Political Resource Allocation: Benefits and Costs of Voter Initiatives

s away from the specialization benefit that is surely an important reason for delegated decision making. We also assume that the voter is capable of determining which of the policy options is in his best interest. A natural starting point to address the issues of expertise and voter ignorance would be to include policy uncertainty along the lines of Gilligan and Krehbiel (1987). This would presumably make resource allocation by representatives more valuable. It would then be possible to study the effect of different institutions on the incentives to acquire information, an important issue addressed recently by Aghion and Tirole (1997). A signaling role for interest groups and endorsements would arise, as in Gerber and Lupia (1995). In such an environment, it would also be possible to think more systematically about the apparent fact that certain types of interest groups have a comparative advantage in lobbying the legislature, while others do better with initiatives (Gerber, 1999). Another issue that deserves attention is how the initiative affects the representative’s probability of reelection. For example, voters may be less inclined to punish representatives for bad policy choices in a jurisdiction where an initiative override is possible than in one where the decision cannot be challenged. In terms of our model, by altering the electoral environment, the initiative may change the representative’s (induced) ideal point—which we take as independent of the decision-making process—and change the way he responds to interest group threats. Although little work exists on this problem, Feldman (1998) reports that reelection probabilities are different in initiative than noninitiative states, which suggests that an effect of this sort may be at work. Although we have focused on public sector decision making, the analysis could be recast to consider resource allocation in corporations. The question becomes, what are the benefits and costs of shareholder initiatives? Here is a sketch of how such an analysis might proceed: In a corporate context, the shareholders are the voters and the directors/managers are the representatives. For most corporate decisions, shareholders are likely to have similar preferences—they want the policy that will maximize value. Because managers face little uncertainty about shareholder preferences, initiatives are unlikely to be valuable for most decisions, which is consistent with the evidence in Karpoff, Malatesta, and Walkling (1996). However, our analysis suggests that shareholder initiatives will be beneficial for particular issues—those where manager and shareholder interests diverge (such as executive compensation and takeover defenses) and those where shareholder 43. Signaling considerations might also introduce a cost to the representative of being overridden. The model assumes that the representative does not mind being overridden by an initiative except to the extent that he dislikes the new policy. Observation suggests that representatives might dislike being overridden per se, however. It is easy to show that if the representative pays a pure utility cost from being overridden, the deterrence region changes, but the comparative statics and comparisons remain qualitatively the same. However, the situation would become more complicated if the representative had a privately observed skill level and an initiative override signaled a poor choice of policy and hence a low skill level. 446 The Journal of Law, Economics, & Organization, V17 N2 interests are heterogeneous (such as an investment in apartheid South Africa in the 1980s). Appendix Proof of Lemma 1. Suppose that i = 0. By Equation (2), x∗ r = 0, and by Equation (1), p x∗ i xr = 1− , independent of the choice of xr . Then E R is maximized with x∗ r = r . Suppose that i = (the case of i =− is symmetric.) Then E R = { R + 1− R xr if xr >−z; 1− R −xr + R xr if xr ≤−z. Consider r = 0. Then x∗ r = 0 if xr > −z, and x∗ r = −z if xr ≤ −z. The expected payoff is R in the first case and R −z in the second. The first exceeds the second when > 1/3. Consider r =− . Then x∗ r →−z if xr >−z, and x∗ r =−z if xr ≤−z. The expected payoff in the first case approaches from below the expected payoff in the second case. Proof of Lemma 2. Suppose r = and i= 0. Then the representative either deters with xr =C/ 1− [see Equation (3)] or allows the initiative, leading to xi = 0 and xr = [from Equation (2) and Lemma 1]. Deterrence is optimal for the representative if R ( C 1− ) > R + 1− R 0 (A.1) Equation (A.1) reduces to C > 1− . The case of r =− and i = 0 is symmetric. Suppose r =− and i= . Then the representative can deter with xr = − C/ , or allow the initiative, giving xi = z and xr = z (Lemma 1). Deterrence is optimal if R −C/ > R −z + 1− R z (A.2) Equation (A.2) reduces to C > 1− 1− 2 / 2− 3 . The case of r = and i =− is symmetric. Suppose r = 0 and i = . Then the representative can deter with xr = −C/ . Without deterrence, the representative’s policy depends on . If > 1/3, then xr = 0 and xi = . Deterrence is optimal if R −C/ > 1− R 0 + R , which reduces to