ON TIME BUSINESS MANAGEMENT, A+ WORK, ON TIME, NO PLAGARIZING; ON TIME

wk6agnnEBSCO-FullText-10_11_2026.pdf

Home >Business & Finance homework help >Management homework help >ON TIME BUSINESS MANAGEMENT, A+ WORK, ON TIME, NO PLAGARIZING; ON TIME

WHAT DRIVES TORT REFORM LEGISLATION? AN ANALYSIS OF STATE DECISIONS TO RESTRICT LIABILITY TORTS

Yiling Deng George Zanjani

ABSTRACT

This article studies the timing of state-level tort reform enactments between 1971 and 2005. Using discrete-time hazard models, we find the level of litigation activity—as measured by incurred liability insurance losses, the number of lawyers, and tort cases commenced—to be the most important and robust determinant of tort reformadoption. Political-institutional factors and regional effects—such as Republican control of the state government, single-party control of the legislature and governorship, and a (relatively) conservative political ideology among a state’s Democrats—are also associated with quicker reform adoption.

INTRODUCTION

Over the past several decades, theUnited States has experienced several waves of tort reform. The first wave came in the mid-1970s, when a number of states enacted reforms such as caps on noneconomic damages, changes to collateral source rules, and limitations on the application of joint and several liability. This was followed by a larger wave in the mid-1980s and another wave in the early 2000s.

Much research has been devoted to estimating the impact of tort reforms, with many studies finding that tort reforms have large negative effects on various measures of

Yiling Deng is at the Department of Accounting, Finance, and Economics, Donald R. Tapia School of Business, Saint Leo University, Saint Leo, FL 33547, 325-588-8190. Deng can be contacted via e-mail: [email protected]. George Zanjani is at the Department of Risk Management and Insurance, J. Mack Robinson College of Business, Georgia State University, Atlanta, GA 30303, 404-413-7464. Zanjani can be contacted via e-mail: [email protected]. The authors thank the editor, an anonymous referee, Charlotte Alexander, Daniel Bauer, Richard Butler, Martin Grace, Tyler Leverty, Nan Zhu, and seminar participants at Georgia State University, 2014 American Risk and Insurance Association (ARIA) conference, and the 2015 ASSA conference for helpful comments. Travel support from the Center for the Economic Analysis of Risk (CEAR) is gratefully acknowledged.

959

litigation activity. However, relatively little attention1 has been paid to the question of why states enact tort reforms, and the propensity to reform varies significantly across the states. For example, as can be seen in Table 1, Florida has been extremely aggressive in instituting reforms, while, nearby, South Carolina has stood pat; similarly, one moves from an aggressive to a passive reform environment when one crosses the border from Idaho to Wyoming. Why did some states enact tort reforms while others stood pat? This articlemaps out the diffusion of tort reforms across states over time and studies the social, economic, and political factors associated with reform decisions. Specifically, we focus on adoptions of four of themost common and prominent liability tort reforms—caps on punitive damages, limitations on joint and several liability laws, caps on noneconomic damages, and changes to collateral source rules—during the 1971–2005 period.

Several theoriesmayhelp explain the political process of legal change. Building on the segmentation offered by Kroszner and Strahan (1999), private interest or economic theory (Stigler, 1971; Becker, 1983; Peltzman, 1989) describes the legislative process as one of competition among interest groups, where well-organized industrial and professional interests capture rents at the expense of more dispersed groups such as consumers. In the case of tort reform, lawyers, physicians, the insurance industry, and businesses are all interest groups to be considered. The public interest theory (Joskow and Noll, 1981) features a benevolent legislature with a primary concern of social welfare. In this view, lawmakers identify failures in the current civil justice system and attempt to correct them. A third group of theories emphasize the impact of political-institutional factors in the legislative process such as Republican versus Democratic control (Dixit, 1996; Irwin and Kroszner, 1999), or the nature of state government ideology (Poole and Rosenthal, 1997; Berry et al., 1998). A fourth group emphasizes the impact of legislative outcomes in nearby polities, seeking to explain the diffusion of law across borders as being facilitated by, for example, the availability of ideas or templates for reform.2

It is important to recognize that these theoretical approaches are by no means diametrically opposed, despite some attempts in the literature to cast them as such. To take but one example, one could imagine interest group lobbying reinforcing a push for reform in the “public interest” that was inspired by legislation in neighboring states and consistent with the ideology of the state legislators. This article recognizes that the factors identified by these theories may work in concert to influence tort reform enactments. To investigate what drives reform enactment, we use a discrete- time proportional hazard model to explain the timing of tort reform adoption. We incorporate proxies for the size of various interest groups, the extent of and costs associated with litigation, political-institutional factors, the liability climate, state insurance regulation, economic conditions, the presence of reforms in nearby states, and regional effects.

1An important exception is Harrington (1994), who studies the adoption of automobile insurance no-fault laws in the early 1970s.

2This diffusion literature is notably diffuse. See Twining (2005) for a review.

960 THE JOURNAL OF RISK AND INSURANCE

TA B LE

1 In iti al

St at e G en

er al

Lia bi lit y To

rt Re

fo rm

Le gi sla

tio n,

by D ec ad

R ef o rm

S ta te s P as si n g

R ef o rm

in 19 70 s

S ta te s P as si n g

R ef o rm

in 19 80 s

S ta te s P as si n g

R ef o rm

in 19 90 s

S ta te s P as si n g

R ef o rm

in 20 00 s

S ta te s P as si n g

N o T o rt

R ef o rm

C ap

s o n n o n ec o n o m ic

d am

ag es

— A L , A K , C O , F L ,

H I, ID

, K S , M D ,

M N , O R , W A

IL M S

A Z , A R , C A , C T , D E , G A , IN

, IA

, K Y , L A , M E , M A , M I, M O ,

M T , N E , N V , N H , N J, N M , N Y ,

N C , N D , O H , O K , P A , R I, S C ,

S D , T N , T X , U T , V T , V A , W V ,

W I, W Y

C ap

s o n p u n it iv e

d am

ag es

— A L , C O , F L , G A ,

K S , M T , N V , N H ,

T X , V A

A K , IL , IN

, L A , N J,

N C , N D , O H , O K ,

ID , M S

A Z , A R , C A , C T , D E , H I, IA

, K Y , M E , M A , M N , M O , N M ,

N Y , O R , P A , R I, S C , S D , T N ,

U T , V T , W V , W I, W Y

C o ll at er al

so u rc e ru le

ch an

g e

F L , N D

A L , A K , C O , C T ,

G A , H I, IL , IN

, IA

, K Y , M I, M N ,

M T , O H , O R

ID –

A Z , A R , C A , D E , K S , L A , M E ,

M A , M S , M O , N E , N V , N H ,

N M , N Y , N C , O K , P A , R I, S C ,

S D , T N , T X , U T , V T , V A , W V ,

W A , W I, W Y

Jo in t an

d se v er al

li ab

il it y

F L , K S , N V , O K ,

O R

A L , A K , A Z , C A ,

C O , C T , G A , H I,

ID , IL , IN

, IA

, K Y , L A , M I, M N ,

M S , M O , M T , N J,

N M ,N

Y ,N

D ,T

X ,

U T , W A , W Y

N E , N H , O H , T X , W I,

A R , P A

D E , M E , M D , M A , N C , R I, S C

N ot e: M I, N E , an

d W A

en ac te d ca p s o n p u n it iv e d am

ag es

b ef o re

19 70 s; S D

an d V T en

ac te d jo in t an

d se v er al

li ab

il it y b ef o re

19 70 s.

WHAT DRIVES TORT REFORM LEGISLATION? 961

Our most robust finding is that various measures of litigation activity are associated with quicker enactment of tort reform. More lawyers, more tort cases in general jurisdiction courts, and higher liability insurance costs are all associated with faster tort reform adoption. Other variables are not consistently associated with tort reforms, with the exceptions of (1) Republican control of the state government and single-party control of the legislature and governorship (both of which are positively associated with tort reform enactment in many specifications) and (2) regional effects (with states in the Northeast being significantly less likely to adopt reforms).

Evidence on other influences is mixed. Reform adoption in neighboring states is associated with quicker adoption in some specifications. We also find some evidence of the influence of concentrated interest groups, that is, insurance industry professionals, on quicker tort reform adoption in some specifications, though it should be noted that seemingly contrary effects are found for lawyers (more lawyers are associated with quicker adoption in most specifications) and physicians (more physicians are associated with slower adoption in some specifications).

The article is organized as follows. The “Tort Crises and Tort Reforms” section provides background on insurance crises and discusses related literature on tort reforms. The “Hypotheses and Variable Definitions” section describes the hypothe- ses, proxy variables, and data sources. The “Methods and Empirical Results” section explains the empiricalmethods and results. The “Robustness Checks and Extensions” section describes various robustness checks and extensions, the details of which are available in the Online Appendix (Deng and Zanjani, 2016). The “Discussion and Conclusion” section concludes.

TORT CRISES AND TORT REFORMS

Three Tort Crises in Liability Since the 1970s, the United States has experienced three “liability crises.” The crises were characterized by sharp rises in insurance premiums, accounting recognition of liability losses by insurance companies, and restrictions in coverage. The first crisis happened in the mid-1970s, and several states enacted tort reforms during that period. The first liability crisis was especially acute in the area of medical malpractice liability, and it led several states to enact tort reforms targeted only at medical malpractice. The second liability crisis occurred in the mid-1980s, and many states passed reforms around this time. Priest (1987) attributes this crisis to the interpretation of modern tort law. He argues that judicial findings of greater levels of liability in insurance contracts, combined with a decline in the interest rate, led to insurers increase prices, and restrict coverage. The third andmost recent crisis started in the late 1990s and continued into the early 2000s.

Four Tort Reforms There are four prominent reforms: caps on punitive damages, limitations on joint and several liability, caps on noneconomic damages, and reforms to collateral source rules. These reforms have been most widely analyzed by other researchers and also

962 THE JOURNAL OF RISK AND INSURANCE

directly impact the determination of awards.3 Table 1 shows the states that passed each of these four tort reforms by decade. Evidently, the largest tort reformwave took place in the 1980s. Table A1 and Figures A1–A7 in the Online Appendix (Deng and Zanjani, 2016) illustrate the history of the four tort reform enactments by state since 1971 in detail.

Prior to1971,most statesdidnot have capsonpunitivedamages and limitationson joint and several liability, and no states had caps on noneconomic damages and collateral source reforms related to general liability.4 Starting in the mid-1970s, however, many states passed various reforms to the common law rules underwhich tort cases are tried in state courts. Most states enacted at least one of the four tort reforms during the 1970s and 1980s. The first wave of tort reform during the 1970s mainly targeted lawsuits related to health care, whereas the second wave of tort reform enactments during the 1980s was more general (Sloan, Mergenhagen, and Bovbjerg, 1989).

Caps on Punitive Damages Reform. Punitive damages are awarded to punish tortfeasors for malicious and reckless behavior and to deter future misconduct (American Tort Reform Association, 2012). Punitive damages awards are infrequent but the awards can be enormous and are often sought in civil lawsuits. Views differ on the predictability and opacity of the awards (see, e.g., the discussion in Viscusi, 2004; Eisenberg et al., 2010). Caps on punitive damages typically limit the amount of award either to a specific dollar amount (e.g., $250,000 in Alabama) or to a multiple of compensatory damages. Some even prohibit punitive damages entirely.

Joint and Several Liability Reform. Joint and several liability permits the plaintiff to recover damages from multiple defendants or from each defendant individually. If one defendant does not have enough resources to pay the tort award, the plaintiff can seek redress from other defendants. Reform of joint and several liability modifies the joint responsibility that two or more defendants carry, typically by limiting a defendants’ financial responsibility for harm to a percent of total damages according to fault. The most common form of the reform is a limit to the application of joint and several liability in awarding noneconomic damages (Lee, Browne, and Schmit, 1994).

