Sorokina 2016

G Model
JEB-5749; No. of Pages 20 ARTICLE IN PRESS

Journal of Economics and Business xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Journal of Economics and Business
Reactions of equity markets to recent financial

reforms夽
Nonna Sorokina a,∗, John H. Thornton Jr. b,1
a
Wake Forest University, United States
b
Kent State University, United States
a r t i c l e i n f o a b s t r a c t
Article history: We conduct event studies of broad equity market reaction to the
Received 31 May 2015 events surrounding introduction and enactment of recent financial
Received in revised form 1 May 2016 reform initiatives. In response to the introduction of the Dodd-
Accepted 4 May 2016
Frank Act, financial firms and firms from a few other industries
Available online xxx
experience a statistically significant decrease in systematic risk,
while a substantial number of industries, representing a broad
JEL classification:
cross-section of the economy, experiences a statistically significant
G14
increase in systematic risk. The systematic risk in some industries
G21
G28 does not change. The increase in risk is concentrated in industries
in which firms are dependent on external capital. The initial market
Keywords: reaction to Dodd-Frank indicates that it may lower the risk in finan-
Dodd-Frank Act cial firms, but the risk for many non-financial firms simultaneously
Systematic risk
increases.
Regulation
© 2016 Elsevier Inc. All rights reserved.
Event study
Dependence on external capital
夽
We are grateful for the very useful comments and suggestions received from Allen Berger, David Hillier, Kenneth Kopecky,
Philip Strahan, two anonymous referees, and the sessions’ participants at the FMA 2013 European Conference, IFABS 2013
Conference, FMA 2013 International Conference, and MFA 2012 Conference. All errors and omissions remain our own.
∗ Corresponding author. Tel.: +1 336 758 6177.
E-mail addresses: sorokiny@wfu.edu (N. Sorokina), jthornt5@kent.edu (J.H. Thornton Jr.).
1
Tel.: +1 330 672 1214.
http://dx.doi.org/10.1016/j.jeconbus.2016.05.001
0148-6195/© 2016 Elsevier Inc. All rights reserved.
Please cite this article in press as: Sorokina, N., & Thornton Jr., J.H. Reactions of
equity markets to recent financial reforms. Journal of Economics and Business (2016),
G Model
2 N. Sorokina, J.H. Thornton Jr. / Journal of Economics and Business xxx (2016) xxx–xxx
1. Introduction
The financial crisis of 2007–2009 demonstrates that problems in the financial system can have a
deleterious impact on the entire economy. Since problems in the financial system extend far beyond
the system itself, financial reform has become an urgent priority. In the United States, the most impor-
tant financial reform arising from the financial crisis is the Dodd-Frank Wall Street and Consumer
Protection Act of 2010 (Dodd-Frank), the most fundamental overhaul of the domestic financial system
since the 1930s.
Given the scope of the changes stemming from Dodd-Frank, it is likely that the effects of the act
will be felt far beyond the financial system. In particular, firms that depend on external financing
will be disproportionately affected. Moreover, market participants should anticipate the effects as
legislative and regulatory actions are being developed and should adjust market prices accordingly.
To test this hypothesis, we conduct event studies of the reactions of the stock market in the United
States to the events surrounding introduction, legislative process and enactment of the Dodd-Frank
Act. In contrast to many event studies of financial regulations, we examine the effects of Dodd-Frank
across the entire economy, not just financial firms. We do this by conducting an event study for each of
the 49 Fama-French industry indices. Our research design allows us to test for changes in systematic
risk as measured by market beta, as well as measure cumulative abnormal returns associated with
important milestones in the legislative process of regulatory reform. We find a statistically significant
decrease in systematic risk for seven industries, including financial services (banking, insurance, and
trading), real estate and construction. More interestingly, fourteen industries experience a statistically
significant increase in systematic risk at the introduction of Dodd-Frank.1
To further explore different industry reactions, we obtain a sample of firm-level data during the
period 2008–2010 from COMPUSTAT. We divide the sample into three groups: (1) firms in industries
that experienced a statistically significant increase in systematic risk; (2) firms in industries that
experienced a statistically significant decrease in systematic risk; and (3) firms in industries with
no statistically significant change in systematic risk. We then estimate a multinomial logit model
and pairwise logit model to examine the characteristics of firms in the different groups. We find
that firms in industries with a statistically significant increase in systematic risk are more profitable,
have higher leverage, and have higher book-to-market ratios than firms in industries that did not
experience a statistically significant change in systematic risk. These findings suggest that the increase
in systematic risk is concentrated in industries that are more dependent on external capital. Compared
to industries without a statistically significant change in systematic risk, firms in industries with a
significant decrease in systematic risk are larger and less tangible, have less cash, invest less in R&D and
grow faster. We further find that firms that experienced an increase in systematic risk are significantly
more dependent on external finance than firms in industries in which systematic risk has decreased.
The intent of Dodd-Frank Act was to decrease systemic risk in the financial system (i.e., the risk that
the failure of one or more financial institutions could endanger the financial system). Our tests detect
changes in systematic risk based on stock market returns (i.e., the covariance between the returns
of an individual stock or portfolio of stocks and the returns of the entire market). As we discuss
more fully in Section 2.2, the more stringent capital requirements and other provisions of Dodd-
Frank are likely to make firms that are dependent on external capital more sensitive to economic
conditions, thereby increasing systematic risk. Also, existing studies show that systematic risk is a
valid component of systemic risk; see, for example, Allen and Saunders (2004) or Acharya, Pedersen,
Philippon, and Richardson (2010). Therefore, the result does raise questions about the unintended
1
In an appendix available from the authors on request, we also test the impact of the domestic and international regulatory
and legislative actions beyond the enacting country’s borders. We conduct event studies of reactions of the broad stock market
indices in 21 of the world’s largest economies to the events surrounding introduction, legislative process and enactment of
Dodd-Frank, Basel III and UK Financial Reform. Our results indicate that broad stock markets do react to significant regulatory
changes. Moreover, for country-specific reforms, we find that the effects extend beyond the borders of the country implementing
the reform. The spillover across borders does not affect all the countries in our study. Countries in closer geographic proximity
to the enacting country experience larger and more statistically significant effects.
G Model
N. Sorokina, J.H. Thornton Jr. / Journal of Economics and Business xxx (2016) xxx–xxx 3
effects of Dodd-Frank on the broad economy in general and on firms dependent on external capital in
particular.
It is important to note that when the study was conducted, financial reform was fairly recent.
As a result, we are only able to perform a short-term event study that essentially measures market
expectations incorporated into equity prices rather than the real economic impact (Ang & Zhang,
2004; Bremer, Buchanan, & English, 2011; Lamdin, 2001; Neale & Peterson, 2005). The global nature
of the financial reforms implies delayed real economic reaction. In the case of the Gramm-Leach-
Bliley Act of 1999 (GLBA), which aimed to deregulate the financial system in the United States by
repealing restrictions imposed by the depression-era Glass-Steagall Act of 1933 and the Bank-Holding
Company Act of 1956, the companies arguably could take advantage of the new rules right away. The
recent financial reforms are so massive that their concepts will not be practically implemented for a
prolonged period of time.
The remainder of this paper is structured as follows. In Section 2, we provide a brief review of the
relevant literature and develop our hypotheses. Section 3 presents our event study model and esti-
mation methodology and describes the data. Section 4 presents our results and discusses robustness
checks we performed. Section 5 concludes the paper.
2. Literature review and hypotheses
2.1. Literature review
While there is a large body of literature concerning Dodd-Frank, we are aware of only two papers
that implement event studies. Gao, Liao, and Wang (2013) examine the market reaction of stocks and
bonds of systematically important financial institutions. They find a negative stock market reaction,
but a positive bond market reaction. Fier and Liebenberg (2013) examine the market reaction in the
insurance industry. They conclude that market participants viewed Dodd-Frank as negative for the
insurance industry.
Many event studies of financial regulatory actions focus only on financial firms. GLBA may be the
most heavily studied of recent regulatory acts (e.g., Akhigbe & Whyte, 2004; Mamun, Hassan, & Lai,
2004; Mamun, Hassan, & Maroney, 2005; Neale & Peterson, 2005; Yildirim, Kwag, & Collins, 2006).
Most of these studies find a reduction in risk for financial institutions around the adoption of GLBA.
A few studies examine the impact of events in the financial sector on the broader economy. Several
papers on the impact of GLBA (e.g., Carow & Kane, 2002) find a negative effect of GLBA on banks’
customers. Their findings raise the question of whether the gain of financial institutions as a result of
the act was a net gain for the economy or simply wealth redistribution toward the financial sector.
The study of the impact of the Japanese banking crises on non-financial companies (Miyajima & Yafeh,
2007) raises the issue of spillover impact of a financial system-related event on non-financial com-
panies. They find a significant reaction in small, leveraged, low-tech, low book-to-market and poor
credit standing companies.
Chava and Purnanandam (2011) study the impact of the financial crisis precipitated by the Russian
sovereign debt default in 1988. They find that a crisis in the financial system has an adverse effect on
the well-being of bank-dependent non-financial firms. Since the 1988 financial crisis was not accom-
panied by a general economic downturn, Chava and Purnanandam claim their findings show a causal
relationship between the financial crisis and adverse effects on non-financial firms.
2.2. Hypotheses
Rajan and Zingales (1998) demonstrate that availability of the resources and services provided by
the financial system affects growth of industries that depend on external financing. Therefore, shocks
to the financial system have effects that are felt well beyond the banking industry and are particularly
pronounced in firms that are dependent on external financing (e.g., Chava & Purnanandam, 2011).
Provisions of the Dodd-Frank Act are likely to affect the cost and availability of external financing
for non-financial firms. The most obvious provision is the requirement for increased capital ratios.
One way for a bank to increase its capital ratios is to shrink (or at least slow the growth of) its loan
G Model
portfolio, thereby limiting bank credit (Ivashina & Scharfstein, 2010). Additionally, Baker and Wurgler
(2015) argue that an increase in bank capital holdings raises the risk premium on their assets and
significantly increases rates for their borrowers. According to DeAngelo and Stulz (2015), the role of
banks as liquidity providers is essential for the broad range of entities in the economy. Banks create
liquid financial claims sourced from leverage. Therefore, limiting their leverage reduces availability
of liquidity for financially constrained high-equity firms and households that rely on the financial
system for their liquidity needs. As a result of increased capital ratios, banks may be less willing to
provide liquidity to financial markets, particularly bond markets, indirectly affecting liquidity needs
of the non-financial sector. Finally, the Volker Rule severely limits banks’ ability to trade on their own
account and could cause financial institutions to reduce liquidity provision in multiple markets.
Risk-based bank capital ratios are more likely to be binding during downturns, as substantial
research on the cyclical effect of the capital requirements shows. For example, according to Repullo’s
(2013) model, when risk-based capital requirements are not lowered in the presence of a negative
shock to the economy, banks remain safer, but choose to reduce aggregate investment significantly.
This implies that firms dependent on external financing would be disproportionately affected during
economic downturns, making them more sensitive to economic conditions and therefore increasing
their systematic risk by definition. As a result, an unintended consequence of Dodd-Frank could be
an increase in the systematic risk of external finance dependent firms. In contrast, by strengthening
capital ratios and restricting some risky activities such as proprietary trading, Dodd-Frank could make
financial institutions less sensitive to economic conditions and lower their systematic risk. Although
most provisions of Dodd-Frank required months or years to be fully implemented, it is likely that
market participants anticipated some of these effects as the legislation process evolved.
Based on this discussion, we develop the following hypotheses.
H1. The systematic risk of financial firms decreased around the adoption of the Dodd-Frank Act.
H2. The systematic risk of non-financial firms dependent on external financing increased around the
adoption of the Dodd-Frank Act.
3. Methodology and data
3.1. Event study methodology
In this study, we implement an extended version of the market model to test the stock market’s
reaction to regulatory reforms.2 This type of models has become increasingly popular in studies of
regulatory actions (e.g., Akhigbe & Whyte, 2004; Mamun et al., 2004; Neale & Peterson, 2005). The
model includes dummy variables as event identifiers and captures both cumulative abnormal returns
and changes in systematic risk in response to the event. We extend previously developed methodology
by measuring the change in systematic risk both when legislation is introduced and when it is enacted.
The adoption of the new variables that measure market reaction at the introduction of the legislation
helps mitigating the effect of information dispersion over the period of time preceding legislation
enactment that usually complicates event studies of the regulation.3 The extended market models,
such as Mamun et al. (2004), employ risk free rate and forex market return as control variables, in
addition to return on equity market. These additional variables allow to control for noise specific to
financial services, and allows obtaining cleaner indication of the event effect.
Our model is defined as follows:
Rit = ˛ + ˛1i INTRO + ˛2i PASS + ˇ1i (INTRO ∗ Rmt ) + ˇ2i (PASS ∗ Rmt ) + ˇ3i EVENT + ˇ4i Rmt
+ ˇ5i Rfxt + ˇ6i Rft + εit (1)
2
See Binder (1998) for a very nice and clear description of the market model based on beta and dummy variables identifying
cumulative abnormal returns associated with the event. Binder cites regulatory studies as a case of portfolios of similarly
affected stocks.
3
See Binder (1998) and Lamdin (2001) for a discussion of issues related to event studies of regulatory actions.
G Model
Table 1
Major news of the Dodd-Frank Act legislative process. In this table we present the list of
the most significant public information releases identified from various media sources.
The events most likely have driven the investors’ reaction to the progress of Dodd-Frank
Act through the legislative process and could be reflected in the equity returns.
Event date Event description
6/17/2009 Initial proposal by president Obama

