RATIONALITY AND INSTITUTIONAL CONTINGENCY: · PDF fileFIRE INSURANCE INDUSTRY TIM BARTLEY*...

Sociological Perspectives Volume 45 Number 1 pages 47ndash79Copyright copy 2002 by Pacic Sociological Association All rights reservedSend requests for permission to reprint to Rights and Permissions University of California Press Journals Division 2000 Center St Ste 303 Berkeley CA 94704-1223ISSN 0731-1214 online ISSN 1533-8673

RATIONALITY AND INSTITUTIONAL CONTINGENCY THE VARYING POLITICS

OF ECONOMIC REGULATION IN THE FIRE INSURANCE INDUSTRY

TIM BARTLEYUniversity of Arizona

MARC SCHNEIBERGReed College

ABSTRACT Are the politics of economic regulation contingent oninstitutions Drawing on arguments about institutional mediation andthe institutional bases of rational action we explore how institutions shapethe dynamics invoked in two theories of regulation We argue that institu-tional arrangements affect both the clarity and the content of group inter-ests in regulation Event history analyses of US statesrsquo passage of reinsurance regulation from 1906 to 1930 support these arguments inseveral ways Market-heterogeneity dynamics specied by cartel-capturetheory affected the passage of regulation only under some conditionsmdashnamely relatively depoliticized settings with little overt conict or uncer-tainty about policy outcomes In addition interest group dynamics wereconditioned by the consolidation of interorganizational elds around a par-ticular model of market order which allowed divergent interest groups toconverge in support of regulation The research suggests ways of thinkingabout theoretical generality and specicity regulatory politics and therelationship between institutional theory and rationality

Political interests and conicts often appear universal and unproblematic Chal-lenging groups ght for political inuence Firms have interests in stability auton-omy and prot maximization While it is clear that interests and conict structurepolitical outcomes less clear are the processes that dene interests and structureconicts or complementarities of interest

An enduring feature of institutional arguments in sociology is that institutionalarrangements shape the processes that drive social action and outcomes Fried-land and Alford (1991245) for instance argue that ldquoutility maximization satis-

Direct all correspondence to Tim Bartley Department of Sociology University of Arizona Tucson AZ 85721 e-mailbartleyuarizonaedu

48 SOCIOLOGICAL PERSPECTIVES Volume 45 Number 1 2002

cing income maximization prot maximization risk power even interest itselfare all institutionally contingentrdquo Similarly Douglas (1986) argues that institu-tions shape the terrain of action by classifying highlighting and obscuring par-ticular features of a situation ldquoNew institutionalistrdquo research in sociology andpolitical science provides an empirical warrant for thinking of institutions as par-tially constituting instrumental action (Hall and Taylor 1996 March and Olsen1989 Powell and DiMaggio 1991) Furthermore recent neoinstitutional researchhas begun to move from analyzing the spread of models of order to examininghow institutional logics and political regimes moderate the dynamics of organiza-tional behaviormdashshowing that causal mechanisms are contingent on institutionalconditions (Dobbin and Dowd 2000 Thornton and Ocasio 1999)

Our approach to institutional contingency considers two theories of economicregulationmdashcartel-capture theory and interest group theorymdashthat provide gen-eral accounts of when and why states intervene in industries to control pricesentry or investment1 These theories use rational choice models of action andassume that actors maximize economic interests We focus on institutional sourcesof interests in regulation and argue that variation in political-institutional condi-tions may produce variation in the clarity and content of preferences for oragainst regulation Cartel-capture theory highlights the market conditions that leadrms to use regulation as a tool for creating or maintaining monopoly but typi-cally ignores the way in which policy shapes competitive behavior and the politi-cization of organizational elds alters interests in ldquostate capturerdquo Interest grouptheory emphasizes the organizing capacities and stakes in regulation of groupsboth internal and external to an industry but typically pays little attention to insti-tutional shifts over time that may clarify and alter group interests These twoapproaches make valuable contributions to understanding economic regulationand the relationship between state policy and private forms of market control Yetthey are more valuable when treated as institutionally contingent

Empirically we use event-history analyses to examine US statesrsquo passage ofrate regulation in the re insurance industry from 1906 to 1930 This industry pro-vides fertile ground for examining the institutionally mediated relationshipsamong market forces interest group politics and state policy First the industrywas central to the economic infrastructure of the United States re insurance wasrequired by creditors and major urban centers frequently had massive res(Brearley 1916 Mowbray 1946) Second the industry was characterized by sub-stantial political and market organization including cartels industry associationsand rate-making bureaus Third starting in 1909 US states began to pass laws toregulate re insurance rate making Typically these laws allowed state insurancecommissioners to order changes in insurance rates or rate-making practices whilealso authorizing insurers to form rating bureaus or rate-making associations Ineffect these laws eschewed antitrust principles in favor of ldquoregulated coopera-tionrdquo By 1930 thirty-three states had enacted rate regulation generating substan-tial variation in outcomes for the purposes at hand

The goal here is quite specicmdashto examine the contingent character of two the-ories of regulation In previous work we have assessed competing theories andfound direct effects of institutional factors on rate regulation (Schneiberg and

Rationality and Institutional Contingency 49

Bartley 2001) Here however we examine how institutional conditions shaped theprocesses by which regulation emerged rather than test theories or construct a com-prehensive account of the outcome This allows us to identify the specic conditionsin which causal mechanisms are most applicable Methodologically it implies afocus on interaction effects rather than direct effects of institutional variables

We begin by considering how institutions mediate political processes and thenconsider one type of mediation in more detail by focusing on the institutionalbases of rationality Next we outline theories of regulation discuss ways in whichinstitutional factors may shape the interests specied by these theories andexplain how these apply to the case of re insurance After a discussion of meth-ods and measurement we turn to the results of event-history analyses whichsupport several ideas about institutional contingency in theories of regulation

VARIETIES OF INSTITUTIONAL MEDIATION

As Amenta (199819) comments ldquoInstitutional theses typically specify the limitson social policy but often ignore what drives it Political theories specify theactors and resources that drive public social provision but often ignore the sys-temic limits on political actionrdquo Integrating these approaches requires conceivingof institutions as mediating political processes or conditioning political dynam-ics Beyond rst-order ldquoinstitutional effectsrdquo on policy institutions shape the rela-tionship between behavioral independent variables and policy outcomes (AmentaDunleavy and Bernstein 1994)

One view of institutional mediation focuses on political opportunities thatshape the organizing capacities and successes of challengers Outsidersrsquo politicalinuence may be limited or amplied by features of the formal political structureand party system like the basis of party support or presence of ldquoveto pointsrdquo inthe legislative process (Amenta 1998 Immergut 1992) In addition ldquopolicy feed-backrdquo arguments suggest that institutionalized policies organize or disorganizeconstituencies for future political action (Pierson 1994) Further institutional con-texts may change over time and thus modify the effects of other variables (Isaacand Grifn 1989 Sutton and Dobbin 1996 Tolbert and Zucker 1983) These argu-ments suggest that institutions shape political contests as much as do traits of thecontestants themselves Institutions do so in part by dening payoffs opportuni-ties and constraints on collective behavior In this conception institutions areldquorules of the gamerdquomdashwhich are themselves variable

Going further literature on the ldquonew institutionalismrdquo in sociology and politi-cal science suggests that institutions are ldquoconstitutive and prescriptiverdquo as well asldquoconstraining and proscriptiverdquo (Clemens and Cook 1999 Hall and Taylor 1996Schneiberg and Clemens forthcoming Thelen and Steinmo 1992) Therefore insti-tutions may mediate the effects of other causal dynamics not only by deningpolitical opportunities but also by shaping the interests of political actors and eventhe actors themselves (Meyer et al 1997) This line of theory uses a broaddenition of institutions as ldquosupraorganizational patterns of human activity bywhich individuals and organizations produce and reproduce their material sub-sistence and organize time and space They are also symbolic systems ways of


ordering reality and thereby rendering experience of time and space meaningfulrdquo(Friedland and Alford 1991243) Thus institutions are not simply ldquorules of thegamerdquo that constrain action but also systems that help to constitute actors andinterests It is this focus on the institutional construction of preferences that moti-vates our analysis of the politics of regulation

INSTITUTIONAL FOUNDATIONS OF RATIONALITY

To specify and delimit the rational choice theories of regulation under consider-ation we focus on the social context that underlies rational action While scholarsoften leave rationality as a ldquoblack boxrdquo we join others (from Weber to culturaltheorists to rational choice institutionalists) in examining how social conditionsshape denitions of interest and create an infrastructure of clarity and calculabil-ity that allows rational choices to be made By unpacking the black box of ratio-nality we seek to explain how the intersection of institutions and interests drivesthe passage of regulation

The basic premise of rational choice theory is as follows ldquoWhen faced with sev-eral courses of action people usually do what they believe is likely to have thebest overall outcomerdquo (Elster 198922) This simple statement presumes a numberof conditions of choosing First if one is to choose from several courses of actionthe options must be commensurablemdashthat is they must be enough alike that onecan compare and rank them (Espeland 1998) Second if choices are made on thebasis of expected outcomes then rationality requires enough evidence or stabilityin the context of action to calculate an expected outcome (Heimer 1985) Forinstance Elster (198933) suggests that ldquobeliefs [or preferences] are indeterminatewhen the evidence is insufcient to justify a judgment about the likelihood of thevarious outcomes of actionrdquo In this situation uncertainty may override rationality

Crucially commensurability predictability and the calculation of risk are notmerely formal requirements of rational decision making they are shaped bysocial context Rational choice explanations fundamentally rest on institutionalconditions for calculation and rationality Rational choice theorists themselveshave recognized cognitive limits on rationality problems of incomplete informa-tion and the context-bound character of rational choice mechanisms (Bates et al1998 Brinton and Nee 1998) In fact this approach has contributed fundamentalinsights into how actors craft institutions in response to uncertainty and incom-plete information (Akerloff 1970 Alchian and Demsetz 1972 Williamson 1985)Furthermore work in this tradition has become increasingly self-conscious aboutexplanations that rest on assumptions of full information and theories that areintended to apply only under certain institutional conditions Yet the institutionalconditions for the two rational choice theories of regulation under considerationhave gone unspecied with little consideration of the social contexts that sup-port inhibit or dene the rational decisions they assume We seek to contribute tothis topic by exploring several ways in which institutions underlie clarify andmodify interests

Institutional arrangements that provide well-articulated standards make theworld more predictable Weber argued that rational exchange relied on legal and


nancial institutions that made the world calculable by enforcing contracts andstandardizing taxation and currencies (Collins 1980) Standardization is also cen-tral to commensuration which is ldquoconcerned with measuring different propertiesnormally represented by different units with a single common standard or unitrdquo(Espeland 199824) Likewise institutions make the world more orderly by main-taining classications and intensely ordered ldquoadministrative gridsrdquo (Douglas1986 Scott 1998) An important effect of institutions then is to order an otherwiseoverwhelmingly chaotic set of objects and to draw attention to some objects overothers This is an approach to rationality in which ldquothe individual is seen as anentity deeply embedded in a world of institutions composed of symbols scriptsand routines which provide the lters for interpretation of both the situation andoneself out of which a course of action is constructedrdquo (Hall and Taylor 1996939)

Conversely situations in which institutions are poorly developed or unsettledcan subvert commensurability predictability and calculability As Berk (1994)argues in periods of ldquoconstitutive politicsrdquo the economic landscape is itself upfor grabs leaving economic and political interests ambiguous In periods ofldquopower politicsrdquo many of the overarching institutional parameters are relativelyxed but in constitutive periods institutional environments may be composed ofcompeting frameworks conicting standards and indeterminacy regardingoptions the consequences of choice or the ldquorules of the gamerdquo Here institutionalconditions complicate rather than simplify the task of interpreting realitymdashpro-ducing intractable uncertainties and undermining the conditions for rationalchoice (Heimer 1985 March and Olsen 1976 1989)

If the assumption that preferences are clear is in need of further specication sotoo are assumptions that conicts or complementarities of interest are structuredby ldquoinitial conditionsrdquo that are relatively stable over time Rational choice theo-rists have begun to address the problem of interests changing over time byexploring how preferences get altered through iterative processes of trial-and-error and learning (Axelrod 1984 Greif 1998) Yet such microlevel ldquobottom-uprdquoaccounts are weaker at explaining higher-level institutions that ldquostructure theexchange and the exchange process that create identities for the subjects involvedin the exchange and that create a subjectively meaningful social context in whichthe exchange takes placerdquo (Hamilton and Feenstra 1998159) Changes in thesehigher-order structures may alter the costs benets and risk factors involved inindividual calculations of interest and therefore create new complementaritiesandor conicts of interest For instance the consolidation of an institutional eldover time may clarify or constitute otherwise inchoate interests and is itself a crit-ical source of initial conditions

These ideas about rationality undergird our exploration of the institutionalfoundations of interests relevant to capture and interest group accounts of regula-tion In discussing these theories we emphasize that industry groups and outsid-ers are likely to have clearer preferences about regulatory policy when institu-tional conditions and disputes over the rules of the game are settled and wheninstitutional environments provide safeguards for predictability or make gover-nance options more clearly commensurable In addition policy regimes thatshape market dynamics may affect the content of preferences as could institu-


tional changes that reduce the risks involved in implementing regulationmdashwhichmay end up aligning the interests of previously antagonistic groups

CARTEL-CAPTURE THEORY

In the nineteenth-century US economy rms commonly sought to organize mar-kets collectively via interrm compacts or associations and to pool informationset standards x prices or develop technical knowledge (Chandler 1977 Fligstein1990 Schneiberg and Hollingsworth 1990) These historical observations motivatecartel-capture theoryrsquos central claim that economic regulation can also be an attrac-tive option for rms trying to organize or stabilize markets The theory rejects thecommonsense notion that regulation is constructed in the public interest torestrict monopoly power or solve market failures Instead it argues that industriesuse regulation as a tool for suppressing competition creating monopolies andextracting rents (Kolko 1963 Posner 1974 Stigler 1971) For instance rms incompetitive industries may be able to better control the market if the stateimposes restrictions on entry or allows associations to set minimum prices

