The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Do Costly Internal Equity Injections Reveal
Bank Expectations about Post-Crisis
Real Outcomes?
WP 23-03
Arun Gupta
Federal Reserve Board
Horacio Sapriza
Federal Reserve Bank of Richmond
Do Costly Internal Equity Injections Reveal Bank Expectations about PostCrisis Real Outcomes?
Arun Gupta 1 and Horacio Sapriza 2
December 21, 2022
Abstract
We construct a novel signal of bank expectations utilizing confidential data and a regulatory constraint
imposed on bank internal capital markets during the 2008 crisis that made internal equity injections to
commercial bank subsidiaries difficult to reverse. When the US government initiated a $176 billion
recapitalization program during the crisis, this constraint made it costly ex-ante for multi-bank holding
companies (MBHC) to use these funds for the purpose of recapitalizing subsidiaries against future
anticipated losses; in contrast, lending the funds to subsidiaries was exempt from the constraint and thus
carried an option value for future reallocations across sibling subsidiaries. Several findings emerge.
First, we show that MBHCs treated internal equity injections as a scarce resource when emergency
funds arrived, whereas single-bank holding companies did not because the constraint was not costly for
them. Second, we find that excess internal equity injections by MBHCs form a signal of their
expectations for post-crisis subsidiary outlook—i.e., future profitability, supervisory ratings, and credit
originations. Third, the geographical aggregation of these individual bank signals predicts the long-run
real effects of the 2008 crisis on small businesses across US states—i.e., post-crisis growth in small
business revenues, employment, establishments, payroll, and wages. Our study provides a more direct
test of “banks as efficient information producers” (e.g., Diamond (1984), Fama (1985)), and is the first
to show that credible signals of this bank knowledge can be extracted from the internal capital markets,
allowing regulators to forecast in real time a geographical rank-order for post-crisis real outcomes at
small firms. This new policy tool can be seen as a potential side benefit of government-sponsored bank
recapitalization programs, of which there have been 33 in the past 40 years worldwide.
JEL Codes: G01, G21, G28, G30
Keywords: Macrofinance, Financial Crises, Banks, Financial Intermediation, Internal Capital Markets,
SME, TARP
Federal Reserve Board (arun.gupta@frb.gov)
Federal Reserve Bank of Richmond (horacio.sapriza@rich.frb.org)
The views expressed in this paper are solely the responsibility of the authors and should not be interpreted as reflecting the views of the
Federal Reserve Board, the Federal Reserve Bank of Richmond, its staff, or any other person associated with the Federal Reserve System.
We would like to thank William English, Gary Gorton, Andrew Metrick, Egon Zakrajšek, Murillo Campello, Heather Tookes, Arvind
Krishnamurthy, Huberto Ennis, David Jenkins, and seminar participants at the Yale Finance seminar series and the Federal Reserve Board
for helpful comments. We would like to also convey a special thanks to Cindy Vojtech and Gazi Kara.
1
2
1. Introduction
The literature on macrofinance has found that some financial indicators can serve as predictive
signals for future economic activity—e.g., King and Levine (1993), Adrian et al. (2019), Gilchrist and
Zakrajšek (2012), Schularick and Taylor (2012). Our paper complements this literature by deriving a
novel empirical signal with three unique features. First, while studies have traditionally focused on
predicting recessions along the country-level time-series dimension, our signal forecasts the long-term,
real transmission effects of a given crisis event, once it has occurred, along the cross section of local
geography. Specifically, at the start of a crisis, it can forecast a rank order of geographies based on the
real effects they will experience post-crisis. Second, our forecasting focuses specifically on real
outcomes in the “small firm” economy—i.e., growth in small business revenues, employment,
establishments, payroll, and wages —which is an important, opaque, and understudied segment of the
economy. Third, this is accomplished by observing costly bank actions during the crisis, which yield a
credible signal of the downside risks that banks foresee in the local economy. This signal can be
constructed by regulators in real time during a given crisis to glean bank information for the purpose of
designing more efficiently targeted local stimulus programs during recovery.
To develop this signal, we combine insights from financial intermediation theory and the internal
capital markets literature. A seminal theory behind the existence of financial intermediaries states that
banks hold a special role in the economy as information producers and delegated monitors for the credit
markets because of cost advantages inherent to their information production function (Diamond, 1984).
We propose a new testable implication of this theory: if this theory is valid, banks’ private information
about their investment opportunities should have predictive power for post-crisis outcomes. Of course, a
key obstacle remains: the superior information set of banks is, by definition, unobserved. Our key
contribution is to show that banks revealed this private information through costly actions undertaken in
the internal capital markets during the 2008 crisis. Using novel data that decompose each bank into its
respective subsidiary-level and parent holding company-level balance sheets, we provide evidence that
costly recapitalizations of subsidiaries by parent holding companies during the 2008 crisis credibly
revealed the banks’ private information about the downside risks facing their post-crisis investment
opportunities in the geographies where they operated. In other words, the magnitudes of these costly
recapitalizations generate cross-sectional signals at the bank-state level. Each signal represents a given
bank’s assessment of a state’s economic performance post-Lehman. We connect individual bank signals
to states using the subsidiary’s pre-crisis, small business credit exposures within each state. To filter out
idiosyncratic information about specific borrowers, we average individual signals of all banks
competing within a given state, and find that these state-level signals have surprising predictive power
for the future growth of small business revenues, employment, establishments, payroll, and wages years
after the crisis. This is the first paper to introduce a signal that predicts post-crisis outcomes for the
“small firm” economy, which represents businesses that are typically financially constrained, opaque,
and whose dynamics tend to diverge from those of large firms. The predictive power of our signal for
small firms suggests that banks have garnered significant information regarding future local demand
conditions, to which small firms are highly sensitive. 3
Our research design requires two key ingredients that were present during the 2008 crisis. First,
to observe subsidiary recapitalization decisions, parent holding companies need to receive new equity
capital to allocate. For this reason, our empirical setup focuses on the largest US government-sponsored
recapitalization program in history, known as the Capital Purchase Program (CPP). This program was
part of the Troubled Asset Relief Program (TARP) announced by the US Treasury in response to the
systemic collapse triggered by the 2008 bankruptcy of Lehman Brothers. The CPP provided parent
holding companies with a $176 billion windfall of fresh emergency capital that simultaneously entered
This is consistent with the findings that small firms find it optimal to be subject to monitoring (Fama (1985), Diamond
(1991), Nakamura (1993)).
3
the internal capital markets of 246 bank holding companies soon after the Lehman bankruptcy, enabling
our analysis of within-TARP variation. Second, internal equity injections must be costly if they are
going to reveal private bank expectations. To this end, our identification strategy exploits a special
regulatory constraint imposed on the internal capital market during the 2008 crisis that made internal
equity injections to bank subsidiaries difficult to reverse by limiting their ability to pay dividends
upstream to parent holding companies. 4 To show that the informativeness of internal equity injections
comes from the costs of this constraint, we identify the intensity by which the regulatory constraint
imposed costs on the internal capital market of each organization by exploiting preexisting differences in
the organizational structure of bank holding companies—namely, multi-bank holding companies
(MBHC) versus single-bank holding companies (SBHC). Specifically, the constraint made it costly exante for MBHCs to use the emergency CPP funds to recapitalize a given subsidiary in anticipation of
future losses, because that portion of TARP CPP funds may subsequently become “trapped” for the
remainder of the crisis and would be unavailable to flexibly reallocate to a sibling subsidiary in case
further shocks arose. Instead, lending these funds to the subsidiary in the form of a short-term cash
deposit (debt) would not be subject to the constraint and thus carried the option value of flexible future
reallocations. This option value is the opportunity cost of injecting equity. Using a revealed preference
argument, an MBHC would thus only find it optimal to inject equity if it deemed that the subsidiary’s
downside risk of future losses outweighed the loss in reallocation option value. Thus, the observed
choice of an MBHC to inject equity credibly reveals its outlook for post-crisis economic outcomes. In
comparison, the option value associated with lending CPP funds is less valuable for SBHCs because, by
definition, they have no other sibling bank subsidiaries to reallocate TARP funds towards. Therefore,
Because bank subsidiaries (not parents) hold FDIC-insured deposits of the public, subsidiary-level regulators were
concerned that legal entities outside the federal safety net (parent holding companies) should not opportunistically attempt to
extract subsidiary equity otherwise meant to provide cushion for insured deposits. As a result, regulators limited the payment
of dividends from the subsidiaries upstream to the parent during the 2008 crisis.
4
internal equity injection choices by SBHCs do not produce credible signals of their private information,
as injections do not carry an opportunity cost for SBHCs. This difference in the cost of subsidiary
recapitalizations across the two types of holding companies allows us to create two signals for analytical
purposes: a “treatment” signal constructed from the internal equity injection decisions of MBHCs that
we hypothesize should reveal bank expectations, and a “placebo” signal constructed from the internal
equity injections of SBHCs that should not. 5
Our empirical analysis begins by documenting how parents of MBHCs and SBHCs internally
allocated CPP funds based on each subsidiary’s confidential supervisory exam rating, known as the
CAMELS score. Our first set of results show that on the first date that TARP CPP funds arrived,
MBHCs provided nearly zero internal equity injections to subsidiaries with the best CAMELS score,
more (16 basis points) to subsidiaries with a medium CAMELS score, and the most (64 basis points) to
subsidiaries with the worst CAMELS scores. 6 In contrast, SBHCs provided the same amount (141 basis
points) of internal equity injections to their sole bank subsidiary, regardless of the subsidiary’s
CAMELS score. That is, the SBHC injection amount is not sensitive to the subsidiary’s distress level.
This result is consistent with the notion that, because the opportunity cost of an internal equity injection
is higher for MBHCs than it is for SBHCs, MBHCs treat internal equity injections of CPP funds as a
scarce resource, whereas SBHCs do not. Unlike MBHCs, SBHCs inject a flat amount because they do
not need to portion injections based on subsidiary “need” (e.g., subsidiary distress ratings), as SBHCs
have no sibling subsidiaries to reallocate to. In addition to this finding that illustrates the scarcity of
MBHC internal injections in the cross section, we also find evidence of the scarcity of MBHC internal
injections in the time series. To limit the amount of “trapped” capital on any given date, MBHCs
SBHCs may have their own private information. However, this information does not get revealed through the internal
equity injection variable because injections have little opportunity cost for SBHCs.
6
The size of an internal equity injection is measured as the number of basis points that the subsidiary’s capital ratio increases
as a result of the internal equity injection.
5
perform all their internal equity injections to the subsidiaries with the worst CAMELS score on the first
date of TARP CPP arrival, while they smooth out injections through time to subsidiaries with the
medium CAMELS score and they smooth out injections even more over time to subsidiaries with the
best CAMELS score. In contrast, SBHCs perform all their injections in one shot on the first date of
TARP arrival without any time-smoothing.
Next, we test whether the realized internal equity injection decisions have predictive power for
financial distress at the recipient subsidiary several years into the future. Because internal equity
injections provide the recipient subsidiary with additional capital cushion to protect against current and
future distress, it is important to distinguish that there are two types of distress that the raw internal
equity injections address. The first is distress associated with a preexisting (pre-crisis) condition.
Supervisory exams, as represented via CAMELS scores, are designed to capture this backwards-looking
distress component. The second is future (forward-looking) distress related to the deterioration of local
investment opportunities that the parent anticipates will affect the subsidiary’s condition in the future
due to the crisis event (e.g., the Lehman bankruptcy). As our goal is to extract the bank’s outlook about
forward-looking outcomes due to the crisis event, we define our signal as the residual of the internal
equity injection after controlling for the CAMELS score. Therefore, this variable, which we label
“excess” internal equity injection, captures the amount of internal equity injections that the parent
decided to provide in excess of what a subsidiary with a given CAMELS score received on average. 7
We find that “excess” internal equity injections reveal and predict which subsidiaries will experience
future distress in the post-crisis period. Specifically, for every 100 basis points of excess internal equity
7
For the purpose of extracting the parent’s information about future outcomes, we use the internal equity decisions as of the
first date of TARP. Even though these injections are smoothed through time for subsidiaries with better CAMELS scores, the
injections on day 1 are the most informative because that is when the scarcity of internal equity injections is highest, as
parents have to deal with the cost emanating from the regulatory constraint of trapped subsidiary capital as well as the fact
that multiple subsidiaries compete to receive injections. Once time passes and some sibling subsidiaries have been satisfied
via recapitalizations, it becomes less costly to inject and thus less informative for our purpose of creating a signal that reveals
private information. Thus, the date of TARP arrival contains all the information content for our signal.
injections provided by an MBHC parent on the date of TARP arrival, the recipient subsidiary
subsequently exhibits 43 basis points lower annual ROA and 329 basis points lower annual ROE postcrisis. These recipient subsidiaries also see future CAMELS scores worsen in the post-crisis period—
i.e., for every 100 basis points of excess internal equity injections provided by an MBHC parent on the
date of TARP arrival, the recipient subsidiary’s CAMELS score worsens each quarter thereafter by 0.08
rating points. Consistent with our hypothesis, the corresponding tests for SBHCs all show no predictive
power, as they do not contain reveal any information about their expectations.
The final set of results test the ability of the excess internal equity injections to predict local real
outcomes. We construct an aggregated excess injection signal by calculating state-level averages of the
individual excess injection signals across competing bank subsidiaries (only MBHCs) weighted by their
respective preexisting small business exposures in the given state. We find that these state-level excess
internal equity injections reveal and predict which states will experience the worst growth in post-crisis
small business revenues, employment, establishments, payroll, and wages. Specifically, for each 100
basis points of aggregated excess internal equity injections, the recipient state subsequently exhibits 229
basis points lower annual growth in nonemployer firm revenues, 97 basis points lower annual
employment growth, 108 basis points lower establishment growth, 172 basis points lower annual payroll
growth, and 336 basis points lower wage growth at small businesses in the post-crisis period.
Coming back to the topic of how individual subsidiary credit decisions respond to this
deterioration in future real conditions, we see that recipient subsidiaries exhibit strong reductions in
future credit originations—i.e., for every 100 basis points of excess internal equity injections provided
by an MBHC parent on the date of TARP arrival, the recipient subsidiary subsequently curtails new
credit originations annually by 48 percentage points for the next four years in the post-crisis period. This
finding is in line with the notion that banks utilize their private information regarding post-crisis
investment conditions by shifting their small business loan portfolio away from these negative outlook
states. Consistent with our hypothesis, the excess injections from SBHCs show no results, as they are
not credible signals of the SBHCs’ expectations.
These findings reveal an added side benefit of government-sponsored recapitalization programs,
of which there have been 33 worldwide in the past 40 years. 8 Specifically, our signal is a useful policy
tool for regulators to extract (in real time) the rank order of local geographies that banks expect to be
most impacted by crisis transmission effects. This knowledge can aid in the efficient design of stimulus
programs for small businesses affected by systemic shocks (e.g., the Paycheck Protection Program and
the Main Street Lending Program).
The paper is organized as follows. Section 2 provides a review of the relevant literature as well
as our paper’s contribution. Section 3 delivers an overview of the institutional details surrounding the
structure of bank internal capital markets as well as details on the TARP CPP program. Section 4
describes the data and summary statistics. Section 5 describes the empirical specifications. Section 6
discusses the results. Section 7 concludes.
2. Literature Review and Contribution
Our study contributes to three strands of research—namely, the literature on financial intermediation,
internal capital markets, and macrofinance.
a. Financial Intermediation
Historically, there have been 33 government-sponsored bank recapitalization programs in the past 40 years: Argentina (1995, 2011),
Bolivia (1994), Brazil (1994), Bulgaria (1996), Chile (1981), Colombia (1982,1998), Croatia (1998), Czech Republic (1996), Ecuador
(1998), Estonia (1992), Finland (1991), Ghana (1982), Indonesia (1997), Jamaica (1996), Japan (1997), Korea (1997), Lithuania (1995),
Malaysia (1997), Mexico (1994), Norway (1991), Paraguay (1995), Philippines (1997), Sri Lanka (1989), Sweden (1991), Thailand (1997),
Turkey (2000), Uruguay (2002), the United Kingdom (2007), the United States (2009), Venezuela (1994), and Vietnam (1997).