12/2/2009 Law proposed in congress
12/11/2009 House bill passed
1/21/2010 Obama proposed Volker’s rule
4/29/2010 Shareholders proxy amendment by Dodd
5/13/2010 Interchange fees amendment by Durbin
5/20/2010 Senate passed
6/25/2010 Conference reconciliation finished
6/30/2010 Conference report
7/15/2010 Senate passed
7/21/2010 Signed into Law
Rit daily return on industry equity;

INTRO before/after the Dodd-Frank Act introduction dummy (0 – before, 1 – after);
PASS before/after the Dodd-Frank Act enactment dummy (0 – before, 1 – after);
EVENT dummy variable of the event windows (1 – during a window, 0 – otherwise);
Rmt market return variable;
Rfxt return on forex market variable;
Rft risk-free rate variable.
We use data from 120 days before the beginning of the first event window (1 day before the initial
proposal of the legislation) to 120 days after the end of the last event window (1 day after the bill was
signed into Law)). The full list of events is presented in the Table 1. Fig. 1 presents the structure of the
event period. The 120 day limit was initially imposed by data availability, but we note that McKinley
(1997) cites 120 trading days as commonly being implemented in event studies. For our reported
results, we use a (−1, +1) day event window for important events during the legislative process. We
also ran our tests with (−3, +3) and (−5, +5) event windows (not tabulated) with similar results.
In the model (1), the components should be interpreted as follows: ˛ – intercept (market model
alpha); ˛1i – coefficient, associated with the Dodd-Frank introduction dummy, captures difference
between industry alpha before/after the Dodd-Frank Act was introduced; ˛2i – coefficient, associated
with the Dodd-Frank enactment dummy, captures difference between industry alpha before/after the
Dodd-Frank Act enacted ˇ1i – coefficient associated with interaction of the Dodd-Frank introduction
dummy and market return, captures change in market model beta after the Dodd-Frank Act was
Introduced Event1 Event 2 … Event n … Enacted
t
120 days 120 days
Event windows (-1;+1) (-3;+3) (-5;+5)
Event Period
Fig. 1. Event period structure.
G Model
Introduced Legislative process Enacted

