Congressional lettermarks,ideology,and member receipt of stimulus awards from the US Department of Labor

Abstract

Recent years have witnessed an increase in members’ use of lettermarks—letters written to the bureaucracy in an effort to secure federal funding for their districts. In spite of the vitriol that has been directed toward the practice, it is unclear whether lettermarks are even effective in aiding legislators’ attempts at funneling dollars toward their districts. In an effort to determine whether lettermarks “work,” we combined content analysis of letters sent by members to the Department of Labor with data on monies awarded to particular districts through the Department of Labor from the stimulus to investigate the relationship between letter writing and awards to members’ districts. We find that, conditional on ideology, lettermarking can have an effect on the likelihood that a district received stimulus money through the Department of Labor. For most legislators, though, lettermarking does not appear to be an effective device for securing federal dollars for their districts. This nuanced set of findings suggests that the bureaucracy may be sensitive to member ideology when deciding whether to fund certain projects. Given the limited time frame and scope of our data, though, we urge researchers to continue to pursue this important line of inquiry.

Introduction and Existing Literature

Earmarking is a time-honored tradition in the US Congress, and pork barrel projects have long been thought to both grease the legislative wheels (Evans, 2004) and aid in members’ efforts at gaining re-election (Crespin and Finocchario, 2013). Under pressure from anti-spending watchdog groups, though, members of Congress have increasingly taken a pass on the pork barrel in recent years. While some in Congress have publicly decried such developments as an abdication of Congress’s responsibility to guard its spending power from executive encroachments, many members have turned to other, more surreptitious, means of securing federal funding for pet projects. That is, members of Congress, including many of the most vocal opponents of particularistic spending, have taken to writing letters to bureaucratic agencies in an effort to fund specific programs (The New York Times, 2010).

This practice, more colloquially known as “lettermarking,” allows legislators to fly beneath the radar when it comes to bringing home federal monies, as they are able to outwardly give the appearance of opposing earmarks while at the same time privately lobbying for spending that benefits narrow interests within their districts. In turn, such spending offers credit claiming opportunities if the member wants to emphasize the project. Some measure of secrecy is afforded to legislators with lettermarks by virtue of the fact that “a lettermark—unlike an earmark—is nearly impossible to track” (Dawson and Kleiner, 2015: 204). And, in spite of the existence of an Executive Order stipulating that “[a]ll written communications from the Congress . . . recommending that funds be committed, obligated, or expended on any earmark shall be made publicly available on the Internet by the receiving agency” (quoted in Dawson and Kleiner, 2015: 205), it is rare for bureaucratic agencies to make copies of lettermarks available to the public (Dawson and Kleiner, 2015).

The opaque nature of the lettermarking process has led to a considerable amount of hand-wringing among observers. Dawson and Kleiner, for instance, describe lettermarking as a “pernicious practice” (2015 201). In spite of the amount of controversy that has stemmed from the practice of lettermarking, though, there is little evidence that it works. Far from being a “cattle prod to agency heads” (quoted in Dawson and Kleiner, 2015: 202), what empirical work exists on the subject suggests that lettermarks from members of Congress alone are not effective in securing particularistic outcomes for individual legislators (Mills and Kalaf-Hughes, 2015; Mills, Kalaf-Hughes, and MacDonald, 2016). But existing evidence on this point is drawn from one federal agency, the Federal Aviation Administration (FAA), and employs one, particularly visible, measure of congressional success: the issuance of a national interest exemption for control towers that were marked for closure (Mills and Kalaf-Hughes, 2015; Mills, Kalaf-Hughes, and MacDonald, 2016). While consistent with a lettermarks “do no harm” or “don’t matter” conclusion, the nature of these requests means that they are likely very visible to organized interests, affected constituents, and ultimately, voters. Furthermore, the potential impact of such exemptions on policy and jobs is both regulatory and substantial. Consequently, while analysis of lettermarking is challenging due to the need for systematic data on requests matched to awards or decisions, drawing more generalized inferences about the effectiveness of lettermarking on the basis of requests for exemption from FAA recommendations for tower closures is difficult.