Caps on Noneconomic Damages Reform. Damages for noneconomic losses are for pain and suffering and are inherently difficult to measure (Sunstein, 2008). The discretion of juries may result in substantial variation in awards. Caps on

3Two additional reforms that are popular in the literature are contingency fee reform and periodic payments reform. We include neither in our analysis since only five states (AR, IL, ME, NH, andWY) enact contingency fee reform on general liability and only three states enact periodic payments reform on general liability (OH, RI, and SD) during the sample period. The results do not change materially if we include these reforms along with the other four reforms when analyzing the propensity of states to pass any type of tort reform.

4Whenwriting “tort reformon general liability” or “general liability tort reform,”we imply that the application of the tort reformwas not restricted tomedicalmalpractice. Please see TableA1 in the Online Appendix (Deng and Zanjani, 2016) for states that enacted tort reform on general liability before 1971.

WHAT DRIVES TORT REFORM LEGISLATION? 963

noneconomic loss typically provide numerical guidelines for the award or provide specific dollar ceilings on awards for noneconomic damages.

Collateral Source Rule Reform. The collateral source rule forbids the use of evidence at trial that the plaintiff is being compensated from alternative sources such as insurance contracts (American Tort Reform Association, 2012). Collateral source rule reform typically requires that court awards be adjusted for compensation from other sources. Thus, total damages awarded at trial are offset by the amount paid to the plaintiff through other sources such as health insurance, auto insurance, andworkers compensation insurance.

Recent Studies of Tort Reform The literature on tort reforms is vast and goes back at least 30 years. Previous studies have generally focused on three issues: whether tort limitations have affected the frequency and severity of claims (e.g., Browne and Puelz, 1999); whether tort reforms have affected insurance market quantities, such as premiums, losses, or loss ratios (e.g., Viscusi et al., 1993); and whether tort reforms have had a direct influence on health market outcomes, including physician supply, the practice of defensive medicine, hospital expenditures, and health insurance market indicators (e.g., Avraham and Schanzenbach, 2010; Avraham, Dafny, and Schanzenbach, 2012).

Analyses of the first issue have provided strong evidence that limitations on tort liability reduce the frequency of claims and the size of claims (Browne and Puelz, 1999; Browne and Schmit, 2008; Paik, Black, and Hyman 2013). Analyses of the effect of tort reforms on insurance market quantities typically indicate that tort reforms reduce insurance losses (Born and Viscusi, 1994, 1998; Born, Viscusi, and Baker, 2009) and insurance premiums (Zuckerman, Bovbjerg, and Sloan, 1990; Born and Viscusi, 1994; Heaton, 2015), although the results on premiums are less strong. When examining the 1980s tort reforms, some researchers have found significant negative effects on general liability insurance costs butmixed evidence onmedical malpractice insurance costs (e.g., Viscusi et al., 1993). Analyses of the last issue provide evidence that medical malpractice tort reforms have hadmodest effects on defensive medicine and physician supply (Kessler and McClellan, 1996; Kessler, Sage, and Becker, 2005; Matsa, 2007), and that reforms lower the cost of health insurance to a certain extent (Avraham and Schanzenbach, 2010; Avraham, Dafny, and Schanzenbach, 2012; Karl, Born, and Viscusi, 2016).5

With respect to the data structure in tort reform studies, there have been a number of articles examining the effect of liability reforms using either state-level data (Blackmon andZeckhauser, 1991; Viscusi et al., 1993) or firm-level data from theNAIC database (Born and Viscusi, 1994, 1998; Viscusi and Born, 1995, 2005). Since liability insurance claims may develop over long periods, Born, Viscusi, and Baker (2009) examine the long-run effects of tort reforms using the developed losses of insurers.

5For example, Avraham, Dafny, and Schanzenbach (2012) find that the enactments of various tort reforms decrease group self-insured health insurance premiums by 1–2 percent.

964 THE JOURNAL OF RISK AND INSURANCE

Although the content of reforms varies greatly across states (e.g., the stringency of a cap is determined by the level and type of the cap), almost all of these articles quantify tort reform by using binary variables equal to 1 for all the years in which reforms are effective, and 0 otherwise (excepting Hyman et al., 2009, who study the impact of various caps). In this article, we adopt this traditional method of using a dichotomous variable to indicate tort reform enactments.

HYPOTHESES AND VARIABLE DEFINITIONS

Our empirical analysis considers the timing of four major tort reforms on general torts from 1971 to 2005: caps on punitive damages, limitations on joint and several liability, capsonnoneconomicdamages, andcollateral source rule reforms.Theeffectiveyearsof tort reforms are obtained by state for the years 1971–2005 from Database of State Tort Law Reforms (Avraham, 2014) and complemented by the American Tort Reform Association (www.atra.org). Among those reforms, caps on punitive damages and limitations on joint and several liabilityusually apply to all torts rather than justmedical malpractice, as shown inFiguresA1andA2 in theOnlineAppendix (DengandZanjani, 2016), whereas caps on noneconomic damages and collateral source rule reforms are often targeted exclusively at medical malpractice torts, as shown in Figures A3–A6.

In the remainder of this section, we describe the variables and how they connect to hypotheses suggested by theory concerning the influence of lobbying, public interest, political-institutional factors, state insurance regulation, and regional effects. All the variables and corresponding sources discussed below are described in detail in Table 2. Table 3 reports means and standard deviations for the explanatory variables used in the analysis and the average waiting time for tort reform adoption. In total, there are 1,750 observations (50 states with 35 years) in the whole sample.

Interest Groups Lawyers, doctors, and insurance companies are all concentrated stakeholders with interests in tort reform. Doctors and insurance companies are typically cast in favor of tort reform, and lawyers are typically cast in opposition.6 Lawyers and injured parties, on the other hand, stand to gain from opposing tort reform. Private interest theory predicts that the legislative outcome will thus be influenced by the relative power of these groups.

We use the number of professionals in each of these groups as an indicator of the power of each group in each state. Specifically, we use per capita employment in the insurance sector (Employment in insurer) as a proxy for the power of the insurance

6These characterizations are obviously debatable generalizations based on broad character- izations of self-interest, as well as the presumption that tort reform will restrict or reduce the effectiveness of litigation. The nature of economic self-interest is evident in the case of doctors, as tort reform typically will involve some reduction in their liability exposures (although one could argue that the ability to sue will increase the value of the doctor’s service to the patient). The logic is less clear in the case of insurers; while they pay claims and legal defense costs emanating from tort lawsuits, their future premium revenue volume is connected to the ongoing threat of lawsuits. Lawyers’ business volume obviously depends on lawsuits.

WHAT DRIVES TORT REFORM LEGISLATION? 965

http://www.atra.org

TA B LE

2 Va

ria bl e D efi

ni tio

ns an

d So

ur ce s

D ef in it io n

S o u rc es

T o rt

re fo rm

s C ap s on

pu ni ti ve

da m ag es

A n in d ic at o r v ar ia b le

eq u al

to 1 if a st at e se ts a ca p o n

th e re co v er y o f p u n it iv e d am

ag es

A m er ic an

T o rt

R ef o rm

A ss o ci at io n

(w w w .a tr a. o rg ) an

d A v ra h am

(2 01 4)

Jo in t an d se ve ra l lia bi lit y

A n in d ic at o r v ar ia b le

eq u al

to 1 if a st at e li m it s a

p ar ty ’s

re sp

o n si b il it y fo r d am

ag es

to a p er ce n t o f

to ta l d am

ag es

co rr es p o n d in g to

th at

p ar ty ’s d eg

re e

o f fa u lt

N on ec on om

ic da m ag e ca ps

A n in d ic at o r v ar ia b le

eq u al

to 1 if a st at e h as

en ac te d

a ca p o n th e si ze

o f co m p en

sa ti o n fo r in ju re d

p er so n s d u e to

in ta n g ib le

b u t re al

in ju ri es

su ch

as p ai n an

d su

ff er in g

C ol la te ra l so ur ce

ru le

A n in d ic at o r v ar ia b le

eq u al

to 1 if a st at e p er m it s

in su

ra n ce

re co v er y fr o m

a v ic ti m ’s

fi rs t p ar ty

in su

re r to

o ff se t th e d am

ag e ju d g m en

t E co n o m ic s v ar ia b le s

Li ab ili ty

in su ra nc e lo ss

A g g re g at ed

d ir ec t lo ss

in cu

rr ed

o f li ab

il it y in su

ra n ce

(i n cl u d in g o th er

li ab

il it y , au

to li ab

il it y , co m m er ci al

m u lt ip le -p er il li ab

il it y , an

d m ed

ic al

m al p ra ct ic e)

at st at e le v el

d iv id ed

b y G S P an

d m u lt ip li ed

b y 1, 00 0

A .M

. B es t C u st o m

R ep

o rt

H ea lt h ca re %

P er so n al

h ea lt h ca re

ex p en

d it u re

o n p h y si ci an

an d

cl in ic

se rv ic es

d iv id ed

b y p er so n al

an n u al

in co m e

in a st at e m u lt ip li ed

b y 10 0

C M S O ff ic e o f th e A ct u ar y , 19 71 –

20 05

C re di t sp re ad

T h e d if fe re n ce

b et w ee n th e A aa

co rp o ra te

b o n d y ie ld

an d B aa

co rp o ra te

b o n d y ie ld

fo r ea ch

y ea r

F ed

er al

R es er v e B an

k o f S t. L o u is ’s

F R E D

E co n o m ic

D at a

G SP

pe r ca pi ta

G ro ss

d o m es ti c p ro d u ct

(G S P ) d iv id ed

b y th e to ta l

p o p u la ti o n sc al ed

b y 1, 00 0 in

a st at e

B u re au

o f E co n o m ic

A n al y si s,

19 71 –2 00 5

(C on ti nu

ed )

966 THE JOURNAL OF RISK AND INSURANCE

http://www.atra.org

TA B LE

2 C on

tin ue

D ef in it io n

S o u rc es

P o li ti cs

an d re g u la ti o n v ar ia b le s

R at io

of R ep ub

lic an

T h e av

er ag

e ra ti o o f th e lo w er

h o u se

an d th e u p p er

h o u se

th at

ar e R ep

u b li ca n

S ta ti st ic al

A b st ra ct

o f th e U n it ed

S ta te s,

19 71 –2 00 5

R ep ub

lic an

co nt ro la

T h e fr ac ti o n o f st at e g o v er n m en

t (l o w er

h o u se ,u

p p er

h o u se , an

d g o v er n o rs h ip ) co n tr o ll ed

b y

R ep

u b li ca n s

Sa m e pa rt y

A n in d ic at o r v ar ia b le

eq u al

to 1 if th e sa m e p ar ty

co n tr o ls

th e g o v er n o r’ s o ff ic e an

d h as

m aj o ri ti es

in th e lo w er

h o u se

an d th e u p p er

h o u se

R ep ub

lic an

id eo lo gy

A m ea su

re o f th e st at e R ep

u b li ca n p ar ty

id eo

lo g y , in

w h ic h 0 is

th e m o st

co n se rv at iv e an

d 10 0 is

th e

m o st

li b er al

B er ry

et al . (1 99 8)

