Informational Lobbying: Theory and Effectiveness

Note: A ver­sion of this doc­u­ment was re­viewed by Hauke Hille­brandt, David Man­heim, Sa­muel Hil­ton, Aaron Gertler, and Tara Law, who I thank profusely for their use­ful and en­courag­ing feed­back.

At the end of this doc­u­ment, I make a pro­posal for high-lev­er­age lob­by­ing for al­tru­is­tic ends, con­duct a pre­limi­nary cost-effec­tive­ness anal­y­sis, and offer some rough sug­ges­tions for poli­cies that might be worth ad­vo­cat­ing in Congress. Sev­eral re­view­ers said this was their fa­vorite part of this piece, so you may want to read it first.


My con­fi­dence in each con­clu­sion leads each bul­let point, in paren­the­ses.

  • (85%) In­for­ma­tional lob­by­ing can shift policy rel­a­tive to the coun­ter­fac­tual.

  • (80%) Well-re­sourced in­ter­est groups are no more or less likely to achieve policy suc­cess, in gen­eral, than their less well-re­sourced op­po­nents. I’ve elimi­nated this con­clu­sion in light of some thought­ful ques­tions from @jackva and @Jamie_Har­ris be­low. In gen­eral, I think that there’s no ev­i­dence that re­source-rich groups are effec­tive ir­re­spec­tive of their de­ploy­ment of those re­sources. But that con­clu­sion is (1) it­self not well-sup­ported enough to in­clude here and (2) not prac­ti­cally use­ful. The most use­ful take­away here is prob­a­bly: re­sources are not a guaran­tor of suc­cess, be­cause suc­cess is con­tin­gent upon many other fac­tors. Ba­si­cally, less well-re­sourced groups can win and of­ten do.

  • (70%) Dol­lar for dol­lar, lob­by­ing against a pro­posal is more effec­tive than lob­by­ing for a pro­posal.

  • (65%) Lob­by­ing is more effec­tive when is­sues are not salient to the pub­lic, or (more par­tic­u­larly) to the lob­bied mem­ber’s con­stituency. After more re­view, I’ve elimi­nated this con­clu­sion as well. I’m now some­where around 5050 on this, with some ex­pec­ta­tion that the di­rec­tion of the effect is ac­tu­ally re­versed. I think this was in fact a se­ri­ous er­ror on my part: I (1) heav­ily over­weighted the the­o­ret­i­cal work (2) in­cluded ap­ples-to-or­anges work that fo­cused on PAC ex­pen­di­ture and (3) did not take into ac­count ex­plicit ev­i­dence to the con­trary. I also over­weighted my own de­duc­tions, which are worth spel­ling out here: There is good ev­i­dence that lob­by­ing ex­pen­di­ture in­creases with is­sue salience. If the be­low con­clu­sion about net di­rec­tion lob­by­ing is true, then lower salience trans­lates into less over­all spend­ing, which trans­lates into less to­tal ex­pen­di­ture for the big­ger spender, and a higher prob­a­bil­ity of suc­cess per dol­lar spent.

  • (60%) The net di­rec­tion of lob­by­ing, e.g. the differ­ence be­tween the num­ber of groups lob­by­ing for or against a pro­posal, or the differ­ence be­tween to­tal ex­pen­di­ture for or against a pro­posal, in­fluences policy out­comes. Or­ga­ni­za­tions that lobby more on an is­sue rel­a­tive to their ad­ver­saries are slightly more likely to win.

The qual­ity of the ev­i­dence in this do­main is con­sid­er­ably worse than in other ar­eas of in­quiry in the so­cial sci­ences. In gen­eral, the ques­tion I tried to ask my­self when as­sess­ing these po­ten­tial con­clu­sions was this: “How likely would I have been to see this ev­i­dence, in ag­gre­gate, if this state­ment was false?”


A few months ago, I started think­ing about the Tul­lock para­dox: “Why is there so lit­tle money in poli­tics?” That is, given the ex­traor­di­nar­ily large re­wards available to an in­ter­est that cap­tures the co­op­er­a­tion of gov­ern­ment, why is out­right cor­rup­tion so rare, and why is in­dus­try’s ex­pen­di­ture on poli­ti­cal in­fluence rel­a­tively small? Tul­lock’s ob­ser­va­tion was made mostly with re­gard to cam­paign con­tri­bu­tions, but it ap­plies just as well to lob­by­ing: the mar­ket for new ve­hi­cles in the U.S. was worth nearly half a trillion dol­lars in 2019, but the en­tire au­to­mo­tive in­dus­try spent only $70 mil­lion on lob­by­ing -- 0.01% of the value of the mar­ket.

Tul­lock’s ob­ser­va­tion sug­gests that, for groups that want some­thing from gov­ern­ment, lob­by­ing is po­ten­tially an ex­traor­di­nar­ily long lever with which to move policy. The ques­tion mo­ti­vat­ing this re­view is whether that lever is also available for al­tru­is­tic ends.

This re­view has three pur­poses:

1) To serve as a refer­ence re­gard­ing the effec­tive­ness of in­for­ma­tional lob­by­ing for groups or in­di­vi­d­u­als con­sid­er­ing fund­ing ad­vo­cacy or in­fluence cam­paigns.

2) To provide a start­ing point for oth­ers who may be in­ter­ested in con­duct­ing fur­ther re­search them­selves.

3) To try to an­swer to my satis­fac­tion the ques­tion of whether lob­by­ing works be­fore we try to un­cover the most effec­tive lob­by­ing tac­tics and tech­niques.

A caveat

It’s clear that lob­by­ing has a nega­tive con­no­ta­tion. Sto­ries about the per­ni­cious im­pacts of lob­by­ing by cigarette com­pa­nies, oil com­pa­nies, and pro­fes­sional or­ga­ni­za­tions are com­mon, and it seems clear that the in­fluence of moneyed in­ter­ests can hurt the pub­lic. But lob­by­ing is just a tool: in­for­ma­tional lob­by­ing, as I’ll dis­cuss fur­ther down, is no more or less than talk­ing to gov­ern­ment. Although the re­search I re­view here is fo­cused on pro­fes­sional lob­by­ists, lob­by­ing in gen­eral can be done by any­one, from K Street pro­fes­sion­als to high school stu­dents. I don’t take a moral stance on whether lob­by­ing in gen­eral is good or bad, or whether it should be re­stricted in some way.

My process


The em­piri­cal sec­tion of this re­view is in­ten­tion­ally nar­row. I fo­cus on lob­by­ing di­rected at leg­is­la­tive bod­ies at the na­tional level (e.g. the Se­nate and House) in the United States. Aside from my be­ing Amer­i­can and there­fore more fa­mil­iar with the Amer­i­can sys­tem, there are a few rea­sons for this nar­row­ness:

  • The struc­ture and effec­tive­ness of lob­by­ing at the fed­eral level is likely differ­ent than at the state or lo­cal level, or at the na­tional level in other coun­tries. It’s not clear to me that we can draw mean­ingful con­clu­sions if we mix these con­texts, so I choose not to. My re­view in­di­cates this is prob­a­bly a good choice: Benned­sen and Feld­mann (2002), for ex­am­ple, write that “ob­servers of the policy pro­cess are of­ten struck by the in­ten­sity of lob­by­ing—or lack thereof—in the sys­tem on the other side of the At­lantic.”

  • Since the U.S. Congress wields con­sid­er­ably more power than other bod­ies that are plau­si­bly sus­cep­ti­ble to lob­by­ing, de­ter­min­ing whether Congress can be in­fluenced seems (by com­par­i­son to oth­ers) to be a top pri­or­ity.

  • Lob­by­ing of fed­eral reg­u­la­tory agen­cies, which also seems like a very promis­ing av­enue for in­ves­ti­ga­tion, is likely sub­stan­tively differ­ent than lob­by­ing leg­is­la­tors. For the sake of clar­ity, I try not to mix these differ­ent set­tings.

Cru­cially, I fo­cus on in­for­ma­tional lob­by­ing: when lob­by­ists com­mu­ni­cate di­rectly with poli­cy­mak­ers to shift policy. Poli­cy­mak­ers in­clude mem­bers of Congress and their staffs, mem­bers of the ex­ec­u­tive branch, bu­reau­crats, and reg­u­la­tors. I ex­pressly do not in­ves­ti­gate the em­piri­cal im­pact of poli­ti­cal cam­paign con­tri­bu­tions or grass­roots ad­vo­cacy here. How­ever, the top­ics are closely tied to­gether, as I de­scribe at the be­gin­ning of the “The­o­ret­i­cal Back­ground” sec­tion.


This is a nar­ra­tive re­view. I have tried to syn­the­size what I have learned about the cur­rent state of aca­demic knowl­edge on in­for­ma­tional lob­by­ing at the level of the U.S. Congress, com­mu­ni­cate my read on the ev­i­dence, and high­light the main points of un­cer­tainty.

I took a snow­ball ap­proach and started with two ori­gin points: (1) a hand­ful of rele­vant Google Scholar queries (e.g. “em­piri­cal lob­by­ing effec­tive­ness”, “in­ter­est group in­fluence in poli­tics”), and (2) the em­piri­cal stud­ies de­scribed in the liter­a­ture re­views I sum­ma­rize be­low. Start­ing from these lists, I fol­lowed both back­ward- and for­ward-cita­tion chains un­til I was satis­fied I had en­coun­tered most of the main themes in the liter­a­ture and had read most of the em­piri­cal work rele­vant to the re­view. Not all ma­jor stud­ies are in­cluded, largely as a func­tion of my own re­source con­straints. John Mark Hansen’s 280-page Gain­ing Ac­cess: Congress and the Farm Lobby, 1919-1981 (1991), for ex­am­ple, does not get the at­ten­tion it de­serves here.