t
Change in Cumulative abnormal Change in

alpha or beta return on index alpha or beta
Fig. 2. Timing the event effect.
introduced; ˇ2i – coefficient associated with interaction of the Dodd-Frank enactment dummy and
market return, captures change in market model beta after the Dodd-Frank Act was enacted; ˇ3i
– coefficient associated with dummy variable of event window that captures cumulative abnormal
returns; ˇ4i , ˇ5i , and ˇ6i are coefficients of control variables; εit – error term that does not contain
abnormal returns associated with the event.
With all components of enhanced market model, dummy variables to measure change in both
systematic risk (beta), and abnormal returns (alpha) at the points of time when legislation is introduced
and enacted, and cumulative abnormal returns during the key points of the legislative process, we are
able not only to capture the full spectrum of the investors’ response, but also to time their response.
Fig. 2 demonstrates how the three stages of the event effect are distinctly identified in the model. The
change in risk as a result of the legislation’s introduction comes first, the cumulative abnormal returns
in response to the key events in the legislation process demonstrate the second stage reaction, and,
finally, change in risk at the legislation enactment, concludes.
We use OLS with heteroscedasticity-consistent standard errors (White, 1980) for the main infer-
ences from our tests. Beginning with Binder (1985), several studies use the seemingly unrelated
regression (SUR) framework of Zellner (1963) in regulatory event studies.4 However, if the regres-
sors are the same in all equations of a system of equations, then the results from OLS, run on an
equation-by-equation basis, are identical to the results obtained using SUR (Wooldridge, 2002 The-
orem 7.6, p. 169).5 Moreover, Brown and Warner (1985), based on their simulation results, find that
in many cases, OLS works as well as more complex regression models. Even though we believe that
OLS is appropriate for our sample, for robustness, we also ran our tests using SUR. The results (not
tabulated) are qualitatively similar to the reported results.
Henderson (1990) supports the power of non-parametric tests over parametric in event studies.6
For robustness, we use a non-parametric regressions based on Huber’s (1973) and Yohai’s (1987)
methods to support our results (not tabulated).
As an additional robustness test for the change in systematic risk identified by the dummy variables
in specification (1), we run the Quandt-Andrews test for breakpoint in relationship between two
variables when the timing of the break is unknown, as introduced by Andrews (1993). We run the test
as a series of Chow tests (Chow (1960)), sequentially setting breakpoint on each day in the period. The
most likely breakpoint is identified by selecting a statistically significant breakpoint with the greatest
value of F-statistic within the period of interest.
3.2. Methodology to examine the determinants of industry risk changes
As we discuss in Section 4, our most interesting finding in the industry event studies is that a
substantial number of industries experiences a statistically significant increase in systematic risk at
4
Example studies using the SUR framework include Carow and Heron (2002), Mamun et al. (2004) and Mamun et al. (2005).
5
Another problem with using the SUR framework is that in SUR, all countries must have a reported return on a given day for
that day to be included in the estimation. Since national holidays vary from country to country (e.g., Thanksgiving in the United
States is not a holiday in other countries), observations are lost in a SUR estimation.
6
Cowan (1992) performs simulation-based tests of the generalized sign test and the ranked test and finds them both useful in
event studies under the different circumstances. MacKinlay (1997) also mentions these tests as efficient tools in event studies.
G Model
the introduction of the Dodd-Frank legislation, while only a few experience a statistically signifi-
cant decrease in systematic risk. To further investigate these results, we obtain a sample of firms from
COMPUSTAT and estimate a pooled multinomial logit regression model to identify the influence of the
various firm-specific factors on the type of response to Dodd-Frank introduction (positive, negative or
none). The response variable is not ordered, so we use a regular multinomial model upon the recom-
mendations in Johnsen and Melicher (1994). Based on the example of firms’ financial health, Johnsen
and Melicher demonstrate that expanding the outcome space from standard binomial (two-outcome)
to multinomial (three outcomes) helps reduce misclassification error significantly. According to their
study, the refined classification carries important economic information regarding firms’ financial
condition. Around the time of Johnsen and Melicher’s study, multinomial models quickly gained
popularity in corporate finance literature. To cite just a few of the numerous multinomial logit appli-
cations, Lawrence and Arshadi (1995) use a similar approach to analyze problem loans resolution in
banking, Helwege and Liang (1996) implement multinomial logit in their study of capital structure
choices, Martin (1996) uses multinomial logit in his exploration of acquisition types, and Jagannathan,
Stephens, and Weisbach (2000) use a multinomial logit model for non-ordered multiple choices in
payout policy analysis.
For multinomial logit to be appropriate, the application must satisfy certain assumptions. We
believe that there is not an independence of irrelevant alternatives problem by definition. We might
be slightly concerned with the attribution of the effect in industries that did not produce a statisti-
cally significant change in systematic risk (since there was some change in risk, although statistically
insignificant) that might relate those industries to the negative/positive change group. To alleviate any
concerns related to the estimation quality, we assess the classification in a pairwise design in specifi-
cation (4). The underliying dataset combines quarterly firm level data for three years 2008–2010. We
measure the effect of the changes in the variables from 2008 to 2009 to 2010, incorporating dummy
variables for years 2009 and 2010 into the regression equation.
Yit = ˛ + ˇ1 ATit + ˇ2 ROAit + ˇ3 Cashit + ˇ4 CapExit + ˇ5 Tangit + ˇ6 Debtit + ˇ7 Growthit
+ ˇ8 R&Dit + ˇ9 BMit + ˇ10 2009 + ˇ11 2010 + εit (2)
Y firm level categorical variable, where -1 indicates that a firm belongs to an industry with a
significant negative change in systematic risk; 0 is an indication of a firm from an industry
with no significant change in risk; +1 identifies a firm associated with an industry with a
significant and positive change in systematic risk;
AT firm size;
ROA profitability;
Cash cash holdings;
CapEx capital intensity;
Tang tangibility;
Debt debt to assets ratio;
Growth actual growth;
R&D growth opportunities;
BM book-to-market;
2009 a dummy variable set to 1 for all records of the year 2009, and to 0 otherwise;
2010 a dummy variable set to one for all records of the year 2010, and to 0 otherwise.
The details of the variable construction are provided in Table 3. The firm characteristics selection
is warranted by the research of Miyajima and Yafeh (2007), which highlights the characteristics of
the non-financial firms most affected by the banking crises, and components of the external finance
dependence variable of Rajan and Zingales (1998).
Furthermore, we implement a similar logit model based on changes in the independent variables
between year 2010 and year 2008 as represented by the vector i to identify any impact of the changes
G Model
Table 2
Change in U.S. industries’ systematic risk in response to the Introduction of Dodd-Frank.
Panel A. Change in systematic risk at the legislation introduction.

This panel presents the change in systematic risk in U.S. industries in response to the introduction of the Dodd-Frank Act.
The change was captured as beta prime in the following model:
Rit = ˛ + ˛1i INTRO + ˛2i PASS + ˇ1i (INTRO * Rmt ) + ˇ2i (PASS * Rmt ) + ˇ3i EVENT + ˇ4i Rmt + ˇ5i Rfxt + ˇ6i Rft + εit
Rit – daily return on industry equity; INTRO – before/after the Dodd-Frank Act introduction dummy (0 – before, 1 – after);
PASS – before/after the Dodd-Frank Act enactment dummy (0 – before, 1 – after); EVENT – dummy variable of the event
windows (1 – during a window, 0 – otherwise); Rmt – market return variable; Rfxt – return on forex market variable; Rft –
risk-free rate variable (see “Methodology and Data” for greater details on variables and models). We use 49 Fama-French
industries and report industry numbers, letter codes and name as identifiers. We also report systematic risk before
introduction of Dodd-Frank Act (ˇ4i ) and change in systematic risk in response to introduction of the Dodd-Frank Act
(ˇ1i ). We code an increase in risk as follows: positive, statistically significant coefficient ˇ1i (increase in risk) as “1”,
negative, statistically significant coefficient ˇ1i (decrease in risk) as “-1”. Detailed results are not reported for the
industries with no statistically significant change in risk.
Industry No Industry Code Industry Name Change ˇ4i ˇ1i p-value

in Risk of ˇ1i
5 Smoke Tobacco Products 1 0.2356 0.3797 0

11 Hlth Healthcare 1 0.6955 0.1280 0.027
12 MedEq Medical Equipment 1 0.6122 0.1399 0.0076
13 Drugs Pharmaceutical Products 1 0.4805 0.1463 0.0005
14 Chems Chemicals 1 1.1430 0.1118 0.0194
18 Cnstr Construction −1 1.5534 −0.1441 0.0428
20 FabPr Fabricated Products −1 1.6541 −0.2058 0.0072
24 Aero Aircraft 1 0.9899 0.1110 0.0368
25 Ships Shipbuilding, Railroad Equipment 1 0.9336 0.6933 0
27 Gold Precious Metals 1 0.2153 0.6799 0
28 Mines Non-Metallic and Industrial Metal Mining 1 1.4623 0.1790 0.0418
31 Util Utilities 1 0.6016 0.1696 0
32 Telcm Communication −1 0.9372 −0.0935 0.0144
33 PerSv Personal Services 1 0.5704 0.3245 0.0002
34 BusSv Business Services 1 0.9435 0.0673 0.017
42 Whlsl Wholesale 1 0.8411 0.0824 0.0157
43 Rtail Retail 1 0.7078 0.0848 0.0487
45 Banks Banking −1 2.1680 −0.8140 0
46 Insur Insurance −1 1.4201 −0.2601 0
47 RlEst Real Estate −1 1.9816 −0.2912 0.0005
48 Fin Trading −1 1.7628 −0.5288 0
The following industries did not experience a statistically significant change in risk: Agriculture (1), Food Products (2),
Candy & Soda (3), Beer & Liquor (4), Recreation (6), Entertainment (7), Printing and Publishing (8), Consumer Goods (9),
Apparel (10), Rubber and Plastic Products (15), Textiles (16), Construction Materials (17), Steel Works Etc (19), Machinery
(21), Electrical Equipment (22), Automobiles and Trucks (23), Defense (26), Coal (29), Petroleum and Natural Gas (30),
Computer Hardware (35), Computer Software (36), Electronic Equipment (37), Measuring and Control Equipment (38),
Business Supplies (39), Shipping Containers (40), Transportation (41), Restaurants, Hotels, Motels (44), Other (49)
Panel B. Robustness check: Breakpoint identification with Quandt-Andrews test.