D em

oc ra ti c id eo lo gy

A m ea su

re o f th e st at e D em

o cr at ic

p ar ty

id eo

lo g y , in

w h ic h 0 is

th e m o st

co n se rv at iv e an

d 10 0 is

th e

m o st

li b er al

G ov er no r id eo lo gy

A m ea su

re o f th e st at e g o v er n o r’ s id eo

lo g y , in

w h ic h

0 is

th e m o st

co n se rv at iv e an

d 10 0 is

th e m o st

li b er al

T or t ca se s in

st at e co ur t

T o rt

ca se s co m m en

ce d in

st at e g en

er al

ju ri sd

ic ti o n

co u rt s d iv id ed

b y p o p u la ti o n an

d m u lt ip li ed

b y

1, 00 0 fo r ea ch

y ea r

N at io n al

C en

te r fo r S ta te

C o u rt s,

19 75 –2 00 5

T or t ca se s in

fe de ra l co ur t

T o rt

ca se s co m m en

ce d in

fe d er al

co u rt s d iv id ed

b y

n at io n al

p o p u la ti o n an

d m u lt ip li ed

b y 10 ,0 00

fo r

ea ch

y ea r

S ta ti st ic al

A b st ra ct

o f th e U n it ed

S ta te s,

19 71 –2 00 5

R at e re gu

la ti on

A n in d ic at o r v ar ia b le

eq u al

to 1 if th e st at e ap

p li es

a fo rm

o f p ri o r ap

p ro v al

ra te

re g u la ti o n , 0 o th er w is e

(C on ti nu

ed )

WHAT DRIVES TORT REFORM LEGISLATION? 967

TA B LE

2 C on

tin ue

D ef in it io n

S o u rc es

P un

it iv e da m ag es

in su ra bl e

A n in d ic at o r v ar ia b le

eq u al

to 1 if st at e al lo w s

p u n it iv e d am

ag es

to b e in su

re d

A m er ic an

In su

ra n ce

A ss o ci at io n ,

S u m m ar y o f S ta te

L aw

s an

d R eg

u la ti o n s re la ti n g to

In su

ra n ce , 19 71 –2 00 5

N o- fa ul t ve rb al

A n in d ic at o r v ar ia b le

eq u al

to 1 if st at e h as

ad o p te d

n o -f au

lt in su

ra n ce

w it h v er b al

th re sh

o ld , an

d 0

o th er w is e

N o- fa ul t hi gh

va lu e

A n in d ic at o r v ar ia b le

eq u al

to 1 if st at e h as

ad o p te d

n o -f au

lt in su

ra n ce

w it h th re sh

o ld

la rg er

th an

$1 ,0 00 , an

d 0 o th er w is e

N o- fa ul t ot he r

A n in d ic at o r v ar ia b le

eq u al

to 1 if st at e h as

ad o p te d

n o -f au

lt in su

ra n ce

w it h th re sh

o ld

le ss

th an

$1 ,0 00

o r h as

a to rt

sy st em

, an

d 0 o th er w is e

O th er

v ar ia b le s

E m pl oy m en t in

in su re r

R at io

o f th e n u m b er

o f em

p lo y ee s in

in su

ra n ce

ca rr ie rs

to th e st at e p o p u la ti o n m u lt ip li ed

b y 1, 00 0

B u re au

o f E co n o m ic

A n al y si s,

19 71 –2 00 5

La w ye r

R at io

o f th e n u m b er

o f la w y er s to

th e to ta l

p o p u la ti o n in

a st at e m u lt ip li ed

b y 10 ,0 00

T h e L aw

y er

S ta ti st ic al

R ep

o rt ,

v ar io u s y ea rs

P hy si ci an

R at io

o f th e n u m b er

o f p h y si ci an

s to

th e to ta l

p o p u la ti o n in

a st at e m u lt ip li ed

b y 10 ,0 00

S ta ti st ic al

A b st ra ct

o f th e U n it ed

S ta te s,

19 71 –2 00 5

St at e sp ill ov er

T h e p ro p o rt io n o f n ei g h b o ri n g st at es

p as si n g to rt

re fo rm

s b ef o re

th e st at e en

ac ts

th e to rt

re fo rm

Id eo lo gi ca l st at e sp ill ov er

a T h e p ro p o rt io n o f id eo

lo g ic al

n ei g h b o rs

p as si n g to rt

re fo rm

s b ef o re

th e st at e en

ac ts

th e to rt

re fo rm

. “I d eo

lo g ic al

n ei g h b o rs ” ar e st at es

w it h id eo

lo g y

sc o re s th at

ar e w it h in

10 p o in ts

o f th e st at e in

q u es ti o n

(C on ti nu

ed )

968 THE JOURNAL OF RISK AND INSURANCE

TA B LE

2 C on

tin ue

D ef in it io n

S o u rc es

U rb an

a T h e p ro p o rt io n o f th e n u m b er

o f u rb an

p eo

p le

to to ta l st at e p o p u la ti o n , m ea su

re d in

ev er y d ec ad

e U n it ed

S ta te s C en

su s B u re au

, 19 71 –2 00 5

So ut h

A n in d ic at o r v ar ia b le

eq u al

to 1 if st at e is A L ,A

R ,F

L ,

G A , K Y , L A , M S , N C , O K , S C , T N , T X , V A , o r W A

M id w es t

A n in d ic at o r v ar ia b le

eq u al

to 1 if st at e is

IA , IL , IN

, K S , M I, M N , M O , N E , N D , O H , S D , o r W I

N or th ea st

A n in d ic at o r v ar ia b le

eq u al

to 1 if st at e is

in C T , M A ,

M D , M E , N H , N J, N Y , P A , R I, V T , an

d W V

W es t

A n in d ic at o r v ar ia b le

eq u al

to 1 fo r o th er

st at es

n o t

li st ed

ab o v e

a T h es e al te rn at iv e m ea su

re m en

ts ar e u se d in

ro b u st n es s ch

ec k s.

WHAT DRIVES TORT REFORM LEGISLATION? 969

TA B LE

3 M ea

ns an

d St an

da rd

D ev ia tio

ns of

Va ria

bl es

in A na

ly sis

fo rG

en er al

Lia bi lit y,

19 71

– 20

V ar ia b le

W h o le

S am

p le

F o u r T o rt

R ef o rm

C ap

s o n

N o n ec o n o m ic

D am

ag es

C ap

s o n

P u n it iv e

D am

ag es

C o ll at er al

S o u rc e

Jo in t an

d S ev

er al

E co n o m ic

ch ar ac te ri st ic s

Li ab ili ty

in su ra nc e lo ss

a 8. 81 1

8. 26 7

8. 72 1

8. 63 9

8. 58 8

8. 34 3

(2 .7 25 )

(2 .9 10 )

(2 .7 75 )

(2 .7 99 )

(2 .7 57 )

(2 .8 53 )

M ed ic al

lia bi lit y lo ss

0. 47 5

0. 38 3

0. 48 7

0. 47 6

0. 42 9

0. 43 7

(0 .3 67 )

(0 .3 21 )

(0 .3 88 )

(0 .3 85 )

(0 .3 50 )

(0 .3 58 )

H ea lt h ca re

ex pe nd

it ur e

2. 76 9

2. 22 1

2. 64 4

2. 59 8

2. 55 8

2. 33 1

(0 .9 41 )

(0 .7 92 )

(0 .9 13 )

(0 .9 12 )

(0 .9 07 )

(0 .8 41 )

C re di t sp re ad

1. 08 2

1. 22 5

1. 11 3

1. 12 7

1. 13 7

1. 20 5

(0 .3 97 )

(0 .4 23 )

(0 .4 08 )

(0 .4 10 )

(0 .4 12 )

(0 .4 22 )

G SP

pe r ca pi ta

20 .4 91

14 .1 32

19 .0 89

18 .5 05

18 .1 41

15 .1 3

(1 1. 39 8)

(9 .4 13 )

(1 1. 07 3)

(1 1. 02 5)

(1 0. 88 7)

(9 .7 67 )

P o li ti ca l ch

ar ac te ri st ic s

R at io

of R ep ub

lic an

0. 41 9

0. 35 6

0. 41 2

0. 40 3

0. 39 9

0. 36 7

(0 .1 79 )

(0 .1 89 )

(0 .1 78 )

(0 .1 84 )

(0 .1 81 )

(0 .1 84 )

Sa m e pa rt y

0. 46 3

0. 49 2

0. 46 5

0. 46 8

0. 48 1

(0 .4 99 )

(0 .5 00 )

(0 .4 99 )

(0 .5 00 )

R ep ub

lic an

id eo lo gy

35 .0 78

36 .7 65

35 .3 47

36 .1 34

35 .3 92

36 .5 89

(8 .0 96 )

(8 .3 73 )

(8 .1 84 )

(8 .4 14 )

(8 .3 49 )

(8 .0 89 )

D em

oc ra ti c id eo lo gy

68 .1 69

68 .0 02

67 .9 29

68 .6 45

67 .8 82

67 .9 8

(7 .3 17 )

(7 .8 70 )

(7 .4 29 )

(7 .4 12 )

(7 .7 04 )

(7 .6 90 )

G ov er no r id eo lo gy

53 .0 73

55 .7 99

52 .7 77

54 .4 26

53 .5 31

55 .5 1

(1 7. 96 9)

(1 6. 54 7)

(1 7. 59 5)

(1 7. 66 4)

(1 7. 78 9)

(1 6. 61 4)

T or t ca se s in

fe de ra l co ur t

1. 53 8

1. 37 7

— —

(0 .3 68 )

(0 .3 06 )

(C on ti nu

ed )

970 THE JOURNAL OF RISK AND INSURANCE

TA B LE

3 C on

tin ue

V ar ia b le

W h o le

S am

p le

F o u r T o rt

R ef o rm

C ap

s o n

N o n ec o n o m ic

D am

ag es

C ap

s o n

P u n it iv e

D am

ag es

C o ll at er al

S o u rc e

Jo in t an

d S ev

er al

T or t ca se s in

st at e co ur t

2. 61 2

2. 60 9

— —

(1 .2 05 )

(1 .2 33 )

R at e re gu

la ti on

0. 58 3

0. 65 5

0. 59 1

0. 54 6

0. 62 4

0. 64 3

(0 .4 93 )

(0 .4 76 )

(0 .4 92 )

(0 .4 98 )

(0 .4 85 )

(0 .4 79 )

P un

it iv e da m ag e in su ra bl e

0. 56 0

0. 60 1

0. 56 6

0. 58 2

0. 57 9

0. 58 7

(0 .4 97 )

(0 .4 90 )

(0 .4 96 )

(0 .4 93 )

(0 .4 94 )

(0 .4 93 )

N o- fa ul t ve rb al

0. 07 2

0. 03 2

0. 07 0

0. 04 9

0. 04 5

0. 05 1

(0 .2 59 )

(0 .1 75 )

(0 .2 55 )

(0 .2 16 )

(0 .2 08 )

(0 .2 20 )

N o- fa ul t hi gh

va lu e

0. 18 5

0. 12 6

0. 15 4

0. 16

0. 12 2

0. 14 3

(0 .3 88 )

(0 .3 32 )

(0 .3 62 )

(0 .3 67 )

(0 .3 27 )

(0 .3 50 )

N o- fa ul t ot he r

0. 74 3

0. 84 2

0. 77 6

0. 79 1

0. 83 3

0. 80 6

(0 .3 35 )

(0 .4 43 )

(0 .5 60 )

(0 .3 11 )

(0 .2 80 )

(0 .3 03 )

O th er

v ar ia b le s

E m pl oy m en t in

in su re r

5. 85 5

5. 17 5

5. 78 2

5. 59 7

5. 41 9

5. 30 1

(3 .3 45 )

(2 .6 87 )

(3 .4 05 )

(3 .3 67 )

(2 .8 33 )

(2 .6 72 )

La w ye r

24 .7 04

21 .5 65

23 .8 35

23 .9 53

23 .3 35

21 .8 75

(9 .3 21 )

(8 .7 44 )

(9 .2 57 )

(9 .5 95 )

(8 .8 05 )

(8 .3 29 )

P hy si ci an

19 .4 35

18 .0 26

19 .0 28

19 .2 22

18 .9 29

18 .4 15

(6 .5 47 )

(7 .1 83 )

(6 .5 35 )

(7 .1 47 )

(6 .8 72 )

(7 .1 53 )

St at e sp ill ov er

— 0. 31 3

0. 05 9

0. 13 4

0. 12 6

0. 25 2

(0 .3 07 )