By contrast, the present study focuses on an outcome that might very well go unnoticed by watchdog groups, unless, of course, the member chooses to advertise it: the receipt of federal monies from the American Recovery and Reinvestment Act (ARRA) as administered through the Department of Labor (DOL). Using content analysis of lettermarks to the DOL obtained through a Freedom of Information Act (FOIA) request by Cause of Action Institute, coupled with data on the allocation of federal dollars from the stimulus, we endeavor in this brief paper to answer a question that has dogged researchers for some time: Are members of Congress successful in securing particularistic spending for their districts through the use of lettermarks? In the following section we describe the data and methods that we employ in our analysis and highlight our expectations. After detailing the results from a regression model predicting the receipt of ARRA dollars from the DOL, we then go on to situate our findings within the nascent literature on lettermarking.

Data, methods, and expectations

The data used in this paper come from several sources. Recipient and agency spending data from the stimulus were obtained from quarterly reports from Recovery.gov—the government’s official website devoted to ARRA spending. ¹ We reduced this dataset just to the awards that were handed down through the DOL. This was done so as to combine our federal awards data with content analysis of lettermarks that were sent to the DOL over the period from February 2009 to August 2010. ² We reduced the stimulus dataset to the same time frame, further cutting it down to feature only awards administered by the DOL that could have possibly been awarded at the behest of a member (or members) of Congress. And while we cannot be certain that these files represent the complete universe of letters sent to the DOL from members of Congress throughout this time period, they comprise the best evidence available to us at the current juncture, pending additional FOIA requests to other government agencies.

Lettermarks sent to the DOL were coded for several items, including whether the letter mentioned the ARRA, mentioned a particular dollar amount (and, if so, what that amount was), and whether multiple legislators signed on to the same letter. In total, 294 unique letters were content-analyzed from 165 individual legislators. The number eventually included in this study is smaller, however, as we focus only on members of the US House in this study in an effort to be as precise as possible about our outcome variable of interest, as it is nearly impossible to determine whether ARRA spending targeted toward, say, a congressional district in New York should also count as spending targeted toward Senator Chuck Schumer. Focusing only on members of the US House leaves us with a total of 175 letters from 124 different legislators.

In addition, information on members’ partisan affiliations and ideological proclivities (as measured by DW-NOMINATE) was merged with our content analysis and data from the stimulus. We also included a measure of district “need” as well—the percent unemployed in each district (see also Mills and Kalaf-Hughes, 2015). ³ This measure was culled from the US Census. The unit of analysis in the final dataset is the individual legislator, leaving us with 446 total cases—the number of unique legislators who served in the US House over the time period under investigation. ⁴ As such, several of our content analysis items were calculated in terms of counts (such as the number of letters to the DOL authored by a particular legislator over this time period) or proportions (for instance, the proportion of all letters that were co-authored by other legislators—typically an entire state delegation).

The dependent variable that we employ in the first part of our analysis is the natural log of the stimulus dollars received from the DOL by the legislator’s district over the period under examination. We obtain the same basic results, though, if we use a binary indicator of whether or not a district received any awards from the stimulus through the DOL (results not shown). Our key independent variables include an indicator for whether a legislator wrote a letter to the DOL requesting funds, a measure of legislator ideology, legislator partisanship, and a series of interaction terms.

We include these interaction terms in an effort to explore the possibility that the effect of lettermarks may be conditional on the ideological leanings of the legislator requesting funding from the DOL and his or her party. For much as the effect of lettermarks appears to be conditioned by the preferences of the granting agency (Mills and Kalaf-Hughes, 2015; Mills, Kalaf-Hughes, and MacDonald, 2016), we expect the political leanings of a bureaucratic agency’s principal—the president—to affect their willingness to bend to pressure from legislators. Since the legislative agendas of Democratic presidents typically fare better among more liberal members of Congress—Democrat and Republican alike—we expect that the bureaucracy is more likely to grant requests for funding from more liberal representatives. Doing so serves to reward members for their support of the president (or a promise of future support) and, in turn, provides legislators with the ability to credit-claim (Grimmer, Messing, and Westwood, 2012).

