FRBSF Economic Letter
2002-15; May 17, 2002
Off-Site Monitoring of Bank Holding Companies
Bank supervisors engage in extensive monitoring of banking organizations
in order to enforce regulations and to guard against systemic risk. In
the United States, primary responsibility for monitoring bank holding
companies (BHCs) falls to the Federal Reserve. The Fed currently uses
a combination of on-site and off-site monitoring to fulfill its supervisory
responsibilities. On-site supervisory visits produce a detailed picture
of an institution's financial condition and risk profile, but they do
absorb considerable resources and are conducted only about once a year.
Given the changing nature of banking, this is more than enough time for
an institution's risk profile to change sharply, so the Fed complements
on-site inspections with off-site monitoring based on analyzing supervisory
data gathered on a quarterly basis through standard regulatory reporting
forms. Off-site monitoring permits more timely supervisory analysis and
hence a potentially more efficient allocation of scarce supervisory resources.
An important component of off-site monitoring is based on using econometric
models. This Economic Letter summarizes recent research on using
econometric models to conduct off-site monitoring of BHCs.
Supervision of bank holding companies
Financial intermediaries such as banks and BHCs are thought to play key
roles in the economic system by creating highly liquid deposit contracts
out of funds that are invested in highly illiquid projects. This benefit
comes with a cost, however. Banking institutions are necessarily fragile
and can be susceptible to runs. Policymakers have seen fit to provide
insurance to protect against the harm created by bank runs. However, this
support in the form of deposit insurance and discount window lending has
had the consequence of giving banks an incentive to take more risks than
they would otherwise. If the government is willing to provide insurance,
banks may choose to increase their risk because, if these risks pay off,
the profit goes to the bank, while the losses go to the government. This
"heads-I-win-tails-you-lose" scenario is a principal justification for
bank regulation in this country.
Since the financial condition of the holding company could affect the
condition of its bank subsidiaries, full-scope on-site inspections of
BHCs are a key element of the supervisory process. They are generally
conducted once a year, particularly for larger and more complex BHCs.
At the conclusion of an inspection, the supervisors assign the BHC a composite
rating summarizing their assessment of the BHC's overall health and financial
condition. This rating is called BOPEC, and it stands for the five key
areas of supervisory concern: the condition of the BHC's Bank subsidiaries,
Other nonbank subsidiaries, Parent company, Earnings, and Capital adequacy.
BOPEC ratings range from one (best) to five (worst). A rating of one or
two indicates that the BHC is not considered to be of supervisory concern.
BOPEC ratings are highly confidential and are not publicly available.
Between on-site inspections, supervisors use an off-site system based
primarily on two key information sources. First is the BHC Performance
Report, which is filed by BHCs and their subsidiaries and is a detailed
summary of their quarterly regulatory reporting forms. The report summarizes
approximately 800 BHC-related variables across several years. From this
report, certain variables are selected as key performance criteria, and
if a BHC fails to meet these criteria in a given quarter, it is noted
as an exception that requires further monitoring.
The second source is the supervisory CAMELS ratings assigned to banks
within the holding company. These ratings are assigned by the various
bank supervisory agencies; the OCC for national banks, the FDIC for state
banks that are not members of the Federal Reserve System, and the Federal
Reserve for state member banks. As with BOPEC ratings, CAMELS ratings
are assigned after bank examinations. The acronym refers to the six key
areas of concern: the bank's Capital adequacy, Asset quality, Management,
Earnings, Liquidity, and Sensitivity to risk. The composite CAMELS rating
is like the BOPEC rating—one to five with one as the best rating. Since
the condition of a BHC is closely related to the condition of its subsidiary
banks, the off-site BHC surveillance program includes monitoring recently
assigned CAMELS ratings.
As with on-site BHC inspections, on-site bank examinations occur about
once a year, which is long enough for the gathered supervisory information
to become less representative of the bank's condition. To address this
issue, in 1993 the Federal Reserve instituted an off-site monitoring system
for banks, known as SEER—System for Estimating Examiner Ratings. An important
component of SEER is a model that forecasts bank CAMELS ratings for the
next quarter. The model is estimated every quarter in order to reflect
the most recent relationship between selected financial ratios and the
two most recent quarters of actual CAMELS ratings. Significant changes
in a bank's CAMELS rating as forecasted by the SEER model could be sufficient
to warrant closer monitoring of the bank. The off-site BHC surveillance
program also explicitly monitors the SEER model's forecasted CAMELS ratings.