8
In the financial intermediation literature, it is commonly argued that banks play a special role in
the financial system because they resolve important information asymmetries. 9 Theoretical models
highlight the unique monitoring functions of banks (e.g., Diamond, 1984, 1991, Ramakrishnan and
Thakor, 1984, Fama, 1985). These studies generally argue that banks have a comparative cost advantage
that enables them to undertake superior debt-related monitoring in an imperfect capital market. If banks
do know more about the prospects of the firms they lend to as compared to nonbank creditors, a standard
test in the literature argues that bank loan agreements should then be “unique” or “special” relative to
other forms of finance—e.g., they should convey useful information to the market. Many empirical
studies show evidence that bank loan announcements yield a significantly positive return for borrowers
(e.g., James 1987, Lummer and McConnell, 1989, Mikkelson and Partch, 1986, Best and Zhang, 1993,
Billett, Flannery, and Garfinkel, 1995). 10 Recently, some studies have questioned this evidence. Maskara
and Mullineaux (2011) suggest that the positive announcement effect of bank loans suffer from sample
selection bias because the borrower’s decision to announce a bank loan is generally discretionary, while
Gande and Saunders (2012) document that the traditional monitoring incentives of banks were
weakened due to the development of the secondary market for loans.
We contribute to the financial intermediation literature by providing a more direct way to test
this theory that does not rely on the indirect implication of borrower stock price reactions but rather
directly shows the predictive power of bank information (revealed via costly bank actions) for
forecasting future distress associated with borrowers. Our results suggest that banks’ costly injection
The literature began with two problems associated with the production of information by a generic economic agent—
specifically, the “reliability problem” and the “appropriability problem”. The returns to producing the information could not
all be captured by the information producer, possibly making the production of the information uneconomic (Grossman and
Stiglitz (1980)). Leland and Pyle (1977) and Campbell and Kracaw (1980) show that the reliability and appropriability
problems can be eliminated if the information producer invests a sufficient minimum amount of its own wealth.
10
Recent papers have explored the relationship between information acquisition and physical distance between borrowers
and lenders (Petersen and Rajan (2002), Agarwal and Hauswald (2010)). Qian et al. (2015) highlight how the quality
(predictive power) of information production varies with incentives and communication costs.
9
decisions can be useful for predicting small business outcomes (such as small business revenues,
employment, establishment, payroll, and wages) across geographical regions. The predictive power of
our signal for small businesses provides the first empirical confirmation of Fama (1985), Nakamura
(1993), and Diamond (1991), who argue that the specialness of banks would be most applicable to small
firms, as these firms subject themselves to bank monitoring and bank loan finance. 11 In addition,
Krishnaswami et al. (1999) find evidence consistent with the intuition that banks’ low information
production costs are more valuable when information asymmetry is severe. Our main results are
consistent with this claim, as we find that banks’ costly internal equity injections were extremely
informative for predicting economic outcomes after the 2008 crisis, a time when information asymmetry
between the market and banks was historically high.
Overall, our findings constitute a unique and more direct test of the special role that banks play
in the economy, suggesting that their information production technology is likely very cost-efficient.
Not only are banks information producers of borrower creditworthiness, but they also possess valuable
expectations about future real prospects at the lower size distribution of the US nonfinancial business
sector.
Disclosing information to banks (rather than public markets) is less costly, since banks have access to the firm’s transaction
accounts and can gather much of the required nonpublic information at low cost (Nakamura, 1993). Because the cost of
producing the information required for public debt financing is high, small firms tend to rely more on bank loans (Fama,
1985). Further, Nakamura (1993) argues that the informational advantage of bank debt is less pronounced for big firms
because their numerous accounts are usually spread over a greater number of banks, and each bank only has access to partial
information. Thus, for big firms the cost advantage in borrowing from a bank is considerably lower. Fama (1985), Nakamura
(1993), and Diamond (1991) predict that small firms use more bank debt. The Diamond (1991) model has some interesting
empirical implications. First, it implies that new borrowers—for example, start-ups or young firms—initially submit to bank
monitoring to acquire a reputation and then switch to public debt. Finally, as smaller and younger firms generally have less
established reputations in financial markets, the information production role of an intermediary is more important to them
relative to larger, more established firms.
11
b. Internal Capital Markets
The theoretical literature argues that control rights (in the sense of Grossman and Hart, 1986,
Hart and Moore, 1990, and Hart, 1995) are one key difference between the study of internal versus
external capital markets, leading to higher monitoring incentives (Gertner, Scharfstein, and Stein, 1994).
In this way, the internal capital market brings a higher quality of information to bear on decisions than
the external market and can create value by actively reallocating scarce resources across projects (e.g.,
Alchian, 1969, Williamson, 1975, Stein, 1997). With regards to empirical work, there have been
numerous studies exploring the operation of internal capital markets in nonfinancial firms, many of
which show evidence of investment spillovers due to capital market frictions in the external markets
(i.e., Shin and Stulz, 1998, Lamont, 1997, Maksimovic and Phillips, 2002, Giroud and Mueller, 2015,
2019). Relatedly, the empirical literature for bank internal capital markets has also identified causal
spillovers of liquidity shocks through the internal market (e.g., Peek and Rosengren, 1997, 2000,
Houston et al., 1997, Campello, 2002, Schnabl, 2012, Cetorelli and Goldberg, 2012a, 2012b, Gilje,
Loutskina, and Strahan, 2016).
Whereas the prior literature (bank or nonfinancial) has focused on showing causal evidence for
the shock transmission role of internal capital markets in propagating spillovers across sibling divisions,
our paper flips the question and shows that frictions or costs inside the internal capital markets can be
exploited to extract the private information of firm managers in order to create novel signals that predict
a variety of future outcomes depending on the empirical context (what we term the revealed information
role of internal capital markets). More specifically, because of large information asymmetries between
managers and external investors, observing costly managerial actions inside the internal capital market
creates credible signals of private managerial information about future investment conditions that
contain local macroeconomic informational content. In our study, we find that due to the regulatory
constraint during the 2008 crisis, the internal equity injection variable becomes a particularly
informative signal for the bank’s future insolvency expectations, leading to a myriad of bank private
information signals at the bank-state level. As we do in our real outcomes analysis, these granular
signals can be aggregated to create local macroeconomic signals with surprising predictive power.
c. Macrofinance
There is a large literature examining the ability of financial variables to predict macroeconomic
outcomes and financial crises. Research has focused on stock and bond markets (Harvey, 1989),
commercial paper spreads (Bernanke, 1990, Friedman and Kuttner, 1992), the slope of the yield curve
(Estrella and Hardouvelis, 1991), systemic risk measures (Giglio, Kelly, and Pruitt, 2016, Allen, Bali,
and Tang, 2012), high-yield bonds (Gertler and Lown, 1999), corporate bond credit spreads (Gilchrist
and Zakrajsek, 2012, Krishnamurthy and Muir, 2020, Philippon, 2009), credit market sentiment (LopezSalido et al., 2017), composite financial conditions indices (Adrian, Boyarchenko, and Giannone, 2019),
corporate loan spreads (Saunders et al., 2021), credit growth (Schularick and Taylor, 2012), financial
development measures (King and Levine, 1993), and mutual fund flows (Ben-Rephael et al., 2021).
Our study complements this literature in several ways. First, while the literature has traditionally
focused on forecasting the arrival of a recession or crisis at the country time-series dimension, our
measure predicts the magnitude by which a given crisis event will transmit macroeconomic shocks to
small firms along the cross-sectional distribution of US states. To the best of our knowledge, there are
no prior studies that have established predictive signals for the geographical distribution of crisis
transmission effects, especially not for small business real outcomes. The study of small firms, whose
economics differ widely from those of large firms, is particularly important (e.g., Gupta, Sapriza, and
Yankov, 2022, Petersen and Rajan, 1994, 2002, Deyoung et al., 2015, Chodorow-Reich et al., 2022,
Gertler and Gilchrist, 1994, Adelino et al., 2015) yet understudied in the broader macroeconomic
literature due to data limitations. Second, the common theme in the empirical research on
macroeconomic time-series forecasting is the use of financial signals from external capital markets,
deriving their predictive power from frictions present in a variety of economic models (e.g., financial
accelerator mechanisms and credit channel theories). In contrast, our financial signal is sourced from the
internal capital markets, deriving its predictive power from the theory that banks hold significant
amounts of private information about borrowing firms due to their economic role as efficient
information producers (e.g., Diamond, 1984, Fama, 1985).
3. Institutional Background
a. Internal Capital Market Structure of Bank Holding Companies
To illustrate the internal allocation of CPP funds through the organization, we provide an
overview of the holding company structure as well as regulatory frictions impacting subsidiaries. Figure
1 denotes a simplified representation of a bank holding company structure comprised of three legally
separate entities—namely, the parent holding company, commercial bank subsidiary 1, and commercial
bank subsidiary 2. 12
Insert Figure 1: Consolidated View of Bank Holding Companies
Insert Figure 2: Disaggregated “Internal Capital Markets” Balance Sheet View of Bank Holding
Companies
Figures 1 and 2 illustrate how to visualize the relationship between a consolidated balance sheet and the
disaggregated “internal capital markets” view of the parent and subsidiary balance sheets within that
firm. Assume that a bank holding company has two subsidiaries. The left panel in Figure 1 represents a
traditional consolidated balance sheet of a bank holding company (similar to what might be presented in
a 10-K annual report). The right panel of this figure then breaks down each asset and liability by legal
Commercial bank subsidiaries come in three varieties. National banks are federally chartered and overseen by the Office of the
Comptroller of the Currency as well as the Federal Reserve. State member banks are chartered and regulated by the local state banking
authority in addition to being members of the Federal Reserve System. Finally, state nonmember banks are state-chartered but have elected
not to be members of the Federal Reserve System; thus, they are overseen by the FDIC.
12
entity ownership. Figure 2 further separates these assets into three individual balance sheets (holding
company-only, bank subsidiary 1, and bank subsidiary 2). It is important to note that the key variable of
our study, the holding company's internal equity injection, only becomes visible in the disaggregated
balance sheet view of Figure 2.
Parent holding companies do not typically perform any external business on their own; instead,
they raise a variety of nondeposit unsecured funding (commercial paper and long-term corporate bonds)
from debt markets as well as issue shares to equity investors. These funds are then typically
downstreamed to subsidiaries as either internal equity injections, cash deposits (debt claim), or longterm loans. Parent holding companies almost always own 100% of the equity in their operating
commercial bank and nonbank subsidiaries. Market participants can own claims on the parent holding
company via long-term corporate bonds, commercial paper, and equity stock, and the commercial bank
subsidiary via deposits insured by the Federal Deposit Insurance Corporation (FDIC), uninsured
deposits, and corporate bonds. External claims on the holding company ultimately derive value from the
interest and dividend income paid upstream from subsidiaries to parents. Parents are dis-incentivized
from allowing strategic failures of their commercial bank subsidiaries due to the FDIC's cross-guarantee
authority introduced through the Financial Institutions Reform, Recovery, and Enforcement Act of 1989.
Because bank subsidiaries (not parents) hold FDIC-insured deposits of the public, the cross-guarantee
authority provides the FDIC unilateral authority to offset losses incurred from the failure of a depository
institution by assessing claims against its healthy, sibling institutions.
The key internal capital market friction in our research design came from a regulatory constraint
specially imposed during the 2008 crisis. Due to regulatory concerns about potential leakages of the
federal insurance safety net after the Lehman collapse, subsidiary-level supervisors temporarily
increased their scrutiny regarding upstream payments of internal dividends out of commercial bank
subsidiaries. This helped to prevent the extraction of subsidiary equity, which provided a crucial cushion
against the default of insured deposits. Whereas banks could attempt to obtain special permissions, the
general enhancement of supervisory scrutiny increased the cost of flexibly reallocating equity once it
had been internally injected during the 2008 crisis. 13
Supervisors assess the financial health of each bank subsidiary through the use of regular, on-site
supervisory exam ratings (known as CAMELS ratings). Performed once every 12 to 18 months, on-site
exams culminate in a private and highly confidential rating, commonly known by acronym as a
CAMELS rating. To produce this single rating, the examiner evaluates the bank subsidiary along six
components: (C) capital adequacy, (A) asset quality, (M) management's ability to ensure the safety,
soundness, and compliance of its bank operations, (E) earnings, (L) adequacy of bank liquidity, and (S)
the bank's sensitivity to market risk. The composite CAMELS rating is not simply an average of the
individual component ratings but instead reflects examiners’ informed judgment as to how the
individual components ratings are combined to provide a summary measure of a bank’s overall
condition. The composite CAMELS rating, and its component ratings, are expressed through a
numerical scale of 1 through 5, with 1 being the best rating, and 5 being the worst rating. Financial
distress is defined by a CAMELS rating of 3, 4, or 5, whereas a rating of 1 or 2 means the subsidiary is
in good standing. It is important to know that the CAMELS rating is not meant to incorporate any
expectations or projections of future distress but rather the subsidiary’s current distress condition.
CAMELS ratings prove to be a strongly incentivizing factor for banks, as they are key inputs for a host
of regulatory costs, such as regulatory fines, FDIC insurance premiums, cease-and-desist enforcement
actions, and access to the Fed's discount window, in addition to regulatory licensing, branching, and
merger approval decisions.
As monitoring and examination resources are limited, supervisors also perform off-site
surveillance through a program known as the Supervision and Regulation Statistical Assessment of
13
This enhanced scrutiny lasted for the duration of the 2008 crisis.
Bank Risk (SR-SABR), which identifies banks with potential vulnerabilities based on an algorithm that
monitors quarterly Call Report information (and incorporates examination data and information about
historical bank failures). In identifying such banks, it uses an econometric model that estimates the
chance that a bank may be adversely rated or fail (or become critically undercapitalized) in the near
future. Because these grades are updated each quarter, SR-SABR provides a useful way of monitoring
bank health between examinations and assessing priorities for examination scheduling.
It is also important to know that the commercial bank subsidiary's ability to provide internal
loans to nonbank parts of the organization (e.g., parent holding company and nonbank siblings) is tightly
limited by Section 23A of the Federal Reserve Act. Enacted in 1933 in the aftermath of the Great
Depression, regulators implemented Section 23A to prevent the leakage of the federal subsidy onto
nondepository financial institutions. This law imposes quantitative limitations and collateral
requirements on commercial bank extensions of credit to nonbank subsidiaries, constituting a kind of
ring fence on funds being lent by commercial bank subsidiaries to nonbank subsidiaries or the parent
holding company.
b. Capital Purchase Program (Troubled Asset Relief Program)
The onset of the 2008 Global Financial Crisis led to the creation of several governmentsponsored emergency funding programs. In October 2008, TARP was created under the Emergency
Economic Stabilization Act and provided the US Treasury with $700 billion to purchase troubled assets
from banks. A sub-program of TARP was CPP, which used $218 billion of taxpayer funds to purchase
newly issued senior preferred shares and warrants of US banks. This constituted the largest
recapitalization program in the history of the US banking system.
The nine largest financial institutions were forced to participate in the capital injection—namely,
JP Morgan Chase & Co, Citigroup Inc., Wells Fargo & Co., State Street Corp., Bank of New York
Mellon, Bank of America Corp., Merrill Lynch, Morgan Stanley, and Goldman Sachs. The rest of the
program involved the voluntary participation of banks through an application and Treasury approval
process. Taliaferro (2009) outlines this procedure. Because our analysis compares the internal allocation
choices of TARP recipients, all of our analysis looks at within-TARP variation. From October 2008 to
June 2009, over 600 financial institutions participated (including bank holding companies, standalone
banks, thrifts, and credit unions). As the paper's focus is exclusively on bank holding companies, our
data sample contains the 246 bank holding companies that participated, which covers $176 billion of the
CPP. It is important to note that a key eligibility criterion for a bank holding company to participate in
the TARP CPP program was that it was not allowed to own any subsidiaries with a composite CAMELS
rating of 4 or worse, as the TARP CPP program was not designed to be a bailout. Although the research
question of our paper is not related to the efficiency of the government recapitalization program, there
have been many studies on this line (e.g., Veronesi and Zingales, 2010, Hoshi and Kashyap, 2010,
Bayazitova and Shivdasani, 2009, Duchin and Sosyura, 2014, and Philippon and Schnabl, 2013).
4. Data and Empirical Framework
a. Data Description and Summary Statistics
Information on the issuance of TARP CPP funds is collected from TARP Transaction Reports
from the US Treasury’s website. These transactions are hand matched to each recipient banking
organization’s identifier, known as RSSDs. In order to trace the flow of these emergency funds through
each organization, we deconstruct each US bank holding company into parent holding company and
bank subsidiary balance sheets. We obtain quarterly parent holding company-only balance sheet
information from the FR Y-9LP dataset and merge this with the quarterly balance sheet data of
commercial bank subsidiaries from the FFIEC 031/041 Call Report. The combination of FR Y-9LP and
FFIEC 031/041 Call Report data enables us to view the parent internal investments in commercial bank
subs as a panel structure. FR Y-9C data are also utilized to construct control variables representing
balance sheet conditions of each consolidated financial institution.