Read­ers may note that the the­o­ret­i­cal sec­tion is heavy on the poli­ti­cal sci­ence liter­a­ture and com­par­a­tively light on rele­vant sub­fields in eco­nomics, such as pub­lic choice the­ory. The pri­mary rea­son for this is that the liter­a­ture in pub­lic eco­nomics, while the­o­ret­i­cally in­struc­tive, is some­what or­thog­o­nal to my main con­cern here, which is this: How effec­tive, as a means of in­fluenc­ing policy, is the pro­vi­sion of in­for­ma­tion di­rectly to leg­is­la­tors, and what cir­cum­stances in­fluence a lob­by­ist’s chances of re­al­iz­ing the policy that s/​he ad­vo­cates?

Back­ground and Terminology

Why in­for­ma­tional lob­by­ing?

I use the defi­ni­tion pro­vided by De Figueiredo and Richter (2013): Lob­by­ing is “the trans­fer of in­for­ma­tion in pri­vate meet­ings and venues be­tween in­ter­est groups and poli­ti­ci­ans, their staffs, and agents.”

Lob­by­ing can be done by the in­ter­ested par­ties them­selves, but the term “lob­by­ist” gen­er­ally refers to in­di­vi­d­u­als and firms that at­tempt to in­fluence gov­ern­ment on be­half of pri­vate clients, such as cor­po­ra­tions or non­prof­its. It is also con­ven­tional to dis­t­in­guish be­tween “in­side” lob­by­ing, which con­sists of di­rect com­mu­ni­ca­tion with poli­cy­mak­ers, and “out­side” lob­by­ing, in which or­ga­nize in­ter­ests at­tempt to shift pub­lic opinion in hopes of af­fect­ing leg­is­la­tive de­ci­sions. This re­view con­cerns only in­side lob­by­ing.

Lob­by­ing and poli­ti­cal cam­paign con­tri­bu­tions, as to PACs, are both types of cor­po­rate poli­ti­cal ac­tivity (CPA), along with in-kind gifts, par­ti­ci­pa­tion in Cham­bers of Com­merce, le­gal ac­tion, grass­roots in­fluence, and other similar ac­tivi­ties. It’s nec­es­sary for me to spec­ify my fo­cus on in­for­ma­tional lob­by­ing be­cause lob­by­ing is of­ten con­flated with other kinds of CPA, par­tic­u­larly PAC con­tri­bu­tions, in both the aca­demic liter­a­ture and the main­stream press, as in this FT ar­ti­cle by the economist Tim Har­ford.

How­ever, not only are these av­enues of in­fluence struc­turally and legally very differ­ent, but spend­ing on lob­by­ing in a given year is typ­i­cally an or­der of mag­ni­tude larger than spend­ing on cam­paign con­tri­bu­tions. Some re­searchers (e.g. Lux, Crook and Woehr 2010) view the spend­ing differ­en­tial be­tween lob­by­ing and PACs as an in­di­ca­tion of the greater rel­a­tive im­por­tance of lob­by­ing.

My fo­cus on in­for­ma­tional lob­by­ing as op­posed to other av­enues of in­fluence is prag­matic. Ev­i­dence about the effi­cacy of in­for­ma­tional ac­tivi­ties is valuable and portable. I see two pos­si­ble take­aways from this re­search:

  • If lob­by­ing is effec­tive un­der cer­tain con­di­tions, well-re­sourced or­ga­ni­za­tions might be jus­tified in tak­ing a port­fo­lio (“hits-based”) ap­proach, pay­ing pro­fes­sional lob­by­ists to ad­vo­cate for a menu of highly im­pact­ful poli­cies meet­ing those con­di­tions, or con­duct­ing such ad­vo­cacy on their own ini­ti­a­tive. For a pro­posal on this front, please see the sec­tion en­ti­tled “Effec­tive Lob­by­ing.”

  • In­for­ma­tion about the cir­cum­stances un­der which poli­ti­cal in­fluence cam­paigns are suc­cess­ful can im­prove re­source al­lo­ca­tion effi­ciency for groups that already fund poli­ti­cal ad­vo­cacy.

How does lob­by­ing work?

In 2019, re­ported lob­by­ing ex­pen­di­tures to­taled around $3.5 billion. This money, spent largely by in­dus­try, pays the salaries of reg­istered lob­by­ists; it doesn’t go di­rectly to mem­bers of Congress or their staff. Lob­by­ists con­duct re­search on rele­vant top­ics, in­clud­ing the poli­ti­cal con­stituen­cies that may sup­port or op­pose a given policy, and set meet­ings with bu­reau­crats and poli­ti­ci­ans, which they use to provide in­for­ma­tion that they hope will sway votes or policy po­si­tions. Many firms keep lob­by­ists on re­tainer in or­der to pro­tect the sta­tus quo, and change their poli­ti­cal in­fluence strate­gies only when some new threat arises. This means that the effect of lob­by­ing can be hid­den: the ab­sence of new leg­is­la­tion can also be a goal of lob­by­ing. Un­like cam­paign con­tri­bu­tions, lob­by­ing ex­pen­di­tures are not limited by law; nev­er­the­less, most firms do not lobby—al­though they are of­ten mem­bers of trade as­so­ci­a­tions and other groups that do en­gage in poli­ti­cal ad­vo­cacy. There is some ev­i­dence that lob­by­ists some­times help leg­is­la­tors write leg­is­la­tion (Hall and Dear­dorff 2006).

In the U.S., lob­by­ists are re­quired by The Lob­by­ing Dis­clo­sure Act of 1995 to reg­ister with the gov­ern­ment and to file reg­u­lar re­ports on their ac­tivi­ties and on their spend­ing. A lob­by­ist is any­one who makes more than one com­mu­ni­ca­tion to a cov­ered offi­cial while on con­tract to a par­tic­u­lar client. Covered offi­cials in­clude not just mem­bers of the Se­nate and House but their em­ploy­ees, em­ploy­ees of Con­gres­sional com­mit­tees, etc. - re­gard­less of their job level or func­tion.

The Ex­ec­u­tive branch is also cov­ered by the Lob­by­ing Dis­clo­sure Act, but is less ex­ten­sive: only the Pres­i­dent, the Vice Pres­i­dent, Sched­ule C poli­ti­cal ap­poin­tees (such as a spe­cial coun­sel), and a hand­ful of other top-level po­si­tions are cov­ered by the Act.

How can lob­by­ists in­fluence policy?

  • Lob­by­ists can help to set the agenda. In other words, they can help to define the things that peo­ple in gov­ern­ment are pay­ing at­ten­tion to: mak­ing leg­is­la­tors aware of is­sues, cre­at­ing pub­lic aware­ness and pres­sure, or pub­li­ciz­ing causes and spe­cific items of leg­is­la­tion.

  • They can provide ev­i­dence and ar­gu­ments to leg­is­la­tors or staffers that ei­ther (a) per­suade them to change their opinion or (b) sup­ply them with rhetor­i­cal am­mu­ni­tion that they can use in com­mit­tee, dur­ing floor de­bate, or in me­dia ap­pear­ances.

  • They can provide out­side re­search that leg­is­la­tive offices or bu­reau­cra­cies may not have the in­ter­nal ca­pac­ity to de­velop on their own (see the “leg­is­la­tive sub­sidy” sec­tion be­low).

  • They can demon­strate al­ign­ment be­tween con­stituents and the lob­by­ists’ preferred policy out­comes. For ex­am­ple, lob­by­ists can con­duct pol­ling show­ing that a policy is preferred by a ma­jor­ity of a Se­na­tor’s con­stituents.

  • They can par­ti­ci­pate in “is­sue net­works” or sub­gov­ern­ments, at­tempt­ing to shift policy over a longer time scale by pub­li­ciz­ing and prop­a­gat­ing their preferred poli­cies and stances within a given policy do­main.

The­o­ret­i­cal background

I don’t claim to in­clude all rele­vant the­o­ries in what fol­lows, but I aim to make some use­ful dis­tinc­tions and syn­the­size some of the main themes in the liter­a­ture. I do not in­clude the siz­able liter­a­ture on the drivers of policy change in gen­eral; for that I re­fer the reader to com­pen­dia such as The­o­ries of the Policy Pro­cess (Sa­batier and Weible 2018).

Ex­change and persuasion

The­o­ret­i­cal work about poli­ti­cal in­fluence has tended to fall into two broad (though not ex­clu­sive) cat­e­gories. In gen­eral, there are “ex­change the­o­ries” and “the­o­ries of per­sua­sion” (Hall & Dear­dorff 2006).

In ex­change mod­els, cam­paign con­tri­bu­tions or in-kind gifts are ex­changed for Con­gres­sional votes. This is, to a first ap­prox­i­ma­tion, the model as­sumed in most lay con­ver­sa­tions about “money in poli­tics.” There is mixed ev­i­dence re­gard­ing this phe­nomenon, also known as vote buy­ing. For a good (if now some­what dated) lit re­view on the topic see An­solabehere, de Figueiredo, and Sny­der (2003).

In per­sua­sion mod­els, in­for­ma­tion is at the cen­ter of the re­la­tion­ship be­tween leg­is­la­tors and lob­by­ists. Lob­by­ists, in this view, can use in­for­ma­tion to ei­ther (1) gen­uinely per­suade leg­is­la­tors or (2) con­vince self-in­ter­ested leg­is­la­tors that their chances of elec­toral suc­cess would be en­hanced by tak­ing the ad­vo­cated po­si­tion. Lob­by­ists have a com­par­a­tive ad­van­tage in their abil­ity to ex­pend sig­nifi­cant re­sources col­lect­ing and pre­sent­ing such in­for­ma­tion, e.g. polls of rele­vant con­stituen­cies.

