Mariano-Florentino Cuellar has posted this draft on SSRN.  Here is the abstract:

In legislation, to earmark means roughly to designate (through a statutory provision or an accompanying committee report) certain appropriated funds for narrow (nearly always geographically-delimited) purposes that appear to benefit particular interests. Policymakers, civil society organizations, and scholarly observers routinely condemn earmarking as a practice putatively tied to corruption, or reflecting abuse of the political process – critiques that have spawned a variety of recent reform efforts. Yet a meticulous prescriptive evaluation of the practice soon raises fairly profound questions encompassing institutional design, legal theory, organizational practice, and the role of a cognitively-overburdened public in a democracy. Upon closer inspection, for example, earmarks seem no more or less likely to be connected to corruption than a host of other highly-targeted outputs of the legislative process, such as private immigration bills or intricate changes to complex regulatory statutes benefiting particular companies or interest groups. Earmarks can also serve as side payments capable of protecting legislative bargains from costlier distortions as lawmakers seek to advance their constituents’ interests. Moreover, not all earmarks constitute ‘pork-barrel’ spending, as principled legislators could use targeted measures in order to manage the enormous analytical difficulty of designing complex legal provisions applying general principles to specific situations, and to protect lawmakers’ role in a system of separated powers.

Given these factors and the highly variable substantive content of earmarking, I reach three conclusions. Some earmarks are substantively defensible. On balance, conventional earmarks are probably more transparent than many other political deals. Hence, although earmarking reformers seem to assume that banishing earmarks would make it harder for lawmakers to advance their narrow personal interests or those of their constituencies, there is little basis for such a conclusion. Even if certain specific earmarks are not desirable, any sensible evaluation of the overall practice of earmarking implicates a broader discussion regarding the merits of the legislative process and the pluralist system in which it exists. A more analytically sound approach to earmarking would recognize the connection between targeted spending and legislative compromise. Such an approach would consider incremental changes promoting greater transparency and focus greater attention on discussions of the merits of individual earmarks. In contrast, aggressive efforts to limit earmarks altogether are exceedingly difficult to defend and may engender wider distortions in otherwise defensible statutes and regulatory policies.