According to the theory rms seek regulation when market dynamics of com-petition and collective action make private means of organizing an industry unten-able Two of the least costly forms of private organization are cartels and price-xing associations Yet these are not always viable market control mechanismssince the voluntary character of participation opens the door to free-riding defec-tion from agreements and other enforcement problems Regulation may then bean effective alternative if rms can create or capture state agencies that willenforce interrm agreements and limit competition

In this view it is market conditions that prompt the shift from cartels to captureMarket size and heterogeneity are particularly important (Bowman 1989 Jac-quemin and Slade 1989 Posner 1974 Stigler 1971 1974) When rms are few innumber and similar in assets costs and strategy it is relatively easy to form andmaintain effective voluntary interrm agreements Here rms opt for private car-tels Yet with increased market size and heterogeneity private organizationalstrategies become more difcult because of problems of managing large numbersand divergent interests Here rms develop interests in state-enforced coopera-tion while also increasing their ability to forge ldquocapture coalitionsrdquo Specicallylarge numbers increase rmsrsquo political clout while diversity creates an ldquoasymme-try of interestrdquo in the industry and increases incentives to join the ldquocapture coali-tionrdquomdashto guard against one industry faction using regulation against a nonpartic-ipating group (Stigler 1971) Accordingly increasing size and heterogeneityprompts a shift from private organization to the state

However increasing size and heterogeneity beyond intermediate levels mayreverse this process At some point greater numbers and diversity may subvertrmsrsquo abilities to form effective lobbies (Stigler 1974) Indeed extremely high lev-els of heterogeneity may make the faction-against-faction incentive for regulationdevolve into inghting that inhibits rmsrsquo political inuence Growing marketheterogeneity may also increase the chances that a subset or ldquocritical massrdquo ofactors will emerge who are interested and capable of organizing markets pri-


vately on their own (Marwell and Oliver 1993)2 Thus cartel-capture theory sug-gests that market size and heterogeneity have inverse-U-shaped effects on the likelihood ofpassing regulation

Institutional Mediator Anticompact Laws

The utility of cartel-capture theory lies in its sophisticated account of howrmsrsquo economic strategies shape political outcomes However cartel-capture the-orists largely ignore the way in which other political conditions shape rmsrsquo gov-ernance options and strategies Institutionalists have found that political regimesconstitute and alter economic behavior Research on the development of Ameri-can capitalism suggests that antitrust laws encouraged the vertical growth of UScorporations by constraining their horizontal growth (Hollingsworth 1991) Simi-larly Dobbin and Dowd (2000) show that the effects of market dynamics on rmacquisitions are conditional on an antitrust regime that limits cartels and enforcescompetition

Drawing on Dobbin and Dowd (2000) one may expect regimes that ban cartelsand enforce competition to create precisely the conditions that cause rms to seekregulation Lacking laws that limit interrm cooperation rms will pursue pri-vate forms of industry organization But if this option is blocked by antitrustenforcement then rmsrsquo organizing efforts shift to publicly regulated coordina-tion Following this logic we expect market conditions to be more strongly related to thepassage of regulation in states with antitrust laws than in states without antitrust lawsIn fact in states that lack policy enforcing competitive behavior the dynamics ofcompetition and collective action specied by cartel-capture theory should beirrelevant to the passage of re insurance rate regulation

A second possibility is that hostile political regimes increase uncertainty forrms in calculating the expected costs and benets of regulation A potential costof regulation is that it may end up serving as a channel that admits powerful out-siders into industry councils mobilizes hostile forces or exposes prices to politics(Bowman 1989 Stigler 1971) When anticompany forces manage to challenge thepower of rms in an industry it becomes more difcult for rms to calculate thisexpected cost Furthermore when policy undermines predominant rm strategieswithout providing alternatives conict and ambiguity over the creation of newstrategies tend to ensue (Dobbin and Dowd 2000) Therefore politicization of theinstitutional environment undermines the conditions for the type of rationalchoice posited by cartel-capture theory For rms hostile political regimes increaseuncertainty while decreasing predictability and the commensurability of publicand private governance options Following this logic we expect industry preferencesfor capture to be ambiguous under conditions of political hostility and clearer in less polit-icized environments

In the re insurance case state ldquoanticompactrdquo laws both enforced competitionand politicized institutional environments Anticompact laws were antitrust lawsthat banned interrm price compacts and collaborative rate making in re insur-ance They were a response to extensive systems of interrm coordination thatcharacterized the nineteenth-century re insurance industry Such systems led


farm groups local boards of trade and local public ofcials to stir up a legitimacycrisis for the industry by arguing that cooperation among insurers produced exor-bitant rates Consumer groups also decried arbitrary treatment by company rate-making associations and ultimately turned to anticompact laws for relief Theselaws typically prohibited collusive price-xing outright while targeting specicpractices such as associations ldquoadvisingrdquo companies about rates

Ohio passed the rst anticompact law in 1885 and Michigan followed in 1887During the 1890s and early 1900s anticompact legislation spread rapidly throughthe populist regions of the country leaving a band of anticompact states centeredlargely in the Midwest and the South (see Figure 1) By 1910 the anticompactmovement had run its course3

Anticompact laws constituted the political-institutional context in which rateregulation emerged (Schneiberg 1999 Schneiberg and Bartley 2001) Once statespassed rate regulation they either repealed anticompact laws or allowed them tolanguish on the books Nevertheless the anticompact movement was a deningfeature of the setting in which rms considered regulation as a possible solutionto market control problemsmdashand thus decisively shaped the context which cartel-capture mechanisms apply4 On one hand anticompact laws enforced marketcompetition limiting rmsrsquo abilities to cooperate openly in making and enforcingrates At the same time such regimes represented a conguration of power withindustry outsiders able to subject the industry and its markets to disruption andpolitical uncertainty They represented not only a defeat of industry interests inassociation but also a legitimacy crisis in which industry organization was sub-jected to public scrutiny controversy and criticism In addition where anticom-pact laws were in place they often served as platforms for political attacks againstinsurers Indeed insurance companies fought state antitrust intervention becauseit subjected the industry and its operations to ldquopoliticsrdquo and to the strategicbehavior of politicians who would ldquoplay to the galleryrdquo and attack the ldquoinsurancecombinerdquo in their search for electoral advantage (Schneiberg 1999) For compa-nies then these interventions created uncertainty about the prospects for viablemarket organization and the meaning and consequences of regulation

At this point we consider it an empirical question whether the primary effect ofanticompact laws was to (1) enforce competition and thus spur the dynamicsspecied by cartel-capture theory or (2) politicize industry governance to such anextent as to foster uncertainty in insurersrsquo interests in regulation We model bothof these possibilities and allow the data to lend support to one or the other Table 1summarizes the predictions of cartel-capture theory and institutionally contin-gent versions I and II

INTEREST GROUP THEORY

Interest group theory looks beyond the rms and market conditions specied bycartel-capture theory to consumers suppliers and subsectors of an industry Inthis view regulation allows consumers and other industry outsiders to share inthe spoils of the regulated industry or to contain concentrated private powerHere regulation reects the relative organizing capacities of producer and con-


Figure 1

Development of Anticompact Measures 1885ndash1910


sumer groups and therefore the passage of regulation is at least a partial victoryof anti-industry forces such as farmers and major consumers

The ability of a group to inuence and benet from regulation depends on itscapacity to organize (Zeigler 1964) which in turn largely depends on economicstakes in an issue and the number of potential participants (Kanazawa and Noll199423)5 Interest group theory suggests that group size can have multiple effectssince theorists couple the analysis of free-ridercollective action dynamics withthe analysis of political inuence Accordingly while increased numbers decreasesthe likelihood of a group becoming organized it also increases the potential polit-ical power of that group In the main however interest group analyses emphasizethe latter effect especially when there is other evidence that a group was orga-nized Their primary prediction about group size is that the larger an interest blocthe more likely it is to inuence the legislative process (Fishback and Kantor 1996Kanazawa and Noll 1994 Meier 1988)

Since regulation represents a victory of outsiders the larger the population of reinsurers in a state and the greater the stakes associated with re insurance (the moreinsurers have to lose) in a state the less likely that state should be to pass rate regulationStanding against the industry were two major consumer groupsmdashfarmers andmanufacturers In the early twentieth century both groups were well organized atthe local level In addition they were the largest consumers of re insurance pro-viding them with high stakes in re insurance regulation and the potential to usethe state to constrain private power Therefore the larger the population of farmersand manufacturers in a state and the greater the stakes each group held in the operation ofre insurance the more likely that state should be to pass regulation

Because they take (potentially hostile) ldquooutsidersrdquo as seriously as producers in

TABLE 1

Predicted Effects of Cartel-Capture Variables

Theory Core Claim Predicted Effects

General Cartel-capture theory

Market factors drive the passage of regulation

Market size and heterogeneity have inverted-U-shaped effects on the likelihood of regulation

Institutionally Contingent Version I

Market dynamics are contingent on policy that enforces competition

Market size and heterogeneity have inverted-U-shaped effects on the likelihood of regulation only in states with anticompact laws Coefcients are weak or nonsignicant in states without anticompact laws

Institutionally Contingent Version II

Politicization of institutional environments creates uncertainty in industry preferences for regulation

Market size and heterogeneity have inverted-U-shaped effects on regulation only in states without anticompact laws Coefcients are weak or nonsignicant in states with anticompact laws


the industry to be regulated interest group theorists are less prone than cartel-capture theorists to ignoring the larger political environment Still interest grouptheories assume a lot by deriving clear stable preferences of groups in regulatoryoutcomes based on a set of initial conditions Some interest group theorists do rec-ognize that goals may change with changes in the environment or the bureaucra-tization of a movement (Zeigler 1964) and others suggest that uncertainty aboutthe effects of regulation may inhibit interest group support ldquoIf the effects of a reg-ulation or the precise identity of the beneciaries and the losers cannot be pre-dicted beforehand with reasonable accuracy the incentive to contribute to a lob-bying effort is obviously reducedrdquo (Noll and Owen 198345) However interestgroup analysts have not typically modeled the determinants of certainty or thestability of interests as factors in the passage of regulation

Institutional Mediator Institutional Consolidation and Administrative Capacities

We argue that the consolidation of interorganizational elds and innovationsin suprastate administrative capacities shape the clarity and content of groupinterests in regulation First preferences become more xed as institutionalconsolidation makes ends more predictable reduces uncertainty about out-comes and articulates stakes In effect groups can better calculate the expectedcost and benets of regulation once settlements are reached over the rules ofthe game once standard models or cognitive frameworks become the center-piece in a eld and once administrative capacities for routine decision makingand conict resolution are established (Scott 1994 1995) Likewise in ldquosettledrdquoperiods stability in the set of available beliefs and models provides resourcesfor purposive action (Swidler 1986) Therefore we expect interest groups to havea clearer inuence in the passage of regulation when institutional conditions are set-tled and decision-making structures and safeguards against uncertainty have beeninstitutionalized

Again there is a second version The development of stable rules safeguardsand rationalized procedures for adjudicating rate conicts may create the condi-tions for compromise between interest groups allowing them to converge on acommon mutually acceptable set of organizational solutions (Skowronek 1982Weir and Skocpol 1985) Both consumer and industry groups are likely to supportrate regulation if models of objective rate making and judgment are available andlegitimated Following this logic we expect interest group coefcients to become posi-tive with the stabilization of the organizational eld and the creation of administrativecapacities

In the re insurance case the years 1914 through 1916 marked a fundamentalsystemwide shift in the institutional context of regulation Three events helped tosettle conicts and consolidate the insurance eld around a specic set of theoriesand governance options (1) a Supreme Court decision endorsing rate regulation(2) the National Convention of Insurance Commissionersrsquo development of amodel law and (3) the creation of an actuarial bureau for gathering classied lossdata As a group these acts instituted ldquoregulated cooperationrdquo and scientic rate


making as the norm for the industry and created the administrative capacity forimplementing these models This shifted insurance from a setting characterizedby political conict and uncertainty about the efcacy of market organization andmeaning of regulation into a more stable settingmdashone where the models of appro-priate organization were more clearly dened and the actions of the state weremore predictable Since each of these events of 1914ndash15 was the culmination ofyears of debate and struggle we discuss the historical development of each keyevent separately

The Supreme Court Decision

Since the nineteenth century insurers had sought to manage markets throughrate-making associations while consumer groups and lawmakers had rejected theldquoinsurance trustrdquo and its ldquocombinesrdquo as nothing more than monopoly predationIn 1909 Kansas passed the rst insurance rate regulation law which granted thestatersquos insurance commissioner the authority to order changes in excessive orunjust rates (Grant 1979 Hobbs 1942363) Almost immediately afterward theKansas Superintendent of Insurance ordered a statewide rate reduction Inresponse companies sought to void the law in the courts taking the issue to theSupreme Court in German Alliance Ins v Lewis

At this point the question of which outcomes would obtainmdashwhat regulatoryor organizing options were possible what they meant and whose interests theyservedmdashwas fraught with uncertainty and conict The period was marked bycompeting models of order and struggles over the balance between public andprivate control Insurance companies generally advocated a private cooperativemodel of insurance markets whereby companies could associate freely withoutstate interference However this model contended directly with antitrust policiesstatist policies of public insurance provision consumer self-insurance throughmutual organizations and ldquoregulated cooperationrdquo that combined private coop-eration with state supervision (Schneiberg 1999)

The Supreme Courtrsquos 1914 decision in the German Alliance case went some waytoward establishing options and settling matters of insurance governance TheCourt ruled that re insurance was ldquoaffected with a public interestrdquo and thatstates had the right to regulate insurance rates (Crane 197256ndash57 Grant 1979Hobbs 1925) In effect it established regulation as a legitimate option for the statesand endorsed ldquoregulated cooperationrdquo as a solution to the problem of monopoly