Subsidiary-level measures of financial distress are obtained from two confidential supervisory
sources, known as the CAMELS exam ratings and the SR-SABR database. Whereas the CAMELS score
represents the subsidiary’s most recent exam rating, the SR-SABR provides official, regulatory
statistical estimates each quarter of the probability that a given subsidiary will fail or will experience a
CAMELS downgrade in the near future, based on historical exam and Call Report information.
Additionally, we collect information on the small business lending originations from subsidiaries to
individual counties and states using data from the FFIEC Community Reinvestment Act database.
Information on small business labor decisions is comprised of state-level employment, establishment,
and payroll growth rates by firm size from the Census Bureau’s Statistics of US Businesses (SUSB)
database. Data for state-level average weekly wages (per employee) at small businesses are accessed
from from the Bureau of Labor Statistics Quarterly Census of Wages and Employment (BLS QCEW)
database. Note that we collect these variables specifically for small businesses, which we define as
establishments with nine or fewer employees. While data are not reported for revenues at these small
businesses, the Census Bureau does collect yearly revenue receipts at nonemployer firms in the Census
Bureau’s Nonemployer Statistics (NES) database. Nonemployer firms are small businesses that have no
paid employees but are subject to federal income tax. They appear in practically every industry and
comprise the majority of nearly 80 percent of industry subsectors. Examples include firms that operate
museums and hobby shops; provide pet care services; manufacture clothing, electrical equipment, and
machinery; sell real estate and process data. 14 Control variables for state-level macro time series on per
capita real GDP growth and personal income growth are retrieved from the Bureau of Economic
Analysis. Growth in housing price indices come from the Federal Housing Finance Agency’s website
(FHFA). Finally, aggregate annual realized returns on state-level stock portfolios are constructed from
individual firm stock price data from the Center for Research in Security Prices (CRSP). This is
constructed by summing the market capitalizations of all firms headquartered within each state and
calculating their respective yearly realized returns.
Insert Table 1
Table 1 presents summary statistics at the subsidiary, bank holding company, subsidiary-county,
and state macro levels. The data are comprised of 55 MBHCs owning 226 bank subsidiaries, and 185
SBHCs. MBHCs in the sample typically own five commercial bank subsidiaries. The median internal
equity injection across all BHCs on the date of TARP CPP arrival was 53 basis points, where this
variable is measured as the percentage point increase in the subsidiary’s capital ratio that results from
the parent’s internal equity injection. Upon controlling for measures of preexisting financial distress (via
CAMELS), excess internal equity injections average roughly zero; however, there are large standard
deviations in the cross section of subsidiaries, pointing to heterogeneity of parents’ outlook for postcrisis distress and future local investment conditions. The median CAMELS score for subsidiaries of
TARP participants is roughly 2, while the regulatory estimates for the probability of subsidiary failure
and CAMELS downgrade are 0.17 percent and 6.19 percent, respectively. The median annual growth in
small business employment, establishments, and payroll (where small business is defined as
14
See https://www.census.gov/programs-surveys/sbo.html
establishments employing nine or fewer workers) are negative 12, negative 305 basis points, and
negative 253 basis points, respectively, from 2004 to 2011. Median annual nonemployer firm revenue
growth in this period was 317 basis points. Control variables, such as per capital real GDP growth,
housing price growth, and state-level stock returns, were 113 basis points, 10 basis points, and 637 basis
points, respectively, through this period.
5. Empirical Analysis
a. The Internal Equity Injection Behavior of Parent Holding Companies
Our empirical framework starts by documenting the decision of parent holding companies to recapitalize
their subsidiaries using TARP CPP funds during the 2008-09 financial crisis.
( 1)
𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡
= 𝛽𝛽0 + 𝛽𝛽1 ∙ 𝟙𝟙�𝑡𝑡 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 �
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑅𝑅𝑅𝑅𝑅𝑅𝑠𝑠,𝑝𝑝,𝑡𝑡−1
+𝛽𝛽2 ∙ 𝟙𝟙�𝑡𝑡 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 � ∙ 𝟙𝟙 �𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝−1 == 2�
+𝛽𝛽3 ∙ 𝟙𝟙�𝑡𝑡 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 � ∙ 𝟙𝟙 �𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝−1 == 3�
+𝛽𝛽4 ∙ 𝟙𝟙�𝑡𝑡 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 � ∙ 𝟙𝟙�𝑀𝑀𝐵𝐵𝐻𝐻𝐻𝐻𝑝𝑝 �
+𝛽𝛽5 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1
+𝛽𝛽6 ∙ 𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑝𝑝,𝑡𝑡−1 + 𝛿𝛿𝑠𝑠 + 𝜋𝜋𝑝𝑝 + 𝜏𝜏𝑡𝑡 + 𝜀𝜀𝑠𝑠𝑠𝑠𝑠𝑠
The left-hand side variable
𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑅𝑅𝑅𝑅𝑅𝑅𝑠𝑠,𝑝𝑝,𝑡𝑡−1
is the flow of new internal equity injections from
the parent p into subsidiary s at time t as a share of the subsidiary’s total risk-weighted assets in the
previous quarter. This ratio represents the increase in the subsidiary’s capital ratio that results from the
internal equity injections provided by the parent. The regression spans 2006:Q1 to 2011:Q4 and isolates
the internal equity injection on the date that CPP funds arrive from the US Treasury to the parent using
an indicator function, 𝟙𝟙�𝑡𝑡 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 �. The coefficients 𝛽𝛽1 , 𝛽𝛽2 , 𝛽𝛽3 , 𝑎𝑎𝑎𝑎𝑎𝑎 𝛽𝛽4 comprise the key
right-hand side coefficients of interest. The coefficient 𝛽𝛽1 captures the average internal equity injection
made on the date of TARP arrival from parents to subsidiaries with a CAMELS rating of 1, while 𝛽𝛽2 and
𝛽𝛽3 capture the same TARP arrival injections for CAMELS 2 and 3 subsidiaries, respectively. 15 The
coefficient 𝛽𝛽4 captures the difference in internal equity injections across MBHCs and SBHCs, all else
being equal. Based on our research strategy, we hypothesize that 𝛽𝛽4 < 0 because the “trapped” capital
constraint is most costly for MBHCs. In this way, MBHCs will treat internal equity injections as a
scarce resource and likely inject less on average. We further look at MBHC-only and SBHC-only
subsample regressions and hypothesize that 𝛽𝛽3 , 𝛽𝛽2 , 𝛽𝛽1 > 0 for the MBHC-only sample. This is consistent
with the notion that MBHCs will inject equity to the subsidiary with the most need. For the SBHC-only
sample, we hypothesize that 𝛽𝛽1 > 0 and that neither 𝛽𝛽2 nor 𝛽𝛽3 will be statistically different from zero.
Because there are no siblings for future reallocations, the SBHC parent does not need to portion
injections based on need. The 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1 and 𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑝𝑝,𝑡𝑡−1 consist of capital
ratios, asset liquidity ratios, deposit funding shares, profitability, and the natural log of total assets at the
subsidiary and bank holding company levels, respectively. We use subsidiary and parent fixed effects to
control for downstreaming “culture", which can be a strong driver when deciding how to allocate TARP
funds. Time fixed effects are also applied to control for macroeconomic shocks. Standard errors are
clustered at the bank holding company level.
Dummy variables for subsidiaries with CAMELS 4 and 5 ratings are not needed in our specifications because none exist in
our sample (a key eligibility criterion for a bank holding company to participate in the TARP CPP program was that it could
not own subsidiaries with a CAMELS rating of 4 or worse.
15
In the next subsection, we explore if this internal equity allocation reveals the parent holding
company’s view of future post-crisis subsidiary outlook. Because internal equity injections provide the
recipient subsidiary with additional capital cushion to protect against current and future distress, it is
important to distinguish that there are two types of distress that the raw internal equity injections
address. The first is distress associated with a preexisting (pre-crisis) condition. Supervisory exams, as
represented via CAMELS scores, are designed to capture this backwards-looking distress component.
The second is future (forward-looking) distress related to the deterioration of local investment
opportunities that the parent anticipates will affect the subsidiary’s condition in the future due to the
crisis event (e.g., the Lehman bankruptcy). As our goal is to extract the bank’s outlook about forwardlooking outcomes due to the crisis event, we define “excess” internal equity injections as the injection
residual to a given subsidiary s, specifically on parent p’s date of TARP arrival, 𝜀𝜀𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝
from equation (1). 16 This quantity represents the amount of internal equity injections that exceed the
average amount of injections expected for a subsidiary with a given CAMELS distress rating,
probability of failure, probability of ratings downgrade, capital ratio, liquidity ratio, size, etc. Using a
revealed information argument, we propose that if a parent decides to allocate an excess amount of the
TARP CPP funds as new internal equity injections (controlling for the subsidiary’s condition preLehman), it is likely that the parent has information that future losses and distress at the recipient
subsidiary will be large.
b. Excess Internal Equity Injections as a Predictor of Subsidiary-Level Post-Crisis
Performance and Financial Distress
This variable is measured as the percentage point increase in the subsidiary’s capital ratio that results from the parent’s
excess internal equity injection as of TARP arrival.
16
Equation (2) explores the informativeness of the parent’s realized excess internal equity injection
decision on the date of TARP arrival for forecasting the future profitability and financial distress of the
recipient subsidiary.
( 2)
𝑌𝑌𝑠𝑠𝑠𝑠𝑠𝑠 = 𝛽𝛽0
+𝛽𝛽1 ∙ 𝟙𝟙�𝑡𝑡 ≥ 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 � ∙ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝
+𝛽𝛽2 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽3 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠,𝑝𝑝,𝑡𝑡−1
+𝛽𝛽4 ∙ 𝟙𝟙�𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1 == 2� + 𝛽𝛽5 ∙ 𝟙𝟙�𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1 == 3�
+𝛽𝛽6 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽7 ∙ 𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑝𝑝,𝑡𝑡−1 + 𝛿𝛿𝑠𝑠 + 𝜋𝜋𝑝𝑝 + 𝜏𝜏𝑡𝑡 + 𝜀𝜀𝑠𝑠𝑠𝑠𝑠𝑠
The dummy variable 𝟙𝟙�𝑡𝑡 ≥ 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 � is a post-TARP indicator that equals 1 for all dates
after TARP funding arrives to a given parent p, and
𝐸𝐸𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥𝑥 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 is the residual from equation (1) on TARP
arrival. The left-hand side variable 𝑌𝑌𝑠𝑠𝑠𝑠𝑠𝑠 represents a set of different post-crisis outcome variables at the
subsidiary-level, where
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑁𝑁𝑁𝑁𝑁𝑁 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡
𝑌𝑌𝑠𝑠,𝑝𝑝,𝑡𝑡 𝜖𝜖 �𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴
𝑠𝑠,𝑝𝑝,𝑡𝑡−1
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑁𝑁𝑁𝑁𝑁𝑁 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡
, 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸
𝑠𝑠,𝑝𝑝,𝑡𝑡−1
, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑔𝑔𝑠𝑠,𝑝𝑝,𝑡𝑡 � .
Because MBHCs incur an opportunity cost when allocating equity capital to subsidiaries (i.e., the option
value of debt deposits), we use a revealed preference argument that the MBHC will only find it optimal
to incur this cost (by injecting equity) if the parent expects future subsidiary losses and distress to be
higher in the post-crisis period. Thus, the parent injects capital in order to cushion against higher losses
and distress anticipated in the future. To this end, the first set of tests explores whether the size of excess
internal equity injections provided by parent holding companies on the day of TARP arrival reveals
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑁𝑁𝑁𝑁𝑁𝑁 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡
information that future subsidiary return-on-assets (𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑁𝑁𝑁𝑁𝑁𝑁 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡
(𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸
𝑠𝑠,𝑝𝑝,𝑡𝑡−1
𝑠𝑠,𝑝𝑝,𝑡𝑡−1
) and return-on-equity
) will be worse ex-post. The second set of tests explores whether subsidiaries
that received higher excess internal equity injections on the arrival of TARP perform worse in future
supervisory exams (𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡 ). In this way, costly internal equity injections forecast future
subsidiary distress.
The coefficient 𝛽𝛽1 isolates the difference in post-crisis future performance and distress across
subsidiaries that were chosen (by parents) to receive more versus lower excess internal equity injections
on the date that TARP CPP funds arrived from the US Treasury. For subsidiaries of MBHCs, we predict
that 𝛽𝛽1 < 0 for future profitability and 𝛽𝛽1 > 0 for future supervisory assessments (higher assessment
scores indicate worse distress), because the “trapped” capital constraint makes internal equity injections
costly. Namely, because MBHCs have multiple subsidiaries to manage, TARP CPP funds that become
trapped as equity in one subsidiary (via injections) will not be available for future reallocations to other
subsidiaries if shocks later arise. Using a revealed preference argument, if we observe that parents
choose to incur the cost of trapping their TARP CPP funds by injecting large amounts, it likely reveals
the parent’s private information about the recipient subsidiary’s future post-crisis outlook. It is the
costliness of the regulatory constraint that makes the MBHC internal equity injection variable
informative. Meanwhile, we predict that 𝛽𝛽1 will be statistically insignificant for subsidiaries of SBHCs
because the cost of trapped capital is low, as there are no bank siblings to reallocate capital towards. In
this way, subsidiaries of SBHCs form a kind of placebo group for our tests because they are not subject
to the same internal capital market cost and thus do not yield a predictive signal. It is important to note
that SBHCs have also have their own expertise and private information, however, this information is not
revealed through the internal equity injection variable because equity injections do not constitute a
costly action for SBHCs. The 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1 , 𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑝𝑝,𝑡𝑡−1, and fixed effects are
the same as in equation (1). Standard errors are clustered at the bank holding company level.
c. Excess Internal Equity Injections as a Predictor of Local Post-Crisis Economic
Outcomes
While section 5.b tests whether excess internal equity injections reveal the parent’s expectations
about the likelihood of future distress and losses related to the subsidiary’s direct lending exposures, this
section explores whether the signal has predictive power for post-crisis local macroeconomic outcomes
as well. We proceed by aggregating all subsidiary-level signals (across MBHCs) by state. Using the
definition below in equation (3), we calculate a weighted average of the excess internal equity injections
across all competing banks that possess direct small business credit exposures in a given state (market
m) pre-crisis.
(3)
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑚𝑚,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴
=
∑𝑠𝑠𝑠𝑠𝑠𝑠(𝑠𝑠) 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑚𝑚,𝑠𝑠,𝑝𝑝,2005≤𝑡𝑡≤2007 ∙ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝑢𝑢𝑖𝑖𝑖𝑖𝑖𝑖 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝
∑𝑠𝑠𝑠𝑠𝑠𝑠(𝑠𝑠) 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑚𝑚,𝑠𝑠,𝑝𝑝,2005≤𝑡𝑡≤2007
We denote the set of all MBHC subsidiaries s that compete within a given state m as m(s). The
state-level measure is 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑚𝑚,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 ,
which is calculated as the weighted average of subsidiary-level injection residuals
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 (calculated from equation 1) for all MBHC
banks competing within a given state m. Specifically, the weighting is done using the total dollar volume
of pre-crisis small business credit originated by the subsidiary s in state m from 2005 to 2007. Note, as a
placebo signal, we also calculate a parallel signal for SBHCs using the weighted average method in
equation (3).
Utilizing this state-level excess injection measure, our local macro forecasting test is captured by
equation (4).
