Broad bills ensure that general state policies remain current and fair, but legislators might avoid them for two reasons. First, they are technically complicated; if you wish to revise the state highway code, for example, you will need expert advice and probably a few studies. Second, they are politically difficult to pass; since they influence the entire state, you’ve got to work to bring a coalition of legislators on board with your proposal. By contrast, “district” bills are technically less complicated; the process of campaigning generally gives legislators all the information they need about some pressing local problem. District bills are also politically easier to pass; since they don’t have any impact outside of a small geographical area, other legislators have no reason to oppose most district bills.

By examining every bill introduced in 13 states in 1881, 1901, 1921, 1941, 1961, 1981, and 1997–that’s over 165,000 bills–Gamm and Kousser try to explain why some states produce so many more district bills than others. In Alabama, only 53% of bills had statewide impact; in Nebraska, 77% did. In general, Gamm and Kousser find that states pass more district bills when legislators have incentives to build up their reelection constituency or to make themselves stand out as an individual, but they pass more statewide bills when legislators have incentives to develop their influence and power within the state legislature. There’s more to it than that, though.

Hypotheses and Findings

Gamm and Kousser identify specific conditions under which legislators might prefer to introduce district bills or statewide bills.

Hypothesis 1 (one-party dominance): When the major parties are evenly balanced in a state, you’ll see more statewide bills; when one party dominates, you’ll see more district bills. This prediction is based on Key’s seminal 1949 book, Southern Politics in State and Nation. Key wrote that two-party competition produces “conditions favorable to government according to rule or general principle” (e.g. statewide bills); by contrast,  “in a loose, catch-as-catch-can [one-party] politics highly unstable coalitions must be held together by whatever means is available…. A loose factional system lacks the power to carry out sustained programs of action [e.g. statewide bills]” (pages 305, 308). In other words, two-party competition creates strong incentives for each party to form a statewide brand name by competing over statewide policies; in a one-party system, unstable factional coalitions within the supermajority party need to be held together with pork and other district bills. This hypothesis is confirmed in the analysis.

Hypothesis 2 (legislative salary): Legislators will introduce more district bills if they are paid more. California’s legislators earn six-figure salaries each year; Montana’s earn less than $10,000. Higher salaries create higher incentives to hold onto your job. Thus, higher salaries create incentives to deliver pork and other particularistic bills to voters in your district, even at the expense of pursuing the state’s general interest. This hypothesis is also confirmed in the analysis.

Hypothesis 3 (session length): Legislators will introduce more statewide bills if they are in session longer. California’s legislators are in session almost year round; in other states, legislators meet only one or two months per year. Because statewide bills are more complicated to write and pass than district bills, you will see fewer statewide bills in states with shorter sessions. As it happens, however, the data do not bear out this expectation; session length appears to have no relationship with the types of bills that pass.

Hypothesis 4 (turnover): Legislators can increase their statute within the legislature by building coalitions and passing statewide bills. As turnover decreases and legislators serve longer, they gain incentives to develop their stature within the legislature. As such, decreased turnover should lead to increased statewide bills. That’s the expectation, anyway. In the analysis, Gamm and Kousser find the opposite: Higher turnover leads to more focus on statewide bills. Perhaps their logic was off. Perhaps what’s really going on is this: Legislators who know they won’t stay long have less need to build up their reelection constituency (with district bills), so they can instead focus on statewide needs. I’m sure Kousser would rather not say this, but…score one for term limits?

Hypothesis 6 (district size): If districts are larger, legislators will focus more on statewide bills. Larger districts include more cities and towns, meaning that an individual district bill will reach a smaller portion of the district. In this situation, a more effective way to show the entire district that their legislator is attentive to their needs might be to pass statewide bills rather than district bills. This hypothesis is confirmed in the analysis.

Yes, I skipped hypothesis 5. I found it unpersuasive, and it ended up being statistically insignificant in the final analysis.

Punchline: If you want your legislators to pass general policies that benefit the state as a whole, pay them less, make districts bigger, and strive for partisan balance. If you want your legislators to pass pork and other district-focused bills, pay them more, make districts smaller, and promote one-party government.

Comment and Criticism

Gamm and Kousser are right to look to state legislatures to answer this question. Too often, folks think that they have to look at Congress if they want to study legislative process. The trouble with that is that there are no cross-sectional variations in the variables that we might expect to matter. In any given year, the majority party margin is the same for every member of the U.S. House; so is the salary, session length, average turnover, and so on. Sure, these things vary over time–but ALL of them vary from year to year, making it hard to tell which over-time variation is influencing the results.

Sometimes, folks who give up on Congress try instead to compare countries to one another. This approach is useful, but it introduces a host of other complications (variations in culture, constitutions, voting rules, etc.). The state legislatures are the perfect venue for this research question. The authors are able to study all the variables they care about, but all their observations are similar by virtue of their status as members of a common political system.

Masket takes this a step further, arguing that local informal party organizations control nominations and through those nominations exert control over the legislative behavior of politicians.  He argues that “parties control the public behavior of their office holders by acting as gatekeepers to political office.”  While agreeing with Aldrich (1995), that parties help organizing the legislature, Masket argues that parties, and in conjunction party nominations, are primarily a mechanism by which concerned citizens hold legislators accountable for their actions.