In the panel A, the date of structural break is forced to fall on the date of the legislation introduction. In this panel, we
present results of the robustness test, where the breakpoint is allowed to vary. We run Quandt-Andrews test as a series of
Chow tests, sequentially setting the breakpoint to each day in the sample. We select the most likely breakpoint within
3-months period preceding the formal introduction of the Dood-Frank Act (6/17/2009) by the highest value of F-statistic
associated with the breakpoint. We also report corresponding p-values to demonstrate statistical significance of the
identified breakpoints.
Industry No Industry Code Industry Name Most Likely Breakpoint Breakpoint
Breakpoint F-Stat p-value
5 Smoke Tobacco Products 05/22/2009 *** 23.53 <0.0001

11 Hlth Healthcare 05/08/2009 ** 3.38 0.0348
12 MedEq Medical Equipment 06/04/2009 *** 7.19 0.00086
13 Drugs Pharmaceutical Products 06/08/2009 *** 10.99 <0.0001
14 Chems Chemicals 03/30/2009 ** 3.98 0.0192
18 Cnstr Construction 04/24/2009 *** 4.53 0.0112
20 FabPr Fabricated Products 05/22/2009 *** 4.61 0.0103
24 Aero Aircraft 03/23/2009 *** 7.09 0.0009
G Model
Table 2 (Continued)
Panel B. Robustness check: Breakpoint identification with Quandt-Andrews test.

In the panel A, the date of structural break is forced to fall on the date of the legislation introduction. In this panel, we
present results of the robustness test, where the breakpoint is allowed to vary. We run Quandt-Andrews test as a series of
Chow tests, sequentially setting the breakpoint to each day in the sample. We select the most likely breakpoint within
3-months period preceding the formal introduction of the Dood-Frank Act (6/17/2009) by the highest value of F-statistic
associated with the breakpoint. We also report corresponding p-values to demonstrate statistical significance of the
identified breakpoints.
Industry No Industry Code Industry Name Most Likely Breakpoint Breakpoint
Breakpoint F-Stat p-value
25 Ships Shipbuilding, Railroad Equipment 06/17/2009 *** 42.54 <0.0001

27 Gold Precious Metals 04/20/2009 *** 16.52 <0.0001
28 Mines Non-Metallic and Industrial Metal Mining 04/15/2009 *** 9.66 <0.0001
31 Util Utilities 05/15/2009 *** 10.49 <0.0001
32 Telcm Communication 06/04/2009 *** 7.52 0.0006
33 PerSv Personal Services 05/21/2009 *** 9.51 <0.0001
34 BusSv Business Services 05/08/2009 ** 4.13 0.0166
42 Whlsl Wholesale 05/06/2009 *** 6.68 0.0014
43 Rtail Retail 05/08/2009 ** 3.49 0.0311
45 Banks Banking 05/19/2009 *** 44.44 <0.0001
46 Insur Insurance 03/24/2009 *** 18.06 <0.0001
47 RlEst Real Estate 06/05/2009 *** 14.53 <0.0001
48 Fin Trading 05/15/2009 *** 53.1 <0.0001
in the firm’s characteristics on the systematic risk over the period of study.
Yit = ˛ + ˇ1 ATit + ˇ2 ROAit + ˇ3 Cashit + ˇ4 CapExit + ˇ5 Tangit + ˇ6 Debtit
+ ˇ7 Growthit + ˇ8 R&Dit + ˇ9 BMit + εit (3)
Yit firm level categorical variable, where -1 indicates that a firm belongs to an industry with a
significant negative change in systematic risk; 0 is an indication of a firm from an industry
with no significant change in risk; +1 identifies a firm associated with an industry with a
significant and positive change in systematic risk;
AT change in firm size between 2008 and 2010;
ROA change in profitability between 2008 and 2010;
Cash change in cash holdings between 2008 and 2010;
CapEx change in capital intensity between 2008 and 2010;
Tang change in tangibility between 2008 and 2010;
Debt change in debt to assets ratio between 2008 and 2010;
Growth change in actual growth between 2008 and 2010;
R & D change in growth opportunities between 2008 and 2010;
BM change in book-to-market between 2008 and 2010;
Our independent variables are computed for all firms in the industry, and we anticipate correlations
in the firm-level independent variables between firms within the same industry. Cameron and Miller
(2015) provide an excellent explanation of the potential issues and suggest using cluster-robust infer-
ences. Following Cameron and Miller (2015), we implement the model with robust clustered standard
errors that should shed light on any potential discrepancies in estimation. We ran the regressions on
year 2009 data (the year when reaction of the industries to Dodd-Frank has been captured), split-
ting our sample pairwise into three subsamples, and compare the positive reaction group with the
no reaction group; the negative reaction group with the no reaction group; and the positive reaction
group with the negative reaction group. In addition, this model provides conveniently interpretable
coefficients that support the results of multinomial model.
G Model
Yit = ˛ + ˇ1 ATit + ˇ2 ROAit + ˇ3 Cashit + ˇ4 CapExit + ˇ5 Tangit + ˇ6 Debtit + ˇ7 Growthit
+ ˇ8 R&Dit + ˇ9 BMit + ˇ10 2009 + ˇ11 2010 + εit (4)
Yit firm level binary variable, where in the first subsample, 1 indicates that a firm belongs to
an industry with a significant negative change in systematic risk; 0 is an indication of a firm
from an industry with no significant change in risk; in the second subsample, 1 is assigned to
firms from industries with a significant positive change in systematic risk, 0 is an indication
of a firm from an industry without change in risk, and, finally, in the third subsample, the
value of 1 is set for firms from industries with a significant increase in systematic risk, while
0 shows that a firm belongs to an industry with a decrease in systematic risk.
Independent variables are the same as in model specification (2).
3.3. Industry data
For the industry analysis of the changes in systematic risk and abnormal returns in response to
the significant events in the legislative process, we use daily returns of the Fama-French 49 indus-
tries obtained from Kenneth French’s website. For the market index, we use the Fama-French market
factor.7 For additional control variables, we use the 6 month LIBOR rate collected from the website of
the mortgage information service Mortgage-X and the Major Currency Index from the Federal Reserve.
To investigate the determinants of industry reactions to Dodd-Frank, we get firm-specific character-
istics for years 2008–2010 from COMPUSTAT. Our sample includes 9266 unique firms. Table 4 shows
summary statistics of the firm-specific data. The average size of the firm is $11.5 billion, while median
firm has $367.36 million in assets. For a median firm in our sample in 2008–2010, ROA is 0.07, cash
to assets ratio is 0.08, capital expenditures reach 0.02 of total assets, tangibility is measured at 0.14,
debt to assets ratio is 0.15, research and development is 0 and book-to-market is 0.5.
3.4. Financial reform
We collect information on the major public news related to the legislative process of the Dodd-
Frank Act from the websites of the Library of Congress, Wall Street Journal, and Law Librarians’ Society
of Washington. Table 1 presents a list of the identified events.
4. Results
4.1. Effects of Dodd-Frank on industry indices
We begin by estimating our event study model (specification (1)) on the 49 Fama-French industries.
The statistically significant results for most industries are on the coefficient beta prime, the change
in systematic risk at the introduction of Dodd-Frank. To conserve space, in Panel A of Table 2, we
only present the results for beta prime (change in systematic risk) and beta (baseline systematic risk,
controlled for changes in interest rates and forex market). We check robustness of our results in several
ways. For our reported results, we measure returns for intermediate steps in the legislative process
using a three-day event window (−1,+1). On selected samples in our studies, we also ran tests using
(−3,+3) and (−5,+5) windows. The results with these alternative windows are generally similar to the
reported results, although there is a slight loss of statistical significance as the length of the period
increases. This finding is consistent with previous literature, which documents that shorter windows
provide the most powerful test.
7
We are grateful to Kenneth French for making the industry and market factor returns available on his website:
http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/index.html.
G Model
Table 3
Firm-specific variables.We have previously identified that some industries reacted to the intro-
duction of Dodd-Frank Act with increase in systematic risk, risk in some industries decreased
and some industries did not experience neither positive nor negative statistically significant
reaction to the legislation. We decide to look into characteristics of the firms, as a driving force
of the change in systematic risk. We compute several firm-specific measures, presented in Panel
B of this table based on the COMPUSTAT items provided in Panel A.
Panel A: Raw COMPUSTAT items
COMPUSTAT code Description
AT Assets – Total
CH Cash
DLC Debt in Current Liabilities – Total
DLTT Long-Term Debt – Total
PPENT Property, Plant and Equipment – Total (Net)
OIBDP Operating Income Before Depreciation
SALE Sales/Turnover (Net)
CAPX Capital Expenditures
BKVLPS Book Value Per Share
CSHO Common Shares Outstanding
XRD Research and Development Expense
MKVALT Market Value – Total – Fiscal
CAPXY Capital Expenditures
OANCFY Operating Activities – Net Cash Flow
Panel B: Computed firm characteristics
Name Formula
Assets AT
ROA [OIBDP]/[AT]
Cash-to-Assets [CH]/[AT]
CapEx-to-Assets [CAPX]/[AT]
Tangibility [PPENT]/[AT]
Leverage ([DLC] + [DLTT])/[AT]
Growth [SALE(Year t)] − [SALE(Year t − 1)])/[SALE(Year t − 1)])
R&D XRD
Book-to-Market BKVLPS* CSHO/MKVALT
External Financing Dependence (CAPXY-OANCFY)/CAPXY
Table 4
Summary Statistics of the firm-specific characteristics. The table provides summary statistics of the firm-specific characteristics
that may be driving changes in the systematics risk identified in the previous analysis. These characteristics serve as independent
variables in the limited dependent variable models (binary and multinomial logit) in the next step of the research.
Variable N Median Mean Std Dev Minimum Maximum
AT 24,687 367.36 11464.12 105,390 0.001 3,510,975