( 4)
𝑌𝑌𝑚𝑚,𝑡𝑡 = 𝛽𝛽0
+𝛽𝛽1 ∙ 𝟙𝟙[𝑡𝑡 ≥ 2009] ∙ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑚𝑚,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴
+𝛽𝛽2 ∙ 𝑃𝑃𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺ℎ𝑚𝑚,𝑡𝑡−1
+𝛽𝛽3 ∙ 𝑃𝑃𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺ℎ𝑚𝑚,𝑡𝑡−1
+𝛽𝛽4 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺ℎ𝑚𝑚,𝑡𝑡−1
+𝛽𝛽5 ∙ 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺ℎ𝑚𝑚,𝑡𝑡−1
+𝛽𝛽6 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑚𝑚,𝑡𝑡−1 + 𝜋𝜋𝑚𝑚 + 𝜏𝜏𝑡𝑡 + 𝜀𝜀𝑚𝑚𝑚𝑚
The left-hand side variable 𝑌𝑌𝑚𝑚,𝑡𝑡 represents a set of different post-crisis macroeconomic outcomes at the
state level, where
𝑌𝑌𝑚𝑚,𝑡𝑡 𝜖𝜖
∆ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑚𝑚𝑚𝑚
∆ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸ℎ𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚
, 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸ℎ𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑚𝑚,𝑡𝑡−1
�
∆ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑚𝑚𝑚𝑚
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑚𝑚,𝑡𝑡−1
𝑚𝑚,𝑡𝑡−1
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑚𝑚𝑚𝑚
, 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝐹𝐹𝑖𝑖𝑟𝑟𝑟𝑟 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅
, ln (𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐴𝐴𝐴𝐴𝐴𝐴𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑚𝑚𝑚𝑚 )
𝑚𝑚,𝑡𝑡−1
If the aggregated excess injection signal is high in a given state m, this means that parents
competing in this market have decided to inject excess levels of internal equity to subsidiaries with
lending exposures to small businesses in state m. Via a revealed preference argument, the fact that
,
�
parents are willing to accept the associated reallocation costs of trapped capital reveals that they likely
anticipate future economic conditions in state m to worsen post-crisis (and therefore find it optimal to
inject excess capital beforehand to protect against these future losses). Specifically, we posit that these
state-level signals aggregate private information across competing banks, where the weighted average
helps to filter out idiosyncratic information (from subsidiary excess injections) about specific small
business borrowers that reside in the state m. Consequently, what remains is average bank expectation
information about future post-crisis investment conditions in the state. 17
The coefficient 𝛽𝛽1 isolates the post-crisis difference in the growth of employment,
establishments, payroll, and wages for small businesses (as well as revenues at nonemployer firms)
located in the states where MBHCs chose to inject excess amounts of internal equity allocations on the
date that TARP CPP funds arrived from the US Treasury. With respect to the signal aggregated from
MBHC excess injections, we predict that 𝛽𝛽1 < 0 because the cost of internal equity injections is high
and thus yields an informative signal. In contrast, we predict that 𝛽𝛽1 will be statistically insignificant if
we instead use the state-level excess injection signal constructed from SBHCs, because SBHC internal
equity injection amounts do not reveal bank expectations about borrower outlook—e.g., the cost of
trapped capital (and the option value of debt) is low for SBHCs, as there are no bank siblings to
reallocate capital towards. In this way, the aggregated signal from SBHCs forms a kind of placebo group
for our tests because they are not subject to the same internal capital markets constraint that MBHCs are.
We use 𝟙𝟙[𝑡𝑡 ≥ 2009] as a post-crisis dummy that equals 1 for all years from 2009 onwards.
𝑃𝑃𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝐺𝐺𝐺𝐺𝐺𝐺 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺ℎ𝑚𝑚,𝑡𝑡−1 , 𝑃𝑃𝑃𝑃𝑃𝑃 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺ℎ𝑚𝑚,𝑡𝑡−1 ,
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑛𝑛𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺ℎ𝑚𝑚,𝑡𝑡−1, 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺𝐺ℎ𝑚𝑚,𝑡𝑡−1, and
Aggregating at the state rather than county allows for large enough numbers of competing subsidiaries to become part of
the weighted average, allowing a better filter for borrower-specific information.
17
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑚𝑚,𝑡𝑡−1 form lagged state-level controls of economic growth.
Standard errors are clustered at the state level.
d. Do Banks Subsequently Divest from the States that Received the Largest Excess
Internal Equity Injections?
Finally, we return to understand the post-crisis lending behavior of banks receiving excess
internal equity injections. Specifically, if these injections do reveal bank expectations for poor postcrisis investment conditions, we next ask whether banks respond by divesting from these problem
markets? This is explored in equation (5) below:
(5)
𝐿𝐿𝐿𝐿(𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂)𝑐𝑐,𝑠𝑠,𝑝𝑝,𝑡𝑡 = 𝛽𝛽0
+𝛽𝛽1 ∙ 𝟙𝟙�𝑡𝑡 ≥ 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 � ∙ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝
+𝛽𝛽2 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠,𝑝𝑝,𝑡𝑡−1
+𝛽𝛽3 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠,𝑝𝑝,𝑡𝑡−1
+𝛽𝛽4 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1
+𝛽𝛽5 ∙ 𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑝𝑝,𝑡𝑡−1
+𝛾𝛾𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝛿𝛿𝑠𝑠 + 𝜋𝜋𝑝𝑝 + 𝜃𝜃𝑐𝑐 + 𝜏𝜏𝑡𝑡 + 𝜀𝜀𝑠𝑠𝑠𝑠𝑠𝑠
The dependent variable of interest is the natural log of small business credit originations issued
by subsidiary s (of parent p) at time t to small businesses in county c. We hypothesize that 𝛽𝛽1<0 for the
MBHC sample. Specifically, if large excess internal equity injections are informative of the parent’s
negative outlook regarding post-crisis investment conditions, we expect that recipient MBHC
subsidiaries would be likely to limit future exposures by curtailing new credit. As before, we expect 𝛽𝛽1
to be statistically insignificant for the SBHC sample. Our regression includes fixed effects for CAMELS
score, subsidiaries, parents, and counties to control for unobserved time-invariant factors. We also use
time fixed effects to control for aggregate national-level shocks to credit. To control for unobserved
business cycle variation at the local level, we also include different combinations of interacted fixed
effects, such as county*year and subsidiary*year, in alternative specifications.
6. Results
a. The Internal Equity Injection Behavior of Parent Holding Companies
Figure 3 presents a cross-sectional plot of the average internal equity injections received by bank
subsidiaries with different CAMELS ratings prior to the date of TARP arrival. The top panel shows that
MBHCs portion the usage of TARP CPP funds by providing higher internal equity injections to
subsidiaries with worse regulatory distress ratings (i.e., a CAMELS rating of 3, 2, and then 1). This
behavior suggests that the internal capital markets constraint was in fact costly for MBHCs during the
crisis. Specifically, as equity injections lead TARP funds to become “trapped” in the recipient subsidiary
ex-post, MBHCs lose the flexibility to reallocate these funds to sibling subsidiaries later in the crisis. In
this way, MBHCs treat internal equity injections as a scarce good ex-ante and, thus, are more willing to
incur this “trapped” capital cost when the subsidiary’s pre-crisis distress condition is worse. In
comparison, the bottom panel of Figure 3 shows that SBHCs provide a uniform amount of internal
equity injections regardless of the subsidiary’s distress rating. This behavior is consistent with the idea
that injecting TARP funds as equity is not costly for SBHCs, as the ability to flexibly reallocate funds in
the future is less valuable given that SBHCs do not have sibling bank subsidiaries to reallocate funds to.
In this way, the SBHCs’ injection decisions are independent of subsidiary distress.
Insert Figure 3
Whereas Figure 3 explored the impact of the constraint on the cross section of internal equity
injections upon TARP arrival, Figure 4 confirms the existence and bindingness of this internal capital
markets constraint in the time series by plotting the average internal equity injections received by bank
subsidiaries with CAMELS ratings of 1, 2, and 3.
The top panel corroborates the notion that MBHCs found this constraint costly, treating internal
equity injections as a scarce resource by prioritizing recapitalizations at the most distressed subsidiaries
sooner in time. Specifically, the top panel shows that subsidiaries with a CAMELS rating of 3 are
immediately recapitalized on the first date that TARP CPP funds arrive at the MBHC parent.
Meanwhile, recapitalizations at MBHC subsidiaries with a CAMELS rating of 2 are lower on date 0 and
are smoothed through time post-TARP. Finally, recapitalizations at MBHC subsidiaries with a
CAMELS rating of 1 are even lower on date 0 and are also smoothed through time. In contrast, the
bottom panel shows the opposite: SBHCs do not treat internal equity injections as a scarce resource
(e.g., “trapped” capital is not costly because there are no other siblings to reallocate to). Accordingly,
they recapitalize their single subsidiary uniformly (regardless of distress level) on date 0, with no
evidence of smoothing through time. The fact that MBHCs withhold injecting equity (on date 0) in
subsidiaries with CAMELS ratings of 2 and 3 suggests they value the ability to reallocate funds in the
future if and when shocks arise.
Insert Figure 4
Controlling for a host of bank characteristics, the results of Table 2 are consistent with the
patterns shown in Figures 3 and 4. Specifically, column 2 finds that MBHCs provided a higher amount
of internal equity injections on the date of TARP arrival to subsidiaries with a CAMELS 2 rating as
compared to subsidiaries with a baseline CAMELS 1 rating—i.e., CAMELS 2 subsidiaries experienced
an increase of 16 basis points in their capital ratios as compared to CAMELS 1 subsidiaries (which
received no injections). Subsidiaries with a CAMELS 3 rating, accordingly, experienced an even higher
relative increase of 64 basis points as compared to the baseline CAMELS 1 subsidiaries on the date of
TARP arrival. In contrast, column 3 shows that SBHCs’ internal equity injections on TARP arrival
raised subsidiary capital ratios uniformly by 141 basis points, regardless of CAMELS rating. In column
1, the key variable of interest is the coefficient for the interaction term, MBHC * TARP Arrival Date.
This column result combines SBHC and MBHC subsamples to show that, holding the CAMELS rating
fixed, subsidiaries owned by MBHCs receive significantly lower internal equity injections than
subsidiaries owned by SBHCs—i.e., capital ratios of MBHC subsidiaries rise 103 basis points less than
those of SBHC subsidiaries on the date of TARP arrival. This point estimate represents an indirect
estimate of the cost of the capital trapping constraint (or, equivalently, the option value associated with
withholding the use of TARP funds for subsidiary recapitalizations). Put differently, controlling for
CAMELS rating, MBHCs chose to withhold 103 basis points of subsidiary capital ratio stimulus in
exchange for retaining the flexibility to inject equity later when future shocks arise (option value).
Insert Table 2
b. Excess Internal Equity Injections as a Predictor of Subsidiary-Level Post-Crisis
Performance and Financial Distress
The results of Table 3 show that, controlling for various measures of ex-ante distress (such as the
subsidiary’s CAMELS rating, probability of failure, and probability of CAMELS rating downgrade), the
amount of excess equity that the MBHC parent decides to internally inject reveals information about
(and predicts) the subsidiary’s ex-post performance after the crisis. As shown in columns 1 and 3, the
recipient subsidiaries perform worse ex-post. Subsidiaries whose capital ratios increased by 100 basis
points more via excess internal equity injections on the date of TARP arrival earned 43 basis points
lower return on assets (and 329 basis points lower return on equity) each quarter in the post-crisis
period. In other words, this result suggests that when an MBHC parent provides equity injections in
excess of that which is warranted by the subsidiary’s pre-existing distress level, this choice reveals
private information about the subsidiary’s future investment opportunities. The predictive nature of the
MBHC’s internal equity injection decision indicate that bank holding companies possess a significant
amount of expertise and private information about the post-crisis performance of each subsidiary’s
preexisting investments as well as the post-crisis prospects for new investment opportunities. Notice that
excess internal equity injections from SBHC parents do not hold any such predictive power for postcrisis subsidiary outcomes (as shown in columns 2 and 4), consistent with the notion that the equity
injections from SBHC parents are not costly and thus are not informative of the SBHC’s expectations
for future subsidiary performance.
Insert Table 3
The next set of results show similar findings for regulatory measures of ex-post subsidiary
distress. Specifically, the amount of equity that the MBHC parent decides to internally inject also
predicts the evolution of post-crisis CAMELS exam scores. Referencing column 1 of Table 4,
subsidiaries whose capital ratios rise by 100 basis points more via excess internal equity injections on
the date of TARP arrival also experience a deterioration in CAMELS ratings by an average of 0.08
points post-crisis (higher CAMELS indicate a worse score). This is consistent with the idea that the
expectations about future distress revealed by the parent’s excess injection decision on the arrival date of
TARP eventually gets captured by future supervisory distress measures once tangible evidence of the
worsening of profitability arrives on subsidiary balance sheets (such as the worsening post-crsisi
profitability evident in Table 3). In this way, regulatory information eventually incorporates the parent’s
private information with a lag. Notice that the excess internal equity injections from SBHC parents do
not hold any such predictive power for post-crisis subsidiary distress (as shown in column 2).
Insert Table 4
Figure 5 provides an illustration of this effect, plotting the time-specific coefficients in a slightly
adjusted version of equation (2). 18 This shows that subsidiaries that received positive an excess amount
of internal equity injections from their MBHC parents on the date of TARP arrival subsequently
demonstrate worse CAMELS exam results several quarters after the crisis. In this way, the predictive
nature of the excess internal equity injection decision suggest that bank holding companies hold a
significant amount of expertise and private information about each subsidiary’s post-crisis type.
c. Excess Internal Equity Injections as a Predictor of Local Post-Crisis Economic
Outcomes
The results in subsection 6.b present evidence that MBHC parents plant equity at subsidiaries in
anticipation of a future deterioration in profitability and distress. The next tables and figures explore
whether the choice of how much TARP CPP funds to provide as excess internal equity injections also
reveal bank expectations about the deterioration of future investment opportunities in geographical
markets that the given subsidiary operates in.
Specifically, this figure plots the 90th percent confidential intervals for coefficients 𝛽𝛽𝛼𝛼 estimated for each Date*Excess
Internal Equity Injection term (𝑓𝑓𝑓𝑓𝑓𝑓 𝛼𝛼𝛼𝛼{−5,5} ) in the following specification: 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠𝑠𝑠𝑠𝑠 = 𝛽𝛽0 + ∑5𝛼𝛼=−5 𝛽𝛽𝛼𝛼 ∙
𝟙𝟙�𝑡𝑡 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 + 𝛼𝛼� ∙ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 + 𝛽𝛽6 ∙
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽7 ∙ 𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑝𝑝,𝑡𝑡−1 + 𝛿𝛿𝑠𝑠 + 𝜋𝜋𝑝𝑝 + 𝜏𝜏𝑡𝑡 + 𝜀𝜀𝑠𝑠𝑠𝑠𝑠𝑠 .
18
We observe that the weighted-average excess amount of internal equity injections associated
with banks that are competing within a state forms an aggregated signal of bank private information
about future revenue growth at very small businesses (nonemployer firms), as displayed in Figure 6.
Specifically, this figure shows a negative cross-sectional relationship between a state’s ex-ante excess
internal equity injection signal on the date of TARP arrival and the state’s post-crisis three-year
cumulative growth in revenues at nonemployer firms from 2009 to 2011.
Insert Figure 6
The regression results of Table 5 corroborate this evidence by showing that if the subsidiaries
competing for small business credit investments within a state experience a weighted average excess
equity injection of 100 basis points on the date of TARP arrival, revenues at nonemployer firms in that
state subsequently grow 233 basis points slower per year during the post-crisis recovery period.
Consistent with findings from the previous tables, this predictability only holds for the signal
constructed using excess internal equity injections from MBHC parents (columns 1 and 2), and not the
signal constructed using excess internal injections from SBHC parents (columns 3 and 4).
Insert Table 5
We also find evidence that the aggregated state-level signal of excess internal equity injections
has predictive power for future employment growth at small businesses located in the state, as displayed
in Figure 7. Specifically, this figure shows a negative cross-sectional relationship between a state’s exante excess internal equity injection signal on the date of TARP arrival and the state’s post-crisis threeyear cumulative growth in small business employment from 2009 to 2011.
Insert Figure 7
Table 6 corroborates this finding. Specifically, if the subsidiaries competing for small business
credit investments within a state experience a weighted average excess internal equity injection of 100
basis points on the date of TARP arrival, employment growth at small businesses (i.e., defined as
establishments with nine or fewer employees) is reduced by 97 basis points more per year during the
post-crisis period. It is important to note that this predictability only holds for the signal constructed
using excess internal injections from MBHC parents (columns 1 and 2), and not the signal constructed
using excess internal injections from SBHC parents (columns 3 and 4), which do not incur costs from
the trapped capital constraint.
Insert Table 6
Figure 8 shows that not only do larger excess internal equity injection signals from MBHCs
predict lower post-crisis employment growth at a given establishment, but we also observe a reduction
in the number of establishments located in these states (extensive margin). Specifically, this figure
shows a negative cross-sectional relationship between a state’s ex-ante MBHC excess internal equity
injection signal on the date of TARP arrival and the post-crisis cumulative growth in small business
establishments from 2009 to 2011.