Money, ac­cess, and information

As Wright (1990) notes, many stud­ies on the im­pact of con­tri­bu­tions on Con­gres­sional vot­ing fail to make a dis­tinc­tion be­tween con­tri­bu­tions and in­for­ma­tional lob­by­ing, which makes it im­pos­si­ble to sep­a­rate the im­pact of lob­by­ing from the im­pact of the con­tri­bu­tions them­selves. This puts a finer point on an im­por­tant way that the ex­change and per­sua­sion mod­els come to­gether: money may not buy votes, but it’s pos­si­ble that money buys ac­cess.

Leg­is­la­tors have limited time, and if they pri­ori­tize con­trib­u­tors when ar­rang­ing meet­ings and phone calls, then these big spenders will have more op­por­tu­ni­ties to in­fluence policy. This is the view ad­vanced by Wright, as well as by Austen-Smith (1995), and sup­ported by some em­piri­cal ev­i­dence (Tri­pathi, An­solabehere and Sny­der 2000 and Lang­bein 1986). Con­tri­bu­tions may also serve a sig­nal­ing func­tion to leg­is­la­tors with refer­ence to pos­si­ble fu­ture lob­by­ing, demon­strat­ing that in­ter­est groups have both the re­sources and the in­tent to provide nec­es­sary in­for­ma­tion.

The­o­ret­i­cal mod­els of lob­by­ist behavior

Plu­ral­ism and the com­mu­ni­ca­tions approach

In the mid-20th cen­tury, a promi­nent strain in poli­ti­cal the­ory was the plu­ral­ist model, which con­ceived of policy as be­ing pri­mar­ily a re­sult of com­pe­ti­tion be­tween di­verse in­ter­est groups (Den­zau and Munger 1986). To plu­ral­ist thinkers like Tru­man (1951) and Latham (1952), com­pe­ti­tion be­tween in­ter­est groups was a source of sta­bil­ity and a fun­da­men­tal com­po­nent of democ­racy. Archety­pal plu­ral­ist works (e.g. Dahl 1956) held that poli­ti­cal out­comes are the re­sult of strug­gle be­tween in­ter­est groups, who com­pete within spe­cific is­sues or policy ar­eas and tend to be in­fluen­tial pri­mar­ily within those do­mains, rather than dom­i­nant in so­ciety in gen­eral. In the plu­ral­ist view, power is dis­persed across so­ciety; policy is a re­sult of me­di­a­tion and healthy dis­cus­sion be­tween a large num­ber of com­pet­ing in­ter­ests. The in­fluence of in­ter­est groups is viewed as a be­nign rep­re­sen­ta­tion of the will of the peo­ple, weighted by its salience to a spe­cific sub­set of cit­i­zens (Low­ery and Gray 2004). The plu­ral­ism hy­poth­e­sis im­plies that in­ter­est groups pri­mar­ily serve the pur­pose of rais­ing leg­is­la­tors’ con­scious­ness of the prefer­ences of ex­ist­ing con­stituen­cies; they can­not make leg­is­la­tors do things that are against their own elec­toral in­ter­ests.

Plu­ral­ism fell from fa­vor un­der the weight of em­piri­cal re­search find­ing (a) that Amer­i­can vot­ers were largely ig­no­rant and ap­a­thetic about poli­tics and (b) that or­ga­nized in­ter­ests seemed to work less like “pres­sure groups” and more like “ser­vice offices” for mem­bers of Congress already agree­ing with them. At the time, lob­by­ists spent con­sid­er­able re­sources com­mu­ni­cat­ing with already-friendly leg­is­la­tors (Bauer, Poole and Dex­ter 1963). Sev­eral other works of the same pe­riod and much later came to similar con­clu­sions, hold­ing broadly that lob­by­ists are at the beck and call of leg­is­la­tors, who have a great deal of agency in how they al­lo­cate their time (Ellis 1992).

In his book The Wash­ing­ton Lob­by­ist (1963), Lester Milbrath pre­sented in­ter­est group poli­tics as a game of com­mu­ni­ca­tion, rather than of in­fluence. The com­mu­ni­ca­tions ap­proach pro­vided a com­pel­ling ex­pla­na­tion for the ten­dency of groups to lobby already-agree­ing leg­is­la­tors: tar­get­ing op­pos­ing leg­is­la­tors con­sti­tuted an at­tempt to pass a mes­sage through a closed chan­nel. Milbrath, along with Bauer et al., is largely re­spon­si­ble for the mod­ern con­cep­tion of lob­by­ing as be­ing largely about in­for­ma­tion. (McGrath 2017).

The sup­ply price for policy

Den­zau and Munger (1986), build­ing partly on ear­lier ideas de­vel­oped by Sti­gler (1971), offered a for­mal model pre­dict­ing that leg­is­la­tors are most likely to re­spond to “or­ga­nized group pres­sures” when con­stituents are un­in­formed or in­differ­ent. In their model, the amount that in­ter­est groups must offer a leg­is­la­tor for his ser­vices (e.g. his vote) de­pends on the leg­is­la­tor’s pro­duc­tivity and the prefer­ences of vot­ers in his dis­trict. The price of policy de­pends on a leg­is­la­tor’s com­par­a­tive ad­van­tage in pro­vid­ing votes. Com­par­a­tive ad­van­tage is de­ter­mined by the hos­tility of vot­ers at home to a given policy and the pro­duc­tive ca­pac­ity of the in­di­vi­d­ual leg­is­la­tor (e.g. se­nior­ity, com­mit­tee as­sign­ment, com­pe­tence). The pri­mary con­clu­sion of this in­fluen­tial pa­per is that “con­tri­bu­tions can have some in­fluence on poli­cies about which vot­ers are di­vided, ig­no­rant, or in­differ­ent.”

Coun­ter­ac­tive lobbying

Austin-Smith & Wright (1994) pro­posed a the­ory of coun­ter­ac­tive lob­by­ing in re­sponse to the by-then main­stream aware­ness that in­ter­est groups of­ten lob­bied leg­is­la­tors pre­dis­posed to agree with them. Coun­ter­ac­tive lob­by­ing oc­curs when groups lobby in or­der to offset the pres­sure ap­plied by op­pos­ing in­ter­ests. Th­ese au­thors pro­pose that lob­by­ing does, in fact, in­fluence the po­si­tions taken by leg­is­la­tors, but that lob­by­ing by friendly in­ter­ests takes place in or­der to pre­vent al­lied leg­is­la­tors from chang­ing their po­si­tions.

Leg­is­la­tive subsidy

Hall & Dear­dorff (2006) have pro­posed a leg­is­la­tive sub­sidy model, in which lob­by­ing op­er­ates with re­spect to bud­get, not to prefer­ences. Lob­by­ists offer a “gift” to like-minded leg­is­la­tors, pro­vid­ing as­sis­tance in the form of leg­is­la­tive in­tel­li­gence, in­for­ma­tion, con­nec­tions to other groups and sup­port­ers, struc­tured ar­gu­ments, and other types of free in­tel­lec­tual la­bor. This more re­cent model, like the coun­ter­ac­tive lob­by­ing the­ory, ad­dresses the demon­strated ten­dency of lob­by­ists to spend time and effort lob­by­ing those who already agree with them. The sub­sidy model and other mod­els are not mu­tu­ally ex­clu­sive; lob­by­ists are free to se­lect from a “menu” of available strate­gies. This model is (for ex­am­ple) con­sis­tent with the Den­zau & Munger model de­scribed above: lob­by­ing un­der the sub­sidy model can be seen as an in-kind con­tri­bu­tion to leg­is­la­tors, so the sup­ply price of policy is coter­mi­nous with lob­by­ing la­bor pur­chased. Hall & Dear­dorff nonethe­less ar­gue that “prefer­ence-cen­tered” lob­by­ing (in which lob­by­ists seek to change leg­is­la­tors’ prefer­ences) is rel­a­tively rare.


Smith (1984) ob­served the lob­by­ing ac­tivi­ties of the Na­tional Ed­u­ca­tion As­so­ci­a­tion in the spring and sum­mer of 1977 and pro­posed a model in which ad­vo­cates—in­clud­ing not just lob­by­ists but also fel­low mem­bers of Congress, staffers and oth­ers— in­fluence mem­bers’ in­ter­pre­ta­tions of the con­se­quences of a policy. Smith pro­poses a two-stage model of policy for­mu­la­tion. In the first stage, mem­bers in­ter­pret the policy; in the sec­ond, they at­tempt to dis­cern how their in­ter­preted policy con­se­quences im­pact their own policy goals. Ad­vo­cates can shift policy by demon­strat­ing to leg­is­la­tors how a given policy could ad­vance their own goals. How­ever, since ba­sic hu­man con­straints (e.g. bounded ra­tio­nal­ity) cause in­ter­pre­ta­tions to be un­sta­ble, ad­vo­cates’ in­fluence on policy is also it­self un­sta­ble.

Poli­ti­cal accountability

Smith (1995) sin­gles out Arnold (1990) as hav­ing a rare com­pre­hen­sive the­ory in which policy out­comes are driven by (1) whether leg­is­la­tors per­ceive con­stituents as at­ten­tive or inat­ten­tive to the is­sue at hand, (2) how salient the is­sue is to con­stituents, and (3) how ca­pa­ble con­stituents are of hold­ing leg­is­la­tors ac­countable. Arnold’s the­ory holds that pres­sure suc­ceeds only when an is­sue is viewed as im­por­tant to at­ten­tive cit­i­zens and not to inat­ten­tive cit­i­zens, and when it has cham­pi­ons in Congress who will do ev­ery­thing “pro­ce­du­rally nec­es­sary” to en­sure that leg­is­la­tors are held ac­countable by con­stituents.