The NCIC Model Law

At the end of 1914 the National Convention of Insurance Commissioners(NCIC) followed the Court by endorsing and further articulating the concept ofregulated cooperation The NCIC promulgated a model law that banned discrim-inatory or unjust rates and empowered state commissioners to order rate changeswhile also authorizing cooperative rate making (NCIC 1915) This quid pro quo ofregulation for authorized association helped to create a basis for convergence inthe interests of insurers and their adversaries The model bills and their explana-


tions endorsed the theories of market failure underlying company efforts to orga-nize insurance markets cooperatively and thus elaborated a rationale for rejectingantitrust Yet the NCIC bills also articulated the problem of monopoly and theo-ries of regulated cooperation advocated by reformers By codifying the model ofregulated cooperation for the states the NCIC made this solution to the problemsof insurance governance available to legislatures across the country

The NBFU Actuarial Bureau

While the actions of the Court and the NCIC made regulated cooperation a via-ble option for the states they left a critical question unresolved By what criteriashould rms or regulators x or order changes in rates As early as 1901 industryreformers had proposed a model of scientic rate making as a solution to the rateproblem (Dean 1901 Merritt Committee 1911) The idea was to let companies setrates through industry bodies but to do so through statistical analyses of the losscosts and hazards associated with a risk class This proposal eventually provedcompelling to consumers legislators and insurance commissioners As oneformer regulator explained to the New York Legislative Investigating Committeein 1910

You should have a classication to base your estimate on and then if a citizencomes to the authorities and said ldquoMy rate is too highrdquo the judge would havesomething to base an opinion on [O]therwise it would be a matter of guesswork (Merritt Committee 1909ndash102421ndash22)

The committee agreed ldquoIt is perfectly certainrdquo it wrote after grilling industryofcials

that the public has a right to demand that in return for the right to combinethe companies shall furnish equitable rates and not only that but that theyshall put themselves in a position to demonstrate [through classied loss data]that they are furnishing equitable rates (Merritt Committee 191172)

The companiesrsquo failure to produce classied data fueled struggles over rate reg-ulation Even as the principle of regulated cooperation emerged its viabilityhinged on rmsrsquo and regulatorsrsquo ability to set and judge rates according to anobjective and generally accepted set of standards Absent such a system compa-nies could not justify their rate-making practices regulators could not evaluaterate conicts and consumers lacked the reliable safeguards necessary for them toendorse rate making by company associations (Schneiberg 1999) For insurersthis meant that regulation was to be the price for private association Driven bythis consideration and the potential resurgence of statism and trust busting theNational Board of Fire Underwriters (NBFU) agreed in 1914 to create a uniformclassication scheme and an actuarial bureau charged with collecting and distrib-uting premium and loss data to each state for a variety of risks and risk classes(Riegel 1916 1917)

This accord transformed the terms of conict at the state level forging a struc-tural basis for companies and their opponents to agree on a rationalized data-


based regime of rate review Once the Actuarial Bureau supplied classied datato state insurance departments regulators could evaluate rate conicts on a rela-tively impartial basis (New York State 1922ndash32) Equally important the rational-ization of rate-making procedures buffered regulation from populist pressuresand made regulation predictable addressing companiesrsquo fears that public ofcialswould play politics with prices and the regulatory process In fact this develop-ment altered rmsrsquo calculations regarding regulation As one company ofcialput it (in an endorsement of regulated cooperation) there is a ldquomiddle wayrdquobetween

the extremes of State rate making [and] complete company control of rate mak-ing without the steadying inuence of State supervision to hold the rule mak-ing body up to a full sense of responsibility for its action to the representativesof the people [This is] leaving to the companies the selection of the person orpersons who shall make rates but giving to the Superintendent of Insurancethe power [to] examin[e] compel removal of discrimination [and] hear com-plaints (Quoted in Riegel 191669 emphasis added)

In sum these three events created a basis for reconciling conicting interestsand helped to consolidate the re and property insurance eld around the modelof a scientically based system of regulated cooperation6 As such they sorted outconicts among competing principles of order and established a single well-understood model as the reference point of cognition calculation and problem-solving behavior We consider it an empirical question whether the primary effectof these reforms was to (1) clarify preferences relating to regulation or (2) align

TABLE 2

Predicted Effects of Interest Group Variables


General Interest group theory

Regulation reects the relative organizing capacities of producer and consumer groups

Increasing the size and stakes of insurers decreases the likelihood of regulation Increasing the size and stakes of consumer groups increases the likelihood of regulation


The development of eldwide models and administrative capacities articulates and clarifies group preferences for regulation

Interest group size and stakes have the expected effects only after 1915 Coefcients are weak or nonsignicant before 1916


The development of eldwide models and administrative capacities aligns the interests of insurers and consumers

Before 1916 insurance industry variables have negative effects and consumer group variables have positive effects on the likelihood of regulation After 1915 both insurance industry and consumer group variables have positive effects on regulation


previously antagonistic interests in support of regulation Table 2 summarizesthe predictions of interest group theory and institutionally contingent versions Iand II

DATA

Dependent Variable Regulation of Fire Insurance Rates

The dependent variable is a binary variable with a value of 1 in the year a staterst passed rate regulation and a value of 0 in all previous years In the frame-work of event-history analysis this coding is equivalent to modeling the rate ofpassage or using the year of passage as the dependent variable (Allison 1995) Wemodel a statersquos rst passage of regulation and therefore exclude a state from therisk set once it passes a rate regulation law which allows us to understand thedynamics that led a state to pass a law in the rst place Since some legislaturesdid not meet every year (US Bureau of the Census 1918) we considered states tobe at risk of passing regulation only in years with legislative sessions (Allison1995227) Figure 2 shows the passage of rate regulation laws from 1906 to 1930

We coded statesrsquo passage of rate regulation using the Spectator Companyrsquos FireInsurance Laws Taxes Fees reports and corroborated the coding with additionalhistorical sources (Hobbs 1925 1942 Merritt Committee 1911 Reigel 1917 1927)To qualify as having passed rate regulation a state must have adopted a law thatgranted an insurance commissioner (or other ofcial charged with regulatinginsurance) power to (a) review insurance rates and either (b) order changes inrates or (c) recommend changes in rates7 The specic powers of insurance com-missioners varied somewhat with some empowered only to judge unfair dis-crimination among risks and others able to make more general judgments aboutexcessive rates In addition to empowering commissioners the vast majority of

Figure 2

Passage of Rate Regulation 1906ndash1930


rate regulation states also required insurers to le their rates with the state andauthorized insurers to form rating bureaus or rate-making associations8

Cartel-Capture Variables

To address cartel-capture theory we constructed state-level measures of marketconditions by using data on re insurance agents companies and the states inwhich they were licensed to do business (Spectator Company 1906ndash40) Thisinformation provided us with a state-level count of agents and allowed us toassign companies to statesmdashcreating for each state a list of the companies licensedto do business the assets of each company and company type (mutual US jointstock foreign) We use the number of re insurance agents in a state as a measureof the number of market actors9 Because the logic of cartel-capture theory impliesdiminishing impacts of increased numbers beyond some point we also includethe square of the number of market actors

We measure market heterogeneity and the prospects for conicts of interestwithin a state using two dimensions of interrm diversity10 We use the coefcientof variation in rm assets to tap asset diversity of rms in a state Based on the stan-dard deviation the coefcient of variation measures the relative amount by whichrms depart from the mean level of re insurance assets in the state Assetsinclude cash real estate mortgage loans stocks bonds collateral loans andreceivables and measure the overall nancial strength of insurers Second we usethe Index of Qualitative Variation to measure diversity by company type (mutualUS joint stock foreign) a potent source of factionalism and conict among insur-ers The measure is based on the percentage of rms of each type and it variesfrom 0 when all rms in a state are of one type to 1 when rms are equally dis-tributed across rm types so that higher values indicate greater heterogeneitySince the effects of heterogeneity may be curvilinear we also include the square ofasset diversity and the square of diversity by company type11

Interest Group Variables

Our interest group measures include both the numerical size of the group and aproxy for its stakes in insurance regulation12 Stakes are based on the economicvalue in a sector with dollar amounts adjusted for ination according to the Con-sumer Price Index (where 1967 5 100) Size and stakes are both measured per cap-ita to control for the size of the state

For insurers themselves stakes are measured as the total assets in re insuranceper capita for each state We measure the numerical strength of re insurers in astate with the number of re insurance agents per capita since agents were theindustryrsquos political shock troops (Grant 1979) For consumer groups we opera-tionalize stakes in regulation with measures of the amount of economic valuepotentially covered by insurance We measure farmersrsquo stakes with the total valueof farm property per capita and the size of the agricultural sector with the total num-ber of farms in a state per capita Manufacturersrsquo stakes are represented by value


added by manufacture per capita and size of the manufacturing sector by the numberof manufacturing establishments per capita13 While conceptually distinct in our datamanufacturersrsquo size and stakes are correlated (r 5 64) Since the measure of man-ufacturersrsquo stakes is also highly correlated with number of farms per capita (r 581) we exclude manufacturersrsquo stakes from our analysis to avoid multicollinear-ity problems Appendix A shows correlations and univariate statistics for theindependent variables

Institutional Variables

We measured the existence of an anticompact law using the same sources usedfor rate regulation laws Anticompact states were considered to be those that atany time during the twentieth century had a law specically prohibiting coopera-tive rate making in re insurance or banning private pricing of re insurance14

The other institutional variation involves the consolidation of the re insuranceeld and the creation of suprastate administrative capacities over time We cap-ture this shift and the institutional regimes it delimits by splitting the analysis intotwo periodsmdashone period (1906ndash15) before and another (1916ndash30) after innova-tions in administrative capacity and standardization were rmly in place Weidentied these periods through extensive study of the history of the industry(Crane 1972 NCIC 1915 New York State 1911ndash1932 1922 Reigel 1917) In thisarticle we approach these two institutional factors as two different perhapscomplementary ways of thinking about the institutional environmentmdashone atthe state level (anticompact laws) and the other at the system level (the shiftover time)15

METHODS

We perform a series of event-history analyses of the passage of rate regulation inforty-eight states from 1906 to 193016 We use discrete-time models with time-varying independent variables These models estimate the effects of independentvariables on the conditional probability of a state passing regulation at time tgiven that the state has not previously passed regulation (Allison 1995) Thediscrete-time specication is appropriate because rating laws were passed onlyduring legislative sessionsmdashthat is at a bounded moment in timemdashand since wecan only measure the time of passage in years In these models the ldquostate-yearrdquo isthe unit of analysis We use 1906 as the start date of the analysis for three reasonsFirst serious proposals for regulating re insurance rates began circulating inabout 1900 (Grant 1979100ndash33) making that the earliest time we could considerstates at risk Second the rst rating laws were not passed until 1909 Third 1906was the earliest year for which good data on all variables exist We use 1930 as ourend date because the format of key data sources changed at that point andbecause thirty-three of the thirty-four states that passed laws had already done soby that time We ran all models using logistic regression procedures in SAS (proclogistic) with maximum likelihood estimation


For both cartel-capture theory and interest group theory we rst use the vari-ables specied by these theories to predict the passage of rate regulation in theentire set of state-years disregarding institutional factors (ldquogeneralrdquo models 1 and4) These models assume general theoretical causal mechanisms and follow thetheories in making no predictions about institutions We then control for anydirect effects of institutional factors (ldquogeneral plus institutionsrdquo models 2 and 5)Our key hypotheses however are about how institutions condition the effects ofother variables Therefore we then use conditional models which identify inter-action effects between institutional context and the variables suggested by theo-ries of regulation (ldquoconditionalrdquo models 3 and 6) Here the effects of variables areallowed to change across social contexts For clarity of presentation we split theset of state-years into institutionally dened segments following Thornton andOcasio (1999)17 This form of presentation is equivalent to using interaction effectswith a dichotomous variable and is often used when one expects the effects ofindependent variables to differ across groupsmdashas is common in straticationresearch (Bobo and Kluegel 1993 England Christopher and Reid 1999 Mariniand Fan 1997)

The conditional models allow us to do two things First we can test for changesin the effects of individual independent variables across institutional conditions bycomparing a model in which a coefcient is constrained to be equal across condi-tions to a model that allows the effect to vary18 Second we can test for the gener-ality of the theories by seeing if the effect of entire blocks of variables depend oninstitutional conditions Here we compare the t of the conditional model to thatof the general model with only direct effects of institutions For both of thesetasks we use a one-tailed Chi-squared test to see if the t of the model improveswith the relevant alteration since the difference in 22 Log Likelihood scores has aChi-squared distribution and since the models for each theory are progressivelynested (ie model 1 is nested in 2 is nested in 3)

RESULTS AND DISCUSSION

Table 3 shows the estimated effects of market size and heterogeneity on the rate atwhich states passed regulation In model 1 cartel-capture theory is only weaklysupported as a general theory Only one market variable has the expected effecton the passage of regulation The positive and signicant coefcient for assetdiversity and the negative and signicant coefcient for the squared term indicatethat heterogeneity among re insurers increases the likelihood of regulation up toa pointmdashafter which further interrm differences begin to decrease the likelihoodof regulation This effect remains when controlling for the existence of anticom-pact laws in model 2 Here up to moderate levels (36 standard deviations abovethe mean) market heterogeneity promotes regulation since it undermines cartelsand increases rmsrsquo incentives for political participation19 But in more heteroge-neous markets interrm differences and inghting appear to impede the successof industry capture-coalitions Model 2 also shows that states with anticompactlaws passed regulation at a rate 264 times higher than other states20 This effect isconsistent with previous ndings on the role of anticompany politics and legiti-

Rationality and Institutional Contingency 65TA

BL

E 3

Eff

ects

of

Car

tel-

Cap

ture

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 1

G

ener

al M

odel

All

Stat

es

Mod

el 2

Gen

eral

pl

us I

nsti

tuti

ons

All

Stat

es

Mod

el 3

C

ondi

tion

al M

odel

b

Ant

icom

pact

St

ates

Non

-ant

icom

pact

St

ates

Diff

acr

oss

cond

itio

nsc

Inte

rcep

t2

334

550

(15

8223

)2

352

841

(16

3797

)2

390

95(1

864

73)