Insert Figure 8
Table 7 corroborates this finding. Specifically, if the subsidiaries competing for small business
credit investments within a state experience a weighted average excess equity injection signal of 100
basis points on the date of TARP arrival, establishment growth at small businesses is reduced by 108
basis points more per year during the post-crisis period. It is important to note that this predictability
only holds for the signal constructed using excess internal injections from MBHC parents (columns 1
and 2), and not the signal constructed using excess internal injections from SBHC parents (columns 3
and 4), which do not incur costs from the trapped capital constraint.
Insert Table 7
Figure 9 complements the labor results on the decline in employment and establishments, as we
also observe a reduction in payroll growth at small businesses located in states that experienced higher
excess internal equity injection signals from MBHCs. Specifically, this figure shows a negative crosssectional relationship between a state’s ex-ante excess internal equity injection on the date of TARP
arrival and the post-crisis cumulative growth in small business payroll from 2009 to 2011.
Insert Figure 9
Tables 8 and 9 corroborate this finding. Specifically, if the subsidiaries competing for small
business credit investments within a state experienced a weighted average excess equity injection signal
of 100 basis points on the date of TARP arrival, payroll growth and average wages at small businesses
grew 174 basis points and 336 basis points slower during the post-crisis period, respectively. It is
important to note that this predictability only holds for the signal constructed using excess internal
injections from MBHC parents (columns 1 and 2), and not the signal constructed using excess internal
injections from SBHC parents (columns 3 and 4), which do not incur costs from the trapped capital
constraint.
Insert Tables 8 and 9
Tables 5 through 9 and Figures 6 through 9 show evidence consistent with the theme that the
MBHC parent’s choice to inject excess internal equity on TARP arrival reveals information about the
general deterioration in the quality of future potential investment opportunities, reflected in the fact that
geographies receiving excess internal injections experience a slowdown in small business revenues,
employment, establishments, payroll, and wages. Table 10 as well as Figures 9 and 10 circle back to the
subsidiary’s post-crisis decisions with respect to new credit allocations in these regions. Specifically,
Table 10 shows that if an MBHC subsidiary received an excess internal equity injection of 100 basis
points, it subsequently reduced its new small business credit originations (flow) in the post-crisis period
by an economically and statistically significant meaningful 48 percent per year. Figure 10 provides an
illustration of this effect, plotting the time-specific coefficients in a slightly adjusted version of equation
(5). 19 Figure 11 illustrates a negative cross-sectional relationship between a state’s ex-ante MBHC
excess internal equity injection signal on the date of TARP arrival and the post-crisis cumulative growth
in small business credit originations. These findings are consistent with the notion that banks act on their
own private information and subsequently curtail new lending exposures to these problematic regions
post-crisis, which coincides with the underperformance of local macroeconomic conditions in these
regions post-crisis demonstrated in Tables 5 through 9 and Figures 6 through 9. As discussed before, the
excess injections signals constructed from SBHCs do not show any relationship with respect to their
post-crisis small business credit originations.
Specifically, this figure plots the 90th percent confidential intervals for coefficients 𝛽𝛽𝛼𝛼 estimated for each Date*Excess
Internal Equity Injection term (𝑓𝑓𝑓𝑓𝑓𝑓 𝛼𝛼𝛼𝛼{2005,2011} ) in the following specification:
𝐿𝐿𝐿𝐿(𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆)𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 = 𝛽𝛽0 + ∑2011
𝛼𝛼=2005 𝛽𝛽𝛼𝛼 ∙ 𝟙𝟙[𝑡𝑡 = 𝛼𝛼] ∙
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 + 𝛽𝛽2 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽3 ∙
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽4 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽5 ∙ 𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑝𝑝,𝑡𝑡−1 + 𝛿𝛿𝑠𝑠 + 𝜋𝜋𝑝𝑝 + 𝜏𝜏𝑡𝑡 + 𝜀𝜀𝑠𝑠𝑠𝑠𝑠𝑠
19
Insert Table 10
7. Conclusion
Our paper shows that frictions or costs associated with banks’ internal capital markets can be
exploited to extract a credible signal of banks’ private information. Our signal has unique predictive
power for post-crisis real outcomes, which we introduce as the revealed information role of internal
capital markets—a new contribution to the literature. Our identification strategy exploits a regulatory
constraint imposed during the 2008 financial crisis on the bank internal capital market that made internal
equity injections to bank subsidiaries difficult to reverse by limiting their ability to pay dividends
upstream to their parent. When the US government initiated a $176 billion recapitalization program
during the crisis, this constraint made it costly ex-ante for MBHCs to use these funds to recapitalize
subsidiaries against anticipated future losses, because lending the funds to subsidiaries as debt instead
was exempt from the constraint and thus carried an option value of future reallocations to sibling
subsidiaries. SBHCs form a placebo group because they did not find this option valuable, as they have
no sibling subsidiaries to reallocate funds towards. We show, first, that MBHCs treated internal equity
injections as a scarce resource when emergency funds arrived, whereas SBHCs did not. Second,
controlling for supervisory ratings, we find that “excess” internal equity injections by MBHCs reveal
their private information about post-crisis subsidiary conditions (i.e., future profitability, supervisory
ratings, default probabilities, credit originations). Third, the geographical aggregation of these individual
bank signals from MBHCs predicts the real transmission effects of the Lehman bankruptcy on the
“small firm” economy at the state level (i.e., growth in small business employment, number of
establishments, payroll, wages, and nonemployer firm revenues). In contrast, excess internal equity
injection signals from SBHCs have no ability to predict either subsidiary distress or real outcomes,
consistent with the notion that these injections were not costly actions and thus not credible signals of
their private information.
Rather than relying on the traditional tests that use borrower stock price reactions to loan
announcements, our study provides a unique, more direct test of the special role that banks play in the
economy as efficient information producers (e.g., Diamond (1984), Fama (1985)) by directly showing
the predictive power of the costly bank internal equity injections for future financial and real outcomes.
Consistent with this theory, our findings suggest that the information production technology of banks is
likely very cost-efficient. Our results suggest that banks hold valuable private information that is useful
for predicting small business outcomes at the state level, e.g., post-crisis growth in small business
revenues, employment, establishments, payroll, and wages. To the best of our knowledge, this is the first
study to establish a predictive signal that can be used in real time to forecast the real transmission effects
of a systemic crisis on small firms across geographies.
One policy implication is that supervisors can potentially use this method to extract granular,
private information in real time from banks about which local geographies they expect employment and
welfare to be most impacted as a result of a systemic crisis. However, aside from imposing regulatory
costs on internal injections, the signal’s construction requires the presence of a large governmentsponsored recapitalization program. Recapitalization programs typically only occur if the crisis event
leads to a bank insolvency issue (which did not occur during the recent COVID crisis). However, when
CPP-like programs do occur (33 worldwide in the past 40 years), our paper finds that there is an
opportunity for regulators to utilize banks’ special role as information producers. Extracting this
information in real time during a downturn may aid in designing more targeted and efficient stimulus
programs for small businesses. This has become particularly valuable given the novel kinds of targeted
stimulus program that were introduced post-COVID (e.g., Paycheck Protection Program, Main Street
Lending Program).
References
Adelino, Manuel, Schoar, Antoinette, and Felipe Severino. 2015. “House Prices, Collateral, and SelfEmployment,” Journal of Financial Economics, Volume 117, Issue 2, Pages 288-306.
Adrian, Tobias, Boyarchenko, Nina, and Domenico Giannone. 2019. “Vulnerable Growth,” American
Economic Review, Volume 109, Issue 4, Pages 1263-1289.
Agarwal, Sumit, and Robert Hauswald. 2010. “Distance and Private Information in Lending,” Review
of Financial Studies, Volume 23, Issue 7, Pages 2757–2788.
Alchian, Armen A. 1969. “Corporate Management and Property Rights,” in H. Manne, ed., Economic
Policy and the Regulation of Corporate Securities (American Enterprise Institute, Washington,
DC).
Allen, Linda, Bali, Turan, and Yi Tang. 2012. “Does Systemic Risk in the Financial Sector Predict
Future Economic Downturns?” Review of Financial Studies, Volume 25, Issue 10, Pages 30003036.
Bayazitova, Dinara, and Anil Shivdasani. 2011. “Assessing TARP,” Review of Financial Studies,
Volume 25, Issue 2, Pages 377-407.
Ben-Rephael, Azi, Choi, Jaewon, and Itay Goldstein. 2021. “Mutual Fund Flows and Fluctuations in
Credit and Business Cycles,” Journal of Financial Economics, Volume 136, Pages 84-108.
Bernanke, Ben. 1990. “On the Predictive Power of Interest Rates and Interest Rate Spreads,” New
England Economic Review (Nov./Dec.), Pages 51-68.
Best, Ronald, and Hang Zhang. 1993. “Alternative Information Sources and the Information Content
of Bank Loans,” Journal of Finance, Volume 48, Pages 1507–1522.
Billett, Matthew T., Flannery, Mark J., and Jon A. Garfinkel. 1995. “The Effect of Lender Identity on a
Borrowing Firm's Equity Return,” Journal of Finance, Volume 50, Pages 699–718.
Campbell, Tim, and William Kracaw. 1980. “Information Production, Market Signalling, and the
Theory of Financial Intermediation,” Journal of Finance, Volume 35, Pages 863-882.
Campello, Murillo. 2002. “Internal Capital Markets in Financial Conglomerates: Evidence from Small
Bank Responses to Monetary Policy,” Journal of Finance, Volume 57, Issue 6, Pages 27732805.
Cetorelli, Nicola, and Linda Goldberg. 2012a. “Liquidity Management of US Global Banks: Internal
Capital Markets in the Great Recession,” Journal of International Economics, Volume 88, Issue
2, Pages 299-311.
Cetorelli, Nicola, and Linda Goldberg. 2012b. “Banking Globalization and Monetary Transmission,”
Journal of Finance, Volume 67, Issue 5, Pages 1811-1843.
Chodorow-Reich, Gabriel, Darmouni, Olivier, Luck, Stephan, and Matthew Plosser. 2022. “Bank
Liquidity Provision across the Firm Size Distribution,” Journal of Financial Economics, Volume
144, Issue 3, Pages 908-932.
Deyoung, Robert, Gron, Anne, Torna, Gokhan, and Andrew Winton. 2015. “Risk Overhang and Loan
Portfolio Decisions: Small Business Loan Supply before and during the Financial Crisis,”
Journal of Finance, Volume 70, Issue 6, Pages 2451-2487.
Diamond, Douglas W. 1984. “Financial Intermediation and Delegated Monitoring,” Review of
Economic Studies, Volume 51, Issue 3, Pages 393–414.
Diamond, Douglas W. 1991. “Monitoring and Reputation: The Choice between Bank Loans and
Directly Placed Debt,” Journal of Political Economy, Volume 99, Issue 4, Pages 689-721.
Duchin, Ray, and Sosyura, Denis. 2014. “Safer Ratios, Riskier Portfolios: Banks Response to
Government Aid,” Journal of Financial Economics, Volume 113, Issue 1, Pages 1-28.
Estrella, Arturo, and Gikas A. Hardouvelis. 1991. “The Term Structure as a Predictor of Real Economic
Activity,” Journal of Finance, Volume 46, Issue 2, Pages 555–576.
Fama, Eugene. 1985. “What's Different about Banks?” Journal of Monetary Economics, Volume
15, Issue 1, Pages 29–39.
Friedman, Benjamin M., and Kenneth N. Kuttner. 1992. “Money, Income, Prices, and Interest Rates,”
American Economic Review, Volume 82, Issue 3, Pages 472-492.
Gande, Amar, and Anthony Saunders. 2012. “Are Banks Still Special When There Is a Secondary
Market for Loans?” Journal of Finance, Volume 67, Issue 5, Pages 1649-1684.
Giglio, Stefano, Kelly, Bryan, and Seth Pruitt. 2016. “Systemic Risk and the Macroeconomy: An
Empirical Evaluation,” Journal of Financial Economics, Volume 119, Issue 3, Pages 457-471.
Gertler, Mark, and Simon Gilchrist. 1994. “Monetary Policy, Business Cycles, and the Behavior of
Small Manufacturing Firms,” Quarterly Journal of Economics, Volume 109, Issue 2, Pages 309–
340.
Gertler, Mark, and Cara S. Lown. 1999. “The Information in the High-Yield Bond Spread for the
Business Cycle: Evidence and Some Implications,” Oxford Review of Economic Policy, Volume
15, Issue 3, Pages 132-150.
Gertner, Robert H., Scharfstein, David S., and Jeremy C. Stein. 1994. “Internal versus External Capital
Markets,” Quarterly Journal of Economics, Volume 109, Issue 4, Pages 1211–1230.
Gilchrist, Simon, and Egon Zakrajšek. 2012. “Credit Spreads and Business Cycle
Fluctuations,” American Economic Review, Volume 102, Issue 4, Pages 1692-1720.
Gilje, Erik, Loutskina, Elena, and Phil Strahan. 2016. “Exporting Liquidity: Branch Banking and
Financial Integration,” Journal of Finance, Volume 71, Issue 3, Pages 1159-1184.
Giroud, Xavier, and Holger M. Mueller. 2015. “Capital and Labor Reallocation within Firms,” Journal
of Finance, Volume 70, Issue 4, Pages 1767–1804.
Giroud, Xavier, and Holger M. Mueller. 2019. “Firms' Internal Networks and Local Economic
Shocks,” American Economic Review, Volume 109, Issue 10, Pages 3617-3649.
Grossman, Sanford J., and Oliver D. Hart. 1986. “The Costs and Benefits of Ownership: A
Theory of Vertical and Lateral Integration,” Journal of Political Economy, Volume 94, Issue 4,
Pages 691–719.
Grossman, Sanford J., and Joseph E. Stiglitz. 1980. “On the Impossibility of Informationally Efficient
Markets,” American Economic Review, Volume 70, Issue 3, Pages 393-408.
Gupta, Arun, Sapriza, Horacio, and Vladimir Yankov. 2022. “The Collateral Channel and Bank
Credit,” Working Paper.
Hart, Oliver. 1995. “Firms, Contracts, and Financial Structure,” Oxford University Press.
Hart, Oliver, and John Moore. 1990. “Property Rights and the Nature of the Firm,” Journal of Political
Economy, Volume 98, Issue 6, Pages 1119-1158.
Harvey, Campbell R. 1989. “Forecasts of Economic Growth from the Bond and Stock Markets,”
Financial Analysts Journal, Volume 45, Issue 5, Pages 38-45.
Hoshi, Takeo, and Anil Kashyap. 2010. “Will the U.S. Bank Recapitalization Succeed? Eight Lessons
from Japan,” Journal of Financial Economics, Volume 97, Issue 3, Pages 398-417.
Houston, Joel, James, Christopher M., and David Marcus. 1997. “Capital Market Frictions and the
Role of Internal Capital Markets in Banking,” Journal of Financial Economics, Volume 46,
Issue 2, Pages 135-164.
James, Christopher M. 1987. “Some Evidence on the Uniqueness of Bank Loans,” Journal of
Financial Economics, Volume 19, Issue 2, Pages 217–235.
King, Robert G., and Ross Levine. 1993. “Finance and Growth: Schumpeter Might Be Right,” Quarterly
Journal of Economics, Volume 108, Issue 3, Pages 717–737.
Krishnamurthy, Arvind, and Tyler Muir. 2020. “How Credit Cycles Across a Financial Crisis,”
Working Paper No. 3579.
Krishnaswami, Sudha, Spindt, Paul A., and Venkat Subramaniam. 1999. “Information Asymmetry,
Monitoring and the Placement Structure of Corporate Debt,” Journal of Financial Economics,
Volume 51, Issue 3, Pages 407-434.
Lamont, Owen. 1997. “Cash Flow and Investment: Evidence from Internal Capital Markets,” Journal
of Finance, Volume 52, Issue 1, Pages 83-109.
Leland, Hayne E., and David H. Pyle. 1977. “Informational Asymmetries, Financial Structure, and
Financial Intermediation,” Journal of Finance, Volume 32, Issue 2, Pages 371-387.
López-Salido, David, Stein, Jeremy C., and Egon Zakrajšek. 2017. “Credit-Market Sentiment and the
Business Cycle,” Quarterly Journal of Economics, Volume 132, Issue 3, Pages 1373–1426.
Lummer, Scott L., and John J. McConnell. 1989. “Further Evidence on the Bank Lending Process and
the Capital-Market Response to Bank Loan Agreements,” Journal of Financial Economics,
Volume 25, Issue 1, Pages 99–122.
Maksimovic, Vojislav, and Gordon M. Phillips. 2002. “Do Conglomerate Firms Allocate Resources
Inefficiently across Industries? Theory and Evidence,” Journal of Finance, Volume 57, Issue 2,
Pages 721–767.
