The Weak Theoretical Justifications for the Single Subject Rule

In yesterday’s post, we explained the basics of the “single subject” rule: initiatives may contain only one “subject” or they are unenforceable. The rule leads to controversy because the concept of a “subject” is not self-defining and state courts have developed different single subject tests.

Regardless of terminology, most courts seek to justify their rules based upon only two reasons: preventing logrolling and minimizing voter confusion. In this post, we explain why both of these justifications are weak.

Logrolling

Without a single subject rule, a person could propose an initiative marrying two completely separate questions, such as a tax break and gun control. The concern of courts is that forcing voters to vote on this package will subvert the majority’s will. The single subject rule, if it treats tax breaks and gun control as separate subjects, would prevent such logrolling.

We do not disagree that preventing the packaging of certain initiated legislation prevents logrolling, but we are far less sure that logrolling is necessarily a social ill. In our article we provide a game theoretic explanation for why this is so. Here, we explain the intuition behind our explanation, and direct readers to the paper if they want more precision and detail.

Logrolling may be undesirable if it subverts the electorate’s will, but it does not necessarily do so. There are clearly situations where allowing logrolling can lead to outcomes that further majority preferences, a point public choice scholars have recognized for a long time.

One concern with logrolling is that by combining two “projects” (such as a tax break and gun control) one that is good and one that is bad, the voters will be forced to adopt the bad project against their interests (what Lowenstein calls a “rider”).

When a bundled law passes with a rider included, this appears to promote, not negate, the public welfare. The voters are better off with the bundle than without it, which is why they approved it in the first place. It is true that voters would be better off if they had the opportunity to vote on the projects separately rather than as a package, but nothing guarantees this would happen if the package is not allowed. Indeed, when a court strikes down a measure on single subject grounds, it does not give voters the opportunity to vote on the separate pieces, but rather forces rejection of both projects, which sometimes makes voters worse off. (Of course, in situations where the bundle as a whole is a bad deal for voters, voters would reject the package and a single-subject limitation is unnecessary.)

The concern over the single subject rule is deepened once we recognize that it may be possible to approve some valuable projects only through a bundle (what Lowenstein calls “coalition-building”). It is easy to construct examples of socially valuable bundled projects that would be defeated if voted on item by item. For example, consider a proposal to build a series of levees, each of which protects a single neighborhood from flooding and is financed by a general tax. If each neighborhood is a small part of the electorate, then none of the levees will be approved if voted on individually (only the neighborhood that would receive the levee would vote for that one), but if the projects are presented to the voters as a package, the protected neighborhoods may combine their votes to gain its approval.

(As an aside, critics of direct democracy often celebrate the give and take of legislatures as compared to the one-shot nature of ballot propositions. It should be recognized that legislatures rely extensively on logrolls to implement their agreements. Indeed, without the ability to logroll it is hard to imagine how complicated legislative bargains could be struck and enforced.)

The previous example is not intended to suggest that logrolling is always beneficial. To the contrary, using game theory we demonstrate in our paper that there are also situations where a logroll can bring about a socially undesirable outcome. The point here is not that logrolls are always beneficial but rather that logrolls can be good or bad. Much of the doctrine and analysis surrounding the single subject rule presumes that logrolls are always bad, so that voters need to be protected against all logrolls. As we have seen, this view is overly simplistic, lacks theoretical justification, and stands a real chance of inhibiting socially desirable policy changes.

Voter Confusion

Another alleged purpose of the single subject rule is to prevent voter confusion The issue of voter competence has been a central concern in thinking about direct democracy for as long as the process has been around, and it is well recognized that voters must have access to information to make wise decisions.

Contrary to simple intuitions, empirical research suggests that citizens are able to vote in a sophisticated manner if they have access to endorsements and other “information cues,: But even accepting the view that voters need to be protected from complexity, it is difficult to see the single subject rule as a vehicle for reducing complexity and alleviating voter confusion. We cannot improve on Lowenstein’s brief-yet-effective argument:

    The rule is ill-suited to prevent voter confusion because no matter how the rule is construed, it will bar some initiatives that are simple and permit others that are hopelessly complex. Consider, for example, an initiative containing two provisions: (1) change the date of the primary election from June to May; and (2) increase the maximum sentence for the crime of rape by one year. While most people would regard it as odd for these two and only these two provisions to be combined in one initiative, and while the measure would presumably violate the single-subject rule, it would also be one of the simplest and most easily understood initiatives ever proposed in California. On the other hand, one can easily imagine a proposal that would contain extensive but more or less technical revisions in a single, specialized area–say, school finance–that could not be understood thoroughly by anyone but a handful of experts, but that would satisfy the single-subject rule under any plausible construction.

    It is no doubt true that, all else being equal, a measure with fewer provisions will be easier to understand than a measure with more provisions. All else is seldom equal, however, and in most cases the complexities of the individual provisions and of the general subject matter are likely to be far more significant factors in the measure’s overall complexity than the mere number of provisions. Furthermore, the correlation between the diversity of the initiative’s subject matter and the number of provisions is likely to be very weak. An outlandishly diverse measure could contain one simple provision per “subject,” whereas a unified measure could contain thousands of provisions.

Since Lowenstein wrote these words, we are unaware of any empirical evidence produced which demonstrates that the single subject rule in practice has reduced complexity or alleviated voter confusion.

Tomorrow, we turn to empirical evidence on how courts actually implement the single-subject rule.