211

20

(4

403

29)

Nu

mbe

r of

age

nts

001

72(

0010

5)0

0065

9(

0011

5)0

0089

8(

0022

9)0

0022

8(

0015

5)x

2 5 1

85(d

f 5

2)

(Nu

mbe

r of

age

nts

)22

000

0005

58(

0000

0044

1)2

000

0001

47(

0000

0046

5)2

000

0001

83(

0000

0108

)2

000

0000

145

(00

0000

585)

Ass

et d

iver

sity

338

2(

1646

)3

632

(17

43)

130

9(

2005

)8

486

(44

24)

x2 5

28

5(d

f 5

2)

(Ass

et d

iver

sity

)22

001

19

(00

0571

)2

001

25

(00

0604

)2

000

44(

0007

1)2

002

91

(00

151)

Div

ersi

ty b

y co

mp

any

typ

e13

472

1(2

584

90)

116

335

(25

5520

)2

276

309

(30

7304

)11

94

(70

0626

)x

2 5 5

361

(df

5 2

)(D

iver

sity

by

com

pan

y ty

pe)

22

604

27(1

673

54)

24

5733

(16

6072

)22

078

9(2

052

39)

273

327

3(4

353

16)

An

tico

mp

act

law

mdash9

720

(44

50)

mdashmdash

N o

f st

ate-

year

s39

839

817

622

2N

of

even

ts33

3321

122

2 L

og L

ikel

ihoo

d20

718

420

216

819

453

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

3 h

as ti

me-

var

yin

g in

terc

epts

an

d t

ime-

var

yin

g ef

fect

s of

cov

aria

tes

all

owin

g th

e ef

fect

s of

cov

aria

tes

to c

han

ge a

cros

s in

stit

uti

onal

con

dit

ion

s C

onst

rain

ing

thes

e ef

fect

sas

a g

rou

p t

o be

equ

al a

cros

s co

nd

itio

ns

doe

s n

ot a

lter

th

e

t of

th

e m

odel

(x

2 5

76

37 d

f 5

7 p

2

5)

cSi

gn

ica

nce

tes

ts f

or a

dif

fere

nce

in c

oef

cien

ts a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

the

mod

el w

ith

th

e co

ef

cien

ts c

onst

rain

ed t

o be

eq

ual

com

par

ed t

o a

mod

el t

hat

all

ows

the

coef

ci

ents

to

vary

Si

gni

can

ce te

sts

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001


macy crises in spurring regulation (Schneiberg and Bartley 2001) but it is the con-ditional model that sheds light on the process by which this institutional factormodies the effect of market dynamics

Model 3 shows that several market factors affect regulation but only in stateswithout anticompact laws In anticompact states none of the market variableshave signicant effects on the passage of regulation Yet in the non-anticompactstates market heterogeneity is an important source of pressure for regulationDiversity by company type has no effect in anticompact states but has theexpected curvilinear effect in non-anticompact states This change across condi-tions is different from zero at the 10 level of signicance (Chi-squared 5 536 2df p 10) Likewise asset diversity has no effect in anticompact states but hasthe expected curvilinear effect in non-anticompact states but this difference ineffects is not statistically signicant (Chi-squared 5 285 2 df p 25)

Overall only in depoliticized settings (non-anticompact states) do market con-ditions affect the likelihood of regulation This supports version II of the institu-tional mediation hypothesismdashthe idea that political environments hostile to theindustry nullify market dynamics as a source of regulation In these environ-ments rms face greater risk and uncertainty about regulation being appropri-ated by anti-industry forces and therefore will not pursue regulation as a solutionto market-control problems In depoliticized settings on the other hand marketheterogeneity provides an impetus for regulation In these settings the positivecoefcient for diversity by company type indicates thatmdashup to a pointmdashincreas-ing the extent to which the insurance market is split between US stock foreignand mutual companies increases the likelihood of regulation As this source ofantagonism increases it becomes more difcult to form private cartels and rmsturn to state regulation Yet the negative squared term indicates that once theextent of diversity by company type exceeds the mean by 45 standard deviationsincreasing heterogeneity decreases the odds of regulation At fairly high levelsthe dynamics of market heterogeneity devolve into inghting and thus impedermsrsquo political action

Support for this version of the anticompact institutional contingency hypothesis(version II) leaves version I without support In that alternative anticompact lawswould magnify the effects of market conditions on regulation by enforcing thecompetitive dynamics specied by cartel-capture accounts of regulation How-ever no market variables reach signicance in anticompact states where lawsenforcing competition existed Therefore the importance of anticompact regimesseems to lie not in their effects on competitive behavior but rather in their effecton rmsrsquo abilities to clearly calculate the consequences of turning from marketsto states

While one might also expect cartel-capture theory as a whole to perform betterwhen made institutionally contingent this idea is not supported Interacting thewhole set of cartel-capture variables with anticompact status does not signi-cantly alter the t of the model (model 3 vs 2 Chi-squared 5 7637 7 df p 25)

Table 4 presents models for interest group theory and its institutionally contin-gent versions When treated as a general theory interest group theory makes onlya marginal contribution to explaining regulation The only notable interest group


TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001


inuence (in model 4) lies in the positive effect of the number of manufacturingestablishments per capita This effect remains when controlling for the periodeffect (1916ndash30) in model 5 So independent of institutional factors the larger therelative numerical size of the manufacturing sector the more likely a state is topass re insurance rate regulation The period also has a positive and signicanteffect indicating that the underlying rate at which states passed regulation was285 times greater after 1915 than it had been previously To understand thedynamics behind this historical change we turn to the conditional model

There is some evidence that states became more likely to pass regulation after1915 because the consolidation of the insurance eld around a model of regulatedcooperation and the development of administrative capacities for implementingthat model created a basis for compromise between insurers and consumergroups Model 6 lends support to this notion in several ways First note that inthe 1906ndash30 period the level of re insurance assets in a state has a negative andsignicant effect on the likelihood of regulation whereas the value of farm prop-erty has a positive and signicant effect From an interest group perspective it isreasonable to interpret these effects as an indicator of divergent interests Beforethe development of systemwide models and capacities that stabilized insurancegovernance nancially strong insurers tended to block the passage of regulationIn contrast farmersrsquo high stakes in the insurance business made them key sup-porters of rate regulation Thus in the 1906ndash15 period regulatory politics tendedto pit insurers against farmers in a battle between the industry and outsiders

Second this antagonism diminished once participants reached a systemwideaccord on a model of ldquoregulated cooperationrdquo and put into place the administra-tive innovations needed to institute this accord in the mid-1910s The effect of reinsurance assets goes from negative in the rst period to nonsignicant in the sec-ond period and this change is statistically signicant (Chi-squared 5 1406 p 001) Whereas large insurers had previously opposed regulation this oppositiondiminished after 1915 In addition one of the key consumers of re insurancemdashthe agricultural sectormdashsupported regulation after 1915 with the number offarms per capita having a positive effect on the likelihood of regulation asignicant change from the previous period (Chi-squared 5 1390 p 001) Theeffect of the size of the manufacturing sector also signicantly changes to a posi-tive effect after 1915 (Chi-squared 5 652 p 05) These last two positive effectsindicate that the size of important consumer groups was particularly signicantfor the passage of regulation after 1915 Notably all interest group effects in the1916ndash30 period are either positive or not signicantly different from zeromdashsug-gesting a lack of organized opposition to regulation Put differently in the secondperiod there is no group for which an increase in size or stakes produces adecrease in the likelihood of regulation

The other institutionally contingent version (version I) of interest group pro-cesses stated that the institutional shift of the mid-1910s around particular modelsof market order and administrative capacities might serve to clarify group inter-ests If the institutional shift did clarify interests we should see an increased num-ber of signicant coefcients in the second period Although the effect of farmsper capita is highly signicant in the 1916ndash30 period this is an exception to the


general pattern of nonsignicant or only mildly signicant effects in this periodThus there is little evidence on the whole that interest group preferences becameclearer with the consolidation of an institutional eld

Still as a whole interest group theory explains more about regulation whenmade institutionally contingent Adding interactions between interest group vari-ables and institutional factors increases the t of the model over the generalmodel even controlling for direct institutional effects (model 6 vs 5 Chi-squared 527109 6 df p 001)21

CONCLUSION

Are the politics of economic regulation contingent on institutions There are sev-eral ways in which institutional factors mediate processes that generate regula-tory intervention Using the re insurance case and arguments about the constitu-tive aspect of institutions we focused on the social contexts that facilitate inhibitor align interests and considered institutional variation occurring across both geo-political space (US states) and historical time (periods) This allowed us to exam-ine the dynamics of regulation in a historically important economic sector and toidentify the conditions under which theories of regulation might apply

We found evidence of two forms of institutional contingency First rms seek-ing to control markets switched from cartels to regulation as market heterogeneityincreased but only under institutional conditions that limited political risks Thisndingmdashthat the dynamics of competition and cartelization affected regulationonly in states without anticompact lawsmdashshows that political environments shapehow market conditions get translated into political interests Relatively depoliti-cized institutional environments facilitate rmsrsquo calculations about the expectedeffects of regulation More hostile settings inhibit clear calculations and subjectcapture strategies to high levels of uncertainty severing connections betweenmarkets and state intervention

Second interest group dynamics drove the passage of rate regulation largelybecause they were conditioned by institutional factors that shaped the content of groupinterests States tended to pass regulation after 1915 because the institutionaliza-tion of regulated cooperation provided a basis for the interests of companies andtheir opponents to converge Rationalized standards and administrative capaci-ties for making and judging the reasonableness of rates allowed insurers to opt forregulation by diminishing the risk that it would be appropriated by hostile out-siders They also allowed farmers and manufacturers to support regulation bydiminishing the risk that collective rate making would foster an ldquoinsurance trustrdquoIn effect standards and rationalized rate administration allowed for compromiseamong interest groups that were previously at odds

Laws regulating re insurance then were not simply the result of characteris-tics of the industry its consumers or some other general process Rather the eco-nomics and politics of regulation rested on specic institutional conditionsdened on the one hand by the state-level political environments and on the otherby historical changes in eld-level administrative and adjudicative capacitiesGiven these contexts regulation emerged at least partly from the dynamics


specied by cartel-capture and interest group theories Crucially however thesecontexts were neither constant nor unproblematic The theories at hand contrib-ute more to the understanding of re insurance regulation when institutional con-text is not assumed but rather is treated as a variablemdashone that sometimesmodies the effects of other variables

This examination of the interplay between institutions and politics contributesto debates about generality and specicity in sociological reasoning to the analy-sis of regulation and to neoinstitutional theory and research In the big pictureour project speaks to questions about causation in social science Do theoriesexplain outcomes over a wide range of social contexts or do their scopes need tobe narrowly dened for their explanatory power to shine through Historicalsociologists have recently debated the role of general theories causal mecha-nisms and contingency in historical sociology (Kiser and Hechter 1991 1998Somers 1998) To simplify on one side of this debate are those who argue thatexplanations should employ causal mechanisms derived from more general theo-ries like rational choice theory On the other side are those who emphasize theneed for epistemologies and causal narratives specic to a historical time andplace The debate hinges in part on divergent philosophies of science and thusthe competing arguments at times appear incommensurable Yet this debateshould also serve as a call for empirical research into the generality or specicityof causal mechanisms and for inquiry into how these mechanisms vary acrossdifferent social contexts As we have demonstrated sensitivity to institutionalcontingency does not necessarily mean jettisoning theory Rather we have fol-lowed Paige (1999) in attempting to develop conditional theory rather than uni-versal theory

At a more concrete level are implications for research on regulation In the pastfew decades this research has expanded its scope outward from intraindustryprocesses centered on rmsrsquo behavior to interest group struggles involving a widerange of actors to broader institutional dynamics and structures Our analysistakes one step further to suggest several possible institutional scope conditions foreconomic and political theories of regulation derived from rational choice the-ory22 An important scope condition for cartel-capture theory may be the lack ofpolitical hostility surrounding an industry If political environments are unthreat-ening to rms then market conditions are important determinants of regulationHowever where oppositional forces have institutional power rms are less likelyto employ regulation as a market control strategy Delimited in this way cartel-capture theory is a powerful way of thinking about how market dynamics shaperegulatory outcomesmdashespecially because it calls attention to market factors notoften measured by sociologists For interest group theory our ndings drawattention to institutional conditions that structure complementarities of interestmdashspecically between rms and consumers Thus our analysis suggests that in-terest group arguments that focus on pluralistic compromise may be most appro-priate in a social context of well-dened models of order and administrativeprocedures Compromise between interest groups may not be as likely when in-formational and administrative standards for judging fairness are themselves atthe center of controversy and political struggle


Yet the most important implications of our results lie at the middle range witheconomic sociology the new institutionalism in organizational analysis and theirrelationships with rational choice theories Organizational forms and strategiesthat appear ldquonaturalrdquo in hindsight are in fact contingent social and historical con-structions Researchers have demonstrated this point in analyses of the moderncorporation (Berk 1994 Roy 1997) organizational succession (Thornton and Oca-sio 1999) and the dynamics of mergers and acquisitions (Dobbin and Dowd2000) In doing so researchers have begun to document how institutions condi-tion mediate or moderate other social processes and thereby facilitate the cre-ation of structures and practices that later become taken for granted We contrib-ute to this project by pointing out the institutionally contingent character ofregulation and the processes and choices that underlie it At a minimum ourndings suggest that two existing rational choice theories of a specic outcomemdashregulationmdashare limited in scope At a maximum our ndings show how eco-nomic and political rationality is itself institutionally constituted

To be clear our results do not necessarily subvert rational choice as an analyticframework In fact rational choice theorists have developed fundamental insightsinto how uncertainty and incomplete information promote institution buildingNevertheless our research draws attention to social contexts characterized notmerely by incomplete information but also by the absence of basic evaluative cri-teria Since it may be the evaluative context of a decisionmdashnot just the informa-tional contentmdashthat is incomplete it seems premature to reduce such periods ofldquoconstitutive politicsrdquo to a state of incomplete information