Maskara, Pankaj, and Donald J. Mullineaux. 2011. “Information Asymmetry and Self-Selection Bias
in Bank Loan Announcement Studies,” Journal of Financial Economics, Volume 101, Issue 3,
Pages 684-694.
Mikkelson, Wayne H., and M. Megan Partch. 1986. “Valuation Effects of Security Offerings and the
Issuance Process,” Journal of Financial Economics, Volume 15, Issues 1-2, Pages 31-60.
Nakamura, Leonard. 1993. “Commercial Bank Information: Implications for the Structure of
Banking,” in Michael Klausner and Lawrence J. White, eds., Structural Change in Banking,
Irwin, New York, Pages 131-160.
Peek, Joe, and Eric S. Rosengren. 1997. “The International Transmission of Financial Shocks: The
Case of Japan,” American Economic Review, Volume 87, Issue 4, Pages 495–505.
Peek, Joe, and Eric S. Rosengren. 2000. “Collateral Damage: Effects of the Japanese Bank Crisis on
Real Activity in the United States,” American Economic Review, Volume 90, Issue 1, Pages 30–
45.
Petersen, Mitchell, and Raghuram Rajan. 1994. “The Benefits of Lending Relationships: Evidence
from Small Business Data,” Journal of Finance, Volume 49, Issue 1, Pages 3–37.
Petersen, Mitchell, and Raghuram Rajan. 2002. “Does Distance Still Matter? The Information
Revolution in Small Business Lending,” Journal of Finance, Volume 57, Issue 6, Pages 25332570.
Philippon, Thomas. 2009. “The Bond Market’s Q,” Quarterly Journal of Economics, Volume 124,
Issue 3, Pages 1011-1056.
Philippon, Thomas, and Phillip Schnabl. 2013. “Efficient Recapitalization,” Journal of Finance,
Volume 68, Issue 1, Pages 1-42.
Qian, Jun, Strahan, Phil, and Zhishu Yang. 2015. “The Impact of Incentives and Communication Costs
on Information Production and Use: Evidence from Bank Lending,” Journal of Finance, Volume
70, Issue 4, Pages 1457-1483.
Ramakrishnan, Ram T. S., and Anjan V. Thakor. 1984. “Information Reliability and a Theory of
Financial Intermediation,” Review of Economic Studies, Volume 51, Issue 3, Pages 415–432.
Saunders, Anthony, Spina, Alessandro, Steffen, Sascha, and Streitz, Daniel. 2021. “Corporate Loan
Spreads and Economic Activity,” Working Paper.
Schnabl, Philipp. 2012. “The International Transmission of Bank Liquidity Shocks: Evidence from an
Emerging Market,” Journal of Finance, Volume 67, Issue 3, Pages 897-932.
Schularick, Moritz, and Alan M. Taylor. 2012. “Credit Booms Gone Bust: Monetary Policy, Leverage
Cycles, and Financial Crises, 1870-2008,” American Economic Review, Volume 102, Issue 2,
Pages 1029-1061.
Shin, Hyun-Han, and Rene M. Stulz. 1998. “Are Internal Capital Markets Efficient?” Quarterly
Journal of Economics, Volume 113, Issue 2, Pages 531–552.
Stein, Jeremy C. 1997. “Internal Capital Markets and the Competition for Corporate Resources,”
Journal of Finance, Volume 52, Issue 1, Pages 111–133.
Taliaferro, Ryan. 2009. “How Do Banks Use Bailout Money? Optimal Capital Structure, New Equity,
and the TARP,” Working Paper.
Veronesi, Pietro, and Luigi Zingales. 2010. “Paulson's Gift,” Journal of Financial Economics, Volume
97, Issue 3, Pages 339-368.
Williamson, Oliver E. 1975. Market and Hierarchies: Analysis and Antitrust Implications. Free Press,
New York, NY.
Tables and Figures
Table 1. Summary Statistics
This table provides summary statistics for key panel variables at the subsidiary, bank holding company, and subsidiarycounty levels, and various state-level macroeconomic data. The table reports the observations, mean, standard deviation, 10th,
25th, median, 75th, and 90th percentiles for each variable. Source: Call Reports, FR Y9C, CAMELS, SR-SABR, FFIEC
Community Reinvestment Act data, US Treasury TARP CPP Transaction Reports, Census Bureau (Nonemployer Statistics,
Statistics of US Businesses), Bureau of Economic Analysis, Bureau of Labor Statistics, FHFA, CRSP.
Percentile
Bank Subsidiary Variables (Subsidiary-Quarter Level)
Internal Equity Injection (%)
Internal Equity Injection on TARP Arrival (%)
MBHC Excess Internal Equity Injection on TARP Arrival
(%)
SBHC Excess Internal Equity Injection on TARP Arrival
(%)
CAMELS Rating
Sup-estimated Probability of Failure (%)
Sup-estimated Probability of Ratings Downgrade (%)
Nat. Log of Bank Sub Total Assets (Thous. USD)
Bank Sub Tier 1 Capital Ratio (%)
Bank Sub Liquid Asset Ratio (%)
Bank Sub Deposit Ratio (%)
Bank Sub ROA, Quarterly (%)
Bank Sub ROE, Quarterly (%)
Bank Holding Company Variables (BHC-Quarter
Level)
Nat. Log of BHC Total Assets (Thous. USD)
BHC Tier 1 Capital Ratio (%)
BHC Liquid Asset Ratio (%)
BHC Deposit Ratio (%)
BHC ROA, Quarterly (%)
Number of Subsidiaries per MBHC
TARP Arrival Date
Obs
Std
Dev
10th
Perc
25th
Perc
50th
Perc
75th
Perc
90th
Perc
9037
9037
4633
0.46
1.04
0.72
0.00
0.00
-0.63
0.00
0.00
-0.44
0.00
0.53
-0.19
0.01
1.92
0.13
0.40
2.63
1.06
4251
1.08
-1.49
-1.18
0.10
1.02
1.38
9036
9036
9036
9037
9016
9037
9037
9037
9037
0.73
11.80
26.96
1.51
15.98
12.51
15.58
0.52
10.56
1.00
0.01
0.71
12.19
9.00
2.90
64.69
-0.39
-4.33
2.00
0.03
1.53
13.03
9.66
5.18
73.03
0.06
0.64
2.00
0.17
6.19
13.80
10.85
8.92
79.51
0.27
3.16
2.00
0.84
24.91
14.76
12.76
14.29
84.19
0.44
5.30
3.00
5.25
65.49
16.27
15.79
22.91
87.43
0.61
7.27
5663
1.53
13.31
13.66
14.28
15.27
16.69
5663
5663
5663
5659
-
2.82
8.15
8.37
0.28
5
-
8.85
5.45
66.33
-0.27
2
08Q4
9.97
7.82
72.32
0.04
2
08Q4
11.39
11.83
78.36
0.15
3
09Q1
13.14
16.95
82.98
0.25
5
09Q1
14.98
23.13
86.06
0.32
10
09Q1
3.22
4.32
5.71
5.81
7.15
8.67
Small Business Lending Variable (Subsidiary-County-Year Level)
191678
Nat. Log of Small Bus. Loan Originations (Thous. USD)
Table 1: continued
Percentile
Local Macro Variables (State-Year Level)
MBHC Excess Internal Equity Injection on TARP Arrival
(%)
SBHC Excess Internal Equity Injection on TARP Arrival
(%)
Annual Small Business Employment Growth (%)
Annual Small Business Establishment Growth (%)
Annual Small Business Payroll Growth (%)
Nat Log of Small Business Avg Weekly Wage
Annual Nonemployer Firm Revenue Growth (%)
Annual Per Capita Real GDP Growth (%)
Annual Per Capita Personal Income Growth (%)
Annual Small Business Credit Origination Growth (%)
Annual House Price Index Growth (%)
Annual Return on State-Based Stock Portfolio (%)
Obs
Std
Dev
10th
Perc
25th Perc
50th
Perc
75th
Perc
90th
Perc
352
0.20
-0.06
0.00
0.17
0.29
0.51
376
0.59
-1.11
-0.86
-0.52
-0.02
0.54
370
370
369
373
311
392
392
392
392
389
1.78
4.41
4.42
6.31
-1.80
2.77
3.26
15.83
7.20
27.29
-2.83
-8.17
-7.70
6.39
1.67
-2.99
-2.01
-29.08
-6.21
-34.05
-1.52
-6.42
-5.64
6.47
3.44
-0.61
2.67
-12.38
-2.63
-6.25
-0.12
-3.05
-2.53
6.49
3.17
1.13
4.23
-0.22
0.10
6.37
0.94
-0.30
0.82
6.59
5.31
2.82
5.67
9.00
5.25
20.16
2.08
3.56
3.52
6.69
7.22
3.99
7.20
14.36
11.45
32.13
Table 2: The Internal Equity Injection of TARP CPP Funds
This table presents regression results using panel data specification of equation (1) in the text. The observations are quarterly
and span 2006:Q1 to 2011:Q4. The dependent variable is the magnitude of internal equity injections provided by the parent
holding company to a given bank subsidiary. This variable is measured by the percentage point increase in the subsidiary’s
capital ratio that results due to the parent’s internal equity injection. TARP Arrival Date is a 0/1 variable denoting the date
when the parent receives TARP CPP funds from the US government. CAMELS 2 Sub and CAMELS 3 Sub are 0/1 variables
denoting if the subsidiary has a CAMELS 2 and 3 rating, respectively, as of the date of TARP arrival. The baseline group are
subsidiaries with a CAMELS 1 rating as of the date of TARP arrival. The CAMELS * TARP Arrival Date interaction
coefficient captures the extra amount of internal equity injections that occur on the day of TARP arrival for subsidiaries with
different CAMELS scores. MBHC is a 0/1 variable that equals 1 if the parent is multi-bank holding company, as opposed to a
single-bank holding company (SBHC). The MBHC * TARP Arrival Date interaction coefficient captures the extra amount of
internal equity injections that MBHC parents provide to a given subsidiary on the day of TARP arrival as compared to SBHC
parents. This coefficient provides an implied estimate of the relative reallocation option value that MBHCs possessed when
they withheld equity injections. Controls include lagged subsidiary and bank holding company level characteristics. All
specifications include fixed effects for bank holding company, bank subsidiary, and date. Standard errors are clustered by
bank holding company. *, **, and *** denote statistical significance at the 10 percent, 5 percent, and 1 percent levels,
respectively. Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP Transaction Reports.
Internal Equity Injection
/ Total Sub Assets (%)
VARIABLES
All BHCs
(1)
MBHCs
(2)
SBHCs
(3)
TARP Arrival Date [0/1]
CAMELS 2 Sub [0/1] * TARP Arrival Date [0/1]
CAMELS 3 Sub [0/1] * TARP Arrival Date [0/1]
MBHC [0/1]
MBHC [0/1] * TARP Arrival Date [0/1]
1.216***
0.050
0.435**
-0.025
-1.031***
0.054
0.159*
0.643**
1.407***
-0.153
0.185
Lag Sup-estimated Prb(Failure)
Lag Sup-estimated Prb(Ratings downgrade)
Lag Ln(Bank Sub Total Assets)
Lag Bank Sub Tier 1 Capital Ratio
Lag Bank Sub Liquid Asset Ratio
Lag Bank Sub Deposit Ratio
Lag Bank Sub ROA
Lag Ln(BHC Total Assets)
Lag BHC Tier 1 Capital Ratio
Lag BHC Liquid Asset Ratio
Lag BHC Deposit Ratio
Lag BHC ROA
Constant
0.001
0.002***
-0.039
-0.005*
0.002
0.000
-0.056***
0.022
0.006
-0.002
0.000
-0.045
0.253
0.003*
0.003***
-0.032
-0.002
0.001
0.000
-0.066***
0.058
0.003
-0.002
0.000
-0.096**
-0.422
-0.003**
0.001***
-0.012
-0.091***
0.003
0.005
0.024
-0.051
0.049***
-0.000
-0.004
-0.080
1.375
9,037
0.309
Y
4,708
0.268
Y
4,315
0.385
Y
Observations
R-squared
BHC FE, BankSub FE, Date FE
Standard Errors Clustered by BHC
*** p<0.01, ** p<0.05, * p<0.1
Table 3: The Predictive Value of Excess Internal Equity Injections for Post-Crisis Subsidiary
Profitability
This table presents regression results using the panel data specification of equation (2) in the text. The observations are
quarterly and span from 5 quarters before and after each BHC receives TARP CPP funds. The dependent variables represent
the subsidiary-level annualized return-on-assets and return-on-equity. Excess internal equity injections represent the excess
amount of internal equity injections provided to a bank subsidiary by parent holding companies on the date of TARP arrival,
after controlling for the subsidiary’s CAMELS exam score. This variable is measured as the percentage point increase in the
subsidiary’s capital ratio that results from the parent’s excess internal equity injection upon TARP arrival. PostTARP is a 0/1
variable that equals 1 for all dates on and after the BHC receives TARP CPP funds. The interaction coefficient captures the
relative difference in post-TARP subsidiary performance across subsidiaries that received different ex-ante amounts of excess
internal equity injections from their respective parents upon TARP arrival. This result shows that the parent’s excess internal
equity injection predicts future subpar post-crisis performance by the subsidiary. MBHC is a 0/1 variable that equals 1 if the
parent is a multi-bank holding company, as opposed to a single-bank holding company (SBHC). Controls include lagged
subsidiary and bank holding company level characteristics. All specifications include fixed effects for bank holding
company, bank subsidiary, and date. Standard errors are clustered by bank holding company. *, **, and *** denote statistical
significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Source: Call Reports, FR Y9C, CAMELS, SRSABR, US Treasury TARP CPP Transaction Reports
Bank Sub ROA
VARIABLES
Bank Sub ROE
MBHC
(1)
SBHC
(2)
MBHC
(3)
SBHC
(4)
PostTARP * Excess Internal Equity Injection on TARP
Arrival
-0.434**
0.036
-3.288**
0.363
Lag Sup-estimated Prb(Fail)
Lag Sup-estimated Prb(Downgrade)
Lag CAMELS score 2 [0/1]
Lag CAMELS score 3 [0/1]
Lag Ln(Bank Sub Total Assets)
Lag Bank Sub Tier 1 Cap Ratio
Lag Bank Sub Liquid Asset Ratio
Lag Bank Sub Deposit Ratio
Lag Ln(BHC Total Assets)
Lag BHC Tier 1 Cap Ratio
Lag BHC Liquid Asset Ratio
Lag BHC Deposit Ratio
Constant
-0.017
-0.021***
-0.502
-0.703
0.507
-0.018*
0.002
0.014
0.634
0.117
-0.008
0.008
-18.372*
0.010
-0.018***
0.106
-0.476*
-1.584
-0.198***
-0.033
-0.135**
1.743*
0.083
0.021
0.128**
0.059
-0.058
-0.195***
-4.913
-5.636
3.106
-0.033
-0.029
0.068
5.885
0.549
-0.068
-0.196
-115.886
0.124
-0.144***
0.281
-3.942*
-10.509
-0.614
-0.005
-0.054
13.966
-0.578
0.065
-0.034
-22.874
2,448
0.605
Y
1,984
0.480
Y
2,240
0.606
Y
1,874
0.539
Y
Observations
R-squared
BHC FEs, Bank Sub FEs, and Date FEs
Standard Errors Clustered by BHC
*** p<0.01, ** p<0.05, * p<0.1
Table 4: The Predictive Value of Excess Internal Equity Injections for Post-Crisis Subsidiary
Distress (Supervisory Ratings)
This table presents regression results using panel data specification of equation (2) in the text. The observations are quarterly
and span 5 quarters before and after each BHC receives TARP CPP funds. The dependent variable represents the subsidiarylevel CAMELS score assessed by supervisory onsite exams. A higher CAMELS rating means more distress. Excess internal
equity injections represent the excess amount of internal equity injections provided to a bank subsidiary by parent holding
companies on the date of TARP arrival, after controlling for the subsidiary’s CAMELS exam score. This variable is
measured as the percentage point increase in the subsidiary’s capital ratio that results from the parent’s excess internal equity
injection upon TARP arrival. PostTARP is a 0/1 variable that equals 1 for all dates on and after the BHC receives TARP CPP
funds. The interaction coefficient captures the relative difference in post-TARP supervisory assessments across subsidiaries
that received different ex-ante amounts of excess internal equity injections from their respective parents upon TARP arrival.