ROA 24,067 0.07 −0.97 23.92 −2106 182.52
CashAssets 24,286 0.08 0.15 0.20 −0.03 1
CapExAssets 24,466 0.02 0.06 0.76 −0.82 79.71
Tang 24,127 0.14 0.27 0.29 0 1
DebtAssets 24,605 0.15 1.35 37.30 −0.05 3770
Growth 23,939 0.05 1.74 180.23 −199.29 27,660.75
RnD 24,687 0 52.40 422.92 −0.31 10,991
BM 20,259 0.5 −5.77 730.23 −89098.6 15,352.39
ExtFinDep 10,319 0 0.28 0.43 0 1
Since several previous studies use SUR framework for regulatory event studies, we ran our cross-
country analysis using SUR (not tabulated). The results are similar to the reported results. We also
ran our tests using non-parametric methods – robust regressions based on the M-estimator and the
MM-estimator (not tabulated). For a majority of samples, the results are generally consistent with our
findings obtained with OLS. Overall, our results are robust to the changes in regression methodology.
G Model
In Panel B of the same table, we present the result of the additional robustness check (Quandt-
Andrews test) for the set of industries that exhibited statistically significant change in systematic
risk. We are able to identify a statistically significant structural break in the relationship between
industry equity returns and market returns within 3-month period preceding the introduction
of Dodd-Frank Act. The deviation of the breakpoint up to three months prior to the date of
the official introduction is expected because the initial introduction of various pieces of legisla-
tion that eventually were rolled into the Dodd-Frank Act white paper presented on 06/17/2009
began on 03/26/2009, according to the website of the Law Librarians’ Society of Washington,
DC.
We find a statistically significant decrease in systematic risk in seven industries: construction, fab-
ricated products, communication, banking, insurance, real estate, and trading. Notice that most of
these industries are either in financial services (banking, insurance and trading) or closely related to
financial services (construction and real estate). For these industries, equity returns represent a reduc-
tion in systematic risk around the Dodd-Frank introduction time. We observe the greatest reduction
in systematic risk of 0.814 in banking, in which baseline beta for the period was very high at 2.168. The
change brings baseline banking beta in line with other industries in the sector. The decrease of 0.814
means that for the 1% change in market return, equity returns in the banking industry will change
by 1.354% instead of 2.168%. Beta of the trading industry declines from 1.7628 to 1.234. The decrease
in the insurance industry is smaller, but the baseline beta was also not that high. The systematic risk
in real estate and construction industries remains more elevated even after corresponding decreases
in beta of 0.2912 and 0.1441. The changes positively affect volatility in shareholders’ values, leverage
ratios and corporate governance of the firm. The significant decrease in systematic risk in the financial
industries is consistent with Hypothesis 1.
While some industries experienced a decrease in systematic risk, substantially more industries
experienced a statistically significant increase in systematic risk at the introduction of Dodd-Frank.
Industries with an increase in risk come from a broad cross-section of the economy: tobacco prod-
ucts, health care, medical equipment, pharmaceutical products, chemicals, aircraft, shipbuilding and
railroad equipment, precious metals, non-metallic and industrial metal mining, utilities, personal
services, business services, wholesale and retail. The greatest increase in the systematic risk was
observed in shipbuilding and railroad industry, rising from 0.9336 to 1.6269. While industry’s risk
was lower than the risk of diversified portfolio prior to the introduction of the financial reform,
in the view of investors, it became much greater after reform. For every 1% of change in market
return, industry’s equity returns change by 1.6269%, compared with the prior value of 0.9336%.
The smallest increase in systematic risk of 0.0673 was observed in business services industries,
while on average, systematic risk increased by a sizable 0.2244 in the negatively affected indus-
tries.
Dodd-Frank was created, at least in part, to limit systemic risk. The systematic risk changes that we
detect are responsible for part of the market participants’ perception of the effects of Dodd-Frank on
systemic risk in the financial system. The increase in systematic risk in many industries suggests that
market participants anticipated substantial effects on the broad economy. The market participants
could be anticipating greater regulatory uncertainty or increased difficulty in raising external capital.
To gain a better understanding of this question, in the next sub-section, we conduct an exploratory
analysis of the firms’ characteristics in the industries with statistically significant changes in systematic
risk.
4.2. Determinants of systematic risk changes
We use multinomial logistic regression to examine the characteristics of the firms in the industries
with statistically significant changes in systematic risk. We obtain a sample of firms from COMPUSTAT
for the period 2008–2010. Definitions of the variables are shown in Table 3, and summary statistics of
the data is presented in Table 4.
The results of the multinomial logit are presented in Table 5. We use firms that belong to industries
that did not demonstrate a statistically significant reaction to the introduction of Dodd-Frank Act as a
benchmark and compare firms in industries with a statistically significant negative or positive reaction
JEB-5749; No. of Pages 20
G Model
Table 5
Firm-specific characteristics as determinants of the industry reaction to the introduction of Dodd-Frank Act.
Panel A
Pooled Multinomial logit Yit = ˛ + ˇ1 ATit + ˇ2 ROAit + ˇ3 Cashit + ˇ4 CapExit + ˇ5 Tangit + ˇ6 Debtit + ˇ7 Growthit + ˇ8 R&Dit + ˇ9 BMit + ˇ10 2009 + ˇ11 2010 + εit , where Y – firm level
categorical variable, where -1 indicates that a firm belongs to an industry with a significant negative change in systematic risk; 0 is an indication of a firm from an industry with no
significant change in risk; +1 identifies a firm associated with an industry with a significant and positive change in systematic risk; and independent variables represent firm
N. Sorokina, J.H. Thornton Jr. / Journal of Economics and Business xxx (2016) xxx–xxx
characteristics: AT- size, ROA- profitability, Cash – cash holdings, CapEx - capital intensity, Tang - tangibility, Debt - debt to assets ratio, Growth - actual growth, R&D - growth
opportunities, BM - book-to-market, and two dummy variables for year 2009 and 2010, where 2009 is a dummy variable set to 1 for all records of the year 2009, and to 0 otherwise;
2010 is a dummy variable set to one for all records of the year 2010, and to 0 otherwise (see “Methodology and Data” for greater details on variables and models).
The Negative vs None part of the results demonstrates that there is a statistically significant differences in the size, cash holdings, tangibility and growth opportunities between firms
that belong to the industries that reacted to the introduction of the Dodd-Frank Act with reduction in systematic risk and firms that belong to industries, which remained indifferent.
There Positive vs None part shows that there are also statistically significant differences in the cash holdings, tangibility and growth opportunities, but not size between firms that
ARTICLE IN PRESS
belong to the industries that reacted to the introduction of the Dodd-Frank Act with increase in systematic risk and firms that belong to industries, which remained indifferent. The
interpretation of the sign of differences relies on the range of the values of corresponding independent variables and is presented in Fig. 3.
Intercept AT ROA Cash CapEx Tang Debt Growth R&D BM 2009 2010
Negative vs None
0.5633 0.000062 0.00698 −2.9200 −0.0310 −5.0806 −0.000319 0.00362 −0.0548 −0.000144 −0.00779 0.0697
(p-value) <0.0001 <0.0001 0.1193 <0.0001 0.4982 <0.0001 0.7677 0.1186 <0.0001 0.2318 0.8862 0.1963
Positive vs None
−0.4971 1.466E−6 −0.00046 1.1310 −0.1189 0.2598 −0.00048 0.00309 −0.00011 −0.00002 −0.00765 0.0222
(p-value) <0.0001 0.4829 0.5163 <0.0001 0.2903 <0.0001 0.2611 0.167 0.0588 0.4295 0.848 0.5783
Panel B
Multinomial logit based on the changes in firm characteristic between 2008 and 2010.
Yit = ˛ + ˇ1 ATit + ˇ2 ROAit + ˇ3 Cashit + ˇ4 CapExit + ˇ5 Tangit + ˇ6 Debtit + ˇ7 Growthit + ˇ8 R&Dit + ˇ9 BMit + εit The dependent variable represents positive and
negative reactions vs no significant reaction to the introduced regulation measured as a change in systematic risk, as explained in Panel A. The independent variables are the
differences between year 2008 and year 2010 for the same firm characteristics, as described in the Panel A. Only changes in capital intensity and tangibility over the period of time
were significant drivers of the changes in systematic risk. In general, the impact of statistically significant determinants of the changes in risk is not related to the variation in firm
characteristics over time.
Intercept AT ROA Cash CapEx Tang Debt Growth R&D BM
Negative vs None
−0.8402 1.52E−06 0.00379 −0.142 2.1301 0.0991 −0.00068 −0.0047 0.000017 −0.00045
(p-value) <0.0001 0.2044 0.3687 0.5705 <0.0001 0.7806 0.7826 0.285 0.9558 0.5444
Positive vs None
−0.2455 1.05E−07 −0.00086 0.1723 0.0402 −0.7088 −0.00043 0.00205 0.000083 0.000786
(p-value) <0.0001 0.9445 0.7824 0.4034 0.9016 0.0135 0.5888 0.4377 0.7436 0.4627
13
14
JEB-5749; No. of Pages 20