A more productive strategy for further inquiry would be to avoid reductionismand to address shifts between (a) the production of specic outcomes within anestablished institutional regime and (b) the creation consolidation or failure toconsolidate an institutional regime The production of outcomesmdashsuch as regula-tionmdashwithin a settled system may often be fruitfully understood in terms of themechanisms specied by existing rational choice theories But in elds character-ized by ambivalence multiple models and ill-dened options actorsrsquo capacitiesto calculate rationally are subverted Under these conditions we would expect ashift away from struggles over particular outcomes and problem solving undergiven constraints to struggles over institutional regimes themselves and efforts tocreate or restore the bases for calculation and political compromise whether bybuilding new behavioral controls (Heimer 1985) by dening and delimitingoptions (Berk 1994 Roy 1997) or by extending processes of commensuration (Espe-land 1998) On this point institutionalists from various camps appear to agree Andon this point rest real possibilities for integration and fruitful cross-fertilization


AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1


Acknowledgments We thank James Noon for his valuable research assistanceFor thoughtful comments and discussions on this article we thank Mark ChavesElisabeth Clemens William Ocasio Patricia Thornton and the anonymous Socio-logical Perspectives reviewers The research reported here was supported in part bythe Social and Behavioral Sciences Research Institute at the University of Arizona

NOTES

1 Importantly our focus is on the dynamics of economic regulation which may be differ-ent from those of ldquosocial regulationrdquo (eg environmental protection equal opportu-nity law)

2 For example in the insurance case a relatively small group of large mostly north-eastern-based national operating companies themselves carried the costs of privatelycartelizing the industrymdashusing their size common location and inuence to gain con-trol over industry councils and conduct price wars and reinsurance boycotts againstdefectors

3 Only one state passed an anticompact law after 1910 Arizona which added a weakversion of a law to its constitution when admitted as a state in 1912

4 To be clear we are examining anticompact laws as the context in which rms try tosolve market problems Thus our key question is about the interaction between anti-compact laws and cartel-capture variables However it is also possible that the legiti-macy crises that anticompact laws represented may have had direct effects on thepassage of regulationmdashan idea that has found support in previous research (Schneibergand Bartley 2001) In a few early cases rate regulation was an extension of anticompactlaws representing a strategy of simply using the state to control rates But for the mostpart regulation was more of a compromise or ldquomiddle wayrdquo in which states allowedassociation but subjected it to public supervision For more on the signicance of anti-compact laws see Schneiberg 1999

5 Interest group theorists also hypothesize that homogeneous populations are morelikely than heterogeneous populations to organize into interest groups These argu-ments are rarely operationalized In this case there is evidence of organization amongseveral distinct producer and consumer groups so we do not address these argumentswhich are largely oriented to explaining why consumer groups are on the whole lesslikely to be organized than producers (Noll and Owen 198344)

6 Institutional innovations continued in 1921 with the ldquo5 year 5rdquo compromise betweenthe National Board of Underwriters and the National Convention of Insurance Com-missioners which set a specic standard for judging rates the use of ve years ofclassied data and a 5 percent underwriting prot (NCIC 192219ndash29)

7 All but one of the rate regulation states granted commissioners the power to reviewrates and order changes in rates The exception was Massachusetts which only allowedcommissioners to recommend rate changes limiting regulators to an advisory role

8 A number of states took the additional step of compelling insurers to either form orjoin associations for the purpose of making and ling rates and a few statutes actuallyformed state rate-making associations In a handful of states especially early adopterslike Kansas the initial legislation subjected rates to regulation without authorizingassociations However by the mid-1910s virtually all of these states had eitheramended their statutes to include such authorization or had recognized associationsde facto through administrative rulings


9 We use agents rather than companies here because agents set prices for policies and arefar more numerous than companies their numbers will be most important for the sec-tor rsquos capacity to monitor and enforce price agreements Further agents are businessowners and resident-voters in their state making them the industryrsquos political shocktroops (Grant 1979) and key players in any political coalition

10 Companies with many assets were relatively low-cost ldquogeneral businessrdquo insurers sell-ing a range of insurance products in multiple states In contrast smaller rms typicallydid business in one or a few states specialized in ldquopreferred risksrdquo such as dwellingsschools or churches and paid high commissions to get those accounts Thus theseldquohome companiesrdquo had different interests than general insurers and vied with themover prices and governance policies within industry associations Regarding companytype stock and mutual companies embodied antithetical operating principles to suchan extent that stock companies excluded mutuals from associations and pushed forlegislation against the mutual form (Schneiberg 1994242ndash70 330ndash64)

11 We recorded all these market conditions every ve years starting in 1905 using linearinterpolation for intervening years The Spectator Insurance Year Books listed data asof December 31 of a year so we lagged all market variables by one year We assignedmultistate rms to each of the states in which they were licensed to do business Weused Bestrsquos Insurance Reports for the same years to ll in missing data on states of oper-ation and for a few small companies had to infer this information based on evidencefrom earlier and later years company size and county-specic company names Vari-ables calculated with and without companies for which data was minimal were corre-lated at 983 or higher

12 Data on farming and manufacturing sectors comes from various years of the StatisticalAbstract and US Census The Census Bureau collected manufacturing data every veyears from 1905 to 1919 and every two years thereafter It collected agricultural dataevery ten years from 1900 to 1920 and every ve years thereafter We used linear inter-polation to ll in intervening years

13 We also attempted to capture differences between small and large manufacturers bymeasuring the percent of manufacturing rms in a state with twenty or fewer employ-ees But we dropped this variable from the analysis because it had a high varianceination factor and was correlated with the value of farm property (72)

14 Three states experimented with ldquostatistrdquo regimes in which the state actually set insur-ance rates Two of these also had anticompact laws For the purpose of the analysis wetreat the context created by a ldquostatistrdquo regime as equivalent to an anticompact regime

15 Here we treat these factors separately but the historical relationships between the twoare complex The institutional reforms of 1914ndash15 represented a systemwide consolida-tion of rate regulationmdasha movement that initially arose in 1909 in the context of anti-compact regimes in some states However once the reforms of 1914ndash15 were in placethe anticompact movement was well past its apogee and was being superseded by rateregulation in many states

16 Since re insurance was regulated at the state level and following established practice(Amenta Dunleavy and Bernstein 1994 Pavalko 1989) we treat states as independentunits of analysis Although a few future states were still territories at the beginning ofthis period these territories tend to resemble other states in terms of laws relating toinsurance

17 Thornton and Ocasio (1999) split their data set into two periods and present ldquopiece-wise exponentialrdquo models As Allison (1995105) explains the idea of a such a model isto ldquodivide the time scale into intervals Assume that the hazard is constant within each


interval but can vary across intervals Thus the intercept in the log-hazard equationis allowed to vary in an unrestricted fashion from one interval to anotherrdquo While weuse discrete time rather than continuous time models our approach to time periods inthe analysis of interest group effects is equivalent to the continuous time piecewisemodel used by Thornton and Ocasio and others (Allison 1995108)

18 To do this we ran additional models and compared for each independent variable thet of a model that included its interaction with the institutional variable to the t of amodel that lacked such an interaction term

19 This and similar ndings were derived by integrating the quadratic equation esti-mated by the model and then transforming the units into standard deviations of theappropriate variable

20 This odds ratio is derived by exponentiating the anticompact coefcient (eb)21 We also explored the converse of the arguments abovemdashthe possibility that cartel-

capture effects are conditioned by the institutional shift over time while interest groupeffects are conditioned by anticompact status We ran an additional ldquofull modelrdquo(results available on request) in which both cartel-capture and interest group coefcientsare allowed to vary across each of these institutional conditions Although we had todelete one variable because of collinearity the results largely conrm the modelingstrategy suggested by our theoretical discussion and used above Cartel-capture effectswere conditioned only by anticompact status Interest group effects were mainly con-ditioned by the institutional shift

22 By ldquoinstitutional scope conditionsrdquo we mean scope conditions that are not necessarilyhistorically specic (ldquohistorical scope conditionsrdquo) but that are also not derived fromthe general theory itself (ldquoabstract scope conditionsrdquo) For a discussion of historicalversus abstract scope conditions see Kiser and Hechter 1998797

REFERENCES

Akerloff George 1970 ldquoThe Market for lsquoLemonsrsquo Qualitative Uncertainty and the MarketMechanismrdquo Quarterly Journal of Economics 84488ndash500

Alchian Armen and H Demsetz 1972 ldquoProduction Information Costs and EconomicOrganizationrdquo American Economic Review 62777ndash95

Allison Paul 1995 Survival Analysis Using the SAS System Cary NC SAS InstituteAmenta Edwin 1998 Bold Relief Institutional Politics and the Origins of Modern American

Social Policy Princeton NJ Princeton University PressAmenta Edwin Kathleen Dunleavy and Mary Bernstein 1994 ldquoStolen Thunder Huey

Longrsquos lsquoShare Our Wealthrsquo Political Mediation and the Second New Dealrdquo AmericanSociological Review 59678ndash702

Axelrod Robert M 1984 The Evolution of Cooperation New York Basic BooksBates Robert Avner Grief Margaret Levi Jean Laurent Rosenthal and Barry Weingast

1998 Analytical Narratives Princeton NJ Princeton University PressBerk Gerald 1994 Alternative Tracks The Constitution of American Industrial Order 1965ndash

1917 Baltimore Johns Hopkins University PressBobo Lawrence and James R Kluegel 1993 ldquoOpposition to Race-targeting Self-Interest

Stratication Ideology or Racial Attitudesrdquo American Sociological Review 58443ndash64Bowman John R 1989 Capitalist Collective Action Competition Cooperation and Conict in

the Coal Industry Cambridge Cambridge University PressBrearly Harry Chase 1916 The History of the National Board of Fire Underwriters New York

Frederick Stokes


Brinton Mary C and Victor Nee 1998 The New Institutionalism in Sociology New YorkRussell Sage Foundation

Chandler Alfred D Jr 1977 The Visible Hand The Managerial Revolution in American Busi-ness Cambridge MA Harvard University Press

Clemens Elisabeth and James Cook 1999 ldquoPolitics and Institutionalism Explaining Dura-bility and Changerdquo Annual Review of Sociology 25441ndash66

Collins Randall 1980 ldquoWeberrsquos Last Theory of Capitalism A Systematizationrdquo AmericanSociological Review 45925ndash42

Crane Frederick G 1972 ldquoRate Regulation The Reason Whyrdquo Journal of Risk and Insurance39511ndash34

Dean A F 1901 The Rationale of Fire Rates Chicago J M MurphyDobbin Frank and Timothy J Dowd 2000 ldquoThe Market That Antitrust Built Public Policy

Private Coercion and Railroad Acquisitions 1825 to 1922rdquo American SociologicalReview 65631ndash57

Douglas Mary 1986 How Institutions Think Syracuse NY Syracuse University PressElster Jon 1989 Nuts and Bolts for the Social Sciences New York Cambridge University

PressEngland Paula Karen Christopher and Lori L Reid 1999 ldquoGender Race Ethnicity and

Wagesrdquo Pp 139ndash82 in Latinas and African American Women at Work edited by IBrowne New York Russell Sage Foundation

Espeland Wendy Nelson 1998 The Struggle for Water Politics Rationality and Identity in theAmerican Southwest Chicago University of Chicago Press

Fishback Price V and Shawn Everett Kantor 1996 ldquoThe Durable Experiment State Insur-ance of Workersrsquo Compensation Risk in the Early Twentieth Centuryrdquo Journal of Eco-nomic History 56809ndash36

Fligstein Neil 1990 The Transformation of Corporate Control Cambridge MA Harvard Uni-versity Press

Friedland Roger and Robert R Alford 1991 ldquoBringing Society Back In Symbols Practicesand Institutional Contradictionsrdquo Pp 232ndash63 in The New Institutionalism in Organiza-tional Analysis edited by W W Powell and P DiMaggio Chicago University of Chi-cago Press

Grant H Roger 1979 Insurance Reform Consumer Action in the Progressive Era Ames Uni-versity of Iowa Press

Greif Avner 1998 ldquoSelf-Enforcing Political Systems and Economic Growth Late MedievalGenoardquo Pp 23ndash63 in Analytical Narratives edited by R Bates A Greif M Levi J LRosenthal and B Weingast Princeton NJ Princeton University Press

Hall Peter A and Rosemary C R Taylor 1996 ldquoPolitical Science and the Three New Insti-tutionalismsrdquo Political Studies 44936ndash57

Hamilton Gary G and Robert Feenstra 1998 ldquoThe Organization of Economiesrdquo Pp 153ndash80 in The New Institutionalism in Sociology edited by M C Brinton and V Nee NewYork Russell Sage Foundation

Heimer Carol 1985 Reactive Risk and Rational Action Managing Moral Hazards in InsuranceContracts Berkeley University of California Press

Hobbs Clarence W 1925 ldquoState Regulation of Insurance Ratesrdquo Proceedings of the CasualtyActuarial Society 11 (June)218ndash75

mdashmdashmdash 1942 ldquoState Regulation of Insurance Rates Part II Regulation of Rates and RatingOrganizationsrdquo Proceedings of the Casualty Actuarial Society 28 (May)344ndash460

Hollingsworth J Rogers 1991 ldquoThe Logic of Coordinating American Manufacturing Sec-torsrdquo Pp 35ndash73 in Governance of the American Economy edited by J L Campbell J RHollingsworth and L N Lindberg Cambridge Cambridge University Press


Immergut Ellen M 1992 ldquoThe Rules of the Game The Logic of Health Policy-making inFrance Switzerland and Swedenrdquo Pp 57ndash89 in Structuring Politics Historical Institu-tionalism in Comparative Analysis edited by S Steinmo K Thelen and F LongstrethNew York Cambridge University Press