This result shows that the parent’s excess internal equity injection predicts future post-crisis distress of the subsidiary.
MBHC is a 0/1 variable that equals 1 if the parent is a multi-bank holding company, as opposed to a single-bank holding
company (SBHC). Controls include lagged subsidiary and bank holding company level characteristics. All specifications
include fixed effects for bank holding company, bank subsidiary, and date. Standard errors are clustered by bank holding
company. *, **, and *** denote statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Source:
Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP Transaction Reports.
Subsidiary CAMELS
Rating
VARIABLES
MBHC
(1)
SBHC
(2)
PostTARP * Excess Internal Equity Injection on
TARP Arrival
0.0792*
-0.0211
Lag Bank Sub Tier 1 Cap Ratio
Lag Bank Sub Liquid Asset Ratio
Lag Ln(Bank Sub Total Assets)
Lag Bank Sub Deposit Ratio
Lag Bank Sub ROA
Lag BHC Tier 1 Cap Ratio
Lag BHC Liquid Asset Ratio
Lag Ln(BHC Total Assets)
Lag BHC Deposit Ratio
Lag BHC ROA
Constant
0.00407
0.00441**
0.152
-0.000342
-0.0769***
-0.0387**
0.00813
-0.295
0.0141***
-0.140*
3.800
0.0290
0.0316**
0.202
0.00586
-0.0344
-0.0567***
-0.0117
-0.930***
0.000844
-0.181
12.28***
Observations
R-squared
BHC FE, Bank FEs, and Date FEs
Standard Errors Clustered by BHC
*** p<0.01, ** p<0.05, * p<0.1
2,458
0.760
Y
2,003
0.685
Y
Table 5: The Predictive Value of Excess Internal Equity Injections for Post-Crisis Nonemployer
Firm Revenue Growth at the State Level
This table presents regression results using panel data specification from equation (4) in the text. The observations are yearly
and span 2004 to 2011. The dependent variable represents the annual growth rate in revenues of nonemployer firms in a
given state and year. Excess internal equity injections represent the excess amount of internal equity injections provided to a
bank subsidiary by parent holding companies on the date of TARP arrival, after controlling for the subsidiary’s CAMELS
rating. This variable is measured as the percentage point increase in the subsidiary’s capital ratio that results from the parent’s
excess internal equity injection upon TARP arrival. In this table, we use an aggregated version of these signals by calculating
state-level averages of the individual bank-level excess injection signals, weighted by each subsidiary’s ex-ante small
business credit exposures in the given state. For the independent variable, the left panel uses the state-aggregated excess
internal equity injection calculated from multi-bank holding companies (MBHCs), whereas the right panel uses the stateaggregated excess internal equity injection calculated from single-bank holding companies (SBHCs). Post Crisis is a 0/1
variable that equals 1 for 2009 onwards. The coefficient of the interaction term Post Crisis * State-Level Excess Internal
Equity Injection captures the relative difference in post-crisis state-level annual revenue growth at nonemployer firms across
states that experienced different ex-ante excess internal equity injection signals. Controls include lagged state-level
macroeconomic variables. All specifications include fixed effects for state and year. Standard errors are clustered by state. *,
**, and *** denote statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Source: Call
Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP Transaction Reports, Census Bureau (Nonemployer
Statistics), Bureau of Economic Analysis, FFIEC Community Reinvestment Act data, FHFA, CRSP.
State-Level Nonemployer Firm Annual Revenue Growth (%)
VARIABLES
Signal from MBHCs
(1)
(2)
Signal from SBHCs
(3)
(4)
Post Crisis * State-Level Excess Internal Equity Injection
on TARP Arrival
-2.292**
-2.325**
-0.718
-0.538
Lag State-Level Per Capita Real GDP Growth
Lag State-Level Per Capita Personal Income Growth
Lag State-Level Small Business Credit Origination Growth
Lag State-Level Housing Price Index Growth
Lag State-Level Stock Price Return
Constant
0.146*
0.0312
-0.00137
0.109
0.0593
0.0100
2.724***
0.139
-0.00179
-0.0151
0.0766**
0.0118**
2.406***
2.422***
0.0979
0.0346
-0.00535
0.0911***
0.00909*
2.098***
344
0.718
Y
341
0.729
Y
352
0.703
Y
349
0.714
Y
Observations
R-squared
State FEs, Year FEs
Standard Errors Clustered by State
*** p<0.01, ** p<0.05, * p<0.1
Table 6: The Predictive Value of Excess Internal Equity Injections for Post-Crisis Small Business
Employment Growth at the State Level
This table presents regression results using panel data specification from equation (4) in the text. The observations are yearly,
spanning 2004 to 2011. The dependent variable represents the annual growth rate in employment in a given state and year
(for establishments with nine or fewer employees). Excess internal equity injections represent the excess amount of internal
equity injections provided to a bank subsidiary by parent holding companies on the date of TARP arrival, after controlling for
the subsidiary’s CAMELS rating. This variable is measured as the percentage point increase in the subsidiary’s capital ratio
that results from the parent’s excess internal equity injection upon TARP arrival. For the independent variable, we use two
aggregated versions of these signals by calculating state-level averages of the individual bank-level excess injection signals,
weighted by each subsidiary’s ex-ante small business credit exposures in the given state. The first version (left panel) uses the
state-aggregated excess internal equity injection calculated from multi-bank holding companies (MBHCs), whereas the
second version (right panel) uses the state-aggregated excess internal equity injection calculated from single-bank holding
companies (SBHCs). Post Crisis is a 0/1 variable that equals 1 for 2009 onwards. The coefficient of the interaction term Post
Crisis * State-Level Excess Internal Equity Injection captures the relative difference in post-crisis small business annual
employment growth across states that experienced different ex-ante excess internal equity injection signals. Controls include
lagged state-level macroeconomic variables. All specifications include fixed effects for state and year. Standard errors are
clustered by state. *, **, and *** denote statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively.
Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP Transaction Reports, Census Bureau
(Statistics of US Businesses), Bureau of Economic Analysis, FFIEC Community Reinvestment Act data, FHFA, CRSP.
Small Business Annual Employment Growth (%)
VARIABLES
Post Crisis * State-Level Excess Internal Equity Injection on
TARP Arrival
Lagged State-Level Per Capita Real GDP Growth
Lagged State-Level Per Capita Personal Income Growth
Lagged State-Level Small Business Credit Origination Growth
Lagged State-Level Housing Price Index Growth
Lagged State-Level Stock Price Return
Constant
Observations
R-squared
State FEs, Year FEs
Standard Errors Clustered by State
*** p<0.01, ** p<0.05, * p<0.1
Signal from MBHCs
(1)
(2)
Signal from SBHCs
(3)
(4)
-0.931*
-0.973**
-0.250
-0.158
0.0251
0.145***
0.0134**
0.0242
0.146***
0.0176***
-0.582***
0.0254
0.128***
0.00653
0.0511***
0.00430
-0.760***
-0.712***
0.0233
0.136***
0.0106*
0.0527***
0.00365
-0.898***
339
0.811
Y
335
0.824
Y
349
0.806
Y
345
0.818
Y
Table 7: The Predictive Value of Excess Internal Equity Injections for Post-Crisis Small Business
Establishment Growth at the State-Level
This table presents regression results using panel data specification from equation (4) in the text. The observations are yearly,
spanning 2004 to 2011. The dependent variable represents the annual growth rate in establishments in a given state and year
(for establishments with nine or fewer employees). Excess internal equity injections represent the excess amount of internal
equity injections provided to a bank subsidiary by parent holding companies on the date of TARP arrival, after controlling for
the subsidiary’s CAMELS rating. This variable is measured as the percentage point increase in the subsidiary’s capital ratio
that results from the parent’s excess internal equity injection upon TARP arrival. In this table, we use two aggregated
versions of these signals by calculating state-level averages of the individual bank-level excess injection signals, weighted by
each subsidiary’s ex-ante small business credit exposures in the given state. The first version (left panel) uses the stateaggregated excess internal equity injection signal calculated from multi-bank holding companies (MBHCs), whereas the
second version (right panel) uses the state-aggregated excess internal equity injection signal calculated from single-bank
holding companies (SBHCs). Post Crisis is a 0/1 variable that equals 1 for 2009 onwards. The coefficient of the interaction
term Post Crisis * State Level Excess Internal Equity Injection captures the relative difference in post-crisis small business
annual establishment growth across states that experienced different ex-ante excess internal equity injection signals. Controls
include lagged state-level macroeconomic variables. All specifications include fixed effects for state and year. Standard
errors are clustered by state. *, **, and *** denote statistical significance at the 10 percent, 5 percent, and 1 percent levels,
respectively. Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP Transaction Reports, Census
Bureau (Statistics of US Businesses), Bureau of Economic Analysis, FFIEC Community Reinvestment Act data, FHFA,
CRSP.
Small Business Annual Establishment Growth (%)
VARIABLES
Post Crisis * State-Level Excess Internal Equity Injection on
TARP Arrival
Lag State-Level Per Capita Real GDP Growth
Lag State-Level Per Capita Personal Income Growth
Lag State-Level Small Business Credit Origination Growth
Lag State-Level Housing Price Index Growth
Lag State-Level Stock Price Return
Constant
Observations
R-squared
State FE, Year FE
Standard Errors Clustered by State
*** p<0.01, ** p<0.05, * p<0.1
Signal from MBHCs
(1)
(2)
Signal from SBHCs
(3)
(4)
-1.085**
-1.077**
-0.233
-0.106
-0.0185
0.173***
0.00406
-0.00790
0.156***
0.00771
-0.678***
-0.0196
0.145***
-0.00560
0.0641***
0.00170
-0.842***
-0.750***
-0.0107
0.134***
-0.00262
0.0677***
0.00185
-0.924***
346
0.786
Y
342
0.803
Y
352
0.778
Y
348
0.798
Y
Table 8: The Predictive Value of Excess Internal Equity Injections for Post-Crisis Small Business
Payroll Growth at the State-Level
This table presents regression results using panel data specification from equation (4) in the text. The observations are yearly,
spanning 2004 to 2011. The dependent variable represents the annual growth rate in small business payroll expenses in a
given state and year (for establishments with nine or fewer employees). Excess internal equity injections represent the excess
amount of internal equity injections provided to a bank subsidiary by parent holding companies on the date of TARP arrival,
after controlling for the subsidiary’s CAMELS rating. This variable is measured as the percentage point increase in the
subsidiary’s capital ratio that results from the parent’s excess internal equity injection upon TARP arrival. In this table, we
use two aggregated versions of these signals by calculating state-level averages of the individual bank-level excess injection
signals, weighted by each subsidiary’s ex-ante small business credit exposures in the given state. The first version (left panel)
uses the state-aggregated excess internal equity injection signal calculated from multi-bank holding companies (MBHCs),
whereas the second version (right panel) uses the state-aggregated excess internal equity injection signal calculated from
single-bank holding companies (SBHCs). Post Crisis is a 0/1 variable that equals 1 for 2009 onwards. The coefficient of the
interaction term Post Crisis * State Level Excess Internal Equity Injection captures the relative difference in post-crisis small
business annual payroll expense growth across states that experienced different ex-ante excess internal equity injection
signals. Controls include lagged state-level macroeconomic variables. All specifications include fixed effects for state and
year. Standard errors are clustered by state. *, **, and *** denote statistical significance at the 10 percent, 5 percent, and 1
percent levels, respectively. Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP Transaction
Reports, Census Bureau (Statistics of US Businesses), Bureau of Economic Analysis, FFIEC Community Reinvestment Act
data, FHFA, CRSP.
Small Business Annual Payroll Growth (%)
VARIABLES
Signal from MBHCs
(1)
(2)
Signal from SBHCs
(3)
(4)
Post Crisis * State-Level Excess Internal Equity Injection on
TARP Arrival
-1.723**
-1.743**
-0.351
-0.181
Lag State-Level Per Capita Real GDP Growth
Lag State-Level Per Capita Personal Income Growth
Lag State-Level Small Business Credit Origination Growth
Lag State-Level Housing Price Index Growth
Lag State-Level Stock Price Return
Constant
0.0291
0.195**
0.00100
0.0163
0.207***
0.00569
1.753***
0.0229
0.154*
-0.00929
0.0880***
0.00767
1.470***
1.540***
0.00690
0.176**
-0.00537
0.0930***
0.00716
1.243***
342
0.714
Y
340
0.725
Y
351
0.702
Y
349
0.714
Y
Observations
R-squared
State FE, Year FE
Standard Errors Clustered by State
*** p<0.01, ** p<0.05, * p<0.1
Table 9: The Predictive Value of Excess Internal Equity Injections for Post-Crisis Small Business
Average Weekly Wage at the State-Level
This table presents regression results using panel data specification from equation (4) in the text. The observations are yearly,
spanning 2004 to 2011. The dependent variable represents the natural log of the average weekly wage in a given state and
year (for establishments with nine or fewer employees). Excess internal equity injections represent the excess amount of
internal equity injections provided to a bank subsidiary by parent holding companies on the date of TARP arrival, after
controlling for the subsidiary’s CAMELS rating. This variable is measured as the percentage point increase in the
subsidiary’s capital ratio that results from the parent’s excess internal equity injection upon TARP arrival. In this table, we
use two aggregated versions of these signals by calculating state-level averages of the individual bank-level excess injection
signals, weighted by each subsidiary’s ex-ante small business credit exposures in the given state. The first version (left panel)
uses the state-aggregated excess internal equity injection signal calculated from multi-bank holding companies (MBHCs),
whereas the second version (right panel) uses the state-aggregated excess internal equity injection signal calculated from
single-bank holding companies (SBHCs). Post Crisis is a 0/1 variable that equals 1 for 2009 onwards. The coefficient of the
interaction term Post Crisis * State Level Excess Internal Equity Injection captures the relative difference in post-crisis small
business average log weekly wages across states that experienced different ex-ante excess internal equity injection signals.
Controls include lagged state-level macroeconomic variables. All specifications include fixed effects for state and year.
Standard errors are clustered by state. *, **, and *** denote statistical significance at the 10 percent, 5 percent, and 1 percent
levels, respectively. Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP Transaction Reports,
Census Bureau (Statistics of US Businesses), Bureau of Economic Analysis, FFIEC Community Reinvestment Act data,
FHFA, CRSP.
Ln(Small Business Avg Weekly Wage)
VARIABLES
Signal from MBHCs
(1)
(2)
Signal from SBHCs
(3)
(4)
Post Crisis * State-Level Excess Internal Equity Injection on
TARP Arrival
-0.0407**
-0.0331**
0.000616
0.00307
Lag State-Level Per Capita Real GDP Growth
Lag State-Level Per Capita Personal Income Growth
Lag State-Level Small Business Credit Origination Growth
Lag State-Level Housing Price Index Growth
Lag State-Level Stock Price Return
Constant
0.000531
0.00146*
0.000386***
0.000649
0.00152*
0.000513***
6.480***
0.000402
0.00152*
0.000188*
0.000952***
-6.03e-05
6.478***
6.477***
0.000500
0.00162*
0.000280**
0.00103***
-6.81e-05
6.476***
395
0.986
Y
391
0.987
Y
387
0.985
Y
383
0.987
Y
Observations
R-squared
State FE, Year FE
Standard Errors Clustered by State
*** p<0.01, ** p<0.05, * p<0.1
Table 10: The Predictive Value of Excess Internal Equity Injections for Post-Crisis Small Business
Credit Originations
This table presents regression results using panel data specification from equation (5) in the text. The observations are yearly,
spanning 2005 to 2011. The dependent variable represents the natural log of new credit originations by a subsidiary to small
businesses located in a given county and year. Excess internal equity injections represent the excess amount of internal equity
injections provided to a bank subsidiary by parent holding companies on the date of TARP arrival, after controlling for the
subsidiary’s CAMELS rating. This variable is measured as the percentage point increase in the subsidiary’s capital ratio that
results from the parent’s excess internal equity injection upon TARP arrival. PostTARP is a 0/1 variable that equals 1 for all
dates on and after the BHC receives TARP CPP funds. The interaction coefficient captures the relative post-TARP difference
in new county-specific credit originations across subsidiaries that received different ex-ante amounts of internal equity
injections from their respective parents. MBHC is a 0/1 variable that equals 1 if the parent is multi-bank holding company,
as opposed to a single-bank holding company (SBHC). Controls include lagged subsidiary and bank holding company level
characteristics. All specifications include fixed effects for bank holding company, bank subsidiary, and year. Standard errors
are clustered by bank holding company. *, **, and *** denote statistical significance at 10 percent, 5 percent, and 1 percent
levels, respectively. Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP Transaction Reports,
FFIEC Community Reinvestment Act data.