Legislators often highlight their ability to channel public dollars toward their own districts in their pursuit of re-election (Hudak, 2014: 19), and we would expect the bureaucracy to be more than happy to offer members opportunities to do so, conditional on their ideological leanings (Alexander, Berry, and Howell, 2016) and partisanship (Berry, Burden, and Howell, 2010; cf. Dynes and Huber, 2015; Gimpel, Lee, and Thorpe, 2012). ⁵ In particular, we might expect pleas for stimulus dollars from liberal Democrats to be especially effective at seeing results. Ideological extremity is often punished at the polls (Canes-Wrone, Brady, and Cogan, 2002), and ideologically extreme members of the majority party may see more in the way of credit-claiming opportunities funneled their direction in an effort to compensate them for any possible losses that they might incur from assembling a voting record that is reliably liberal in nature (see, for instance, Carroll and Kim, 2010). While ideological purism from co-partisans in Congress is likely a boon to the administration’s legislative agenda, victories in the legislative arena can only be sustained if the president’s core supporters keep their jobs in the next election. Alternatively, the executive may have simply sought to reward good partisans for their loyalty when they saw it fit to ask for something over which the bureaucracy held dominion.

As for our expectations regarding specific features of the letters themselves that might track with the receipt of stimulus dollars, it seems likely that several might matter. In particular, we suspect that mentioning a specific monetary amount should correlate with the amount received given the voluminous literature on policy implementation that points to specificity as a key determinant of bureaucratic compliance (Huber and Shipan, 2002; Van Horn and Van Meter, 1977). Bureaucrats may be unwilling to write members a blank check for initiatives in their district, but discrete amounts that are (presumably) coupled with more detailed plans for the funds are likely to be effective. The number of legislators signing on to a particular letter may also serve as a heuristic for bureaucrats who are seeking to determine the support for particular initiatives, thereby increasing the possibility that funds will be allocated for such a purpose.

Results

In testing our expectations we began by estimating a model predicting the logged dollars received by a representative’s district through the DOL from stimulus funds. Table A1 in the online Appendix displays estimates from this model, which includes measures of party, ideology, and district need in addition to a series of interaction terms. It is important to note that we cannot reject the null that lettermarking, in isolation, has no effect on the receipt of stimulus dollars. The inclusion of a triple interaction necessitates, though, that we visually inspect the substantive effect of lettermarking, conditional on both party and ideology (Brambor, Clark, and Golder, 2005). Figure 1 therefore displays the marginal effect of writing a letter to the DOL over the observed range of DW-NOMINATE for both Republican and Democratic members. For Republicans, writing a letter to the DOL has no discernible effect on the amount of money that they received through the DOL from the stimulus (see the right panel in Figure 1). While the effect of letter writing is not statistically significant (with 90 percent confidence intervals) throughout the observed range of DW-NOMINATE for Republicans, the overall pattern suggests that those who lobbied the bureaucracy for funds actually did worse than those who did not. We are reluctant, however, to suggest that the bureaucracy may punish conservative legislators who request money for their districts via lettermarking. Rather, we suspect that Republican (GOP) members of the House who engaged in such activities were attempting to overcome some kind of a deficit by contacting the DOL about the need to fund projects in their districts.

Figure 1.

The effect of lettermarking on members’ awards conditional on ideology and party.

On the Democratic side of things (see the left panel in Figure 1), it appears that lettermarking had a positive and statistically significant (p<0.1) effect on the receipt of funds through the DOL from the stimulus, but only for extreme liberals. Democratic members who engaged in lettermarking and are -0.5 or lower on our measure of ideology (scores that correspond with more liberal roll call voting records) received more ARRA dollars than did similarly situated Democrats who did not write a letter to the DOL. Although the effect is not statistically significant, the pattern displayed in Figure 1 suggests that more moderate Democrats who wrote a letter to the DOL actually received less money than other moderate Democratic members who did not attempt to lobby the bureaucracy. This result mimics the pattern observed among GOP members. Our results therefore demonstrate that only liberal Democrats—those who are the most likely to support the president’s legislative agenda—reaped any benefit from contacting the bureaucracy in an effort to secure a slice of the federal pie for their constituents. Although we do not show them here, these same basic results obtain when we substitute raw DW-NOMINATE scores for the ideological distance between each member and the estimated ideology of the DOL under Obama—as calculated by Chen and Johnson (2015). That is to say that Democratic legislators to the left of the DOL were the only ones who appear to have benefitted from engaging in lettermarking. ⁶

Controlling for district partisanship (as proxied by the Democratic share of the vote for president in the previous election) pushes even this conditional effect into the realm of statistical insignificance (results not shown). Thus, while there is some evidence that lettermarking “works” when undertaken by liberal Democrats, on balance, our results suggest that, for most members across both sides of the aisle, letter writing was epiphenomenal or served some other interest beyond securing dollars. Still, the presence of a marginal effect for the most loyal Democrats is worthy of additional research, if and when additional lettermarking data were to be available as a means to analyze vote trading, partisan reward, and/or bureaucratic bias in conjunction with the distribution of federal discretionary grants.