(0 .1 13 )

(0 .2 15 )

(0 .2 86 )

W es t

0. 28 0

0. 26 2

0. 24 6

0. 28 8

0. 27 9

0. 23 8

(0 .4 49 )

(0 .4 40 )

(0 .4 31 )

(0 .4 53 )

(0 .4 49 )

(0 .4 26 )

M id w es t

0. 24 0

0. 16 1

0. 24 3

0. 21 7

0. 19 4

0. 19 8

(0 .4 27 )

(0 .3 68 )

(0 .4 29 )

(0 .4 12 )

(0 .3 96 )

(0 .3 98 )

(C on ti nu

ed )

WHAT DRIVES TORT REFORM LEGISLATION? 971

TA B LE

3 C on

tin ue

V ar ia b le

W h o le

S am

p le

F o u r T o rt

R ef o rm

C ap

s o n

N o n ec o n o m ic

D am

ag es

C ap

s o n

P u n it iv e

D am

ag es

C o ll at er al

S o u rc e

Jo in t an

d S ev

er al

So ut h

0. 28 0

0. 31 4

0. 29 5

0. 25 6

0. 29 8

0. 31 9

(0 .4 49 )

(0 .4 64 )

(0 .4 56 )

(0 .4 36 )

(0 .4 58 )

(0 .4 66 )

N or th ea st

0. 20 0

0. 26 3

0. 21 6

0. 23 9

0. 22 8

0. 24 5

(0 .4 00 )

(0 .4 41 )

(0 .4 11 )

(0 .4 27 )

(0 .4 20 )

(0 .4 31 )

A v er ag

e w ai ti n g ti m e

— 18 .3 33

30 .8 60

29 .2 61

27 .5 80

20 .4 58

O b se rv at io n s

17 50

82 5

15 43

13 46

13 68

98 2

a L ia b il it y in su

ra n ce

lo ss

in cl u d es

th e O th er

L ia b il it y , A u to

L ia b il it y , M ed

ic al

M al p ra ct ic e L ia b il it y , an

d P ro d u ct

L ia b il it y li n es .

972 THE JOURNAL OF RISK AND INSURANCE

industry, and the number of lawyers in the state as a proxy for the power of the legal professionals. Lawyers per capita (Lawyer) are obtained from the Lawyer Statistical Report and from theAmericanBarAssociation’s annual report.7 Physicians per capita (Physician) is taken from the U.S. Statistical Abstract for each state and year. The data are not available for each year, sowe interpolate values for the years inwhich the data are not reported using a method from previous studies (e.g., Browne and Schmit, 2008).8

In addition to serving as proxies for private interest group power, both Lawyer and Physician may also reflect other influences, which will be described below.

Litigation Activity In practice, litigation costs are often fingered as a prime motivation for reform adoption. In theory, litigation activity could either accelerate or delay the reform process. Litigation produces benefits for plaintiffs and lawyers and costs for defendants. The welfare impact of restricting litigation requires balancing those costs and benefits, along with the indirect benefits and costs associated with the tort environment’s impact on business activity and other behaviors. The prediction of public interest theory thus depends on the outcome of the welfare calculation and could go in either direction. Private interest theory offers similar nuance: more litigation activity means more benefits and more costs, raising the stakes for interest groups on both sides. Thus, the direction of influence of high litigation activity is theoretically ambiguous, although its prominence in policy debates suggests that it plays an important role and is usually cast as a motivation for reform action.

Two variables are used to proxy the litigation activity at the state level: liability insurance losses and tort cases commenced at the state level. We define Liability insurance loss as the ratio of directed loss incurred in the state to gross state product. The source for the information is loss data aggregated by state and year over the period 1971–2005 provided by A.M. Best. The analyses include aggregate loss incurred in Medical Malpractice, Auto Liability, Other Liability (including Product Liability before 1992), and Product Liability.9 The figures are on a calendar-year basis, so that for each line, the incurred loss is calculated as losses paid during the year plus the change in the insurance company’s reserves. This measure has pros and cons. On the one hand, the calendar-year incurred loss reflects the full financial impact of the year on the insurance industry, as the loss payments and reserve changes directly determine the change in the reported balance sheet. The measure can thus be a valuable indicator of current insurance market conditions, to the extent that such

7Though both would seem to benefit from litigation, it is possible that plaintiffs’ lawyers and defense lawyersmight have different attitudes toward tort reforms. Our data, however, do not permit us to explore this distinction.

8We have 21 years of reported data (1971, 1980, 1985, and 1988–2005). Similar to previous studies,weprovide fitted values for themissing years, based on anOLS regressionusing the 21 data points with the specification: lawyers per capita ¼ aþ b� yearþ e:

9We also analyze three subgroups of liability insurance losses (i.e., Other Liability, Medical Malpractice, and Auto Liability) separately. We get similar results in unreported tables.

WHAT DRIVES TORT REFORM LEGISLATION? 973

conditions follow from financial status of the insurance industry. On the other hand, calendar-year incurred loss is not necessarily tightly connected to litigation activity initiated in the current year, as both loss payments and reserve adjustments may be related to cases that had been opened prior to the start of the year. Moreover, the reserves reported by insurers are estimates of future loss payments and are thus subject to future revision. Indeed, some evidence suggests that the estimates are strategically manipulated by individual insurers (e.g., Grace, 1990).

While it is not possible to obtain the accident-year incurred loss data (at the state level) that would enable us to reflect subsequent revisions to initial reserve estimates, we do have alternative measures of current litigation activity. Specifically, we use tort filings in general jurisdiction courts per capita (Tort cases in state court) as an alternative proxy of litigation activity at the state level. State court data are available for states from 1975 to 2005 from theNationalCenter for StateCourts.10Wealsouse thenumberof tort cases commencedper capita in theU.S. federal courtsnationwide (Tort cases in federal court) as another proxy for litigation activity. Jurisdictional rules only allow cases that involve questions of state law, but are (1) between citizens of different states or betweenUnited States and foreign citizens and (2) involve more than $75,000 in losses, to be filed in federal court. The vastmajority of tort cases are filed in state courts. Both of these proxy measures exhibit significant correlation with Liability insurance loss.

It should also be noted that the number of lawyers (as captured in Lawyer), in addition to measuring the power of the legal professionals, could also proxy for state-level litigation activity and related costs to the extent that such activity is not fully captured by the tort case or insurance loss variables. To elaborate, tort filings do not fully capture the all of the activity associated with compensable injuries, as many injury cases are settled without a lawsuit being filed; insurance loss variables also do not capture the full extent of expenses related to compensable injuries as some costs may be uninsured.

Political-Institutional Arrangements Tort reforms are often associated with Republicans as opposed to Democrats, suggesting that states controlled by Republicans may favor tort reform (e.g., Rubin, 2005). Moreover, one party controlling the legislature and the governorship could also be an advantage for getting legislation of any kind passed (see, e.g., Kroszner and Strahan, 1999). The political climate and economic environment could also be correlated within regional areas in the United States. For example, the labels “Conservative” and “Liberal” may suggest different ideologies in different regions of the country.

We use two political variables from the U.S. Statistical Abstract. We measure the influence of Republicans in the legislature by the average ratio of the fraction of the Republicans in the lower house and the upper house (Ratio of Republican). Second, we create a dummy variable that equals 1 if the same party controls the governor’s office and hasmajorities in both houses. Finally, we use threemeasures of ideology in

10Data are not available for SD, NE, VT, IA, IL, LA, KY, MT, OK, SC, VA, and PA.

974 THE JOURNAL OF RISK AND INSURANCE

the state developed by Berry et al. (1998): Republican ideology,Democratic ideology, and Governor ideology. These variables capture the political ideology of the legislators and governor in the state, using the ideologies of the respective congressional delegations as a proxy. The indices have high values if the state’s representatives in Congress are liberal and low values if the state’s representatives in Congress are conservative (Berry et al., 2013).

Insurance Regulation Environment Auto liability insurance premiums account for a significant proportion of liability insurance premiums. No-fault systems provide first party coverage for personal injury protection (medical cost, loss of income, etc.) while limiting the tort liability of negligent drivers. No-fault laws can reduce auto insurance costs if there are strong limitations on the right to sue (Harrington, 1994). In this sense, no-fault legislation could reduce cost pressures associated with tort environments that are in other respects permissive. However, the presence of a no-fault system could also create spillover effects to displace tort activity into other liability markets, or could be reflective of a societal willingness to experiment with tort reform generally. Thus, the predicted association between the presence of no-fault systems and the propensity to enact other tort reforms is not clear.

Rate regulation is also an important aspect of the insurance environment. More stringent regulation may lead to more severe insurance market availability problems in the wake of a shock—accelerating the legislative process in tort reform.

To measure the effect of state insurance laws, we create five variables. First, we use a dichotomous variable that equals 1 if punitive damages are insurable (Punitive damage insurable). Second, we construct an indicator variable that equals 1 if the state has any form of “prior approval” rate regulation (Rate regulation). “Prior approval” rate regulation is the toughest type of rate regulation, reflecting circumstances in which rates cannot be used until approved by the insurance commissioner. Last, we include three variables relating to no-fault automobile insurance laws. No-fault systems with value thresholds make the right to sue conditional on compensatory damages exceeding a designated dollar amount. No-fault systems with a small dollar threshold may thus be ineffective in limiting lawsuits. The benchmark variable No-fault other equals 1 for a tort law state or for a no-fault state with a low value threshold (less than $1,000). Another variable, No-fault high value, equals 1 if a state has a dollar threshold greater than $1,000, and 0 otherwise. Stateswith verbal thresholds have the strictest no- fault system, which allows the right to sue only in the cases where victims have severe damages such as death, disfigurement, or permanent loss of body function. The variable No-fault verbal equals 1 if the state has a verbal threshold, and 0 otherwise.

Spatial Diffusion When states pass laws, it may influence the likelihood that neighboring states will pass similar legislation. Such a relation is called a spatial diffusion effect or state spillover effect. Diffusion researchers suggest that policymakers often look to the policies adopted by nearby states to help them draft andmake decisions about certain policies (Soule and Earl, 2001).

WHAT DRIVES TORT REFORM LEGISLATION? 975

A state’s tort reform enactment may affect the timing of its neighbors’ tort reform enactments for several reasons. First, states may be concerned about businesses (e.g., entrepreneurs or physicians) moving out: if a state’s liability costs are higher than those of its neighbors, businessesmay be driven away. Under this logic, a state ismost influenced by the actions of those states to which its businesses may move. This suggests that a tort reform in one state could produce competitive pressures for corresponding tort reforms in neighboring states. Second, voters may judge politicians’ performance relative to that of politicians in nearby states. An example of this is given by Besley and Case (1995), who provide evidence of “yardstick competition.” All of this suggests that tort reforms in one state may have positive spillover effects on neighboring states. However, some have documented significant negative spillover effects associated with geographic proximity as well, finding that states are less likely to pass controversial laws when they are in close spatial proximity to other states that enact such laws (e.g., Soule and Earl 2001, McMahon- Howard, Clay-Warner, and Renzulli, 2009). The direction of influence is thus an empirical question.

We create a continuous variable, State spillover, to capture the influence of tort reforms in nearby states. The variable captures the proportion of neighboring states that had adopted tort reforms by the end of the previous year. Neighboring states are defined as states sharing a border with the state in question.11

Economic Variables Overall economic conditions, as well as the conditions in industries susceptible to litigation, may create pressures for or against reform. We use gross state product per capita (GSP per capita) as a general economic indicator connected with urbanization and business activity. The GSP data are from the Bureau of Economic Analysis for each state for each year. We use a credit spread variable (Credit spread) to proxy the investment environment each year, measured as the difference between the Moody’s seasoned BAA corporate bond yield and the AAA corporate bond yield, obtained from the Federal Reserve Bank of St. Louis’s Federal Reserve Economic Database.