Isaac Larry W and Larry J Grifn 1989 ldquoAhistoricism in Time-Series Analyses of Histori-cal Process Critique Redirection and Illustrations from US Labor Historyrdquo Ameri-can Sociological Review 54873ndash90

Jacquemin Alexis and Margaret E Slade 1989 ldquoCartels Collusion and HorizontalMergerrdquo Pp 415ndash73 in Handbook of Industrial Organization vol 1 edited by R Schal-manesee and R D Willig Amsterdam Elsevier

Kanazawa Mark and Roger Noll 1994 ldquoThe Origins of State Railroad Regulation The Illi-nois Constitution of 1870rdquo Pp 13ndash54 in The Regulated Economy edited by C Goldinand G Libecap Chicago University of Chicago Press

Kiser Edgar and Michael Hechter 1991 ldquoThe Role of General Theory in Comparative-Historical Sociologyrdquo American Journal of Sociology 971ndash30

mdashmdashmdash 1998 ldquoThe Debate on Historical Sociology Rational Choice Theory and Its CriticsrdquoAmerican Journal of Sociology 104785ndash816

Kolko Gabriel 1963 The Triumph of Conservatism Glencoe IL Free PressMarch James G and Johan P Olsen 1976 Ambiguity and Choice in Organizations Bergen

Universitetsforlagetmdashmdashmdash 1989 Rediscovering Institutions The Organizational Basis of Politics New York Free

PressMarini Margaret Mooney and Pi-Ling Fan 1997 ldquoThe Gender Gap in Earnings at Career

Entryrdquo American Sociological Review 62588ndash604Marwell Gerald and Pamela Oliver 1993 The Critical Mass in Collective Action A Micro-

Social Theory Cambridge Cambridge University PressMeier Kenneth 1988 The Political Economy of Regulation The Case of Insurance Albany State

University of New York PressMerritt Committee 1909ndash11 ldquoHearings before the Joint Committee of the Senate and

Assembly 1909ndash1911rdquo New York Assembly Documents No 30 Albany J B Lyonmdashmdashmdash 1911 ldquoReport of the Joint Committee of the Senate and Assembly of the State of

New York Appointed to Investigate the Corrupt Practices in Connection with Legis-lation and the Affairs of Insurance Companies Other than Those Doing Life Insur-ance Businessrdquo New York Assembly Documents No 30 Albany J B Lyon

Meyer John W John Boli George Thomas and Francisco O Ramirez 1997 ldquoWorld Societyand the Nation Staterdquo American Journal of Sociology 1031444ndash81

Mobray Albert 1946 Insurance New York McGraw HillNational Convention of Insurance Commissioners (NCIC) 1915 Proceedings of the Forty-

sixth Session of the National Convention of Insurance Commissioners Del Monte Califor-nia September 21ndash24 (Includes adjourned meetings in New York on December 81914 and in Chicago on April 12 1915) Columbia SC R L Bryan

mdashmdashmdash 1922 Proceedings of the Fifty-third Annual Session of the National Convention of Insur-ance Commissioners Swampscot Massachusetts September 5ndash8 (Includes adjournedmeetings in New York on December 6ndash8 1921 and in St Louis on May 1ndash2 1922)Columbia SC R L Bryan

New York State Insurance Department 1911ndash32 52nd through 73rd ldquoAnnual Reports ofthe Superintendent of Insurance of the State of New Yorkrdquo and ldquoReports on OfcialExaminationsrdquo New York Senate Documents New York Assembly Documents and NewYork Legislative Documents Albany J B Lyon

New York State Joint Legislative Committee on Housing 1922 ldquoIntermediate Report of


the Joint Legislative Committee on Housingrdquo New York Legislative Documents No 60Albany J B Lyon

Noll Roger G and Bruce M Owen 1983 The Political Economy of Deregulation InterestGroups in the Regulatory Process Washington DC American Enterprise Institute

Paige Jeffrey M 1999 ldquoConjuncture Comparison and Conditional Theory in MacrosocialInquiryrdquo American Journal of Sociology 105781ndash800

Pavalko Eliza 1989 ldquoState Timing of Policy Adoption Workmenrsquos Compensation in theUnited States 1909ndash1929rdquo American Journal of Sociology 95592ndash615

Pierson Paul 1994 Dismantling the Welfare State Reagan Thatcher and the Politics ofRetrenchment New York Cambridge University Press

Posner Richard A 1974 ldquoTheories of Economic Regulationrdquo Bell Journal of Economics andManagement Science 5(2)335ndash58

Powell Walter W and Paul DiMaggio eds 1991 The New Institutionalism in OrganizationalAnalysis Chicago University of Chicago Press

Riegel Robert 1916 Fire Underwriters Associations in the United States New York ChronicleCo

mdashmdashmdash 1917 ldquoRate-making Organizations in Fire Insurancerdquo and ldquoProblems of Fire Insur-ance Rate Makingrdquo Annals of the American Academy of Political and Social Science70(159)172ndash219

mdashmdashmdash 1927 ldquoThe Regulation of Fire Ratesrdquo Annals of the American Academy of Political andSocial Science 219(March)114ndash20

Roy William G 1997 Socializing Capital The Rise of the Large Industrial Corporation in Amer-ica Princeton NJ Princeton University Press

Schneiberg Marc 1994 ldquoPrivate Order and Inter-Firm Governance in the American FireInsurance Industry 1820ndash1950rdquo PhD dissertation University of Wisconsin

mdashmdashmdash 1999 ldquoPolitical and Institutional Conditions for Governance by Association Pri-vate Order and Price Controls in American Fire Insurancerdquo Politics and Society 2767ndash103

Schneiberg Marc and Tim Bartley 2001 ldquoRegulating American Industries Markets Poli-tics and the Institutional Determinants of Fire Insurance Regulationrdquo American Jour-nal of Sociology 107101ndash46

Schneiberg Marc and Elisabeth Clemens Forthcoming ldquoThe Typical Tools for the JobResearch Strategies in Institutional Analysisrdquo In Bending the Bars of the Iron Cageedited by W W Powell and D Jones Chicago University of Chicago Press

Schneiberg Marc and J Rogers Hollingsworth 1990 ldquoCan Transaction Cost EconomicsExplain Trade Associationsrdquo Pp 233ndash46 in The Firm as a Nexus of Treaties edited byM Aoki B Gustafsson and O Williamson London Sage

Scott James C 1998 Seeing Like a State How Certain Schemes to Improve the Human ConditionHave Failed New Haven Yale University Press

Scott W Richard 1994 ldquoInstitutions and Organizations Toward a Theoretical SynthesisrdquoPp 55ndash80 in Institutional Environments and Organizations Structural Complexity andIndividualism edited by W R Scott and J W Meyer Thousand Oaks CA Sage

mdashmdashmdash 1995 Institutions and Organizations Thousand Oaks CA SageSkowronek Stephen 1982 Building a New American State The Expansion of National Admin-

istrative Capacities 1877ndash1920 New York Cambridge University PressSomers Margaret R 1998 ldquo lsquoWersquore No Angelsrsquo Realism Rational Choice and Relationality

in Social Sciencerdquo American Journal of Sociology 104722ndash84Spectator Company 1901ndash55 Fire Insurance Laws Taxes Fees Philadelphia Spectator Com-

panymdashmdashmdash 1906ndash40 Insurance Year Book Fire and Marine Philadelphia Spectator Company


Stigler George J 1971 ldquoThe Theory of Economic Regulationrdquo Bell Journal of Economics andManagement Science 2(1)3ndash21

mdashmdashmdash 1974 ldquoFree Riders and Collective Action An Appendix to Theories of EconomicRegulationrdquo Bell Journal of Economics and Management Science 5(2)359ndash65

Sutton John R and Frank Dobbin 1996 ldquoThe Two Faces of Governance Responses toLegal Uncertainty in US Firms 1955 to 1985rdquo American Sociological Review 61794ndash811

Swidler Ann 1986 ldquoCulture in Action Symbols and Strategiesrdquo American SociologicalReview 51273ndash86

Thelen Kathleen and Sven Steinmo 1992 ldquoHistorical Institutionalism in Comparative Pol-iticsrdquo Pp 1ndash32 in Structuring Politics Historical Institutionalism in Comparative Analy-sis edited by S Steinmo K Thelen and F Longstreth New York CambridgeUniversity Press

Thornton Patricia H and William Ocasio 1999 ldquoInstitutional Logics and the HistoricalContingency of Power in Organizations Executive Succession in the Higher Educa-tion Publishing Industry 1958ndash1990rdquo American Journal of Sociology 105801ndash43

Tolbert Pamela and Lynne G Zucker 1983 ldquoInstitutional Sources of Change in the FormalStructure of Organizations The Diffusion of Civil Service Reform 1880ndash1935rdquoAdministrative Science Quarterly 2822ndash39

U S Bureau of the Census 1918 Financial Statistics of States 1917 Washington DC Gov-ernment Printing Ofce

Weir Margaret and Theda Skocpol 1985 ldquoState Structures and the Possibilities for lsquoKeynes-ianrsquo Responses to the Great Depression in Sweden Britain and the United StatesrdquoPp 106ndash68 in Bringing the State Back In edited by P Evans D Rueschemeyer and TSkocpol Cambridge Cambridge University Press

Williamson Oliver E 1985 The Economic Institutions of Capitalism New York Free PressZeigler Harmon 1964 Interest Groups in American Society Englewood Cliffs NJ Prentice-

Hall


cing income maximization prot maximization risk power even interest itselfare all institutionally contingentrdquo Similarly Douglas (1986) argues that institu-tions shape the terrain of action by classifying highlighting and obscuring par-ticular features of a situation ldquoNew institutionalistrdquo research in sociology andpolitical science provides an empirical warrant for thinking of institutions as par-tially constituting instrumental action (Hall and Taylor 1996 March and Olsen1989 Powell and DiMaggio 1991) Furthermore recent neoinstitutional researchhas begun to move from analyzing the spread of models of order to examininghow institutional logics and political regimes moderate the dynamics of organiza-tional behaviormdashshowing that causal mechanisms are contingent on institutionalconditions (Dobbin and Dowd 2000 Thornton and Ocasio 1999)

Our approach to institutional contingency considers two theories of economicregulationmdashcartel-capture theory and interest group theorymdashthat provide gen-eral accounts of when and why states intervene in industries to control pricesentry or investment1 These theories use rational choice models of action andassume that actors maximize economic interests We focus on institutional sourcesof interests in regulation and argue that variation in political-institutional condi-tions may produce variation in the clarity and content of preferences for oragainst regulation Cartel-capture theory highlights the market conditions that leadrms to use regulation as a tool for creating or maintaining monopoly but typi-cally ignores the way in which policy shapes competitive behavior and the politi-cization of organizational elds alters interests in ldquostate capturerdquo Interest grouptheory emphasizes the organizing capacities and stakes in regulation of groupsboth internal and external to an industry but typically pays little attention to insti-tutional shifts over time that may clarify and alter group interests These twoapproaches make valuable contributions to understanding economic regulationand the relationship between state policy and private forms of market control Yetthey are more valuable when treated as institutionally contingent

Empirically we use event-history analyses to examine US statesrsquo passage ofrate regulation in the re insurance industry from 1906 to 1930 This industry pro-vides fertile ground for examining the institutionally mediated relationshipsamong market forces interest group politics and state policy First the industrywas central to the economic infrastructure of the United States re insurance wasrequired by creditors and major urban centers frequently had massive res(Brearley 1916 Mowbray 1946) Second the industry was characterized by sub-stantial political and market organization including cartels industry associationsand rate-making bureaus Third starting in 1909 US states began to pass laws toregulate re insurance rate making Typically these laws allowed state insurancecommissioners to order changes in insurance rates or rate-making practices whilealso authorizing insurers to form rating bureaus or rate-making associations Ineffect these laws eschewed antitrust principles in favor of ldquoregulated coopera-tionrdquo By 1930 thirty-three states had enacted rate regulation generating substan-tial variation in outcomes for the purposes at hand

The goal here is quite specicmdashto examine the contingent character of two the-ories of regulation In previous work we have assessed competing theories andfound direct effects of institutional factors on rate regulation (Schneiberg and










































Figure 1







TABLE 1
























The NCIC Model Law















TABLE 2














DATA




Figure 2















METHODS








BL

E 3

Eff

ects

of

Car

tel-

Cap

ture

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 1

G

ener

al M

odel

All

Stat

es

Mod

el 2

Gen

eral

pl

us I

nsti

tuti

ons

All

Stat

es

Mod

el 3

C

ondi

tion

al M

odel

b

Ant

icom

pact

St

ates

Non

-ant

icom

pact

St

ates

Diff

acr

oss

cond

itio

nsc

Inte

rcep

t2

334

550

(15

8223

)2

352

841

(16

3797

)2

390

95(1

864

73)

211

20

(4

403

29)

Nu

mbe

r of

age

nts

001

72(

0010

5)0

0065

9(

0011

5)0

0089

8(

0022

9)0

0022

8(

0015

5)x

2 5 1

85(d

f 5

2)

(Nu

mbe

r of

age

nts

)22

000

0005

58(

0000

0044

1)2

000

0001

47(

0000

0046

5)2

000

0001

83(

0000

0108

)2

000

0000

145

(00

0000

585)

Ass

et d

iver

sity

338

2(

1646

)3

632

(17

43)

130

9(

2005

)8

486

(44

24)

x2 5

28

5(d

f 5

2)

(Ass

et d

iver

sity

)22

001

19

(00

0571

)2

001

25

(00

0604

)2

000

44(

0007

1)2

002

91

(00

151)

Div

ersi

ty b

y co

mp

any

typ

e13

472

1(2

584

90)

116

335

(25

5520

)2

276

309

(30

7304

)11

94

(70

0626

)x

2 5 5

361

(df

5 2

)(D

iver

sity

by

com

pan

y ty

pe)

22

604

27(1

673

54)

24

5733

(16

6072

)22

078

9(2

052

39)

273

327

3(4

353

16)