How lob­by­ing is stud­ied empirically

Re­search about poli­ti­cal in­fluence has tended to fo­cus over­whelm­ingly on cam­paign con­tri­bu­tions, to the ne­glect of in­for­ma­tional lob­by­ing. Sev­eral of the pa­pers I re­viewed (e.g. Evans 1996, Bom­bar­dini and Trebbi 2019, Caldeira and Wright 1998) sug­gest that the dis­pro­por­tionate at­ten­tion af­forded to cam­paign con­tri­bu­tions is mostly a func­tion of greater data availa­bil­ity; at least one (Milyo, Primo and Grose­close 2000) sug­gests that this ten­dency in the liter­a­ture is a case of “look­ing un­der the lamp­post”: data about cam­paign con­tri­bu­tions has his­tor­i­cally been eas­ier to come by.

Data sources

While in­for­ma­tion about fed­eral poli­ti­cal con­tri­bu­tions has been available since 1974, data about fed­eral lob­by­ing ex­pen­di­tures has only been available since 1999 as a con­di­tion of the Lob­by­ing Dis­clo­sure Act. The Hon­est Lead­er­ship and Open Govern­ment Act of 2007 made elec­tronic filing manda­tory, and data since 2015 is now available from the Se­nate. Se­nate data is for­mat­ted as XML; more helpfully struc­tured data is available from the Cen­ter for Re­spon­sive Poli­tics (CRP). A con­sid­er­able por­tion of re­cent re­search on this topic uses data from CRP.

A sig­nifi­cant chunk of the re­search I con­sider be­low fo­cuses on the re­la­tion­ship be­tween lob­by­ing ex­pen­di­ture and firms’ fi­nan­cial out­comes; these stud­ies of­ten merge CRP data with busi­ness in­for­ma­tion from Com­pu­s­tat. Some sources also use lob­by­ (now wash­ing­ton­rep­re­sen­ta­ for bi­o­graph­i­cal in­for­ma­tion about lob­by­ists and about their re­la­tion­ships.

Some en­ter­pris­ing re­searchers (e.g. Evans 1996) have pur­sued much more la­bor-in­ten­sive data col­lec­tion op­er­a­tions, with sig­nifi­cant amounts of hand-cod­ing and man­ual re­view. Baum­gart­ner et al 2009, one of the most im­pres­sive re­cent works on this topic, fol­lowed a ran­dom sam­ple of is­sues over a four-year pe­riod, col­lect­ing in­for­ma­tion via large num­bers of in­ter­views with lob­by­ists and poli­cy­mak­ers and hand-cod­ing their re­sponses.

Re­search designs

To my knowl­edge, there has never been a con­trol­led ex­per­i­ment about in­for­ma­tional lob­by­ing— this kind of thing would be tricky to pull off, though it’s prob­a­bly not out of the ques­tion. In gen­eral, work at­tempt­ing to as­sess poli­ti­cal in­fluence through com­mu­ni­ca­tions chan­nels takes one or more of the fol­low­ing ap­proaches:

  • Cross-sec­tional studies

    • e.g. Alexan­der, Scholz and Mazza (2009), who model tax sav­ings due to repa­tri­ated earn­ings from the Amer­i­can Jobs Creation Act of 2004 (AJCA) as a func­tion of lob­by­ing ex­pen­di­tures by firms.

  • Panel studies

    • e.g. Meade and Li (2015), who fol­low a group of firms over a 13-year pe­riod and model effec­tive tax rate as a func­tion of lob­by­ing ex­pen­di­ture with year and firm fixed effects.

  • In­stru­men­tal vari­ables (IV) /​ 2-stage least squares (2SLS)

    • e.g. Gawande, Maloney and Montes-Ro­jas (2009), who find a source of out­side vari­a­tion in lob­by­ing ex­pen­di­ture (an in­stru­men­tal vari­able, in econo­met­rics-speak), and use it to es­ti­mate the effect of pro-tourism lob­by­ing by for­eign coun­tries on tourism from the U.S. to the lob­by­ing na­tions. (See here for a good ex­pla­na­tion of the IV de­sign)

  • In­ter­views and Surveys

    • Baum­gart­ner et al (2009) con­tains a con­sid­er­able statis­ti­cal el­e­ment, but the re­search is heav­ily in­formed by qual­i­ta­tive in­ter­views that provide a lot of back­ground con­text (and were also put into struc­tured form and an­a­lyzed)

  • Case studies

    • In ad­di­tion to in­ter­views, King­don (1984) re­ports and codes policy case stud­ies in or­der to iden­tify where in­ter­est groups seem to be im­por­tant.

Pre­vi­ous Liter­a­ture Reviews

Smith (1995)

Smith iden­ti­fies a body of work that con­cludes, from for­mal mod­els of the lob­by­ist-leg­is­la­tor re­la­tion­ship, that a pri­mary per­sua­sive virtue of lob­by­ists is their abil­ity to af­fect de­ci­sion-mak­ing by re­duc­ing leg­is­la­tors’ un­cer­tainty about bill con­se­quences. Smith finds lit­tle em­piri­cal sup­port for these mod­els, with the ex­cep­tion of Austin-Smith and Wright’s (1992) the­ory of coun­ter­ac­tive lob­by­ing, for which the au­thors find em­piri­cal sup­port in Austin-Smith and Wright (1994).

Smith iden­ti­fies eight stud­ies from the pre­ced­ing decade that as­sess the im­pact of lob­by­ing on roll call votes, and finds that the re­search pre­sents a “mixed pic­ture” of the pos­si­ble effects of lob­by­ing. Smith finds that most em­piri­cal stud­ies suffer from mea­sure­ment and speci­fic­ity er­rors, and draws no firm con­clu­sions about the effec­tive­ness of lob­by­ing in gen­eral.

Pot­ters and Sloof (1996)

This re­view ex­plic­itly fo­cuses on quan­ti­ta­tive data around “how and when” in­ter­est groups in­fluence pub­lic policy, but does not dis­t­in­guish be­tween differ­ent types of in­ter­est group in­fluence (in­for­ma­tional lob­by­ing is among those types stud­ied). The au­thors make a use­ful dis­tinc­tion be­tween two types of vari­ables in em­piri­cal in­ter­est group re­search: (1) the be­hav­ior of in­di­vi­d­ual poli­cy­mak­ers and (2) policy out­comes. Pot­ters and Sloof find that em­piri­cal re­search on in­ter­est group in­fluence out­side of cam­paign con­tri­bu­tions is rare. A key ob­ser­va­tion of this pa­per is that sin­gle-equa­tion mod­els are of­ten used to es­ti­mate in­ter­est group in­fluence, but that in re­al­ity causal­ity flows in both di­rec­tions: poli­ti­ci­ans’ po­si­tions in­fluence group be­hav­ior, as well as the re­verse.

Pot­ters and Sloof con­clude that there is ev­i­dence for the effec­tive­ness of lob­by­ing, par­tic­u­larly when bills have “a nar­row fo­cus and low pub­lic visi­bil­ity.” In ad­di­tion, they find that in­ter­est groups have greater suc­cess in­fluenc­ing policy when poli­ti­ci­ans are un­der low demo­cratic pres­sure (e.g. low in­tra-party com­pe­ti­tion) and less suc­cess when there is a well-in­formed elec­torate.

Burstein and Lin­ton (2002)

Burstein and Lin­ton an­a­lyze ev­ery ar­ti­cle on policy change in the top three so­ciol­ogy and top three poli­ti­cal sci­ence jour­nals. Their unit of anal­y­sis is the effect of a pre­dic­tor on a de­pen­dent vari­able, e.g. be­tween a mea­sure of ei­ther the re­sources or ac­tivi­ties of an or­ga­ni­za­tion and a policy out­come. They find 53 ar­ti­cles con­tain­ing a to­tal of 230 effects. They find that statis­ti­cal sig­nifi­cance is a more com­mon stan­dard in this liter­a­ture than sub­stan­tive sig­nifi­cance, and re­viewed find­ings us­ing a scale that takes both stan­dards into ac­count. Burstein and Lin­ton find that the im­pact of in­ter­est groups is statis­ti­cally sig­nifi­cant roughly half the time, and sub­stan­tively im­por­tant (ac­cord­ing to the au­thors’ own as­sess­ment) in about a fifth of cases.

Burstein and Lin­ton con­clude that the hy­poth­e­sis that in­ter­est groups in­fluence policy sig­nifi­cantly is “not as well sup­ported by the data as we might ex­pect.” How­ever, they offer three po­ten­tial sources of bias in their es­ti­mate.

First, they sug­gest that their es­ti­mate of the fre­quency with which poli­ti­cal or­ga­ni­za­tions in­fluence policy may be too high rel­a­tive to its ac­tual effect—they note that few stud­ies in­clude vari­ables for pub­lic opinion, and sug­gest that the in­clu­sion of such a vari­able might cause other ap­par­ently sig­nifi­cant re­la­tion­ships to dis­ap­pear.

Se­cond, they sug­gest that their es­ti­mate may be too low: they note that the­ory pre­dicts that it is the way in which re­sources are de­ployed, not the size of the re­sources them­selves, that in­fluences policy; nonethe­less, most stud­ies rely on data about re­sources, re­gard­less of de­ploy­ment. This the­sis is sup­ported by their data: mea­sures of re­sources alone im­pacted policy 45% of the time, but re­sources and ac­tivi­ties to­gether im­pacted policy 65% of the time.