Another potential proxy of interest is health care expenditure as a percentage of household income (Health care expenditure). If tort reforms are perceived as ameans of reducing health care costs, public interest theory would predict a positive relationship between this proxy and tort reforms, especially in the area of medical malpractice; again, a similar argument could bemade for private interest theory if the business community is conceived as the relevant interest group. Similar logic could be used to hypothesize a negative coefficient on Physician, as an inverse indicator of the degree of competition in medical services. Tort reform on medical malpractice could be a potential means of addressing cost problems in medical services and encouraging more physicians to practice in the state.

11For example, Florida shares borders with Georgia and Alabama. Georgia and Alabama enacted joint and several liability in 1988 and after 2005, respectively. Thus, State spillover for joint and several liability in Florida is 0 before 1988 and .5 from 1988 to 2005.

976 THE JOURNAL OF RISK AND INSURANCE

Finally, we create regional indicator variables for the North, South, Midwest, and West.

METHODS AND EMPIRICAL RESULTS

In this section, we first use the nonparametric survivalmodel to study the shape of the baseline hazard function. Next, we investigate how the economic and political variables described above are associated with the timing of tort reform enactments using discrete-time survival models with a Weibull baseline hazard. We start by treating reforms as “competing risk events,” so that the duration time ends if any of the four tort reforms is enacted. Then we study each of the four reforms independently.

Hazard Analysis With Nonparametric Estimates The Kaplan–Meier product-limit estimator12 offers a nonparametric estimate of the hazard function over time. Figure 1 graphs survival estimates for the data and shows that each survival function is relatively flat in the 1970s before dropping steeply in the mid-1980s and then becoming flat again. Joint and several liability limits have the shortest waiting time durations, followed by caps on punitive damages, collateral

FIGURE 1 Kaplan–Meier Estimate of Survival Rate, by Tort Reform

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

19 71

19 72

19 73

19 74

19 75

19 76

19 77

19 78

19 79

19 80

19 81

19 82

19 83

19 84

19 85

19 86

19 87

19 88

19 89

19 90

19 91

19 92

19 93

19 94

19 95

19 96

19 97

19 98

19 99

20 00

20 01

20 02

20 03

20 04

20 05

Four tort reform Caps on puni�ve damages Joint and several liability

Collateral source rule Caps on noneconomic damages

12TheKaplan–Meier product limit formula to estimate the survivorship function for the jth year

is Ŝ tð Þ ¼ Y j

P̂r T � tjjT � tj � 1 � � ¼ Y

1� dj rj

� � , where dj is the number of states enacting tort

reforms during jth year and rj is the number of states entering the jth year.

WHAT DRIVES TORT REFORM LEGISLATION? 977

source rule reforms, and caps on noneconomic damages. Figure 1 also shows the survival rate for states based on the enactment of any of the four tort reforms during the sample period (87 percent of states enact one ormore of those tort reforms) aswell as for each of the individual tort reforms during the sample period: about 40 percent of states enacted caps on punitive damages, 79 percent enacted limitations on joint and several liability, 35 percent enacted collateral source rule reforms, and 24 percent enacted caps on noneconomic damages. Table A2 in the Online Appendix (Deng and Zanjani, 2016) presents the results of the state “waiting period spells” by Kaplan– Meier product limit estimates in detail. Consistent with Figures A1–A7 (in the Online Appendix), the null hypotheses of equalities of survival functions across the four tort reforms are rejected by log-rank tests. Survival rates when considering caps on noneconomic damages and collateral source rule reforms are significantly higher than when considering caps on punitive damages and joint and several liability, since fewer states enact the former two tort reforms. Figure 1 also shows that the survival rates when considering the enactment of any type of reform drops dramatically during the years of 1985–1988. However, the nonparametric method is limited because it cannot provide the impact of independent variables on the likelihood of events.

Discrete-Time Proportional Hazard Model Tort reform enactments and most of the independent variables are only observed once a year.13 This implies that the observed durations of tort reform should be grouped into yearly intervals. Moreover, since many of the tort reforms occurred in the mid-1970s or the mid-1980s (see Table 1), there exists the substantial problem of tied duration times. The Cox model (or any fully parameterized continuous-time model) is inappropriate in this case (Cox and Oakes, 1984), because the Cox model is based on the assumption that duration times can be any real number (rather than certain discrete values corresponding to the number of years), and recorded duration times need to be ordered chronologically. The high number of ties and the discrete- time property lead to estimation bias in regression coefficients and in the corresponding covariance matrix.

That the duration variable of interest (time taken to enact the tort reform) is measured yearly means that the feasible approach to modeling the duration is the discrete-time proportional hazard model (also called the discrete-time historical event model). The four tort reforms can be considered as “competing events” in the sense that the state can change its common law civil justice system by enacting different tort reforms. We consider the period from the beginning of the sample (1971) until any tort reform occurs as the duration of interest in the analysis. The states that enacted tort reform before 1971 are excluded from the analysis as left censored data, but the states that had not enacted tort reform by 2005 remain in the sample as right-censored data. The data structure of discrete-time duration models is time-series cross-section, which is organized with as many observations for each state in the sample as there are time periods over which the state is at risk of experiencing the event of interest (Jenkins,

13We recognize that some states are repeat adopters of tort reforms, and we address the problem in the “Robustness Checks and Extensions” section.

978 THE JOURNAL OF RISK AND INSURANCE

1995)—the enactment of a tort reform. The core variable is discrete elapsed time, and event occurrence is a series of binary outcomes denoting whether or not the event occurred at the observation point.

A survival model that can accommodate this structure is a discrete-time proportional hazard model with Weibull baseline hazard. Consider an instantaneous force of mortality given by:

hðt;XitÞ ¼ h0ðtÞ exp Xit 0bð Þ or log h t;Xitð Þ½ � ¼ log h0 tð Þ½ � þ Xit

0b; ð1Þ

where h(t,Xit) is the hazard rate function, and h0(t) is the baseline hazard function, and Xit is a vector of covariates. In this setting, the discrete-time hazard rate (see, e.g., Allison, 1982) of enacting tort reform for state i in year k, with a vector of time-varying covariates, Xik, having spent k–1 years in the same common law civil justice system, can be given by:

hk Xikð Þ ¼ 1� exp �exp g kð Þ þ Xit 0bð Þ� �

; and g kð Þ ¼ log

Zak

ak�1

h0 tð Þdt; ð2Þ

where g(k) represents the baseline hazard function of time, which can be estimated either parametrically or nonparametrically. Equation (2) can also be written in complementary log–log transformation:

log �log 1� hk Xikð Þð Þ� � ¼ g kð Þ þ Xik 0b: ð3Þ

To specify the baseline hazard function g(k), we consider the discrete-time spification similar to theWeibullmodelwith a shape characterized by p: g(k)¼ ulog(k), where u in the discrete-time case approximately corresponds to p–1 in the continuous-time Weibull model. If u is greater than zero, the hazard increasesmonotonically, and if it is less than zero the hazard decreases monotonically.14

The coefficients can be interpreted as the effect of covariates on the hazard rate of enacting tort reform. Positive coefficients indicate an increase in the hazard rate and thus a reduction in the duration of the “waiting” period. The exp(b) represent the hazard ratio of tort reform enactments for a one-unit change in the covariates because we use the proportional hazard model. The economic importance of coefficients on covariates can be evaluated by taking the exponential transform of the coefficients multiplied by the standard deviation of the explanatory variables.

14We use a complementary log-logistic hazard function rather than a logistic function since the process of enacting tort reform is, in reality, continuous although the observations are in discrete-time. Steele (2008) argues that the choice of binarymodel often has little impact on the results. To check the correctness of the proportional hazard specification, we have also estimated a discrete-time logistic model and the results are very similar to complementary log–log.

WHAT DRIVES TORT REFORM LEGISLATION? 979

We first investigate how economic and political variables influence the timing of general liability tort reform enactments by considering the four reforms together. In other words, an “enactment” happens, and a state “dies,” whenever it enacts any of the four reforms. Our first analysis considers the association between tort reform adoption and a set of “core” explanatory variables, which will subsequently be expanded to include additional covariates.

Table 4 shows the correlation of selected political economy factors. Highly correlated explanatory variables pose a common problem in empirical studies, especially when the sample size is limited. Multicollinearity may cause regression results to be unstable and inflate standard errors. As shown in Table 4, variables representing private interest groups are correlated with Liability insurance loss to varying degrees, some significantly so. We address this by using alternative measures of litigation activity (see Table 6) and also by adding explanatory variables to regressions sequentially (see, e.g., Table 5) to investigate the stability of the empirical estimates. We use stepwise regression to select the core variables in Table 5 from all independent variables shown in Table 3, with the exceptions of the tort case variables (which we viewed as substitutes for the insurance loss variable).

Table 5 shows the estimated effects of core variables on the hazard rate for the discrete-time model with Weibull baseline hazard. We have 825 observations in Table 5. Model 1 tests litigation activity costs by including only one variable, Liability insurance loss; Model 2 adds the interest group proxies Physicians and Lawyer; and Model 3 further adds all other core variables selected by stepwise regression.15

Overall, we document significant positive coefficients for Liability insurance loss, Ratio of Republican, Same party, and Lawyer, and a significant negative coefficient for Physician. Pressures in the form of higher litigation costs and a relative shortage of physicians seem to be associated with quicker adoptions. An abundance of lawyers is also associatedwith quicker reform,which could be another reflection of a connection between litigation costs and tort reform—to the extent that greater numbers of lawyers are indicative of high-litigation-cost environments. Private interest theory could be reconciled with the results to the extent that the general business community—which is presumably concernedwith litigation andhealth care costs—is interpreted as a private interest group. On the other hand, evidence concerning other interest groups is weak and even contrary: greater numbers of lawyers, who presumably would oppose reform, are associated with faster adoptions; greater numbers of physicians, who presumably support reform, are associated with slower adoptions. Political-institutional variables reflecting Republicans and same-party control of the government also are statistically significant, with the expected signs.

Table 6, Model 1 shows the estimated effects of all time-varying covariates on the hazard rate for the discrete-time model with Weibull baseline hazard. We have 825

15The variables selected by the stepwise regression also include the dummies for the Northeast and Midwest regions. We do not show any regional dummies in Table 5 as the stepwise method excluded the other two regional dummies from the regression. Results with all variables, including the regional dummies, are shown in Table 6.