An

tico

mp

act

law

mdash9

720

(44

50)

mdashmdash

N o

f st

ate-

year

s39

839

817

622

2N

of

even

ts33

3321

122

2 L

og L

ikel

ihoo

d20

718

420

216

819

453

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

3 h

as ti

me-

var

yin

g in

terc

epts

an

d t

ime-

var

yin

g ef

fect

s of

cov

aria

tes

all

owin

g th

e ef

fect

s of

cov

aria

tes

to c

han

ge a

cros

s in

stit

uti

onal

con

dit

ion

s C

onst

rain

ing

thes

e ef

fect

sas

a g

rou

p t

o be

equ

al a

cros

s co

nd

itio

ns

doe

s n

ot a

lter

th

e

t of

th

e m

odel

(x

2 5

76

37 d

f 5

7 p

2

5)

cSi

gn

ica

nce

tes

ts f

or a

dif

fere

nce

in c

oef

cien

ts a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

the

mod

el w

ith

th

e co

ef

cien

ts c

onst

rain

ed t

o be

eq

ual

com

par

ed t

o a

mod

el t

hat

all

ows

the

coef

ci

ents

to

vary

Si

gni

can

ce te

sts

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall










































Figure 1







TABLE 1
























The NCIC Model Law















TABLE 2














DATA




Figure 2















METHODS








BL

E 3

Eff

ects

of

Car

tel-

Cap

ture

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 1

G

ener

al M

odel

All

Stat

es

Mod

el 2

Gen

eral

pl

us I

nsti

tuti

ons

All

Stat

es

Mod

el 3

C

ondi

tion

al M

odel

b

Ant

icom

pact

St

ates

Non

-ant

icom

pact

St

ates

Diff

acr

oss

cond

itio

nsc

Inte

rcep

t2

334

550

(15

8223

)2

352

841

(16

3797

)2

390

95(1

864

73)

211

20

(4

403

29)

Nu

mbe

r of

age

nts

001

72(

0010

5)0

0065

9(

0011

5)0

0089

8(

0022

9)0

0022

8(

0015

5)x

2 5 1

85(d

f 5

2)

(Nu

mbe

r of

age

nts

)22

000

0005

58(

0000

0044

1)2

000

0001

47(

0000

0046

5)2

000

0001

83(

0000

0108

)2

000

0000

145

(00

0000

585)

Ass

et d

iver

sity

338

2(

1646

)3

632

(17

43)

130

9(

2005

)8

486

(44

24)

x2 5

28

5(d

f 5

2)

(Ass

et d

iver

sity

)22

001

19

(00

0571

)2

001

25

(00

0604

)2

000

44(

0007

1)2

002

91

(00

151)

Div

ersi

ty b

y co

mp

any

typ

e13

472

1(2

584

90)

116

335

(25

5520

)2

276

309

(30

7304

)11

94

(70

0626

)x

2 5 5

361

(df

5 2

)(D

iver

sity

by

com

pan

y ty

pe)

22

604

27(1

673

54)

24

5733

(16

6072

)22

078

9(2

052

39)

273

327

3(4

353

16)

An

tico

mp

act

law

mdash9

720

(44

50)

mdashmdash

N o

f st

ate-

year

s39

839

817

622

2N

of

even

ts33

3321

122

2 L

og L

ikel

ihoo

d20

718

420

216

819

453

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

3 h

as ti

me-

var

yin

g in

terc

epts

an

d t

ime-

var

yin

g ef

fect

s of

cov

aria

tes

all

owin

g th

e ef

fect

s of

cov

aria

tes

to c

han

ge a

cros

s in

stit

uti

onal

con

dit

ion

s C

onst

rain

ing

thes

e ef

fect

sas

a g

rou

p t

o be

equ

al a

cros

s co

nd

itio

ns

doe

s n

ot a

lter

th

e

t of

th

e m

odel

(x

2 5

76

37 d

f 5

7 p

2

5)

cSi

gn

ica

nce

tes

ts f

or a

dif

fere

nce

in c

oef

cien

ts a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

the

mod

el w

ith

th

e co

ef

cien

ts c

onst

rain

ed t

o be

eq

ual

com

par

ed t

o a

mod

el t

hat

all

ows

the

coef

ci

ents

to

vary

Si

gni

can

ce te

sts

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall






TABLE 1
























The NCIC Model Law















TABLE 2














DATA




Figure 2















METHODS








BL

E 3

Eff

ects

of

Car

tel-

Cap

ture

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 1

G

ener

al M

odel

All

Stat

es

Mod

el 2

Gen

eral

pl

us I

nsti

tuti

ons

All

Stat

es

Mod

el 3

C

ondi

tion

al M

odel

b

Ant

icom

pact

St

ates

Non

-ant

icom

pact

St

ates

Diff

acr

oss

cond

itio

nsc

Inte

rcep

t2

334

550

(15

8223

)2

352

841

(16

3797

)2

390

95(1

864

73)

211

20

(4

403

29)

Nu

mbe

r of

age

nts

001

72(

0010

5)0

0065

9(

0011

5)0

0089

8(

0022

9)0

0022

8(

0015

5)x

2 5 1

85(d

f 5

2)

(Nu

mbe

r of

age

nts

)22

000

0005

58(

0000

0044

1)2

000

0001

47(

0000

0046

5)2

000

0001

83(

0000

0108

)2

000

0000

145

(00

0000

585)

Ass

et d

iver

sity

338

2(

1646

)3

632

(17

43)

130

9(

2005

)8

486

(44

24)

x2 5

28

5(d

f 5

2)

(Ass

et d

iver

sity

)22

001

19

(00

0571

)2

001

25

(00

0604

)2

000

44(

0007

1)2

002

91

(00

151)

Div

ersi

ty b

y co

mp

any

typ

e13

472

1(2

584

90)

116

335

(25

5520

)2

276

309

(30

7304

)11

94

(70

0626

)x

2 5 5

361

(df

5 2

)(D

iver

sity

by

com

pan

y ty

pe)

22

604

27(1

673

54)

24

5733

(16

6072

)22

078

9(2

052

39)

273

327

3(4

353

16)

An

tico

mp

act

law

mdash9

720

(44

50)

mdashmdash

N o

f st

ate-

year

s39

839

817

622

2N

of

even

ts33

3321

122

2 L

og L

ikel

ihoo

d20

718

420

216

819

453

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

3 h

as ti

me-

var

yin

g in

terc

epts

an

d t

ime-

var

yin

g ef

fect

s of

cov

aria

tes

all

owin

g th

e ef

fect

s of

cov

aria

tes

to c

han

ge a

cros

s in

stit

uti

onal

con

dit

ion

s C

onst

rain

ing

thes

e ef

fect

sas

a g

rou

p t

o be

equ

al a

cros

s co

nd

itio

ns

doe

s n

ot a

lter

th

e

t of

th

e m

odel

(x

2 5

76

37 d

f 5

7 p

2

5)

cSi

gn

ica

nce

tes

ts f

or a

dif

fere

nce

in c

oef

cien

ts a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

the

mod

el w

ith

th

e co

ef

cien

ts c

onst

rain

ed t

o be

eq

ual

com

par

ed t

o a

mod

el t

hat

all

ows

the

coef

ci

ents

to

vary

Si

gni

can

ce te

sts

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall













The NCIC Model Law















TABLE 2














DATA




Figure 2















METHODS








BL

E 3

Eff

ects

of

Car

tel-

Cap

ture

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 1

G

ener

al M

odel

All

Stat

es

Mod

el 2

Gen

eral

pl

us I

nsti

tuti

ons

All

Stat

es

Mod

el 3

C

ondi

tion

al M

odel

b

Ant

icom

pact

St

ates

Non

-ant

icom

pact

St

ates

Diff

acr

oss

cond

itio

nsc

Inte

rcep

t2

334

550

(15

8223

)2

352

841

(16

3797

)2

390

95(1

864

73)

211

20

(4

403

29)

Nu

mbe

r of

age

nts

001

72(

0010

5)0

0065

9(

0011

5)0

0089

8(

0022

9)0

0022

8(

0015

5)x

2 5 1

85(d

f 5

2)

(Nu

mbe

r of

age

nts

)22

000

0005

58(

0000

0044

1)2

000

0001

47(

0000

0046

5)2

000

0001

83(

0000

0108

)2

000

0000

145

(00

0000

585)

Ass

et d

iver

sity

338

2(

1646

)3

632

(17

43)

130

9(

2005

)8

486

(44

24)

x2 5

28

5(d

f 5

2)

(Ass

et d

iver

sity

)22

001

19

(00

0571

)2

001

25

(00

0604

)2

000

44(

0007

1)2

002

91

(00

151)

Div

ersi

ty b

y co

mp

any

typ

e13

472

1(2

584

90)

116

335

(25

5520

)2

276

309

(30

7304

)11

94

(70

0626

)x

2 5 5

361

(df

5 2

)(D

iver

sity

by

com

pan

y ty

pe)

22

604

27(1

673

54)

24

5733

(16

6072

)22

078

9(2

052

39)

273

327

3(4

353

16)

An

tico

mp

act

law

mdash9

720

(44

50)

mdashmdash

N o

f st

ate-

year

s39

839

817

622

2N

of

even

ts33

3321

122

2 L

og L

ikel

ihoo

d20

718

420

216

819

453

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

3 h

as ti

me-

var

yin

g in

terc

epts

an

d t

ime-

var

yin

g ef

fect

s of

cov

aria

tes

all

owin

g th

e ef

fect

s of

cov

aria

tes

to c

han

ge a

cros

s in

stit

uti

onal

con

dit

ion

s C

onst

rain

ing

thes

e ef

fect

sas

a g

rou

p t

o be

equ

al a

cros

s co

nd

itio

ns

doe

s n

ot a

lter

th

e

t of

th

e m

odel

(x

2 5

76

37 d

f 5

7 p

2

5)

cSi

gn

ica

nce

tes

ts f

or a

dif

fere

nce

in c

oef

cien

ts a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

the

mod

el w

ith

th

e co

ef

cien

ts c

onst

rain

ed t

o be

eq

ual

com

par

ed t

o a

mod

el t

hat

all

ows

the

coef

ci

ents

to

vary

Si

gni

can

ce te

sts

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall














TABLE 2














DATA




Figure 2















METHODS








BL

E 3

Eff

ects

of

Car

tel-

Cap

ture

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 1

G

ener

al M

odel

All

Stat

es

Mod

el 2

Gen

eral

pl

us I

nsti

tuti

ons

All

Stat

es

Mod

el 3

C

ondi

tion

al M

odel

b

Ant

icom

pact

St

ates

Non

-ant

icom

pact

St

ates

Diff

acr

oss

cond

itio

nsc

Inte

rcep

t2

334

550

(15

8223

)2

352

841

(16

3797

)2

390

95(1

864

73)

211

20

(4

403

29)

Nu

mbe

r of

age

nts

001

72(

0010

5)0

0065

9(

0011

5)0

0089

8(

0022

9)0

0022

8(

0015

5)x

2 5 1

85(d

f 5

2)

(Nu

mbe

r of

age

nts

)22

000

0005

58(

0000

0044

1)2

000

0001

47(

0000

0046

5)2

000

0001

83(

0000

0108

)2

000

0000

145

(00

0000

585)

Ass

et d

iver

sity

338

2(

1646

)3

632

(17

43)

130

9(

2005

)8

486

(44

24)

x2 5

28

5(d

f 5

2)

(Ass

et d

iver

sity

)22

001

19

(00

0571

)2

001

25

(00

0604

)2

000

44(

0007

1)2

002

91

(00

151)

Div

ersi

ty b

y co

mp

any

typ

e13

472

1(2

584

90)

116

335

(25

5520

)2

276

309

(30

7304

)11

94

(70

0626

)x

2 5 5

361

(df

5 2

)(D

iver

sity

by

com

pan

y ty

pe)

22

604

27(1

673

54)

24

5733

(16

6072

)22

078

9(2

052

39)

273

327

3(4

353

16)

An

tico

mp

act

law

mdash9

720

(44

50)

mdashmdash

N o

f st

ate-

year

s39

839

817

622

2N

of

even

ts33

3321

122

2 L

og L

ikel

ihoo

d20

718

420

216

819

453

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

3 h

as ti

me-

var

yin

g in

terc

epts

an

d t

ime-

var

yin

g ef

fect

s of

cov

aria

tes

all

owin

g th

e ef

fect

s of

cov

aria

tes

to c

han

ge a

cros

s in

stit

uti

onal

con

dit

ion

s C

onst

rain

ing

thes

e ef

fect

sas

a g

rou

p t

o be

equ

al a

cros

s co

nd

itio

ns

doe

s n

ot a

lter

th

e

t of

th

e m

odel

(x

2 5

76

37 d

f 5

7 p

2

5)

cSi

gn

ica

nce

tes

ts f

or a

dif

fere

nce

in c

oef

cien

ts a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

the

mod

el w

ith

th

e co

ef

cien

ts c

onst

rain

ed t

o be

eq

ual

com

par

ed t

o a

mod

el t

hat

all

ows

the

coef

ci

ents

to

vary

Si

gni

can

ce te

sts

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall



DATA




Figure 2















METHODS








BL

E 3

Eff

ects

of

Car

tel-

Cap

ture

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 1

G

ener

al M

odel

All

Stat

es

Mod

el 2

Gen

eral

pl

us I

nsti

tuti

ons

All

Stat

es

Mod

el 3

C

ondi

tion

al M

odel

b

Ant

icom

pact

St

ates

Non

-ant

icom

pact

St

ates

Diff

acr

oss

cond

itio

nsc

Inte

rcep

t2

334

550

(15

8223

)2

352

841

(16

3797

)2

390

95(1

864

73)

211

20

(4

403

29)

Nu

mbe

r of

age

nts

001

72(

0010

5)0

0065

9(

0011

5)0

0089

8(

0022

9)0

0022

8(

0015

5)x

2 5 1

85(d

f 5

2)

(Nu

mbe

r of

age

nts

)22

000

0005

58(

0000

0044

1)2

000

0001

47(

0000

0046

5)2

000

0001

83(

0000

0108

)2

000

0000

145

(00

0000

585)