G Model
N. Sorokina, J.H. Thornton Jr. / Journal of Economics and Business xxx (2016) xxx–xxx
Table 5 (Continued)
Panel C
Logit with clustered standard errors. Yit = ˛ + ˇ1 ATit + ˇ2 ROAit + ˇ3 Cashit + ˇ4 CapExit + ˇ5 Tangit + ˇ6 Debtit + ˇ7 Growthit + ˇ8 R&Dit + ˇ9 BMit + ˇ10 2009 + ˇ11 2010 + εit Y – firm level
binary variable, where in the first subsample, 1 indicates that a firm belongs to an industry with a significant negative change in systematic risk; 0 is an indication of a firm from an
ARTICLE IN PRESS
industry with no significant change in risk; in the second subsample, 1 is assigned to firms from industries with a significant positive change in systematic risk, 0 is an indication of a
firm from an industry without change in risk, and, finally, in the third subsample, the value of 1 is set for firms from industries with a significant increase in systematic risk, while 0
shows that a firm belongs to an industry with a decrease in systematic risk. The independent variables in this model are described in Panel A. When clustering is taken into account,
we uncover significant additional determinants of the change in systematic risk. Firms that reacted with decrease in risk are large, hold more cash, are more tangible, have greater
Book-to-Market ratio and have greater growth opportunities, but are less profitable and grow more slowly than firms from the industries, where systematic risk increased.
Intercept AT ROA Cash CapEx Tang Debt Growth R&D BM
Negative vs None
0.5537 0 −0.0007 −0.6941 −0.0129 −0.6471 0 0.0002 −0.0001 0
(p-value) 0.0089 0.0494 0.5367 0.06 0.5944 0.0309 0.7834 <0.0001 0.0078 0.3034
Positive vs None
0.5182 0 0.0029 −1.2304 0.601 −0.3118 0.0007 −0.002 0.0002 0.0001
(p-value) 0.3898 0.7659 0.0949 0.1343 0.6204 0.7856 0.0863 0.7288 0.6555 0.038
Positive vs Negative
0.3797 0 −0.0005 0.6784 0.2354 0.6663 −0.0003 −0.0001 0.0001 0
(p-value) 0.0666 0.0819 0.0351 0.0131 0.5862 0.0141 0.1123 0.0083 0.0032 <0.0001
In this table we present the estimated relationship between firm-specific factors and change of industry’s systematic risk in response to the introduction of the Dodd-Frank Act. We use
three specifications of the logit model: multinomial logit with and without year dummies, and binary logit with clustered errors.
G Model
Fig. 3. Probability of the industry positive/negative/none reaction to the introduction of Dodd-Frank Act as a function of the
each independent predictor when others are set at their respective mean levels.
to the no reaction control group. In Panel A of Table 5 (pooled multinomial logistic regression with
year-dummy variables based on three years of firm-level data), the year dummies are not significant.
There does not appear to be a market-wide change in systematic risk as a result of a change in the
independent variables in years 2009 and 2010 compared to the base year 2008. We can safely rely on
the results based on any or all three years in our study of the determinants of a change in systematic risk.
Since there are no significant changes in the data in years 2008–2010, we do not expect autocorrelation
to be an issue. However, we run the same regression separately on the data for years 2008, 2009 and
2010 for robustness. The results for all three years (not tabulated, but available upon request from the
contact author) are very similar and do not materially differ from the results of the pooled regression
presented in this paper.
From the coefficients in Panel A of Table 5, we can conclude that firm size, cash holdings, tangibil-
ity and R&D expenditures are all significant determinants of the industry reaction to the introduction
of Dodd-Frank. To aid further interpretation of the results, in Fig. 3, we plot, for the statistically sig-
nificant variables, the probabilities of the industries’ change in market beta as a function of each
determinant while setting the other independent variables at their mean levels. The figure allows
us to interpret the effect of variability in levels of each variable on the industry reaction to the
legislation.
In Panel A of Fig. 3, larger companies are clearly reacting with a decrease in risk. The probability of
the decrease in risk grows sharply as the company’s size increases and then flattens out at the highest
G Model
level. A median ($367.36 million) or mean ($11.5 billion) company in our sample is very unlikely
to react with a decrease in systematic risk. Those companies are most likely to exhibit no reaction
(with probability around 60%) or their systematic risk increases (with probability about 40%). Only the
largest companies in the sample are likely to react with decrease in systematic risk – the probability
of decrease in risk becomes sizable, above 30% – as a company’s assets surpass the $50 billion mark. In
Panel B, cash rich companies tend to react with an increase in risk and are less likely to be indifferent to
the regulation. Median company or mean company in our sample with respective cash-to-assets ratios
of 0.08 and 0.15 tend to produce no reaction with probability of 50–55% or react with the increase in
systematic risk with probability of 45–50%. In Panel C, companies with high levels of tangible assets
are more likely to react with increase in systematic risk. Mean and median companies on the list with
tangibility ratios of 0.14 and 0.27 have about 5–7% probability to react with decrease in systematic
risk and are more likely to demonstrate no reaction (with 55% probability), while probability to react
with increase in risk is about 40–43% for them. In Panel D, R&D intensive companies are less likely to
react with a risk increase and are more likely to produce no reaction. For median or mean company in
our sample, R&D expenses remain very small, and they are likely to produce no reaction (45–55%) or
react with increase in risk (35–40%), while reaction with decrease in risk varies between 0% and 20%
for them.
Although we did not see any significant changes of the relationship between systematic risk and
firm-specific determinants in years surrounding the Dodd-Frank legislation process, as measured by
the dummy variables in the pooled multinomial regression, we decided to explore the potential differ-
ences further by running a multinomial logit regression on the differences in each firm’s characteristics
between year 2008 and 2010. The results are shown in Panel B of Table 5. We find that an increase in
systematic risk was significantly related to the change in capital expenditures as a fraction of assets.
This may be due to the expected effect of Dodd-Frank on the costs of raising new capital. There was
also a significant relationship between differences in tangibility and changes in systematic risk. This
may be a result of the contemporaneous recession. There was not a significant change in relationship
between changes in the majority of firms’ characteristics over the period of study and a change in
systematic risk.
Panel C of Table 5 provides the results for the specification with robust clustered estimation
of the pairwise relationship between groups based on the change in risk. The signs of the sta-
tistically significant coefficients correspond to the signs of the effect of respective predictors on
the probability of the firm to belong to a particular reference group. When we compare firms
that decreased systematic risk to those with no response, the former are larger, hold less cash,
are less tangible, invest less in R&D and experience higher growth. Companies that belong to
industries with increased systematic risk, compared to those with no change, are more profitable,
more leveraged and have higher book-to-market ratios. Companies that reacted with a system-
atic risk increase, compared to those with decreased risk, are larger, less profitable, more tangible,
hold more cash, invest more in R&D, experience slower growth and have higher book-to-market
ratios.
We believe that the results raise questions about market participants’ initial assessment of the
eventual impact of Dodd-Frank on the economy. Firms in industries with a significant increase
in systematic risk in response to the introduction of Dodd-Frank appear to have characteristics
that would make them more dependent on external capital than other firms. This finding suggests
that market participants anticipate negative consequences for firms that rely on external capi-
tal.
To test our hypothesis that reaction to the Dodd-Frank Act introduction is linked to the reliance on
external capital, we compute the measure of dependence on external capital introduced by Rajan and
Zingales (1998). The results, presented in the Table 6, support Hypothesis 2. Clearly, all industries that
reacted to the introduction of Dodd-Frank with a decrease in systematic risk, except real estate, are less
dependent on external capital, while the majority of industries that reacted with increase in systematic
risk are more dependent on external capital. The aggregate measures of external dependence (mean
and median) for firms in industries that increased risk are twice as high as corresponding measures
for the firms in industries that decreased risk.
G Model
Table 6
External financing dependence and change in systematic risk. We computed the external finance dependence measure of Rajan
and Zingales (1998) for the firms in our sample. The results in this table show that majority of industries that reacted to the
Dodd-Frank introduction with decrease in risk, with exception of real Estate depend less on the external financing, while most of
the industries that reacted with increase in systematic risk depend greater on external finance. In summary, mean and median
external finance dependence for the firms in industries that increased systematic risk was twice of that for the industries that
decreased systematic risk.
Panel A. External financing and change in systematic risk: industries
Ind No Industry Name Change in risk External fin