980 THE JOURNAL OF RISK AND INSURANCE

TA B LE

4 C or re la tio

ns of

C or e

Po lit ic al

Ec on

om y

Fa ct or s In fl ue

nc in g

th e

Ti m in g

of St at e

En ac

tm en

ts of

G en

er al

Lia bi lit y

To rt

Re fo rm

s, 19

71 – 20

1 2

3 4

5 6

7 8

9 10

11 12

1. Li ab ili ty

in su ra nc e lo ss

1. 00 0

2. T or t ca se s in

st at e co ur ts

0. 32 4

1. 00 0

3. T or t ca se s in

fe de ra l co ur ts

0. 51 7

0. 03 6

1. 00 0

4. H ea th

ca re

ex pe nd

it ur e

0. 55 9

0. 08 0

0. 63 8

1. 00 0

5. R at io

of R ep ub

lic an

0. 02 7

0. 01 0

0. 03 7

�0 .0 28

1. 00 0

6. Sa m e pa rt y

�0 .0 99

�0 .0 01

�0 .1 32

�0 .1 10

�0 .4 33

1. 00 0

7. R ep ub

lic an

id eo lo gy

0. 23 3

0. 19 4

0. 04 1

�0 .0 18

�0 .2 54

0. 11 3

1. 00 0

8. D em

oc ra ti c id eo lo gy

0. 16 5

0. 04 1

0. 07 3

0. 02 5

0. 29 5

�0 .2 37

0. 41 4

1. 00 0

9. G ov er no r id eo lo gy

�0 .0 39

0. 01 3

�0 .1 14

�0 .1 31

�0 .0 10

0. 37 6

0. 24 0

0. 28 6

1. 00 0

10 . E m pl oy m en t in

in su re r

0. 37 3

0. 31 8

0. 15 1

0. 13 1

�0 .0 53

0. 49 1

0. 39 5

0. 01 4

1. 00 0

11 . P hy si ci an

0. 51 0

0. 24 6

0. 32 8

0. 37 1

0. 01 3

�0 .1 57

0. 43 1

0. 36 6

�0 .0 15

0. 57 3

1. 00 0

12 . La w ye r

0. 32 0

0. 12 4

0. 28 3

0. 25 0

0. 03 1

�0 .2 14

0. 42 6

0. 47 7

0. 05 3

0. 57 1

0. 68 9

1. 00 0

WHAT DRIVES TORT REFORM LEGISLATION? 981

observations in Model 1, which extends models in Table 5 by including all variables. Higher Liability insurance loss is found to be associated with a higher hazard rate of tort reform enactment: a one-standard-deviation increase in the Liability insurance loss leads to a 159 percent increase in the hazard of tort reform enactment. With regard to the political-institutional variables, the ratio of Republicans in the legislature, single-party control of the government, and the nature of Democratic ideology all have statistically significant associations with the timing of tort reform. Notably, a one-standard-deviation increase in Ratio of Republican leads to an 86percent increase in the hazard rate. The negative coefficient on Democratic ideology indicates that a more liberal character in the state’s Democrats delays tort reform, with a one-standard-deviation increase in Democratic ideology decreasing thehazard rate by 45percent. In addition,Lawyerhas a positive effect on the hazard rate of reform enactment, and Physician has a negative but statistically insignificant effect on the hazard rate of tort reform. A one-standard-deviation

TABLE 5 Discrete-Time Hazard Model of Core Political Economy Factors Influencing the Timing of State Enactments of General Liability Tort Reforms, 1971–2005

Model 1 Model 2 Model 3

Liability insurance loss 0.131�� 0.230�� 0.231��

(0.055) (0.051) (0.047) Physician — �0.233�� 0.222��

(0.041) (0.043) Lawyer — 0.109�� 0.140��

(0.029) (0.032) Ratio of Republican — — 2.560��

(0.857) Same party — — 0.833��

(0.400) Democratic ideology — — �0.046

(0.044) Republican ideology — — �0.025

(0.031) Log duration time 0.487�� 0.902�� 0.842��

(0.247) (0.358) (0.353) Constant �5.410�� 5.604�� 3.796

(0.594) (0.776) (2.619) Log pseudolikelihood �147.29 �137.36 �128.922 Observations 825 825 825

Note: The dependent variable for Models 1–3 is binary that equals 1 if a state enacts any type of general liability tort reform in that year, and 0 otherwise. Observations of each state are included in the analyses until the state passes the tort reform. All the variables are described as in Table 2. Stepwise regression using the full set of explanatory variables is used to choose core variables in this table. The table reports results of the discrete-time hazard model withWeibull baseline hazard. Standard errors robust to clustering by year are shown in parentheses. �� and �� denote statistical significance at the 1 and 5 percent levels.

982 THE JOURNAL OF RISK AND INSURANCE

TA B LE

6 D isc

re te -T im

e H az

ar d M od

el of

A ll Po

lit ic al

Ec on

om y Fa

ct or s In fl ue

nc in g th e Ti m in g of

St at e En

ac tm

en ts

of G en

er al

Lia bi lit y To

rt Re

fo rm

s, 19

71 – 20

M o d el

1 M o d el

2 M o d el

3 G L _C

N G L _C

P G L _C

S G L _J S

Li ab ili ty

in su ra nc e lo ss

0. 32 7�

�� —

— 0. 43 5�

�� 0. 46 3�

�� 0. 66 2�

�� 0. 30 1�

��

(0 .1 16 )

(0 .1 26 )

(0 .1 02 )

(0 .0 76 )

(0 .0 70 )

T or t ca se s in

st at e co ur t

— 0. 48 8�

� —

— —

(0 .2 08 )

T or t ca se s in

fe de ra l co ur t

— —

2. 37 9�

�� —

— —

— (0 .7 53 )

St at e sp ill ov er

0. 07 4

0. 41 2

0. 43 8

0. 45 7

�1 .2 34

2. 17 0

1. 35 0�

(1 .2 78 )

(2 .3 89 )

(1 .3 32 )

(4 .7 81 )

(1 .6 11 )

(1 .4 36 )

(0 .7 57 )

H ea th

ca re

ex pe nd

it ur e

�0 .3 28

0. 13 0

0. 07 7

�0 .4 16

�0 .4 71

�1 .7 65

��

-0 .0 43

(0 .7 36 )

(0 .8 80 )

(0 .0 69 )

(0 .5 31 )

(1 .0 23 )

(0 .5 23 )

(0 .8 11 )

C re di t sp re ad

�0 .1 54

�0 .3 88

0. 54 9

1. 05 6

�0 .3 38

�0 .4 13

0. 11 4

(0 .5 13 )

(0 .7 18 )

(0 .5 63 )

(0 .7 26 )

(0 .7 82 )

(0 .7 84 )

(0 .4 76 )

G SP

pe r ca pi ta

�0 .0 20

�0 .0 89

�0 .0 55

�0 .0 35

0. 05 0

�0 .0 76

�0 .0 07

(0 .0 47 )

(0 .0 61 )

(0 .0 39 )

(0 .0 83 )

(0 .0 65 )

(0 .0 79 )

(0 .0 31 )

R at io

of R ep ub

lic an

3. 28 4�

� 4. 85 8�

� 3. 11 7�

� 2. 14 9�

2. 32 7

2. 93 9

3. 27 6�

�

(1 .5 10 )

(1 .9 68 )

(1 .4 46 )

(1 .3 07 )

(2 .0 12 )

(2 .1 42 )

(1 .2 89 )

Sa m e pa rt y

1. 00 4�

1. 24 0�

� 0. 53 3

0. 86 8

0. 53 4

�0 .1 66

1. 34 7�

��

(0 .5 59 )

(0 .6 26 )

(0 .4 65 )

(0 .5 81 )

(0 .3 55 )

(0 .4 95 )

(0 .4 79 )

R ep ub

lic an

id eo lo gy

0. 01 9

0. 00 4

0. 03 1

0. 08 4�

�0 .1 08

� 0. 16 3�

�� 0. 04 2

(0 .0 63 )

(0 .0 77 )

(0 .0 61 )

(0 .0 48 )

(0 .0 60 )

(0 .0 56 )

(0 .0 49 )

D em

oc ra ti c id eo lo gy

�0 .0 77

� �0

.0 26

�0 .0 62

�� 0

.0 51

�0 .1 33

�� 0

.0 88

� �0

.0 53

(0 .0 46 )

(0 .0 64 )

(0 .0 30 )

(0 .0 87 )

(0 .0 60 )

(0 .0 46 )

(0 .0 56 )

G ov er no r id eo lo gy

0. 00 2

�0 .0 13

0. 00 5

0. 02 6

0. 00 8

0. 03 1

�0 .0 05

(0 .0 11 )

(0 .0 12 )

(0 .0 19 )

(0 .0 16 )

(0 .0 21 )

(0 .0 13 )

E m pl oy m en t in

in su re r

0. 02 5

0. 06 7

0. 11 1

�0 .1 14

�0 .1 89

0. 27 8�

� 0. 02 5

(0 .2 37 )

(0 .2 86 )

(0 .1 83 )

(0 .1 29 )

(0 .1 86 )

(0 .1 18 )

(0 .1 37 )

P hy si ci an

�0 .1 93

�0 .2 45

�0 .1 42

�0 .0 64

�0 .1 27

�0 .5 32

��

�0 .2 71

��

(0 .1 34 )

(0 .2 11 )

(0 .1 33 )

(0 .0 57 )

(0 .0 90 )

(0 .1 20 )

(0 .0 57 )

(C on ti nu

ed )

WHAT DRIVES TORT REFORM LEGISLATION? 983

TA B LE

6 C on

tin ue

M o d el

1 M o d el

2 M o d el

3 G L _C

N G L _C

P G L _C

S G L _J S

La w ye r

0. 17 8�

�� 0. 15 1

0. 12 6�

0. 14 6�

� 0. 12 2�

�� 0. 30 1�

�� 0. 20 6�

��

(0 .0 63 )

(0 .1 08 )

(0 .0 68 )

(0 .0 60 )

(0 .0 29 )

(0 .0 76 )

(0 .0 32 )

N o- fa ul t ve rb al

0. 96 8

2. 56 3�

� 1. 10 2

�2 .0 53

�� 0. 39 7

�2 .6 63

0. 77 0

(0 .7 72 )

(1 .1 33 )

(1 .4 60 )

(0 .9 79 )

(1 .1 18 )

(2 .0 66 )

(0 .7 87 )

N o- fa ul t hi gh

va lu e

�0 .0 42

1. 17 5

�0 .0 55

0. 70 4

1. 02 6

2. 68 8�

�� -0 .1 91

(0 .8 77 )

(1 .2 97 )

(0 .8 97 )

(0 .6 61 )

(1 .1 19 )

(0 .7 00 )

(0 .6 02 )

P un

it iv e da m ag es

in su ra bl e

0. 36 4

0. 17 6

0. 14 7

�0 .4 59

1. 27 3

1. 71 5�

� 0. 03 7

(0 .6 57 )

(0 .8 84 )

(0 .7 97 )

(0 .7 85 )

(0 .8 69 )

(0 .7 97 )

(0 .5 71 )

R at e re gu

la ti on

�0 .1 07

0. 31 5

�0 .1 66

�0 .1 61

0. 93 4

�1 .2 24

��

�0 .1 33

(0 .5 61 )

(1 .2 54 )

(0 .6 27 )

(0 .6 52 )

(0 .7 42 )

(0 .4 15 )

(0 .5 24 )

So ut h

�0 .2 58

2. 83 4

0. 62 2

�0 .7 45

0. 49 2

0. 55 7

�0 .5 40

(0 .8 46 )

(2 .1 98 )

(0 .5 57 )

(0 .8 41 )

(1 .3 18 )

(0 .5 47 )

(0 .5 24 )

M id w es t

0. 78 7

�0 .0 71

�0 .0 69

0. 17 2

3. 62 6�

� 1. 27 4

0. 08 5

(0 .8 01 )

(0 .6 84 )

(0 .7 48 )

(0 .8 64 )

(1 .4 81 )

(0 .8 66 )

(1 .2 90 )

N or th ea st

�2 .0 96

�� 0. 53 8

0. 23 3

�4 .5 70

��

�1 .3 03

� �3

.2 19

��

�2 .1 34

(1 .0 62 )

(0 .8 68 )

(0 .9 15 )

(1 .1 81 )

(0 .7 58 )

(1 .1 85 )

(1 .7 74 )

Lo g du

ra ti on

ti m e

0. 98 7�

�1 .3 89

�1 .8 43

1. 53 1

1. 77 4

2. 71 0�

� 0. 55 1

(0 .5 82 )

(1 .8 57 )

(1 .5 97 )

(1 .0 36 )

(1 .4 94 )

(1 .1 83 )

(0 .9 48 )

C o n st an

t �5

.1 51

�9 .8 90

�� 7

.5 68

�� 1

5. 73 3�

� �4

.3 18

�1 4. 41 4�

� �6

.8 94

(3 .7 85 )

(4 .8 49 )

(3 .2 76 )

(6 .7 56 )

(5 .9 50 )

(5 .8 65 )

(4 .4 72 )

L o g p se u d o -l ik el ih o o d

�1 21 .6 62

�9 1. 15 5

�1 24 .3 37

�6 3. 01 8

�8 0. 76 2

�7 0. 41 8

�1 19 .5 93

O b se rv at io n s

82 5

52 1

82 5

15 43

13 46

13 68

98 2

N ot e: T h e d ep

en d en

t v ar ia b le fo r M o d el s 1– 3 is b in ar y th at

eq u al s 1 if a st at e en

ac ti n g an

y ty p e o f g en

er al li ab

il it y to rt re fo rm

in th at

y ea r, an

d 0 o th er w is e.