Ln(Small Business Credit Originations)
VARIABLES
(1)
MBHC
(2)
(3)
Post TARP * Excess Internal Equity
Injection on TARP Arrival Date
-0.460***
-0.432***
Lag Sup-estimated Prb(Fail)
Lag Sup-estimated Prb(Downgrade)
Lag Ln(Bank Sub Total Assets)
Lag Bank Sub Tier 1 Cap Ratio
Lag Bank Sub Liquid Asset Ratio
Lag Bank Sub Deposit Ratio
Lag Bank Sub ROA
Lag Ln(BHC Total Assets)
Lag BHC Tier 1 Cap Ratio
Lag BHC Liquid Asset Ratio
Lag BHC Deposit Ratio
Lag BHC ROA
Constant
0.007
-0.002
-0.146
3.984
-2.299**
-0.006**
0.155***
0.093
-2.578
3.105
0.011
-38.764***
5.790
201,689
Observations
0.538
R-squared
BHC FE
Y
CAMELS FE
Y
BankSub FE
Y
County FE
Y
Year FE
Y
County*Year FE
BankSub*County FE
Standard Errors Clustered by County and BHC
*** p<0.01, ** p<0.05, * p<0.1
(4)
SBHC
(5)
(6)
-0.475***
-0.003
0.014
-0.012
0.006
-0.001
-0.174
3.807
-2.429***
-0.006**
0.154***
0.091
-3.551
3.643**
0.014*
-37.990***
6.206
0.001
-0.002
-0.030
4.443*
-2.895***
-0.006**
0.152***
0.131
-2.619
3.540
0.025***
-41.178***
2.084
0.001
0.000
0.254**
8.247***
-0.806
0.002
0.046
-0.374
-2.642
-0.010
-0.016
-5.415
8.654**
0.001
0.001
0.187
9.179***
-0.930
-0.004
0.073*
-0.285
-1.938
-0.179
-0.010
-9.731*
8.220**
-0.002
-0.000
0.587***
1.601
-0.774
0.007
-0.013
0.171
1.448
-0.779
0.008
2.149
-7.131***
201,578
0.561
Y
Y
Y
191,678
0.892
Y
Y
59,712
0.356
Y
Y
Y
50,943
0.873
Y
Y
Y
Y
Y
63,534
0.307
Y
Y
Y
Y
Y
Y
Y
Y
Figure 1. Consolidated View of Bank Holding Companies
Note: This figure denotes a simplified representation of a bank holding company structure comprised of three legally-separate
entities - namely, the parent holding company and two commercial bank subsidiaries. The left panel represents a traditional
consolidated balance sheet of a bank holding company (similar to what might be presented in a 10-K annual report). The
right panel shows that the consolidated balance sheet can be broken down by which assets and liabilities are owned by each
legal entity.
Figure 2. Disaggregated “Internal Capital Markets” View of Bank Holding Companies.
Note: This figure further disaggregates the balance sheet from the right panel in Figure 1 into three individual balance sheets
(parent holding company-only, bank subsidiary 1, and bank subsidiary 2). It is important to note that the key variable of our
study, the parent holding company's internal equity investment becomes visible in this disaggregated “internal capital
markets” view of the banking organization.
Figure 3: Scarcity of Internal Equity Injections in the Cross Section
Note: This figure presents a cross-sectional plot of the average internal equity injections received by bank subsidiaries with a
CAMELS ratings of 1, 2, and 3 on the date of TARP arrival. During the 2008 crisis, regulators imposed a constraint on the
internal capital markets by making internal equity injections difficult to reverse. The top panel shows that multi-bank holding
companies (MBHCs) portioned the use of TARP CPP funds by providing higher internal equity injections to subsidiaries
with worse regulatory distress ratings (i.e., a CAMELS rating of 3, 2, and then 1). This behavior corroborates that this
constraint was costly for MBHCs during the crisis, as injected equity could potentially have gotten “trapped” ex-post and
thus be unavailable for reallocation to sibling subsidiaries if future shocks arose. In this way, MBHCs treated internal equity
injections as a scarce good ex-ante and, thus, were more willing to incur this “trapped” capital cost when the subsidiary’s precrisis distress rating was worse. In comparison, the bottom panel shows that single-bank holding companies (SBHCs)
provided a uniform amount of internal equity injections regardless of the subsidiary’s distress rating. This behavior is
consistent with the notion that the regulatory constraint was not costly for SBHCs. SBHCs likely did not value the option to
flexibly reallocate funds in the future, given that they had no sibling subsidiaries to reallocate funds to. Source: Call Reports,
FR Y9C, CAMELS, SR SABR, US Treasury TARP CPP Transaction Reports.
Figure 4: Scarcity of Internal Equity Injections in the Time Series
Note: This figure presents a time series plot of the average internal equity injections received by bank subsidiaries with a
CAMELS ratings of 1, 2, and 3. During the 2008 crisis, enhanced scrutiny from regulators imposed a constraint on the
internal capital markets by making internal equity injections difficult to reverse. This constraint created an option value for
lending funds to the subsidiary instead of injecting equity because internal debt exposures were exempt and could be
reallocated anytime. The top panel suggests multi-bank holding companies (MBHCs) found this constraint costly (because
the possibility of “trapped” equity meant it could not be reallocated to sibling subsidiaries ex-post) and, thus, treated internal
equity injections as a scarce resource by prioritizing recapitalizations at the most distressed subsidiaries sooner in time.
Specifically, the top panel shows that subsidiaries with a CAMELS rating of 3 were immediately recapitalized on the first
date that TARP CPP funds arrived at the MBHC parent. Meanwhile, recapitalizations at subsidiaries with a CAMELS rating
of 2 were lower on date 0 and were smoothed after the arrival of TARP. Finally, recapitalizations at subsidiaries with a
CAMELS rating of 1 were even lower on date 0 and also smoothed through time after the arrival of TARP. In contrast, the
bottom panel shows the opposite: single-bank holding companies (SBHCs) did not treat internal equity injections as a scarce
resource (e.g., the potential for “trapped” equity was less costly for SBHCs because they had no other sibling subsidiaries to
reallocate to). Thus, SBHCs recapitalized their sole subsidiary uniformly (regardless of distress level) all in one shot on date
0, with no evidence of smoothing through time. The fact that MBHC internal equity injections to subsidiaries with CAMELS
ratings of 2 and 3 were withheld on date 0 suggests MBHCs likely valued the flexibility to reallocate these funds as equity
injections in the future in case shocks arose. Source: Call Reports, FR Y9C, CAMELS, SR SABR, US Treasury TARP CPP
Transaction Reports.
Figure 5: Excess Internal Equity Injections Predicts the Subsidiary’s Post-Crisis Distress (Future
CAMELS Scores)
Note: This figure shows evidence that the amount of excess equity that the MBHC parent decided to internally inject upon
TARP arrival revealed its private information about the subsidiary’s ex-post distress post-crisis, as measured by future
CAMELS ratings. Specifically, this figure plots the 90th percent confidential intervals for the estimated coefficients 𝛽𝛽𝛼𝛼 for
each Date*Excess Internal Equity Injection term (𝑓𝑓𝑓𝑓𝑓𝑓 𝛼𝛼𝛼𝛼{−5,5} ) in the following specification: 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡 = 𝛽𝛽0 +
∑5𝛼𝛼=−5 𝛽𝛽𝛼𝛼 ∙ 𝟙𝟙�𝑡𝑡 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 + 𝛼𝛼� ∙ 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑡𝑡=𝑇𝑇𝐴𝐴𝐴𝐴𝐴𝐴 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 + 𝛽𝛽6 ∙
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽7 ∙ 𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑝𝑝,𝑡𝑡−1 + 𝛿𝛿𝑠𝑠 + 𝜋𝜋𝑝𝑝 + 𝜏𝜏𝑡𝑡 + 𝜀𝜀𝑠𝑠,𝑝𝑝,𝑡𝑡 . S stands for subsidiary, p stands for parent
holding company, and t stands for time. Note that a higher CAMELS score indicates more distress. The predictive nature of
its excess internal equity injection decision reveals that bank holding companies possessed a significant amount of expertise
and private information about each subsidiary’s post-crisis type. Source: Call Reports, FR Y9C, CAMELS, SR SABR, US
Treasury TARP CPP Transaction Reports.
Figure 6: Predictive Value of Excess Internal Equity Injections for Post-Crisis Nonemployer
Revenue Growth (Cross Section)
Note: This graph illustrates the predictive value of excess internal equity injections for the future post-crisis investment
opportunities of very small businesses in the cross-section of states. Specifically, the x-axis shows the state-aggregated excess
internal equity injections made by multi-bank holding companies (MBHCs) on the date of TARP arrival (during the 2008
crisis), while the y axis shows the post-crisis three-year cumulative growth in revenues at nonemployer firms from 2009-2011
within each respective state. Nonemployer firms are small businesses that have no paid employees but are subject to federal
income tax. Nonemployer businesses appear in practically every industry and comprise the majority of nearly 80 percent of
industry subsectors. Examples include firms that operate museums and hobby shops; provide pet care services; manufacture
clothing, electrical equipment, and machinery; sell real estate and process data. Excess internal equity injections represent the
excess amount of internal equity injections provided to a bank subsidiary by its parent holding company on the date of TARP
arrival, after controlling for the subsidiary’s CAMELS rating. This variable is measured as the percentage point increase in
the subsidiary’s capital ratio that results from the parent’s excess internal equity injection upon TARP arrival. The stateaggregation of this variable (x axis) is constructed by calculating the average of the individual excess injection signals across
banks that competed within a state, weighted by each subsidiary’s ex-ante small business credit exposures in the given state.
Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP Transaction Reports, Census Bureau
(Nonemployer Statistics).
Figure 7: Predictive Value of Excess Internal Equity Injections for Future Small Business
Employment Growth (Cross Section)
Note: This graph illustrates the predictive value of excess internal equity injections for the post-crisis employment growth of
small businesses in the cross-section of states. Specifically, the x-axis shows the state-aggregated excess internal equity
injections made by multi-bank holding companies (MBHCs) on the date of TARP arrival (during the 2008 global financial
crisis), while the y axis shows the post-crisis 3-year cumulative growth in small business employment from 2009-2011 within
each respective state. For small business employment, we focus on changes in employment at establishments with fewer than
or equal to nine workers. Excess internal equity injections represent the excess amount of internal equity injections provided
to a bank subsidiary by parent holding companies on the date of TARP arrival, after controlling for the subsidiary’s
CAMELS rating. This variable is measured as the percentage point increase in the subsidiary’s capital ratio that results from
the parent’s excess internal equity injection upon TARP arrival. The state-aggregation of this variable (x axis) is constructed
by calculating the average of the individual excess injection signals across bank competing within a state, weighted by each
subsidiary’s ex-ante small business credit exposures in the given state. Source: Call Reports, FR Y9C, CAMELS, SR-SABR,
US Treasury TARP CPP Transaction Reports, Census Bureau (Statistics of US Businesses).
Figure 8: Predictive Value of Excess Internal Equity Injections for Future Small Business
Establishment Growth (Cross Section)
Note: This graph illustrates the predictive value of excess internal equity injections for the post-crisis establishment growth of
small businesses in the cross-section of states. Specifically, the x-axis shows the state-aggregated excess internal equity
injections made by multi-bank holding companies (MBHCs) on the date of TARP arrival (during the 2008 crisis), while the y
axis shows the post-crisis three-year cumulative growth in small business establishments from 2009-2011 within each
respective state. We define small business establishments as establishments employing fewer than or equal to nine workers.
Excess internal equity injections represent the excess amount of internal equity injections provided to a bank subsidiary by
parent holding companies on the date of TARP arrival, after controlling for the subsidiary’s CAMELS rating. This variable is
measured as the percentage point increase in the subsidiary’s capital ratio that results from the parent’s excess internal equity
injection upon TARP arrival. The state-aggregation of this variable (x axis) is constructed by calculating the average of the
individual excess injection signals across banks competing within a state, weighted by each subsidiary’s ex-ante small
business credit exposures in the given state. Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP
Transaction Reports, Census Bureau (Statistics of US Businesses).
Figure 9: Predictive Value of Excess Internal Equity Injections for Future Small Business Payroll
Growth (Cross Section)
Note: This graph illustrates the predictive value of excess internal equity injections for the post-crisis annual payroll growth
of small businesses in the cross-section of states. Specifically, the x-axis shows the state-aggregated excess internal equity
injections made by multi-bank holding companies (MBHCs) on the date of TARP arrival (during the 2008 crisis), while the y
axis shows the post-crisis three-year cumulative growth in small business payroll from 2009-2011 within each respective
state. For small business employment, we focus on changes in payroll at establishments with fewer than or equal to nine
workers. Excess internal equity injections represent the excess amount of internal equity injections provided to a bank
subsidiary by parent holding companies on the date of TARP arrival, after controlling for the subsidiary’s CAMELS rating.
This variable is measured as the percentage point increase in the subsidiary’s capital ratio that resulted from the parent’s
excess internal equity injection upon TARP arrival. The state-aggregation of this variable is constructed by calculating the
average of the individual excess injection signals across banks competing within a state, weighted by each subsidiary’s exante small business credit exposures in the given state. Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury
TARP CPP Transaction Reports, Census Bureau (Statistics of US Businesses).
Figure 10: Predictive Value of Excess Internal Equity Injections for Future Changes in Small
Business Credit Originations
Note: This figure shows the time-series growth in new small business credit originations across subsidiaries that received
positive (versus negative) amounts of excess internal equity injections from the MBHC parent upon TARP arrival. Consistent
with findings in previous charts and tables that excess internal equity injections revealed the banks’ expectation for the postcrisis deterioration in investment opportunities, this chart suggests that banks accordingly divested from these regions by
curtailing post-crisis new small business lending. Specifically, this figure plots the 90th percent confidential intervals for
estimated coefficients 𝛽𝛽𝛼𝛼 for each Date*Excess Internal Equity Injection term (𝑓𝑓𝑓𝑓𝑓𝑓 𝛼𝛼𝛼𝛼{2005,2011} ) in the following
specification: 𝐿𝐿𝐿𝐿(𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙)𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 = 𝛽𝛽0 + ∑2012
𝛼𝛼=2005 𝛽𝛽𝛼𝛼 ∙ 𝟙𝟙[𝑡𝑡 = 𝛼𝛼] ∙
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝐼𝑠𝑠,𝑝𝑝,𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑝𝑝 + 𝛽𝛽2 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽3 ∙
𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽4 ∙ 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑠𝑠,𝑝𝑝,𝑡𝑡−1 + 𝛽𝛽5 ∙ 𝐵𝐵𝐵𝐵𝐵𝐵 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑝𝑝,𝑡𝑡−1 + 𝛿𝛿𝑠𝑠 + 𝜋𝜋𝑝𝑝 + 𝜏𝜏𝑡𝑡 + 𝜀𝜀𝑠𝑠,𝑝𝑝,𝑡𝑡 .
S stands for subsidiary, p stands for parent holding company, and t stands for time. Source: Call Reports, FR Y9C, CAMELS,
SR-SABR, US Treasury TARP CPP Transaction Reports, FFIEC Community Reinvestment Act data.
Figure 11: Predictive Value of Excess Internal Equity Injections for Future Change in Small
Business Credit Originations (Cross Section)
Note: This graph illustrates the predictive value of excess internal equity injections for the post-crisis credit originations at
small businesses in the cross-section of states. Specifically, the x-axis shows the state-aggregated excess internal equity
injections made by multi-bank holding companies (MBHCs) on the date of TARP arrival (during the 2008 crisis), while the y
axis shows the post-crisis growth in new credit originations to small businesses (flows) within each state during 2011 with
respect to 2008. We define small business establishments as establishments employing fewer than or equal to nine workers.
Excess internal equity injections represent the excess amount of internal equity injections provided to a bank subsidiary by
parent holding companies on the date of TARP arrival, after controlling for the subsidiary’s CAMELS rating. This variable is
measured as the percentage point increase in the subsidiary’s capital ratio that resulted from the parent’s excess internal
equity injection upon TARP arrival. The state-aggregation of this variable is constructed by calculating the average of the
individual excess injection signals across banks competing within a state, weighted by each subsidiary’s ex-ante small
business credit exposures in the given state. Source: Call Reports, FR Y9C, CAMELS, SR-SABR, US Treasury TARP CPP
Transaction Reports, FFIEC Community Reinvestment Act data.
@FedFRASER