Fi­nally, they sug­gest that the range of ex­tant lob­by­ing re­search is very nar­row— it cov­ers only a small num­ber of is­sues and coun­tries, and may sim­ply not be rep­re­sen­ta­tive enough to draw mean­ingful gen­eral con­clu­sions.

Low­ery (2013)

Low­ery ad­dresses him­self to the find­ings of three pre­vi­ous liter­a­ture re­views—Smith (1995), Baum­gart­ner and Leech (1998) and Burstein and Lin­ton (2002)—which held that ev­i­dence about the effec­tive­ness of lob­by­ing is in­con­clu­sive or in­con­sis­tent. Low­ery starts from the premise that null find­ings are com­mon—as­sum­ing that, due to the file drawer effect, sig­nifi­cant find­ings are even rarer than they at first ap­pear—and offers some ex­pla­na­tions for why that might be so.

Low­ery sug­gests sev­eral pos­si­ble ex­pla­na­tions for null find­ings:

  • The most im­por­tant in­stances of in­fluence may not be observable

  • Is­sues com­pete for space on the agenda, and ex­oge­nous forces, not in­ter­est groups, ul­ti­mately de­ter­mine what ends up on the agenda.

  • Lob­by­ing is two-sided, and a failure on one side is a suc­cess on an­other. Null find­ings may sim­ply leave out the in­fluence of ma­jor in­di­rect de­ter­mi­nants of policy, such as pub­lic opinion.

  • Selec­tion bias: The most com­mon type of re­search is an anal­y­sis of a small num­ber of groups work­ing on one is­sue over a spe­cific amount of time, and most in­fluence stud­ies fo­cus on “sub­stan­tive large and con­tro­ver­sial is­sues on which many or­ga­ni­za­tions are en­gaged.” How­ever, “these are pre­cisely the is­sues on which in­fluence might be least ex­pected be­cause pub­lic opinion is likely es­pe­cially con­strain­ing.” In point of fact, most policy pro­pos­als see lit­tle or no lob­by­ing.

    • A quote that gave me a lot to think about: “It could well be that lob­by­ing is far more suc­cess­ful on these kinds of pro­pos­als. But we will not know this for sure un­less we stop con­fus­ing study­ing lob­by­ing in­fluence with study­ing lob­by­ing in­fluence on con­tro­ver­sial is­sues.”

De Figueiredo and Richter (2013)

De Figueiredo and Richter list “em­piri­cal reg­u­lar­i­ties” in the study of lob­by­ing. The un­con­tro­ver­sial topline sum­mary is that lob­by­ing is per­va­sive, that lob­by­ing in­creases when is­sues are more salient or when stakes are high for an or­ga­nized in­ter­est, and that both ex­per­tise and con­nec­tions seem to mat­ter for lob­by­ist effec­tive­ness, but the mag­ni­tude of these effects is un­clear.

They also list three main challenges in the em­piri­cal study of lob­by­ing:

  • Time se­ries per­sis­tence: very lit­tle within-in­ter­est-group variation

  • Omit­ted vari­ables: datasets typ­i­cally used to study lob­by­ing may not con­tain im­por­tant data nec­es­sary to ad­just for these variables

  • En­do­gene­ity: groups may be more likely to lobby when they be­lieve they are more likely to succeed

In gen­eral, the au­thors find that lob­by­ing “can be valuable in some con­texts” but that its suc­cess is heav­ily de­pen­dent on timing, tar­gets, and tac­tics.

Stud­ies that were in­stru­men­tal in my conclusions

You can read sum­maries and key find­ings for 25 of the stud­ies I re­viewed in a Google sheet here. What fol­lows are brief sum­maries of the stud­ies that I found most use­ful and con­vinc­ing.

Caldeira and Wright (1998)

Caldeira and Wright model in­di­vi­d­ual Se­nate votes in the Bork, Souter, and Thomas Supreme Court nom­i­na­tions as a func­tion of lob­by­ing. They use an in­stru­men­tal vari­ables ap­proach in which votes are mod­eled (in the sec­ond stage) as a func­tion of par­ti­san af­fili­a­tion, ide­olog­i­cal ori­en­ta­tion, con­stituency sup­port for the nom­i­nee, and a lob­by­ing in­dex con­structed (in the first stage) as a func­tion of party, ide­ol­ogy, con­stituency, cam­paign con­tri­bu­tion, and or­ga­ni­za­tional strength (e.g. the num­ber of in­ter­est groups lob­by­ing for and against the nom­i­nee) that con­tribute in­de­pen­dently to the value of the lob­by­ing in­dex. They find a sig­nifi­cant re­la­tion­ship be­tween lob­by­ing and Bork vote; the au­thors find coun­ter­fac­tu­ally that if the op­po­si­tion lob­by­ing effort had been 10% smaller, the op­po­si­tion to Bork would have de­creased by two votes. In the case of the Thomas nom­i­na­tion, the au­thors find that “a 25% re­duc­tion in op­po­si­tion lob­by­ing effort would have gar­nered an ad­di­tional 33 votes for Thomas.”

De Figueiredo and Silver­man (2006)

De Figueiredo and Silver­man model aca­demic ear­marks for uni­ver­si­ties against lob­by­ing ex­pen­di­ture by those in­sti­tu­tions in an in­stru­men­tal vari­ables set­ting. They find a marginal $5 re­turn to an ad­di­tional dol­lar of lob­by­ing ex­pen­di­ture. How­ever, this re­la­tion­ship only holds for uni­ver­si­ties with rep­re­sen­ta­tion on Con­gres­sional ap­pro­pri­a­tion com­mit­tees.

Baum­gart­ner et al (2009)

This book, Lob­by­ing and Policy Change: Who Wins, Who Loses, and WHY is a con­tem­po­rary clas­sic. Baum­gart­ner et al take a ran­dom sam­ple of is­sues and con­duct a cor­re­la­tional anal­y­sis, fol­low­ing the is­sues over sev­eral years and col­lect­ing data pri­mar­ily via in­ter­views with lob­by­ists. Though this pro­cess has its weak­nesses, the au­thors pre­sent a num­ber of con­clu­sions that I view as broadly con­vinc­ing cor­re­la­tional ev­i­dence when viewed in light of the full scope of re­lated re­search.

Key find­ings:

  • Re­sources in gen­eral have no sig­nifi­cant cor­re­la­tion with a pos­i­tive policy outcome

  • “When the mo­bi­liza­tion of re­sources is un­bal­anced, the wealthy side tends to win.” In­creased com­par­a­tive re­source ad­van­tage is as­so­ci­ated with suc­cess main­tain­ing the sta­tus quo and greater suc­cess achiev­ing a policy change. How­ever, in most cases there is no such im­bal­ance.

  • Policy change hap­pens over a long time frame.

  • “The most con­sis­tent find­ing through­out our book is that defen­ders of the sta­tus quo usu­ally get what they want.”

  • Lob­by­ing fol­lows a power law dis­tri­bu­tion: most lob­by­ing oc­curs in a small num­ber of cases.

  • High-level gov­ern­ment al­lies are more im­por­tant for suc­cess than fi­nan­cial re­sources: par­tic­u­larly at the level of Con­gres­sional lead­er­ship and the pres­i­dency—when they get in­volved, they will likely win.

  • Baum­gart­ner et al find sup­port for the coun­ter­ac­tive lob­by­ing thesis

Montes-Ro­jas (2013)

Montes-Ro­jas mod­els for­eign aid dis­bursed by the U.S. as a func­tion of lob­by­ing and finds that a 1% in­crease in lob­by­ing ex­pen­di­tures in a given year is as­so­ci­ated with a 0.04% in­crease in for­eign aid the fol­low­ing year, and up to 0.075% in the long run. I want to note that this effect is huge, given the differ­ences in mag­ni­tude be­tween lob­by­ing ex­pen­di­ture and for­eign aid. For­eign lob­by­ing ex­pen­di­tures to­tal about half a billion dol­lars, but the for­eign aid bud­get is around $40 billion. Montes-Ro­jas uses an IV strat­egy in which lagged val­ues of the pa­ram­e­ter of in­ter­est are used as in­stru­men­tal vari­ables; this is not as crazy as it sounds, and I recom­mend you read the pa­per, which leans heav­ily on the work of David Rood­man.

Gold­stein and You (2017)

In a pretty novel ap­proach, these au­thors ex­am­ine the im­pact of lob­by­ing ex­pen­di­ture by U.S. cities with pop­u­la­tions over 25,000 on dis­burse­ments via ear­marks and the Amer­i­can Re­cov­ery and Rein­vest­ment Act. They also use a novel in­stru­ment: the ex­is­tence of a di­rect flight be­tween a given city and Wash­ing­ton, D.C, which the au­thors ar­gue is a source of ex­oge­nous vari­a­tion in lob­by­ing ex­pen­di­ture. They pre­sent sev­eral con­vinc­ing ro­bust­ness checks on this in­stru­ment in their Ap­pendix, but an un­ad­dressed area of un­cer­tainty is whether the ex­is­tence of a di­rect flight can cause leg­is­la­tors to be more fa­mil­iar with a city and there­fore more likely to dis­burse funds to it. If so, this could com­pro­mise the in­stru­ment.

Gold­stein and You find that a 10% in­crease in lob­by­ing ex­pen­di­tures is as­so­ci­ated with in­creases in ear­marks and ARRA grants of 10% and 4%, re­spec­tively. As in the case of Montes-Ro­jas (2013), this fram­ing can ob­scure the mag­ni­tude of the im­pact. Cities spent about $20k on lob­by­ing on av­er­age, and re­ceived an av­er­age of around $1m and $130m in ear­marks and ARRA funds re­spec­tively, so the co­effi­cients sug­gest marginal re­turns of 50x and 2600x, re­spec­tively.