O b se rv at io n s o f ea ch

st at e ar e in cl u d ed

in th e an

al y se s u n ti l th e st at e p as se s th e to rt re fo rm

.C re d it sp

re ad

an d to rt ca se s in

fe d er al

co u rt ar e m ea su

re d

y ea rl y fr o m

19 71

to 20 05 .A

ll o th er

v ar ia b le s ar e m ea su

re d fo r ea ch

st at e in

ea ch

y ea r fr o m

19 71

to 20 05 .A

ll th e v ar ia b le s ar e d es cr ib ed

as in

T ab

le 2. M o d el

1 re p o rt s re su

lt s o f th e d is cr et e- ti m e h az ar d m o d el

w it h W ei b u ll b as el in e h az ar d .M

o d el

2 re p la ce s li ab

il it y in su

ra n ce

lo ss

w it h to rt ca se

fi li n g s in

st at e

co u rt s an

d h as

52 1 o b se rv at io n s si n ce

(1 ) it u se s d at a af te r 19 74

an d (2 ) d at a fo r S D ,N

E ,V

T ,I A ,I L ,L

A ,K

Y ,M

T ,O

K ,S C ,V A ,a n d P A ar e m is si n g .G

L _C

N ,

G L _C

P , G L _C

S , an

d G L _J S re p o rt re su

lt s o f th e d is cr et e- ti m e h az ar d m o d el

w it h W ei b u ll b as el in e h az ar d fo r ca p s o n n o n ec o n o m ic

d am

ag es , ca p s o n

p u n it iv e d am

ag es ,c h an

g es

to co ll at er al

so u rc e ru le s an

d li m it s o n jo in t an

d se v er al

li ab

il it y ,r es p ec ti v el y .T

h e d ep

en d en

t v ar ia b le fo r th e fo u r m o d el s is

b in ar y th at

eq u al s 1 if a st at e en

ac ts

th e to rt

re fo rm

o n g en

er al

li ab

il it y in

th at

y ea r, an

d 0 o th er w is e.

S ta n d ar d er ro rs

cl u st er ed

b y y ea r ar e sh

o w n in

p ar en

th es es . ��

� , �� , an

d � d en

o te

st at is ti ca l si g n ifi ca n ce

at th e 1,

5, an

d 10

p er ce n t le v el s.

984 THE JOURNAL OF RISK AND INSURANCE

increase in lawyers per capita results in a 3.7 times higher hazard of enacting any kind of tort reform. The regional dummies indicate that states in Northeast are associated with delayed timing of tort reform.

Model 2 replaces Liability insurance losswith tort case filings in state courts and has 521 observations since it uses only data after 1974, and data from SD, NE, VT, IA, IL, LA, KY, MT, OK, SC,VA, and PA are now omitted due to the lack of information on tort case filings. The coefficient on tort cases in state general jurisdiction courts inModel 2 indicates a one-standard-deviation rise in tort cases in state court increases the hazard by 83 percent. Model 3 uses tort cases filings in federal courts and the coefficient is significantly positive. A one-standard-deviation increase in tort cases in federal courts increases the hazard by 107 percent. The coefficients on State spillover and Health care expenditure are not statistically significant in any of the three models.

We then perform the analysis for each of the four reforms individually. The final four columns of Table 6 show the results for the timing of the enactments of each of the four tort reforms using the same method. We use Liability insurance loss to proxy the state liability environment in all individual tort reform models.

Starting with caps on punitive damages (GL_CP) and limits on joint and several liability (GL_JS), there are 1,346 observations in the regression for caps on punitive damages and 982 observations in the regression for limitations on joint and several liability. The positive and statistically significant coefficients on the Liability insurance loss in both models indicate the association between higher insurance losses and enactments of both tort reforms, and these effects are economically important. The coefficient is 0.463 in the punitive damage cap regression, which means a one- standard-deviation increase in liability insurance loss per GSP results in a 2.7 times higher hazard of enacting caps on punitive damages. Similarly, the GL_JS column estimates indicate that hazard of joint and several liability reform enactment is higher by 136 percent with a one-standard-deviation rise in Liability insurance loss. Lawyer has a positive effect on the hazard rates in both models; for example, a one-standard- deviation increase in Lawyer is associated with a 2.2 times higher hazard of caps on punitive damages reform enactments.

The structure of the state government influences the timing of joint and several liability reform as well. Consistent with the political-institutional theory, split control of the branches of government tends to delay limitations on joint and several liability. The negative and statistically significant coefficient of Physician indicates that fewer physicians accelerate the timing of joint and several liability reform. States with fewer physicians are found to have a higher hazard of enacting tort reform compared to their counterparts: the effect of a one-standard-deviation decrease in Physician is to increase the hazard by 86 percent. Regional effects are also present. The coefficients on theNortheast regional indicators in both regressions are large inmagnitude and have negative signs, which imply the two types of tort reform occur later in the Northeast.

Turning to caps on noneconomic damages (GL_CN) and changes to collateral source rules (GL_CS), there are 1,543 observations in the regression for the GL_CN model and 1,368 observations in the regression for the GL_CSmodel. Both tort reforms occur earlier in states with larger Liability insurance loss. A one-standard-deviation increase

WHAT DRIVES TORT REFORM LEGISLATION? 985

in the liability insurance loss per GSP leads to a 2.3 times higher hazard rate for caps on noneconomic damages. A one-standard-deviation increase in Liability insurance loss results in a 5.2 times higher hazard rate for collateral source rule reform. The coefficients onDemocratic ideology are negative and statistically significant in the case of collateral source rule reform. The coefficient ofEmployment in insurer is significantly positive for the GL_CS model, a possible indication that insurer lobbying power accelerates the reform of collateral source rules. A one-standard-deviation increase in the Employment in insurer leads to a 120 percent increase in the hazard of changes to collateral source rules. This finding, however, does not appear in the other models of Table 6.

To summarize the results from the discrete-time proportional hazard model in Table 6, litigation activity is strongly associated with reform adoptions. The most consistent positive and significant results are those connecting insurance losses and tort cases with reform adoptions. The two variables could also be connected to the insurance industry and general business interest groups in the context of private interest theory. The number of lawyers exhibits a positive associationwith tort reform enactments, which is also consistent with the hypothesized connection between litigation activity and tort reform. On the other hand, we find little evidence of the influence of lobbying by special interest groups. On the contrary, as noted above, the number of lawyers exhibits a positive and significant association with enactments, and the number of physicians exhibits a negative association with enactments. These findings could be interpreted as supportive of the public interest theory, or as indicators of higher liability/medical costs that couldmobilize businesses to lobby for reform as predicted by the economic interest group theory.16 Political and institutional factors also help to explain the timing of tort reforms to some extent, as Republican influence, one-party governments, and conservative Democratic ideology are associated with quicker action in all specifications. Aside from one statistically significant result in the analysis of joint and several liability reform,we do not find strong linkages between the timing of tort reforms and spatial diffusion effects in Table 6.

ROBUSTNESS CHECKS AND EXTENSIONS

We test the robustness of the results and extend the analysis in a variety of ways. We offer a brief overview of these checks and extensions here and refer the reader to the Online Appendix (Deng and Zanjani, 2016) for further details.

The Online Appendix segments the robustness checks into five areas. First, we check the robustness of the Weibull baseline hazard by using a discrete-time proportional hazard model with a nonparametric baseline hazard. Second, we conduct analyses aimed at omitted variables problems, specifically by controlling for the presence of

16It is also possible that the number of lawyers and the number of physicians are inadequate proxies for the size and power of the relevant special interest groups. One might argue that better proxies would be the subpopulations of lawyers and physicians that are most sensitive to tort reform, such as lawyers specializing in personal injury cases or physicians engaged in high-risk specialties. Unfortunately, we do not have data on such subpopulations.

986 THE JOURNAL OF RISK AND INSURANCE

medical malpractice tort reforms. Third, we redo the analysis by restricting our attention to “permanent” reforms (i.e., discarding reforms that are repealed or declared to be unconstitutional; see Grace and Leverty, 2013). Fourth, we use different estimation techniques to incorporate the possible dependence of different tort reform enactments within a state. We do so because some states may pass tort reforms gradually while others pass reforms as a “package.” We also incorporate state unobservable heterogeneity into our model. Fifth, we experiment with alternative measurements to further test the robustness of our findings. Specifically, we consider an alternative definition of “neighborliness” in constructing a state spillover variable that incorporates “ideological proximity” as well as geographic proximity, alternative measurements of political variables, the degree of urbanization as an additional measure of the state tort reform environment, a different categorization of tort reform effective dates based on the dates of their enactments, and also lagged independent variables to address the possibility that the legislative process may be slow.

The results of robustness checks are largely consistent with the results reported in Table 6. One notable difference is the State spillover coefficient, which becomes large and statistically significant in some specifications, suggesting specifically that states respond to the actions of their competitors with tort reforms, and that this response is sensitive to the perceived constitutional strength of the reform in the competing state.

The Online Appendix also contains details on three extended analyses. To isolate the roles of time-series variation and cross-sectional variation in explaining tort reform enactments, we run a pooled time-series cross-sectional OLS regression with state- level dummies and a fractional logit model with initial values for key explanatory variables. Our results show that Liability insurance loss is positive and statistically significant, which indicates that this association persists even after controlling for state effects. The coefficients on Liability insurance loss, Ratio of Republican, Physician, and Lawyer are generally consistent with signs of previous models and statistically significant in the fractional logit model. The results suggest that the effects of these variables may also be driven by cross-sectional variation.

We make another attempt to detect interest group effects by studying differences in reform responses among the subsample of states experiencing shocks in the 1980s insurance crisis.We find Lawyer andRepublican control to be positively associatedwith states’ enactment of tort reform. Insurance company employment is found to be positively associatedwith collateral source rule reforms only. Physicians are found to be negatively associated with reform efforts, significantly so in the case of collateral source rules and punitive damage caps.

Finally, we apply the discrete-time model to medical malpractice tort reforms. The signs of the significant coefficients on theMedical malpractice insurance loss and Ratio of Republican are largely consistent with the results from the models of tort reform on general liability. The coefficients on Health care expenditure are now significantly positive in almost every specification. We find Same party to be positive and statistically significant in two of the specifications. Overall, the analysis of medical malpractice tort reforms has directionally similar results to the general liability analysis, thus providing a consistency check.

WHAT DRIVES TORT REFORM LEGISLATION? 987

DISCUSSION AND CONCLUSION

This article evaluates the factors associated with state adoptions of tort reforms from 1971 to 2005. The study contributes to at least two lines of literature. First, we add to the literature on the diffusion of laws (e.g., Romano, 2006; Smythe and Bird, 2008; McMahon-Howard, Clay-Warner, and Renzulli, 2009). To our knowledge, our article is the first study to examine the diffusion pattern of tort reforms and the relative importance of social, economic, and political factors in that process. Second, a huge literature focuses on the influence of tort reforms after their enactment but largely ignores the question of what motivated the reform in the first place.17

Our most robust and economically significant findings are that measures of litigation activity—such as insurance losses and tort cases filed—are associated with quicker adoption of tort reforms. These findings are broadly consistent with public interest theories of regulation, and generally with the idea that states respond to liability cost problems with tort reforms. The findings can also be interpreted as being consistent with economic interest theories to the extent that the litigation activity affects the interests of the business community and of the insurance industry.

