I have discussed the theory that the passage of the 17th Amendment was a serious blow to federalism on this blog before, and I have been supportive of the argument. Although I have not seen an explicit public choice theory for why the indirect election of senators institutionally encourages federalism, the implicit logic is obvious enough: rationally self-interested state legislators, interested in their own power and influence, jealously guard and work to expand the power of state the legislatures by voting for national legislators who defer to the states as much as possible.
Initially, I found this mechanism very convincing, and I thought that the passage of the 17th Amendment explained essentially all of the decline of Federalism in the US since 1917, but I have changed my mind, and I want to explain why.
Rational self-interest does not imply that the indirect election of national legislators strongly encourages federalism because of vote dilution. Like voters, state legislators have only a very small impact on the outcome of the vote and therefore only a very small impact on their own welfare. This means that even though legislators are rationally self-interested, their political action, like that of voters, will be largely sociotropic (perceived “good” or welfare maximizing).
Because legislators face weak incentives to protect the power of state legislatures, they are likely to follow voter preferences in this domain, even if they are weak, and it may be the case that voters increasingly want more involvement from the national government. Communication and transportation technologies continually get better, which makes the world “smaller,” and I would guess that this makes people think on a more geographically broad level. People also increasingly have a national identify instead of a state level identity, before the union of the states, people often identified with their state, but now most people identify with their country; few people are more loyal to their state than the nation. This may lead people to think of national level instead of the state level as the appropriate level for politics to take place and therefore lead to less federalism. If this is the case, it would lead to reduced federalism no mater the method used to elect the national legislature. Riker’s (1955) account of the passage of the 17th Amendment seems to support the idea that direct election is inherently popular. Before the passage of the 17th Amendment, many states passed laws requiring the legislature to elect the candidate who won the popular vote in the state, in response to the growing trend of state legislators being forced to campaign primarily on who they would vote for for national senate instead of what they would accomplish in state.
However, there are some potential mechanisms that could motivate state legislators to actively protect the power of the state legislatures.
Indirect election of the national legislature could still systemically encourage federalism if people are biased in certain ways. For example, if people are generally cognitively biased towards accepting ideas which are perceived as in their self or group interest, then legislators will have a skewed perception of what is “good for society.” This will lead state legislators will elect national legislators that favor stronger state legislative power than if they legislators were unbiased.
Indirect election of the national legislature might encourage federalism relative to direct election of the national legislature if state legislators have biased cost/benefit information. For example, state legislators have only the viewpoint and experience of state legislators, so they will be far more aware of the costs and benefits to the state legislatures of proposed structural changes than the costs and benefits to the national government. This would lead state legislators to make decisions based about what they support at the national level largely on what the costs and benefits would be to state legislatures. I am aware of a little empirical support for this proposition: some evidence suggests that people avoid gaining information about the impact of their potential choices on others.
In addition, direct election may actively encourage the transfer of political power from the state legislatures to the national legislature because of the incentives that directly elected legislators have in their attempt to appeal to voters. Directly elected legislators (whether they are national or state) have incentives to expand the control of their legislature both “down” into areas already covered by state bodies and “out” into areas not currently regulated. Legislators have an incentive to work to expand the power of the legislature into new areas in order to claim credit for working to resolve real or perceived problems. This incentive will operate even when those areas are already controlled by the state legislatures because voters are generally ignorant of their own state laws. For example: national legislative candidates in a directly elected system might successfully tout their support for stronger national gun laws, even though the states already good gun laws. This would appeal to voters who think that strong gun laws are a good idea that are ignorant of gun laws in their own state.
Under indirect election, however, national legislators do not have the same incentives to expand the power of their legislature “down” and “out” because, state legislators will generally be far more informed about what issues state law already covers and more often realize that the national legislature does not have any advantage in solving the problem over them. Moreover, while state legislators do have similar incentives to expand both “up” into areas regulated by the national legislature and “out” into unregulated areas, the political system usually prohibits them from expanding upward, taking over areas filled by the national government.
I can’t really comment on whether these mechanisms are true or not, because, while they seem intuitively appealing to me, they mostly are empirical questions for psychology, and I have very little knowledge of the relevant literature.
Overall, It seems to me that there are good reasons to suspect that the 17th Amendment eliminated significant forces in favor of federalism, and that, more generally, the indirect election scheme encourages power sharing between national and state governments. There also seem to be few reasons to believe that indirect election of the national legislature would be a bad political institution. However, the 17th Amendment was probably not an all-important force preserving federalism. Therefore, I think that indirect election of the national legislature is overall a good idea. Unfortunately, it seems that it is a challenge to preserve institutional features like indirect elections, even if they have positive effects, when they can be replaced by popular alternatives. If through some magic, the 17th Amendment were repealed, I doubt that it would stay that way for very long because direct election is popular. While indirect election of the national legislature seems like a moderately good feature of a political system, I have trouble seeing it as a permanent one in normal political systems.