Ludema, Mayda and Mishra (2018)

Ludema, Mayda, and Mishra ex­am­ine tar­iff sus­pen­sion bills, of which 1400 were in­tro­duced in Congress be­tween 1999 and 2006. They model pas­sage of these bills as a func­tion of lob­by­ing ex­pen­di­ture for the Con­gresses span­ning that time pe­riod and use a set of in­stru­ments that they ar­gue (con­vinc­ingly) are not read­ily available to leg­is­la­tors and are there­fore un­likely to in­fluence tar­iff sus­pen­sions. They find that in­creases in effec­tive lob­by­ing ex­pen­di­ture (ex­pen­di­tures rel­a­tive to a min­i­mum baseline) in op­po­si­tion to sus­pen­sions are sig­nifi­cantly as­so­ci­ated with sus­pen­sion, ex­pen­di­ture in sup­port is like­wise as­so­ci­ated with no sus­pen­sion (with a smaller effect size); like­wise num­ber of op­po­nents.


Does lob­by­ing work?

A very heav­ily qual­ified yes. I think that there are enough good-qual­ity quan­ti­ta­tive stud­ies, with enough con­vinc­ing the­o­ret­i­cal back­ing, to con­clude that in­for­ma­tional lob­by­ing can effec­tively shift policy. There are a num­ber of rea­son­able con­cerns with the state of the re­search: much of the ex­ist­ing work is not causal in na­ture, and I found al­most no pub­lished pa­pers on this topic with null re­sults, sug­gest­ing that a bit of the file drawer effect may be at play here. For this rea­son, it’s not ad­vis­able to at­tempt to ex­tract any quan­ti­ta­tive es­ti­mate of the effec­tive­ness of lob­by­ing from the pub­lished work.

Two sep­a­rate fac­tors give me some con­fi­dence in this con­clu­sion. The first is the rel­a­tively high qual­ity of many of the quasi-ex­per­i­men­tal stud­ies and the ev­i­dence these higher qual­ity stud­ies rep­re­sent in ag­gre­gate. The sec­ond, per­haps more de­bat­able, fac­tor is my sense that some of the re­peat­ing pat­terns in cor­re­la­tional and ob­ser­va­tional work on this topic are best ex­plained by a causal story rather than omit­ted vari­ables. Th­ese pat­terns in­clude the ob­served as­so­ci­a­tion be­tween the net gap in lob­by­ing (sup­port­ing minus op­pos­ing) and policy suc­cess, con­sis­tently higher co­effi­cients on nega­tive lob­by­ing rel­a­tive to pos­i­tive lob­by­ing, and the broad agree­ment on con­di­tions for lob­by­ing suc­cess that I dis­cuss be­low.

When does lob­by­ing work?

Pre­vi­ous liter­a­ture re­views have iden­ti­fied a num­ber of stylized facts about the de­ter­mi­nants of lob­by­ing suc­cess that seem to have a strong em­piri­cal ba­sis. Please note that the scope of my re­view cen­tered on leg­is­la­tive lob­by­ing at the na­tional level, and that the fol­low­ing ar­eas of agree­ment should be con­sid­ered to be re­stricted to the leg­is­la­tive do­main.

The first area of broad agree­ment is that lob­by­ing suc­cess seems to be more likely on is­sues of low salience to the elec­torate, is­sues with low pub­lic visi­bil­ity, or is­sues about which the pub­lic is largely ig­no­rant. This is a the­o­ret­i­cal pre­dic­tion made by Arnold (1990) and Den­zau and Munger (1986) and ob­served em­piri­cally by e.g. Evans (1986). Liter­a­ture re­views con­ducted by de Figueiredo and Richter (2013) and Pot­ters and Sloof (1996) con­sider it well-es­tab­lished.

EDIT: I’ve in­cluded the above para­graph be­cause it ac­cu­rately re­flects the re­search refer­enced, but af­ter dig­ging in more de­tail into to the stud­ies refer­enced in the liter­a­ture re­views, I don’t have con­fi­dence that it’s ac­tu­ally the case.

The sec­ond area of broad agree­ment is that, all other things be­ing equal, lob­by­ing against poli­cies seems to be more effec­tive than lob­by­ing for poli­cies (see e.g. McKay 2011, Caldeira and Wright 1998, Ludema, Mayda and Mishra 2018); in other words, lob­by­ists for the sta­tus quo are like­lier to suc­ceed.

The third area of agree­ment is that hav­ing high-level al­lies, ei­ther in com­mit­tee lead­er­ship, the ex­ec­u­tive branch, or among Con­gres­sional staffers, is con­sis­tently as­so­ci­ated with policy suc­cess (see e.g. Baum­gart­ner et al 2009, de Figueiredo and Silver­man 2006, Evans 1996).

Some other ar­eas for which I found some em­piri­cal sup­port but no broad con­sen­sus are as fol­lows:

  • In­ter­est groups with a larger mem­ber­ship are more in­fluen­tial (Pot­ters and Sloof 1996)

  • In­ter­group con­flict dam­ages a group’s chances of reg­u­la­tory suc­cess. In ad­di­tion, con­flict is more likely to oc­cur on is­sues where there are win­ners and losers, e.g. reg­u­la­tory and re­dis­tribu­tive is­sues (Evans 1996).

  • Poli­cies that fail in a given Congress have a poor shot in sub­se­quent Con­gresses (Kang 2016)

Case Studies

I think it’s worth list­ing a few in­stances when minor in­ter­ven­tions have sig­nifi­cantly im­pacted policy on a large scale. I’m not picky about whether these ex­am­ples would have been quan­tifi­able by any of the meth­ods re­viewed here; I just want to give an idea of the set­tings in which lob­by­ing (ex­ter­nal or in­ter­nal) can in­fluence policy (in the bu­reau­cracy or the leg­is­la­ture) and some sense of the mag­ni­tude of its po­ten­tial effects. Th­ese cases demon­strate in­stances in which policy was mean­ingfully im­pacted sim­ply by com­mu­ni­ca­tion, even though they do not all re­sem­ble a typ­i­cal pic­ture of lob­by­ing.

  • Terry Daw­son, an ad­vi­sor to the House Science Com­mit­tee, was ap­par­ently re­spon­si­ble for in­fluenc­ing com­mit­tee chair Ge­orge Brown to re­quire, in a piece of 1991 leg­is­la­tion, that NASA in­ves­ti­gate as­ter­oid de­tec­tion and deflec­tion strate­gies (Mor­ton 2017 and Mor­ri­son 2005). The vast ma­jor­ity of such as­ter­oids are now known and tracked as part of the re­sult­ing pro­gram.

  • The fa­mous Ein­stein-Szilárd let­ter, though per­haps not com­monly thought of as lob­by­ing, di­rectly in­fluenced FDR’s de­ci­sion to ini­ti­ate the Man­hat­tan Pro­ject.

  • OPP’s His­tory of Philan­thropy in­cludes sev­eral salient ex­am­ples in­clud­ing the fol­low­ing:

    • Nu­clear arms con­trol re­search funded by the Carnegie and MacArthur Foun­da­tions re­sulted in a re­port co-au­thored by Ash­ton Carter, the Har­vard pro­fes­sor (and fu­ture Sec­re­tary of Defense). Carter briefed Se­na­tors Nunn and Lu­gar on the re­port and ap­par­ently helped to write the text of the re­sult­ing bill.

    • The Robert Wood John­son Foun­da­tion funded the SmokeLess States ini­ti­a­tive, which shifted in 2000 to fo­cus en­tirely on policy change. Ben Soskis’s OPP-funded re­search re­ports that var­i­ous eval­u­a­tions have found that SmokeLess States suc­cess­fully lob­bied for in­creased ex­cise taxes on to­bacco and other state-level poli­cies in­tended to com­bat smok­ing.

  • Cor­po­rate lob­by­ists have had con­sid­er­able suc­cess en­gag­ing in the type of leg­is­la­tive sub­sidy known as leg­is­la­tive draft­ing: they liter­ally write sec­tions of bills that be­come law. Ellis and Groll (2019) re­port that lob­by­ists from Ge­nen­tech and Citi­group have pro­vided talk­ing points or pro­posed lan­guage to law­mak­ers that found their way, ver­ba­tim, into leg­is­la­tion.

  • Stokes & Breetz (2018) re­port that dur­ing the early 2000s, the so­lar in­dus­try tar­geted Repub­li­can mem­bers of Congress “whose dis­tricts benefit­ted [sic] from so­lar em­ploy­ment.” Suc­cess came in 2005 with a Busi­ness So­lar In­vest­ment Tax Credit, which was meant to ex­pire at the end of 2007, but which was ex­tended and ex­panded for eight years as a part of the 2008 eco­nomic stim­u­lus. In 2016, so­lar leas­ing com­pa­nies again lob­bied suc­cess­fully for an ex­ten­sion of the tax credit. Some sources see the ITC as an im­por­tant fac­tor be­hind the re­cent growth of so­lar power.

“Effec­tive lob­by­ing”

As ob­served by many of the pa­pers refer­enced here, most is­sues face no lob­by­ing. If the coun­ter­ac­tive lob­by­ing the­ory is true, then the sim­ple act of in­tro­duc­ing lob­by­ing on one of these is­sues can stim­u­late op­po­si­tion; if the op­po­si­tion is from an op­po­nent will­ing to “out­bid” on policy, then they are likely to win, and lob­by­ing can fail, or fail to be cost-effec­tive. Yet in some cases there sim­ply is no or­ga­nized op­po­si­tion, and a rel­a­tively small in­vest­ment can mean­ingfully al­ter word­ing, put a policy on the agenda, or cause a bill to be­come law.