We find no evidence of the influence of medical or legal lobbies in the context of private interest theories. Lawyers per capita are consistently associated with quicker reform enactments, while physicians per capita are associated with slower enact- ments. We also find very little evidence of an insurance industry lobby. We also find that political-institutional factors, such as Republican control and control of the government by the same party, and insurance regulation environments affect tort reform, although these variables are not significant in all specifications.

If the main motivation for tort reform lies in litigation costs, this naturally leads to the question ofwhat drives litigation costs. Indeed, if tort reform is a response to litigation costs, the deeper question about tort reform may not concern its impact, but rather why it was needed in the first place. Why did costs reach critical levels in some states but not others? Such questions are well beyond the scope of this article, but they deserve more attention from future researchers.

17We checkwhether the ex post consequences of tort reform are consistentwith previous studies by measuring the effects of all four tort reforms on the insurance loss across different liability lines (Other Liability, Auto Liability, and Medical Malpractice). The dependent variables representing insurance loss are the direct loss incurred of different insurance liability lines scaled by gross state product and multiplied by 1,000 for each state in that year. The four tort reform indicators are equal to 1 in the years when the tort reforms are effective, and 0 otherwise. We find coefficients on noneconomic damage caps and joint and several liability reforms to be of the expected negative signs in all four of the equations, and they are statistically significant in the case of Other Liability. The coefficient on noneconomic damage caps is also significant in the case of Auto Liability. The coefficients on noneconomic damage caps are larger in absolute value than any of the other coefficients for those two lines. Our results are roughly in line with prior literature studying the effect of tort reforms on the insurance market. However, coefficients on collateral source rule reforms and caps on punitive damages are not statistically significant in any of the regressions, and the signs are mixed. See Table A3 in the Online Appendix (Deng and Zanjani, 2016) for details.

988 THE JOURNAL OF RISK AND INSURANCE

REFERENCES

Allison, P. D., 1982, Discrete-Time Methods for the Analysis of Event Histories, Sociological Methodology, 13: 61-98.

American Tort ReformAssociation, 2012, State and Federal Tort Reform Issue, Available at http://www.atra.org/legislation/states

Avraham, R., 2014,Database of State Tort Law Reforms, 5th edition, Available at http:// papers.ssrn.com/sol3/papers.cfm?abstract_id=902711

Avraham, R., L. Dafny, and M. Schanzenbach, 2012, The Impact of Tort Reform on Employer Sponsored Health Insurance Premiums, Journal of Law, Economics, and Organization, 28: 657-686.

Avraham, R., and M. Schanzenbach, 2010, The Impact of Tort Reform on Private Health Insurance Coverage, American Law and Economics Review, 12: 319-355.

Becker, G. S., 1983, A Theory of Competition Among Pressure Groups for Political Influence, Quarterly Journal of Economics, 98: 371-400.

Berry,W.D., R.C. Fording, E. J. Ringquist, R. L.Hanson, andCarlKlarner, 2013,ANew Measure of State Government Ideology, and Evidence That Both the New Measure and an Old Measure Are Valid, State Politics and Policy Quarterly, 13: 164-182.

Berry,W. D., E. J. Ringquist, R. C. Fording, and R. L. Hanson, 1998, Measuring Citizen and Government Ideology in the American States, 1960–93, American Journal of Political Science, 42: 327-348.

Besley, T., and A. C. Case, 1995, Incumbent Behavior: Vote Seeking Tax Setting and Yardstick Competition, American Economic Review, 85: 25-45.

Blackmon, G., and R. Zeckhauser, 1991, State Tort Reform Legislation: Assessing Our Control of Risks, in: P.H. Schuck, ed., Tort Law and the Public Interest (NewYork, NY: W. W. Norton & Co.).

Born, P. H., and W. K. Viscusi, 1994, Insurance Market Responses to the 1980s Liability Reforms: An Analysis of Firm-Level Data, Journal of Risk and Insurance, 61: 192-218.

Born, P. H., andW. K. Viscusi, 1998, The Distribution of the Insurance Market Effects of Tort Liability Reforms, in: M. N. Baily, C.Winston, and P. C. Reiss, eds., Brookings Papers on Economic Activity, Microeconomics: 1998 (Washington, DC: Brookings Institution Press), pp. 55-105.

Born, P. H., W. K. Viscusi, and T. Baker, 2009, The Effect of Tort Reform on Malpractice Insurers’ Ultimate Losses, Journal of Risk and Insurance, 76(1): 197-219.

Browne, M. J., and R. Puelz, 1999, The Effect of Legal Rules on the Value of Economic andNon-EconomicDamages and theDecision to File, Journal of Risk andUncertainty, 18: 189-213.

Browne, M. J., and J. T. Schmit, 2008, Litigation Patterns in Automobile Bodily Injury Claims 1977–1997: Effects of Time andTort Reforms, Journal of Risk and Insurance, 75: 83-100.

Cox, D. R., and D. Oakes, 1984, Analysis of Survival Data (London: Chapman & Hall).

Deng, Y. and G. Zanjani, Supplementary Material to ‘What Drives Tort Reform Legislation? An Analyisis of State Decisions to Restrict Liability Torts’, Journal of Risk and Insurance. https://doi.org/10.1111/jori.12186

WHAT DRIVES TORT REFORM LEGISLATION? 989

http://www.atra.org/legislation/states

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=902711

https://doi.org/10.1111/jori.12186

Dixit, A., 1996, The Making of Economic Policy: A Transactions-Cost Politics Perspective (Cambridge, MA: MIT Press).

Eisenberg, T., M. Heise, N. L. Waters, and M. T. Wells, 2010, The Decision to Award Punitive Damages: An Empirical Study, Journal of Legal Analysis, 2: 577-620.

Grace, E. V., 1990, Property-Liability Insurer Reserve Errors: A Theoretical and Empirical Analysis, Journal of Risk and Insurance, 57(1): 28-46.

Grace, M., and J. T. Leverty, 2013, How Tort Reform Affects Insurance Markets, Journal of Law, Economics, and Organization, 29: 1253-1278.

Harrington, S. E., 1994, StateDecisions to Limit Tort Liability:AnEmpiricalAnalysis of No-Fault Automobile Insurance Laws, Journal of Risk and Insurance, 61: 276-294.

Heaton, P., How Does Tort Law Affect Consumer Auto Insurance Costs? Journal of Risk and Insurance, Advance online publication. doi:10.1111/jori.12095

Hyman, D. A., B. Black, C. Silver, and W. M. Sage, 2009, Estimating the Effect of Damages Caps in Medical Malpractice Cases: Evidence From Texas, Journal of Legal Analysis, 1: 355-409.

Irwin, D., and R. S. Kroszner, 1999, Interests, Institutions, and Ideology in Securing Policy Change: The Republican Conversion to Trade Liberalization After Smoot- Hawley, Journal of Law and Economics, 42: 643-673.

Jenkins, S. P., 1995, Easy Estimation Methods for Discrete-Time Duration Models, Oxford Bulletin of Economics and Statistics, 57: 129-138.

Joskow, P., and R. Noll, 1981, Regulation in Theory and Practice: An Overview, in: G. Frommed, ed., Studies in Public Regulation (Cambridge, MA: MIT Press), pp. 1-65.

Karl, J. B., P.H. Born, andW.K.Viscusi, 2016, TheRelationship Between theMarket for Health Insurance and Medical Malpractice Insurance, Applied Economics, 48: 1-16.

Kessler, D. P., and M. B. McClellan, 1996, Do Doctors Practice Defensive Medicine? Quarterly Journal of Economics, 111: 353-390.

Kessler, D. P., W. M. Sage, and D. Becker, 2005, Impact of Malpractice Reforms on the Supply of Physician Services, Journal of the American Medical Association, 293(21): 2618-2625.

Kroszner, R. S., and P. E. Strahan, 1999, What Drives Deregulation? Economic and Politics of the Relaxation of Bank Branching Restrictions, Quarterly Journal of Economics, 114: 1437-1469.

Lee, H., M. J. Browne, and J. T. Schmit, 1994, HowDoes Joint and Several Tort Reform Affect the Rate of Tort Filings? Evidence From the State Courts, Journal of Risk and Insurance, 61: 295-316.

Matsa, D. A., 2007, DoesMalpractice Liability Keep the Doctor Away? Evidence From Tort Reform Damage Caps, Journal of Legal Studies, 36: 143-182.

McMahon-Howard, J., J. Clay-Warner, and L. Renzulli, 2009, Criminalizing Spousal Rape: The Diffusion of Legal Reforms, Sociological Perspectives, 52(4): 505-531.

Paik, M., B. Black, and D. Hyman, 2013, The Receding Tide of Medical Malpractice Litigation Part 2: Effect ofDamageCaps, Journal of Empirical Legal Studies, 10: 639-670.

990 THE JOURNAL OF RISK AND INSURANCE

Peltzman, S., 1989, Toward a More General Theory of Regulation, Journal of Law and Economics, 19: 109-148.

Poole, K., and H. Rosenthal, 1997, Congress: A Political-Economic History of Roll Call Voting (Oxford: Oxford University Press).

Priest, G. L., 1987, The Current Insurance Crisis and Modern Tort Law, Yale Law Journal, 96: 1521-1590.

Romano, R., 2006, The States as a Laboratory: Legal Innovation and State Competition for Corporate Charters, Yale Journal on Regulation, 23: 209-247.

Rubin, P. H., 2005, Public Choice and Tort Reform, Public Choice 124: 223-236.

Sloan, F. A., P. M. Mergenhagen, and R. R. Bovbjerg, 1989, Effects of Tort Reforms on the Value of Closed Medical Malpractice Claims: A Microanalysis, Journal of Health Politics, Policy, and Law, 14(4): 663-689.

Smythe, D. J., and R. C. Bird, 2008, The Structure of American Legal Institutions and the Diffusion of Wrongful-Discharge Laws, 1978–1999, Law & Society Review, 42(4): 833-864.

Soule, S.A., and J. Earl, 2001,TheEnactment of State-LevelHateCrimeLaw in theUnited States: Intrastate and Interstate Characteristics, Sociological Perspectives, 44(3): 281-305.

Steele, F., 2008, Multilevel Models for Longitudinal Data, Journal of the Royal Statistical Society: Series A (Statistics in Society), 171(1): 5-19.

Stigler, G. J., 1971, The Theory of Economic Regulation, Bell Journal of Economics and Management Science, 2: 3-21.

Sunstein, C. R., 2008, Illusory Losses, Journal of Legal Studies, 37: S157-S194.

Twining, W., 2005, Social Science and Diffusion of Law, Journal of Law and Society, 32: 203-240.

Viscusi, W. K., 2004, The Blockbuster Punitive Damage Awards, Discussion Paper, No. 473, Harvard Law School, John M. Olin Center for Law, Economics, and Business.

Viscusi, W. K., and P. H. Born, 1995, The General-Liability Reform Experiments and the Distribution of Insurance-Market Outcomes, Journal of Business & Economic Statistics, 13: 183-198.

Viscusi,W.K., andP.H. Born, 2005, DamagesCaps, Insurability, and the Performance of Medical Malpractice Insurance, Journal of Risk and Insurance, 72: 23-43.

Viscusi,W. K., R. J. Zeckhauser, P. H. Born, andG. Blackmon, 1993, The Effect of 1980s Tort Reform Legislation on General Liability and Medical Malpractice Insurance, Journal of Risk and Uncertainty, 6: 165-186.

Zuckerman, S, R. R. Bovbjerg, and F. A. Sloan, 1990, Effect of Tort Reforms and Other Factors on Medical Malpractice Insurance Premiums, Inquiry, 27: 167-182.

SUPPORTING INFORMATION

Additional supporting information may be found in the online version of this article at the publisher’s website.

WHAT DRIVES TORT REFORM LEGISLATION? 991

Copyright of Journal of Risk & Insurance is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.