dependence
5 Tobacco Products 1 0.00

48 Trading −1 0.03
45 Banking −1 0.11
46 Insurance −1 0.13
20 Fabricated Products −1 0.13
24 Aircraft 1 0.18
33 Personal Services 1 0.18
42 Wholesale 1 0.19
18 Construction −1 0.19
11 Healthcare 1 0.20
32 Communication −1 0.21
43 Retail 1 0.21
34 Business Services 1 0.24
31 Utilities 1 0.30
25 Shipbuilding, Railroad Equipment 1 0.34
14 Chemicals 1 0.37
47 Real Estate −1 0.43
12 Medical Equipment 1 0.50
13 Pharmaceutical Products 1 0.67
28 Non-Metallic and Industrial Metal Mining 1 0.78
27 Precious Metals 1 0.79
Panel B. External financing and change in systematic risk: summary
External financing dependence
Change in risk Mean Median
Increase 0.35 0.18

Decrease 0.27 0.13
5. Conclusion
In this paper, we conduct event studies of equity market reactions to the events surrounding finan-
cial reform initiatives that were adopted in response to the recent financial crisis. We conjecture that
due to the importance of the financial system to the overall economy, the effects of significant financial
reforms will be felt far beyond financial institutions.
We perform our study in three steps. First, we identify the reaction of the Fama-French 49 industry
indices to the Dodd-Frank Act legislative process. The most statistically significant finding is that
systematic risk, as measured by market beta, changes with the introduction of the act. Financial firms
and a few industries closely related to the financial sector experience a statistically significant decrease
in systematic risk. However, a substantial number of industries representing a broad cross-section of
the economy experiences a statistically significant increase in systematic risk. The initial reaction of
market participants to Dodd-Frank is mixed. While it may lower systematic risk in financial firms, at
the same time, it may increase systematic risk for the many non-financial firms in the economy. Since
systematic risk is a part of systemic risk, the changes in market beta imply corresponding increases or
decreases in systemic risk and changes in the magnitude of volatility of returns associated with the
stages of economic cycle.
G Model
We further determine the characteristics of firms in industries that experience significant changes
in systematic risk. Firms in industries that reacted with increase in systematic risk are more profitable,
more leveraged and have higher book-to-market value than industries that demonstrated no reaction.
Firms in industries that experienced a reduction in risk carry more assets, possess less cash, have less
R&D, have less tangible assets and grow faster. These results are consistent with the finding that the
firms in the group of industries with investors who reacted negatively to the introduction of Dodd-
Frank Act are much more dependent on external capital than the firms in industries in which systematic
risk decreased.
Overall, our results confirm that while financial and related industries are expected to benefit from
the reduction of the risk, as a result of the new regulation, industries reliant on external sources of
capital may potentially suffer because of their investors’ immediate expectations, which are incor-
porated into equity prices. Our study should provide a note of caution to legislators and regulatory
authorities. Actions taken to reduce risk in the financial industry may have the unintended spillover
consequence of increasing risk for non-financial firms.
Exhibit A. Brief summary of the Dodd-Frank Act
The Dodd-Frank Act, at more than 2000 pages, contains a large number of provisions to correct
many perceived problems with the financial system. Some of the more important provisions include:
• Consumer Protection: Creates the Consume Financial Protection Bureau and centralizes respon-
sibility for consumer protection within this single entity. The new entity is very independent as
compared to most government agencies. It is led by an independent director, functioning on with an
independent budget paid by the Federal Reserve. The bureau consolidates functions previously per-
formed my several regulatory bodies. It will include an Office of Financial Literacy and a consumer
complaint hotline.
• Too Big to Fail: The Act limits the formation and functioning of the institutions that may pose sys-
temic risk. It restricts proprietary trading by banks and non- banking institutions supervised by
Federal Reserve. These institutions are also prohibited from investing in hedge funds and private
equity funds. It extends the right of Federal Reserve to control non-bank institutions that may pose
an economy-wide risk. It promotes uniform risk-management rules for systemically important insti-
tutions; and, requires them to create comprehensive exit strategies. It sets up orderly liquidation
procedures and tightens requirements for the FDIC guarantee release. The Act prohibits bailouts at
the taxpayers’ expense and specifies that costs should be borne by the industry.
• Systematic Risk: The Act creates the Financial Stability Oversight Council, consisting of 10 federal
financial regulators, an independent member, and 5 non- voting members. The council is expected
to identify systemic risks at the early stage and respond to the threat. It will make recommendations
to Federal Reserve on the financial standards to limit the formation of Too Big to Fail institutions;
and, authorize the Federal Reserve to regulate non-bank financial companies. The Council will be
supported by a new Office of Financial Research.
• Regulation of Derivatives: The Act calls for SEC and CFTC regulation of over- the-counter deriva-
tives; establishes central clearing and exchange trading of certain derivatives. It requires significant
disclosures, increases standards of conduct and establish financial safeguards for participants in the
derivatives markets.
• Mortgage Lending: The Act reforms the mortgage market. It requires lenders to ensure borrower’s
sustainability, controls fairness of the lending practices, establishes non-compliance penalties for
the lenders and brokers; and, creates the Office of Housing Counseling as a part of HUD.
• Corporate governance: Gives shareholders rights to participate in director- nominating and pay-
related issues. It eliminate compensation-targeted business strategies and reporting manipulation,
tightens control over the compensation in financial industry, and establishes SEC review of the
compensation tied to stock performance.
• Credit Ratings: The Act creates an Office of Credit Ratings at the SEC. It requires methodology dis-
closure and independent information collection so as to prevent conflicts of interest. It requires
G Model
rating analyst’s to pass a qualification exam; and, requires credit rating agencies to carry liability for
non-diligent rating process.
• Regulatory Agencies: The Act reforms the Federal Reserve system; improves bank and thrift regula-
tion by closing the Office of Thrift Supervision; and creates the Federal Insurance Office. I increases
the funding of the SEC and increases its scale and scope of control.
References
Acharya, V. V., Pedersen, L. H., Philippon, T., & Richardson, M. P. (2010). Measuring systemic risk. FRB of Cleveland Working Paper
No. 10-02. Available at SSRN: http://ssrn.com/abstract=1595075 Accessed 10.05.15.
Akhigbe, A., & Whyte, A. M. (2004). The Gramm-Leach-Bliley Act of 1999: Risk implications for the financial services industry.
Journal of Financial Research, 27, 435–466.
Allen, L., & Saunders, A. (2004). Incorporating systemic influences into risk measurements: A survey of the literature. Journal of
Financial Services Research, 26(2), 161–191.
Andrews, D. W. K. (1993). Tests for parameter instability and structural change with unknown change point. Econometrica:
Journal of the Econometric Society, 821–856.
Ang, J. S., & Zhang, S. (2004). An evaluation of testing procedures for long horizon event studies. Review of Quantitative Finance
& Accounting, 23, 251–274.
Baker, M. P., & Wurgler, J. (2015). Do strict capital requirements raise the cost of capital? Bank regulation, capital structure, and
the low risk anomaly. American Economic Review, 105, 315–334.
Binder, J. J. (1985). Measuring the effects of regulation with stock price data. RAND Journal of Economics, 16, 167–183.
Binder, J. J. (1998). The event study methodology since 1969. Review of Quantitative Finance and Accounting, 11, 111–137.
Bremer, R., Buchanan, B., & English, P. (2011). The advantages of using quarterly returns for long-term event studies. Review of
Quantitative Finance and Accounting, 36, 491–516.
Brief summary of the Dodd-Frank Wall Street reform and consumer protection Act. http://www.google.com/url?sa=t&
source=web&cd=1&ved=0CBcQFjAA&url=http%3A%2F%2Fbanking.senate.gov%2Fpublic%2F files%2F070110 Dodd Frank
Wall Street Reform comprehensive summary Final.pdf&ei=-3P-TPKUO5SinQe5uLTECQ&usg=AFQjCNE8970z3jYt7WIxTV
qwbBDT7hSYFg Accessed 05.12.10.
Brown, S. J., & Warner, J. B. (1985). Using daily stock returns: The case of event studies. Journal of Financial Economics, 14, 3–31.
Cameron, A. C., & Miller, D. L. (2015). A practitioner’s guide to cluster-robust inference. Journal of Human Resources, 50(2),
317–372.
Carow, K. A., & Heron, R. A. (2002). Capital market reactions to the passage of the Financial Services Modernization Act of 1999.
Quarterly Review of Economics and Finance, 42, 465–485.
Carow, K. A., & Kane, E. J. (2002). Event-study evidence of the value of relaxing long-standing regulatory restraints on banks,
1970–2000. Quarterly Review of Economics and Finance, 42, 439–463.
Chava, S., & Purnanandam, A. (2011). The effect of banking crisis on bank-dependent borrowers. Journal of Financial Economics,
99(1), 116–135.
Chow, G. C. (1960). Tests of equality between sets of coefficients in two linear regressions. Econometrica: Journal of the Econometric
Society, 591–605.
Cowan, A. R. (1992). Nonparametric event study tests. Review of Quantitative Finance and Accounting, 2, 343–358.
DeAngelo, H., & Stulz, R. M. (2015). Liquid-claim production, risk management, and bank capital structure: Why high leverage
is optimal for banks. Journal of Financial Economics, 116, 219–236.
Dodd-Frank Wall Street Reform and Consumer Financial Protection Act: A Brief Legislative History with Links, Reports and Summaries.
http://www.llsdc.org/dodd-frank-legislative-history Accessed 22.12.15.
Fier, S. G., & Liebenberg, A. P. (2013). Market reaction to potential federal regulation in the insurance industry. Journal of Insurance
Issues, 36, 1–34.
Gao, Y., Liao, S., & Wang, X. (2013). The Economic Implications of the Dodd-Frank Act on Systemically Important Financial Firms:
Evidence from Market Reactions.. Available at SSRN: http://ssrn.com/abstract=1919630
Helwege, J., & Liang, N. (1996). Is there a pecking order? Evidence from a panel of IPO firms. Journal of Financial Economics, 40(3),
429–458.
Henderson, G. V., Jr. (1990). Problems and solutions in conducting event studies. Journal of Risk & Insurance, 57, 282–306.
Huber, P. J. (1973). Robust regression: Asymptotics, conjectures and Monte Carlo. Annals of Statistics, 1, 799–821.
Ivashina, V., & Scharfstein, D. (2010). Bank lending during the financial crisis of 2008. Journal of Financial economics, 97(3),
319–338.
Jagannathan, M., Stephens, C. P., & Weisbach, M. S. (2000). Financial flexibility and the choice between dividends and stock
repurchases. Journal of Financial Economics, 57(3), 355–384.
Johnsen, T., & Melicher, R. W. (1994). Predicting corporate bankruptcy and financial distress: Information value added by
multinomial logit models. Journal of Economics and Business, 46(4), 269–286.
Lamdin, D. J. (2001). Implementing and interpreting event studies of regulatory changes. Journal of Economics and Business, 53,
171–183.
Lawrence, E. C., & Arshadi, N. (1995). A multinomial logit analysis of problem loan resolution choices in banking. Journal of
Money, Credit and Banking, 202–216.
MacKinlay, A. C. (1997). Event studies in economics and finance. Journal of Economic Literature, 35, 13–39.
Mamun, A. M., Hassan, K., & Lai, S. (2004). The impact of the Gramm-Leach-Bliley act on the financial services industry. Journal
of Economics and Finance, 28, 333–347.
Mamun, A. M., Hassan, K., & Maroney, N. (2005). The Wealth and Risk Effects of the Gramm-Leach-Bliley Act (GLBA) on the US
banking industry. Journal of Business Finance and Accounting, 32, 351–388.
G Model
Martin, K. J. (1996). The method of payment in corporate acquisitions, investment opportunities, and management ownership.
Journal of Finance, 1227–1246.
Miyajima, H., & Yafeh, Y. (2007). Japan’s banking crisis: An event-study perspective. Journal of Banking & Finance, 31, 2866–2885.
Neale, F. R., & Peterson, P. P. (2005). The effect of the Gramm-Leach-Bliley Act on the insurance industry. Journal of Economics
and Business, 57, 317–338.
Rajan, R. G., & Zingales, L. (1998). Financial dependence and growth. American Economic Review, 88, 559–586.
Repullo, R. (2013). Cyclical adjustment of capital requirements: A simple framework. Journal of Financial Intermediation, 22(4),
608–626.
The Library of Congress: Thomas. Legislative information. http://thomas.loc.gov/ Accessed 05.12.10.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econo-
metrica: Journal of the Econometric Society, 817–838.
Wooldridge, J. M. (2002). Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: MIT Press.
Yildirim, H. S., Kwag, S.-W., & Collins, M. C. (2006). An Examination of the Equity Market Response to the Gramm-Leach-Bliley Act
across Commercial Banking, Investment Banking, and Insurance Firms.
Yohai, V. J. (1987). High breakdown-point and high efficiency robust estimates for regression. Annals of Statistics, 15, 642–656.
Zellner, A. (1963). Estimators for seemingly unrelated regression equations: Some exact finite sample results. Journal of the
American Statistical Association, 58(304), 977–992.