But are there any features of lettermarks that make them especially effective? Using measures constructed from the content analysis described above, we estimated a model predicting the (logged) funds received from the stimulus (via the DOL) using data on only those members who wrote at least one letter. Results from this model are displayed in Table A2 in the online Appendix. What is striking about this segment of the analysis is that remarkably few features of the letters that the members sent to the DOL have any discernible impact on the receipt of stimulus awards. Consistent with the patterns displayed in Figure 1, legislator ideology is statistically significant (p<0.05), with more conservative members of the House receiving fewer stimulus dollars throughout the period under observation. But only one feature of members’ letters correlates with the amount received from the ARRA. The proportion of members’ letters that mentioned a particular dollar amount is a positive and statistically significant (p=0.05) predictor of the amount allocated to their district. Specificity, it seems, pays dividends when it comes to lettermarking. ⁷

Of course, this finding squares with a broader constellation of results from earlier studies exploring principal-agent relationships across a range of different contexts. Boyd (2015), for instance, finds that federal district judges are more likely to change their behavior in response to specific, unambiguous signals from their direct principal—circuit court judges. Similarly, the literature on policy implementation suggests that “policy precision,” to use Huber and Shipan’s (2002) terminology, is a powerful tool that legislators can use to exert control over the bureaucracy. Along these same lines, Van Horn and Van Meter (1977) maintain that bureaucratic compliance with the policy standards set forth by their principals requires that legislators communicate their expectations with a level of clarity that promotes effective policy implementation. Effective communication of what, exactly, they would like to see in the form of disbursements from the DOL through their letters appears to have aided legislators in their pursuit of credit-claiming opportunities.

Discussion and conclusion

Although our study is limited in scope to a short timeframe and draws upon evidence from a single entity within the bureaucracy, the findings outlined in this brief note build upon existing research on lettermarking and effectively qualify normative critiques of the practice that have characterized communications from legislators as having undue influence on the bureaucracy. By analyzing a different policy area across a broader range of potential projects, our analysis takes a significant step forward in advancing our understanding of lettermarks as a replacement for earmarks in an era that is increasingly characterized by Congress’s willingness to delegate authority over the power of the purse to the president and the federal bureaucracy.

We find that lettermarks can be effective when they include specific details such as a dollar amount. Furthermore, the effect of lettermarking is conditioned by political attributes of the legislator making the request—such as partisanship and ideology. It therefore appears that the bureaucrats responsible for doling out federal funds are sensitive to the ideological leanings of legislators as they go about making decisions about which districts receive awards (see also Alexander, Berry, and Howell, 2016). That the effectiveness of lettermarking may be moderated by political factors (notably increasing toward the direction of the president) mirrors some of the politics of earmarks where the victors get the spoils and the loyalists get even more. This suggests, consistent with other recent work by Mills and colleagues (2016), that the bureaucracy is now the ultimate arbiter of (to channel Lasswell) who gets what, when, and how. Even though certain members of Congress may be able to guide the process from time to time, it appears that the bureaucracy is very much in the driver’s seat when it comes to the distribution of stimulus dollars. As such, if Congress wishes to assert control over the purse strings, then a return to good old-fashioned earmarking may be necessary.

In closing, then, it is important to note that most legislators do not appear to have benefitted from writing lettermarks to the bureaucracy. This raises the question, of course, of why members across the partisan and ideological spectrum engage in such efforts. In his seminal work, Mayhew (1974: 59–60) emphasized that members’ claims regarding their efforts at providing particularized benefits to their constituents must be credible. The existence of letters to the bureaucracy requesting funds or special consideration for pet projects in their home districts likely afford legislators with a measure of credibility in this arena, even in cases in which their actions were not determinative. As such, lettermarks may provide members with ample credit-claiming opportunities in the absence of a direct linkage between their actions and the geographic targeting of federal dollars (see also Grimmer, Messing and Westwood, 2012). We therefore urge future researchers to continue pursuing this line of inquiry.