = 0.255).
- Operator-aligned: independence raised prices (N = 18, p = 0.001) and the deviation from 1.56 to 2.31 (p = 0.066 by permutation test; significant at 1% in the Tobit regression of Table 5).
- Under flexible regulation, operator-aligned prices (1.78) lay below the no-pressure equilibrium and above the strong-pressure one, in which alignment would not matter: voters’ pressure worked, but only partly. Voters punished higher past prices; only 36% of newly elected operator-aligned policymakers lowered their predecessor’s price.
What could have come out otherwise: voters could have pushed operator-aligned prices to the consumer-aligned level, as the strong-pressure equilibrium predicts, which would have left independence nothing to entrench. They did not.
What this does not say
- Not how capture comes about. Alignment is assigned. Nothing in the game influences a policymaker, and the result is silent on whether independence makes capture more likely.
- Not about independent agencies. “Independence” is one policymaker’s price fixed for the horizon. Mandates, appointments, budgets and review are not modelled.
- Not beyond its scale. Three voters, two candidates, one issue. The authors think real electoral pressure is weaker, which would strengthen the entrenchment, but that is their conjecture.
- Not the general policy conclusion. That independence helps consumers only when its designers serve them is THEORY-858, Proposed.
- Not robust in every measure. With the 45.8% of groups whose operator invested in period 1 excluded (the authors’ Appendix F, which this reading did not check), the capture-cell price effect is significant only at 10%, the deviation effect by permutation test is not significant (p = 0.415), and by regression it is marginal (p = 0.07).
Connections
- CLAIM-404, last clause. Formal independence of a procedure does not make it independent: here independence from voters, installed by a captured designer, insulated the capture from the one correction left.
- CLAIM-342. The voters were the check whose arrival did not pass through the policymakers’ interest. Independence removed it.