Sorokina 2016

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Sorokina 2016

Transféré par

Droits d'auteur :

Formats disponibles

G Model

JEB-5749; No. of Pages 20 ARTICLE IN PRESS

Contents lists available at ScienceDirect

Journal of Economics and Business

Reactions of equity markets to recent ﬁnancial

2. Literature review and hypotheses

2.1. Literature review

3. Methodology and data

3.1. Event study methodology

+ ˇ5i Rfxt + ˇ6i Rft + εit (1)

Event date Event description

6/17/2009 Initial proposal by president Obama

Rit daily return on industry equity;

Introduced Event1 Event 2 … Event n … Enacted

Introduced Legislative process Enacted

Change in Cumulative abnormal Change in

Fig. 2. Timing the event effect.

3.2. Methodology to examine the determinants of industry risk changes

Yit = ˛ + ˇ1 ATit + ˇ2 ROAit + ˇ3 Cashit + ˇ4 CapExit + ˇ5 Tangit + ˇ6 Debtit + ˇ7 Growthit

+ ˇ8 R&Dit + ˇ9 BMit + ˇ10 2009 + ˇ11 2010 + εit (2)

Panel A. Change in systematic risk at the legislation introduction.

Industry No Industry Code Industry Name Change ˇ4i ˇ1i p-value

5 Smoke Tobacco Products 1 0.2356 0.3797 0

Panel B. Robustness check: Breakpoint identiﬁcation with Quandt-Andrews test.

5 Smoke Tobacco Products 05/22/2009 *** 23.53 <0.0001

Panel B. Robustness check: Breakpoint identiﬁcation with Quandt-Andrews test.

25 Ships Shipbuilding, Railroad Equipment 06/17/2009 *** 42.54 <0.0001

Yit = ˛ + ˇ1 ATit + ˇ2 ROAit + ˇ3 Cashit + ˇ4 CapExit + ˇ5 Tangit + ˇ6 Debtit

+ ˇ7 Growthit + ˇ8 R&Dit + ˇ9 BMit + εit (3)

Yit = ˛ + ˇ1 ATit + ˇ2 ROAit + ˇ3 Cashit + ˇ4 CapExit + ˇ5 Tangit + ˇ6 Debtit + ˇ7 Growthit

+ ˇ8 R&Dit + ˇ9 BMit + ˇ10 2009 + ˇ11 2010 + εit (4)

Independent variables are the same as in model speciﬁcation (2).

3.3. Industry data

3.4. Financial reform

4.1. Effects of Dodd-Frank on industry indices

Panel A: Raw COMPUSTAT items

COMPUSTAT code Description

Panel B: Computed ﬁrm characteristics

Variable N Median Mean Std Dev Minimum Maximum

AT 24,687 367.36 11464.12 105,390 0.001 3,510,975

4.2. Determinants of systematic risk changes

Intercept AT ROA Cash CapEx Tang Debt Growth R&D BM

JEB-5749; No. of Pages 20

Intercept AT ROA Cash CapEx Tang Debt Growth R&D BM

Panel A. External ﬁnancing and change in systematic risk: industries

Ind No Industry Name Change in risk External ﬁn

5 Tobacco Products 1 0.00

Panel B. External ﬁnancing and change in systematic risk: summary

External ﬁnancing dependence

Change in risk Mean Median

Increase 0.35 0.18

Exhibit A. Brief summary of the Dodd-Frank Act

Vous aimerez peut-être aussi