I want to pro­pose some­thing along the lines of “effec­tive lob­by­ing”: a rigor­ous ap­proach to in­sti­tu­tional-level change, start­ing with the leg­is­la­ture, that would take a port­fo­lio ap­proach to policy ad­vo­cacy.

Although my re­view has fo­cused on lob­by­ing Congress, any ap­proach to effec­tive lob­by­ing would have to op­er­ate with an eye to all pos­si­ble venues: state and lo­cal gov­ern­ments, reg­u­la­tory agen­cies, and the ex­ec­u­tive branch.

How would this work?

1) Iden­tify a set of high-im­pact poli­cies that meet the fol­low­ing crite­ria:

  • Very high ex­pected value (e.g. in DALYs)

  • Mass of dis­tri­bu­tion of po­ten­tial out­comes is mostly above zero

2) Iden­tify the poli­cies least likely to face or­ga­nized op­po­si­tion (e.g. from en­trenched and well-re­sourced in­ter­ests in­side or out­side the gov­ern­ment, or from the pub­lic)

3) Iden­tify what av­enues of lob­by­ing are most ap­pro­pri­ate for en­abling the policy goal

  • Deter­mine the level (e.g. reg­u­la­tory, con­gres­sional, state, mu­ni­ci­pal) at which a policy would be implemented

  • Iden­tify the ap­pro­pri­ate lever by which this policy could be brought about (e.g. via leg­is­la­tion, rules change, bud­get pro­cesses, or ex­ec­u­tive or­der)

4) Iden­tify the prox­i­mate course of ac­tion that would be re­quired to in­crease the like­li­hood of each such policy change in the near term:

  • Re-word­ing of an ex­ist­ing bill

  • Bring­ing the at­ten­tion of in­fluen­tial leg­is­la­tors to the policy (e.g. putting it on the agenda)

  • Nar­row lob­by­ing to shift votes for the pas­sage of a bill already on the floor or in com­mit­tee, or to al­low a bill to come to a vote in ei­ther setting

  • Broad ad­vo­cacy of bills already on the floor

  • Writ­ing model leg­is­la­tion in co­op­er­a­tion with leg­is­la­tive staffers

5) Fund re­search nec­es­sary to max­i­mize the effec­tive­ness of lob­by­ing for each policy

  • Low-cost pub­lic opinion polls in rele­vant con­stituen­cies to iden­tify pub­lic sup­port or opposition

  • Iden­ti­fi­ca­tion of key leg­is­la­tors or leg­is­la­tive staff who can serve as champions

  • In­de­pen­dent anal­y­sis, as by think tanks or ex­ter­nal re­search groups

  • Indi­rect lob­by­ing to en­courage poli­ti­cal al­lies to ex­pend re­sources on an is­sue.

6) For each policy in the port­fo­lio, fund lob­by­ing and ad­vo­cacy in­formed by the re­search pro­duced in (4).

7) Mea­sure impact

  • Any such pro­gram needs to clearly define policy suc­cess, sys­tem­at­i­cally track re­source ex­pen­di­ture, and —ideally—con­duct ex­per­i­ments in or­der to draw causal con­clu­sions about strat­egy choices.

The bulk of the work here needs to hap­pen in (1) and in (4). What would make this ap­proach “effec­tive” is the rigor­ous re­stric­tion of the policy port­fo­lio to only po­ten­tially very high-value poli­cies, and the de­ploy­ment of a sys­tem­atic ap­proach to track­ing and fa­cil­i­tat­ing policy suc­cess.

Does effec­tive lob­by­ing already ex­ist?

Tra­di­tional lob­by­ists are hired by clients, ei­ther for- or non-profit, to ad­vo­cate for spe­cific poli­cies. Once hired, their suc­cess or failure within the speci­fied do­main is only par­tially un­der their con­trol. As the re­search de­tailed here has found, policy suc­cess or failure is de­ter­mined mostly ex­oge­nously, and lob­by­ing suc­cess is heav­ily con­text-de­pen­dent. The ap­proach I’ve sug­gested is fo­cused on find­ing the over­lap be­tween poli­cies that have high so­cial value and those that face fa­vor­able con­texts for lob­by­ing suc­cess. So the first differ­ence is in is­sue se­lec­tion: tra­di­tional lob­by­ists ad­vo­cate as they’re hired to do, whether or not they’re likely to be suc­cess­ful, but fa­vor­able con­text is a nec­es­sary con­di­tion (though not a suffi­cient one) for effec­tive lob­by­ing.

There’s still the ques­tion of the de­gree to which lob­by­ists work­ing to­day ac­tively seek to max­i­mize their effi­ciency, and the de­gree to which they make use of the moun­tain of data that is presently available to as­sist in this end. To con­clu­sively an­swer these ques­tions would re­quire a whole sep­a­rate liter­a­ture re­view, but I did re­view some ex­ist­ing stud­ies that paint a broad pic­ture. Ho­j­nacki and Kim­ball (1998) find ev­i­dence that in­ter­est groups take into ac­count the de­gree of sup­port they be­lieve they have in a leg­is­la­tor’s dis­trict when de­cid­ing whether to lobby her and that in­ter­est groups do tar­get in­fluen­tial com­mit­tee mem­bers. Vic­tor (2007) re­views the liter­a­ture and finds that (a) in­ter­est group per­cep­tions of leg­is­la­tor knowl­edge about a topic im­pact their choice of whether to lobby di­rectly or in­di­rectly, that (b) in­ter­est groups con­sider how the pub­lic views an is­sue be­fore start­ing to lobby, that (c) lob­by­ists ex­pend more re­sources when there is less con­sen­sus on a topic, and that (d) groups lobby more when they per­ceive that there is a pro­ce­du­rally high like­li­hood of suc­cess.

So there is ev­i­dence that lob­by­ists are aware of the fac­tors that seem to en­hance suc­cess. The de­gree to which they take a sys­tem­atic ap­proach is un­clear; sev­eral of the pa­pers I read for this re­view hinted at a be­lief in con­sid­er­able het­ero­gene­ity in effec­tive­ness across lob­by­ists. If this het­ero­gene­ity is sig­nifi­cant, then there is prob­a­bly a lot of value to be gained by iden­ti­fy­ing, en­gag­ing, and fa­cil­i­tat­ing the work of only the most skil­led lob­by­ists.

Foun­da­tions and so­cial change or­ga­ni­za­tions already en­gage in this type of ac­tivity, fund­ing ad­vo­cacy across a wide range of top­ics. This is very close to what I am propos­ing; the only differ­ence is that I am sug­gest­ing a si­mul­ta­neously more di­ver­sified and more tar­geted ap­proach. Rather than plow­ing mil­lions into ad­vo­cacy groups that fight for a broad ros­ter of changes on a small num­ber of hotly con­tested is­sue ar­eas (e.g. cli­mate change), my pro­posal is to in­vest smaller amounts on tar­geted ad­vo­cacy for spe­cific, high-ex­pec­ta­tion poli­cies across a wide range of pos­si­ble fo­cus ar­eas.

Some ideas

Just to give a fla­vor of some of the po­ten­tially high-im­pact (in ex­pec­ta­tion) poli­cies that one might want to lobby for. Please note that some of these are likely to face sig­nifi­cant op­po­si­tion, so they’re not ideal can­di­dates for effec­tive lob­by­ing, but they’re not ipso facto bad can­di­dates (the ex­pected value of avoid­ing nu­clear war, for ex­am­ple, is very high).

Cost-effec­tive­ness analysis

I con­ducted a limited cost-effec­tive­ness anal­y­sis via Monte Carlo simu­la­tion by “back­ing out” dis­tri­bu­tion pa­ram­e­ters from (in my view, con­ser­va­tive) 95% con­fi­dence in­ter­vals. Here are my as­sump­tions:

  • The per-policy lob­by­ing cost lies with 95% con­fi­dence be­tween $20,000 and $10,000,000 dol­lars, and is dis­tributed log-nor­mally.

  • The change in prob­a­bil­ity of policy im­ple­men­ta­tion lies with 95% con­fi­dence be­tween 0 and 5%, and is dis­tributed nor­mally. Note that (1) the baseline prob­a­bil­ity of policy im­ple­men­ta­tion is very low to start with and that (2) this setup also in­cludes a 2.5% chance that lob­by­ing could ac­tu­ally de­crease the like­li­hood of policy im­ple­men­ta­tion.

  • The ex­pected im­pact of policy im­ple­men­ta­tion lies with 95% con­fi­dence be­tween 1 and 100,000,000 DALYs, and is dis­tributed log-nor­mally. I think I’m us­ing a pretty con­ser­va­tive up­per end here, given the po­ten­tial value to be gained by gen­uinely effec­tive longter­mist poli­cies.

  • I also as­sume that in 10% of cases, effec­tive lob­by­ists get the sign of im­pact wrong, so the sign of im­pact is dis­tributed bino­mi­ally, with a 10% chance of a nega­tive sign.

This re­sults in the fol­low­ing setup:

The punch­line here is that the cost per DALY of this ap­proach lies with 95% con­fi­dence be­tween $-157,632 and $27,974,720, with a me­dian value of $879, and with 70% con­fi­dence be­tween $0 and $247,117. There’s also a 20% chance that the cost per DALY lies in the “sweet spot” of be­tween $0 and $50. So the me­dian value is semi-promis­ing, but there’s a lot of un­cer­tainty and there’s a 10% chance effec­tive lob­by­ists will pay as much as $100,000 to make peo­ple’s lives one year shorter. How­ever, it’s worth not­ing again that I tried to be both very con­ser­va­tive and very un­cer­tain here, and that fur­ther ra­tio­nal­iza­tion of this setup could re­sult in more pre­cise (and per­haps more op­ti­mistic) re­sults.