Ass

et d

iver

sity

338

2(

1646

)3

632

(17

43)

130

9(

2005

)8

486

(44

24)

x2 5

28

5(d

f 5

2)

(Ass

et d

iver

sity

)22

001

19

(00

0571

)2

001

25

(00

0604

)2

000

44(

0007

1)2

002

91

(00

151)

Div

ersi

ty b

y co

mp

any

typ

e13

472

1(2

584

90)

116

335

(25

5520

)2

276

309

(30

7304

)11

94

(70

0626

)x

2 5 5

361

(df

5 2

)(D

iver

sity

by

com

pan

y ty

pe)

22

604

27(1

673

54)

24

5733

(16

6072

)22

078

9(2

052

39)

273

327

3(4

353

16)

An

tico

mp

act

law

mdash9

720

(44

50)

mdashmdash

N o

f st

ate-

year

s39

839

817

622

2N

of

even

ts33

3321

122

2 L

og L

ikel

ihoo

d20

718

420

216

819

453

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

3 h

as ti

me-

var

yin

g in

terc

epts

an

d t

ime-

var

yin

g ef

fect

s of

cov

aria

tes

all

owin

g th

e ef

fect

s of

cov

aria

tes

to c

han

ge a

cros

s in

stit

uti

onal

con

dit

ion

s C

onst

rain

ing

thes

e ef

fect

sas

a g

rou

p t

o be

equ

al a

cros

s co

nd

itio

ns

doe

s n

ot a

lter

th

e

t of

th

e m

odel

(x

2 5

76

37 d

f 5

7 p

2

5)

cSi

gn

ica

nce

tes

ts f

or a

dif

fere

nce

in c

oef

cien

ts a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

the

mod

el w

ith

th

e co

ef

cien

ts c

onst

rain

ed t

o be

eq

ual

com

par

ed t

o a

mod

el t

hat

all

ows

the

coef

ci

ents

to

vary

Si

gni

can

ce te

sts

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall














METHODS








BL

E 3

Eff

ects

of

Car

tel-

Cap

ture

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 1

G

ener

al M

odel

All

Stat

es

Mod

el 2

Gen

eral

pl

us I

nsti

tuti

ons

All

Stat

es

Mod

el 3

C

ondi

tion

al M

odel

b

Ant

icom

pact

St

ates

Non

-ant

icom

pact

St

ates

Diff

acr

oss

cond

itio

nsc

Inte

rcep

t2

334

550

(15

8223

)2

352

841

(16

3797

)2

390

95(1

864

73)

211

20

(4

403

29)

Nu

mbe

r of

age

nts

001

72(

0010

5)0

0065

9(

0011

5)0

0089

8(

0022

9)0

0022

8(

0015

5)x

2 5 1

85(d

f 5

2)

(Nu

mbe

r of

age

nts

)22

000

0005

58(

0000

0044

1)2

000

0001

47(

0000

0046

5)2

000

0001

83(

0000

0108

)2

000

0000

145

(00

0000

585)

Ass

et d

iver

sity

338

2(

1646

)3

632

(17

43)

130

9(

2005

)8

486

(44

24)

x2 5

28

5(d

f 5

2)

(Ass

et d

iver

sity

)22

001

19

(00

0571

)2

001

25

(00

0604

)2

000

44(

0007

1)2

002

91

(00

151)

Div

ersi

ty b

y co

mp

any

typ

e13

472

1(2

584

90)

116

335

(25

5520

)2

276

309

(30

7304

)11

94

(70

0626

)x

2 5 5

361

(df

5 2

)(D

iver

sity

by

com

pan

y ty

pe)

22

604

27(1

673

54)

24

5733

(16

6072

)22

078

9(2

052

39)

273

327

3(4

353

16)

An

tico

mp

act

law

mdash9

720

(44

50)

mdashmdash

N o

f st

ate-

year

s39

839

817

622

2N

of

even

ts33

3321

122

2 L

og L

ikel

ihoo

d20

718

420

216

819

453

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

3 h

as ti

me-

var

yin

g in

terc

epts

an

d t

ime-

var

yin

g ef

fect

s of

cov

aria

tes

all

owin

g th

e ef

fect

s of

cov

aria

tes

to c

han

ge a

cros

s in

stit

uti

onal

con

dit

ion

s C

onst

rain

ing

thes

e ef

fect

sas

a g

rou

p t

o be

equ

al a

cros

s co

nd

itio

ns

doe

s n

ot a

lter

th

e

t of

th

e m

odel

(x

2 5

76

37 d

f 5

7 p

2

5)

cSi

gn

ica

nce

tes

ts f

or a

dif

fere

nce

in c

oef

cien

ts a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

the

mod

el w

ith

th

e co

ef

cien

ts c

onst

rain

ed t

o be

eq

ual

com

par

ed t

o a

mod

el t

hat

all

ows

the

coef

ci

ents

to

vary

Si

gni

can

ce te

sts

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall







BL

E 3

Eff

ects

of

Car

tel-

Cap

ture

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 1

G

ener

al M

odel

All

Stat

es

Mod

el 2

Gen

eral

pl

us I

nsti

tuti

ons

All

Stat

es

Mod

el 3

C

ondi

tion

al M

odel

b

Ant

icom

pact

St

ates

Non

-ant

icom

pact

St

ates

Diff

acr

oss

cond

itio

nsc

Inte

rcep

t2

334

550

(15

8223

)2

352

841

(16

3797

)2

390

95(1

864

73)

211

20

(4

403

29)

Nu

mbe

r of

age

nts

001

72(

0010

5)0

0065

9(

0011

5)0

0089

8(

0022

9)0

0022

8(

0015

5)x

2 5 1

85(d

f 5

2)

(Nu

mbe

r of

age

nts

)22

000

0005

58(

0000

0044

1)2

000

0001

47(

0000

0046

5)2

000

0001

83(

0000

0108

)2

000

0000

145

(00

0000

585)

Ass

et d

iver

sity

338

2(

1646

)3

632

(17

43)

130

9(

2005

)8

486

(44

24)

x2 5

28

5(d

f 5

2)

(Ass

et d

iver

sity

)22

001

19

(00

0571

)2

001

25

(00

0604

)2

000

44(

0007

1)2

002

91

(00

151)

Div

ersi

ty b

y co

mp

any

typ

e13

472

1(2

584

90)

116

335

(25

5520

)2

276

309

(30

7304

)11

94

(70

0626

)x

2 5 5

361

(df

5 2

)(D

iver

sity

by

com

pan

y ty

pe)

22

604

27(1

673

54)

24

5733

(16

6072

)22

078

9(2

052

39)

273

327

3(4

353

16)

An

tico

mp

act

law

mdash9

720

(44

50)

mdashmdash

N o

f st

ate-

year

s39

839

817

622

2N

of

even

ts33

3321

122

2 L

og L

ikel

ihoo

d20

718

420

216

819

453

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

3 h

as ti

me-

var

yin

g in

terc

epts

an

d t

ime-

var

yin

g ef

fect

s of

cov

aria

tes

all

owin

g th

e ef

fect

s of

cov

aria

tes

to c

han

ge a

cros

s in

stit

uti

onal

con

dit

ion

s C

onst

rain

ing

thes

e ef

fect

sas

a g

rou

p t

o be

equ

al a

cros

s co

nd

itio

ns

doe

s n

ot a

lter

th

e

t of

th

e m

odel

(x

2 5

76

37 d

f 5

7 p

2

5)

cSi

gn

ica

nce

tes

ts f

or a

dif

fere

nce

in c

oef

cien

ts a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

the

mod

el w

ith

th

e co

ef

cien

ts c

onst

rain

ed t

o be

eq

ual

com

par

ed t

o a

mod

el t

hat

all

ows

the

coef

ci

ents

to

vary

Si

gni

can

ce te

sts

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall









TA

BL

E 4

Eff

ects

of

Inte

rest

Gro

up

Var

iabl

es o

n th

e P

assa

ge o

f Fi

re I

nsu

ran

ce R

ate

Reg

ula

tion

190

6ndash19

30a

Mod

el 4

G

ener

al M

odel

All

Year

s

Mod

el 5

G

ener

al p

lus

Inst

itut

ions

All

Year

s

Mod

el 6

Con

diti

onal

Mod

elb

Diff

acr

oss

cond

itio

nsc

1906

to 1

915

1916

to

1930

Inte

rcep

t2

371

14

(

9999

)2

468

36

(1

044

9)3

2378

(26

887)

28

0923

(18

034)

Fire

insu

ran

ce a

sset

s p

er c

apit

a2

000

02(

0000

72)

20

0006

(00

0082

)2

001

31

(00

0584

)0

0009

5(

0000

75)

x2 5

14

06

(d

f 5

1)

Fire

insu

ran

ce a

gen

ts p

er c

apit

a2

872

9(1

588

8)2

1032

3(1

615

7)2

3302

3(2

562

9)26

773

(278

11)

x2 5

25

6(d

f 5

1)

Farm

pro

per

ty v

alu

e p

er c

apit

a1

458

(14

14)

165

0(

1396

)5

063

(26

10)

21

038

(20

45)

x2 5

33

9(d

f 5

1)

Farm

s p

er c

apit

a8

3122

(69

896)

113

689

(69

373)

229

282

2(1

571

60)

400

316

(1

238

72)

x2 5

13

90

(d

f 5

1)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a32

791

(212

3)

494

6(2

164

)2

818

0(5

974

)83

73

(362

7)

x2 5

65

2(d

f 5

1)

Per

iod

(19

16ndash3

0)mdash

104

67

(40

47)

mdashmdash

N o

f st

ate-

year

s39

839

824

415

4N

of

even

ts33

3316

172

2 L

og L

ikel

ihoo

d22

178

921

509

018

798

1

Not

es

aSt

and

ard

err

ors

in p

aren

thes

es

bM

odel

6 h

as t

ime-

vary

ing

inte

rcep

ts a

nd

tim

e-va

ryin

g ef

fect

s of

cov

aria

tes

all

owin

g t

he

effe

cts

of c

ovar

iate

s to

ch

ange

acr

oss

inst

itu

tion

al c

ond

itio

ns

As

a gr

oup

th

ese

effe

cts

can

not

be

con

stra

ined

to

be e

qual

acr

oss

per

iod

s w

ith

out

sig

ni

can

tly

deg

rad

ing

the

t

of t

he

mod

el ( x

2 5

27

109

df

5 6

p

001

)c

Sign

ica

nce

tes

ts f

or a

dif

fere

nce

in a

coe

fci

ent a

cros

s in

stit

uti

onal

con

dit

ion

s ar

e co

mp

ute

d t

hro

ugh

a x

2 te

st o

f th

e

t of

th

e m

odel

wit

h t

he

coef

ci

ent

con

stra

ined

to

be e

qual

com

par

ed to

a m

odel

that

all

ows

the

coef

ci

ent

to v

ary

Sig

ni

can

ce t

ests

(on

e-ta

iled

) 1

p

10

p

0

5

p

01

p

001









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall









CONCLUSION














AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall










AP

PE

ND

IXA

Cor

rela

tion

s an

d U

niv

aria

te S

tati

stic

s fo

r In

dep

end

ent

Var

iabl

es

Var

iabl

e(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)(9

)(1

0)(1

1)

(1)

Nu

mbe

r of

age

nts

10

mdashmdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(2)

Ass

et d

iver

sity

[C

V]

420

421

0mdash

mdashmdash

mdashmdash

mdashmdash

mdashmdash

(3)

Div

ersi

ty b

y co

mp

any

typ

e [I

QV

]0

1926

359

471

0mdash

mdashmdash

mdashmdash

mdashmdash

mdash(4

) A

nti

com

pac

t law

059

872

185

582

247

011

0mdash

mdashmdash

mdashmdash

mdashmdash

(5)

Fire

insu

ran

ce a

gen

ts p

er

cap

ita

251

732

042

801

8156

20

7115

10

mdashmdash

mdashmdash

mdashmdash

(6)

Fire

insu

ran

ce a

sset

s p

er

cap

ita

24

6596

21

1412

089

612

346

521

7956

10

mdashmdash

mdashmdash

mdash(7

) Fa

rm p

rop

erty

val

ue

per

ca

pit

a0

8703

22

4608

20

6034

121

685

7451

097

891

0mdash

mdashmdash

mdash(8

) Fa

rms

per

cap

ita

22

2315

24

4369

22

7162

494

452

186

932

213

544

0674

10

mdashmdash

mdash(9

) M

anu

fact

uri

ng

val

ue

add

ed p

er c

apit

a2

6593

535

105

2163

23

6200

148

560

0471

24

2810

28

0627

10

mdashmdash

(10)

Man

ufa

ctu

rin

g es

tabl

ish

men

ts p

er c

apit

a2

9459

144

133

6955

22

9362

265

222

111

572

201

552

514

306

3679

10

mdash(1

1) T

ime

per

iod

[19

16ndash3

0]2

092

784

9498

281

492

156

852

052

132

1819

20

9809

21

4178

153

582

255

641

0

Mea

n63

696

9613

699

907

447

442

20

0039

2710

185

080

695

314

50

024

386

9St

and

ard

dev

iati

on64

307

5022

993

21

043

497

30

0018

3222

185

470

376

237

80

011

487

7M

inim

um

331

429

766

636

513

40

000

1018

338

060

914

004

80

301

000

50

Max

imu

m30

5520

983

929

652

10

0088

2440

611

387

71

527

108

050

054

1



NOTES


























REFERENCES











Frederick Stokes















































































Hall



NOTES


























REFERENCES











Frederick Stokes















































































Hall


















REFERENCES











Frederick Stokes















































































Hall















































































Hall

RATIONALITY AND INSTITUTIONAL CONTINGENCY: · PDF fileFIRE INSURANCE INDUSTRY TIM BARTLEY*...

Documents

Transcript of RATIONALITY AND INSTITUTIONAL CONTINGENCY: · PDF fileFIRE INSURANCE INDUSTRY TIM BARTLEY*...