Testing and Results

To test these arguments, Masket uses roll-call analysis of the California Assembly prior and subsequent to the 1953 decision to mandate party labels on primary ballots, effectively eliminating the ability of legislators to cross-file in both political primaries.  His analysis of roll call votes as well as votes for the Speaker of the Assembly shows that the imposition of these new regulations on primary elections had a marked effect of increasing a the partisan nature of the legislature, as politics became less bipartisan after the change, which Masket argues was not the desire of those within the legislature.

Inside the Smoke-Filled Rooms and Thoughts on the How

In addition to his quantitative analysis of legislator behavior, Masket also details the structure of 5 different local party organizations: Orange County Republicans, South Los Angeles Democrats, East Side LA County Democrats, West LA Democrats, and the local party organizations of Fresno County.  Through interviews, Masket details the ways in which party organizations exercise their influence using things such as donor networks, sample ballot mailings, and the mobilization of activist networks.

Unfortunately, none of these items that come out through interviews are easily testable, and while Masket presents ample evidence of local influence on the nomination process, it is difficult to determine how much influence these organizations have on the outcome of nominations.

Small Qualms

While Masket’s work shows clearly the change in the influence of parties after the change in primary election law, he doesn’t give any solid quantitative evidence as to how exactly those mechanisms work.  Is it the fundraising network?  Is it the power of mobilization?   While the interviewees claimed to have influence in all of these aspects, the heads of a campaign or a campaign organization has an incentive to make their role as significant as possible in order to increase their status as the gatekeeper.  Masket clearly demonstrates to the reader that local party organizations influence nominations, but falls a little short on convincing the reader as to the mechanism through which these organizations control nominations.

Previous Work

Because most of the states that adopted the initiative in this period were agrarian, western, and mostly white, historians have traditionally answered this question by claiming that the initiative was most likely to be adopted in “younger states with a strong history of populism and antimonopoly sentiments, more homogeneous [white] populations…, more radical (but weaker) political parties, and stronger interest groups” (pg 334). Smith and Fridkin contend that these conclusions have been based on deeply flawed logic and methods.¹

In particular, scholars have erred in asking why some states adopted the initiative and others did not; such an approach focuses on voter demand for direct democracy. But since the initiative is usually adopted via legislative referendum, in which legislators invite voters to choose whether to adopt the initiative, scholars should instead be asking why legislators in some states chose to offer the initiative to voters, while legislators elsewhere did not.²

Theory and Findings

Smith and Fridkin argue that legislators were most likely to offer the initiative when states experienced close partisan competition. When a minority party gains tenuous control of the legislature, or when a majority feels that it might become the minority soon, then legislators gain an incentive to weaken the legislature by empowering the median voter. This is their main causal argument, which is supported by an event history analysis.

Other conditions also play a role. In particular, legislatures in newer states are more like to offer the initiative than legislatures in older states. The authors attribute this difference to the weaker party organizations in younger states–organizations that might have less ability to rein in independent-minded legislators. Likewise, legislators were more likely to offer the initiative if there were more third-party legislators or if there was a strong interest group pressure in a particular state for instituting the initiative.

Contrary to previous work, the authors do not find any evidence of a contagion/diffusion effect. Nor do they find that initiative adoption was more common in racially and ethnically homogeneous states. Significantly, they find that dummies for the western and southern region are not statistically significant in their models; controlling for other variables apparently eliminates the regional bias in initiative adoption.

Critique

This paper’s methodological rigor makes it a welcome addition. As the authors point out, most previous work on this topic has been somewhat impressionistic and ad hoc. Event history analysis is the right way to study adoption of the initiative. Their shift from studying adoption to studying legislative support for the initiative is very smart.

That being said, a few things seem off to me. First, why did initiative adoption largely cease after 1918? Imperfect as previous work may be, its emphasis on western culture and populism seems to have an answer to this question. Smith and Fridkin don’t so much as speculate; they simply restrict their analysis to 1898-1918. But if some omitted variable caused initiative adoption to cease in 1918, is it possible that the same omitted variable, if measured, would change the results in Smith and Fridkin’s analysis?

Also, it seems that the role of governors may be downplayed. The authors start well (p 339) by listing several Democratic and Republican governors who pressured their legislators to pass an initiative reform; they even note that Republican Hiram Johnson of California made this a major goal. Yet inexplicably, the authors “control” for these considerations by including a dummy for whether the governor belonged to a third party. They provide no justification of this measure which, unsurprisingly, is not statistically significant–leading them to conclude (without basis) that gubernatorial action plays little role in adoption of the initiative. But if we believe that executives can pressure legislators by “going public” or through other means, then we should be looking more at the role governors may have played.

Third, they take no account (empirically) of variations in initiative institutions. Initiatives vary widely in (1) how hard they are to use and (2) how easily the legislature can ignore them (Bowler and Donovan 2004). Why did some states (California, Oregon) pass powerful initiatives, but other states (Utah, Idaho) passed weak ones? The authors acknowledge this variation, but do not seek to explain it. My speculation: The “party competition” variable that the authors discuss might be two different things. If a minority (perhaps allied with third parties) seizes control for a short while, or if a majority genuinely fears becoming a minority, then we might see an effort to create a truly powerful initiative. But if a majority sees its power slipping and wants to do something popular to build up support, then we might see a weak initiative created purely for its symbolic value. Thoughts?

Every study has weaknesses, and the flaws in this one are by no means fatal, nor are they necessarily more numerous or severe than what is typical for an article in a top journal. But I think they point to some avenues for further research.