Sensitivity analysis for lobbying CE

In ad­di­tion to the ba­sic cost-effec­tive­ness calcu­la­tion out­lined above, I pre­sent in this graph four vari­a­tions with differ­ent parametriza­tions. The ini­tial calcu­la­tion is the red dis­tri­bu­tion (la­beled “Origi­nal”). In Vari­a­tions 1 through 4, I ad­just only one pa­ram­e­ter rel­a­tive to the origi­nal.

The key take­away of this anal­y­sis is that there’s an im­por­tant trade­off in the way I’ve set this up: lower me­dian val­ues (and lower, nar­row ranges) for pos­i­tive costs per DALY come at the price of a higher prob­a­bil­ity of a nega­tive cost per DALY. The in­ter­pre­ta­tion of a nega­tive cost per DALY would be, as I men­tioned above, that sub­stan­tial ex­pen­di­tures are made lob­by­ing for a policy that ends up hav­ing a sig­nifi­cant nega­tive im­pact.

This is ob­vi­ously a func­tion of the bino­mial term in the setup de­scribed above. I think this term is defen­si­ble: for any policy that could plau­si­bly af­fect large num­bers of peo­ple in a sig­nifi­cant way, there’s a small but non­triv­ial pos­si­bil­ity of se­vere, nega­tive, un­in­tended con­se­quences.

One of the coun­ter­in­tu­itive con­se­quences of this setup is that re­duc­ing the like­li­hood of harm in­creases the price per DALY: that’s be­cause re­duc­ing harm doesn’t stop us from spend­ing: we get pos­i­tive value that oth­er­wise would be missed, but we still end up spend­ing a lot of money on poli­cies at the low end of the im­pact dis­tri­bu­tion. So (with this setup, at least) we can’t get a bet­ter cost dis­tri­bu­tion just by re­duc­ing harm.

So there’s two main pre­con­di­tions to get­ting a good cost/​DALY in lob­by­ing-based policy change. The first is to ad­vo­cate poli­cies, as I in­di­cated above, that have some prob­a­bil­ity of pro­duc­ing very large-scale benefits; the sec­ond is (ob­vi­ously) to spend less (see Vari­a­tion 3).

I think this sug­gests a cou­ple things. First, the bun­dle of lob­bied poli­cies should in­clude as many gen­uinely longter­mist poli­cies, or poli­cies ad­dress­ing global catas­trophic risks, in or­der to shift as much mass as pos­si­ble onto the right side of the im­pact dis­tri­bu­tion with as lit­tle risk of catas­trophic er­ror as pos­si­ble. Se­cond, if you ac­cept the premise that ac­ci­den­tal er­ror in the form of a “sign re­ver­sal” is a plau­si­ble model, then re­duc­ing the er­ror rate be­comes pro­por­tion­ally more im­por­tant as poli­cies that af­fect larger num­bers of peo­ple are ad­vo­cated. One pos­si­ble av­enue here is to prefer­ence poli­cies that en­hance de­ci­sion-mak­ing in the long term, that im­prove the func­tion­ing of in­sti­tu­tions, over poli­cies that en­courage near-term changes to gov­ern­ment de­ci­sions on spe­cific top­ics.


How does fed­eral leg­is­la­tion work in the U.S.?

Lob­by­ists in­ter­ested in ad­vanc­ing a policy pro­posal must first find a bill spon­sor. After a bill is in­tro­duced, it is con­sid­ered by the stand­ing com­mit­tee that has ju­ris­dic­tion over the bill, which de­pends on its con­tent. Most bills fail at this stage. Com­mit­tee chair­men have enor­mous power, and (as things cur­rently stand) they can de­cide unilat­er­ally whether or not to bring a bill to a sub­com­mit­tee, which would hold a pub­lic hear­ing on the bill’s con­tent. On some oc­ca­sions, these pub­lic hear­ings are held in front of the en­tire com­mit­tee.

After a pub­lic hear­ing, the bill moves up to the com­mit­tee level, where it is sub­ject to “markup”: amend­ments, rewrites, and votes that can make or break the bill. The Con­gres­sional lead­er­ship then de­cides whether or not to sched­ule a floor vote on the bill. At this point, the pro­cesses in the Se­nate and House di­verge.

In the Se­nate, the Se­nate Ma­jor­ity Leader can im­me­di­ately sched­ule a vote—or re­fuse to sched­ule one. Once on the floor, amend­ments can be offered in­definitely un­til de­bate is cut off by a clo­ture vote (sixty sen­a­tors).

In the House, if a ma­jor­ity of the com­mit­tee ap­proves the bill, it is sent to the House Rules Com­mit­tee, which re­ceives amend­ments from mem­bers and de­ter­mines which amend­ments will be con­sid­ered on the floor. For any par­tic­u­lar bill, the Rules com­mit­tee can in­sti­tute an “open rule,” which al­lows con­sid­er­a­tion of all amend­ments, a “closed rule,” which al­lows none, or a rule defin­ing speci­fi­cally which amend­ments can be con­sid­ered. On the floor, House mem­bers must abide by the rules set by the Rules Com­mit­tee.

Differ­ences be­tween House and Se­nate ver­sions of the same bill are rec­on­ciled, and then fi­nal, iden­ti­cal ver­sions of the bill are ap­proved or re­jected by floor votes in each cham­ber.

Lob­by­ists can at­tempt to in­ter­vene at any stage dur­ing this pro­cess: they can at­tempt to mod­ify the word­ing of a mea­sure, to en­courage votes at the sub­com­mit­tee, com­mit­tee, or floor level, to per­suade com­mit­tee or Con­gres­sional lead­er­ship to bring a bill to a vote, or to in­fluence the way in which House and Se­nate ver­sions of the same mea­sure are rec­on­ciled.

One of the pri­mary ways in which cor­po­rate lob­by­ists can im­pact their clients’ earn­ings is via ear­marks—spe­cific ap­pro­pri­a­tions within a bill that define pre­cisely where and how some seg­ment of the ap­pro­pri­ated funds are to be spent: this is what many em­piri­cal re­searchers are con­sid­er­ing when they at­tempt to de­ter­mine the effects of lob­by­ing ex­pen­di­ture on firm fi­nan­cial perfor­mance.

For some classes of is­sue, sig­nifi­cant policy changes are pos­si­ble via lob­by­ing at other lev­els and branches of gov­ern­ment. Poli­cies of fed­eral agen­cies, for ex­am­ple, can have a sig­nifi­cant im­pact: think here of rule­mak­ing by the FDA, EPA, and other reg­u­la­tory agen­cies. Con­sid­er­able lob­by­ing takes place at the state level as well, and can in­di­rectly in­fluence fed­eral policy: in ma­jor states such as Cal­ifor­nia and New York, state reg­u­la­tions can be a blueprint for fed­eral ones.

Code for fu­ture in-depth analyses

As part of my re­search, and to sup­port the deriva­tion of pa­ram­e­ters for the above cost-effec­tive­ness model, I went through the lob­by­ing dis­clo­sure data in de­tail and ex­plored a bit. I dis­cov­ered that it’s hard to get your hands on it in a use­ful for­mat, so I wrote some code to con­vert it from XML to tab­u­lar, as well as to pro­duce some hope­fully use­ful sum­mary anal­y­sis like those be­low. That code is available pub­li­cly on my GitHub. It is clean enough, but is mostly meant to serve as a start­ing place for oth­ers who want to pick up where I have left off here.

The three sum­mary graph­ics be­low are ex­tracted from lob­by­ing dis­clo­sures for 2018. Each filing con­tains a num­ber of is­sue-spe­cific re­ports, some of which con­tain refer­ences to spe­cific bills and some of which do not. In or­der to get thes plots be­low plots, I took re­ported amounts from quar­terly lob­by­ing re­ports and al­lo­cated them across the bills and is­sues ex­tracted from the text of each re­port.

  • For the red and green plots, I ex­tracted men­tions of all House Re­s­olu­tions from the text of the quar­terly state­ments and di­vided the re­ported amounts evenly be­tween them and re­ported lob­by­ing is­sues, then plot­ted the den­sity. If a quar­terly filing in the amount of $10,000 con­tained 6 is­sues, 2 of which refer­enced 3 spe­cific bills each, then each bill was al­lo­cated $1,000, and each is­sue with­out a bill was al­lo­cated $1000.

    • The green plot in­cludes all the bills that were refer­ences in dis­clo­sure filings. This is sev­eral thou­sand bills, 99% of which never came to the floor for a vote.

    • The red plot in­cludes only the ~100 bills that came up for a floor vote in 2018.

  • For the blue plot, I divvied the amounts up only at the is­sue level, with­out refer­ence to spe­cific bills. The unit of ob­ser­va­tion here is an in­di­vi­d­ual is­sue re­port, of which there are four an­nu­ally, so the dis­tri­bu­tion pic­tured is the dis­tri­bu­tion of (al­lo­cated) ex­pen­di­ture per is­sue, per lob­by­ist, per quar­ter. So the me­dian amount that a lob­by­ist spent on any given is­sue in any given quar­ter of 2018 was about $200.

Summary statistics on lobbying data

Bibliog­ra­phy, etc.

My bibliog­ra­phy is on Google Sheets, along with a tab pro­vid­ing short sum­maries, re­search de­signs, and key re­sults of about 25 of the em­piri­cal stud­ies that I re­viewed.