A model for off-site BHC monitoring
and Lopez (2001), we explored the usefulness of a model similar to SEER
for monitoring BHCs off-site. Specifically, the model forecasts the BOPEC
ratings to be assigned at an upcoming on-site inspection using the most
recent data available to supervisors. The data sample for our analysis
includes the supervisory BOPEC ratings assigned over the period from 1990
to 1999. We chose to analyze only BOPEC ratings assigned after an on-site,
full-scope inspection. Our sample of BOPEC ratings is further refined
to include only inspections of BHCs with a lead bank, four quarters of
reported supervisory data, and prior BOPEC ratings. Figure 1 presents
the distribution of the 3,963 BOPEC ratings assigned over our sample period
to 1,440 different BHCs. About 84% of the ratings fall in the upper two
categories, which indicate little supervisory concern.
Our proposed BOPEC off-site monitoring (BOM) model is similar in structure
to the SEER model for CAMELS ratings. The choice of which supervisory
variables to include in the model is challenging; as mentioned, there
are more than 800 variables at the supervisors' disposal for this purpose.
For this study, we selected nine explanatory variables that are reasonable
proxies for the five components of the BOPEC rating.
The first variable is the natural log of total BHC assets, which is our
control variable for firm size. The next three variables capture the supervisory
concerns regarding the BHC's bank subsidiaries, as summarized in the "B"
component of the rating: the CAMELS rating of the BHC's lead bank, the
ratio of the BHC's "problem loans" (i.e., nonperforming loans, nonaccrual
loans, and other real estate owned) to its total assets, and the ratio
of the BHC's allowances for losses on loans and leases (ALLL) to its total
loans, another proxy for the health of the BHC's loan portfolio.
The fifth variable is an indicator of whether the BHC has a securities
subsidiary, which during our sample period is the Section 20 subsidiary
that engages in securities activities that BHCs were not permitted to
engage in before the Gramm-Leach-Bliley Act of 1999. This variable is
a proxy for the types of nonbank activities the BHC is engaged in and
speaks to the "O" component of the BOPEC rating. We also include as the
sixth variable the ratio of a BHC's trading assets to its total assets
as a proxy of its nonbanking activities.
The seventh variable is the so-called "double leverage" ratio between
the BHC and its lead bank, which is the ratio of the lead bank's equity
capital to that of the parent's equity capital. This variable provides
a measure of the soundness of the parent BHC, and we use it as a proxy
for the condition of the parent BHC as summarized in the "P" component
of the BOPEC rating. The eighth variable is the BHC's return on average
assets (ROAA), defined as the ratio of the four-quarter average of the
BHC's net income to the four-quarter average of its assets. This variable
is used to proxy for the "E" component of the BOPEC rating. The ninth
variable is the BHC's ratio of equity capital to its total assets, which
is a proxy for the "C" component of the BOPEC rating. We also include
additional variables related to whether the BHC is publicly traded or
Our empirical results are generally in line with our expectations. We
found that larger BHCs tend to have better BOPEC ratings, which could
be due to larger banks having more diversified asset portfolios. We found
that an improvement in a BHC's lead bank CAMELS rating tends to cause
the parent's BOPEC rating to improve as well. Our results indicate that
an increase in a BHC's problem loans or ALLL tends to cause its BOPEC
rating to worsen. We found that the presence of a Section 20 subsidiary
and increased trading assets tends also to cause BOPEC ratings to worsen.
Our results indicate that an increase in the BHC's equity capital ratio
tends to cause its BOPEC rating to improve. Our results indicate that
a BHC's double leverage and ROAA do not seem to affect its BOPEC rating.
In order to be useful for supervisory purposes, the BOM model must be
able to forecast BOPEC ratings accurately. In order to mimic actual practice,
we re-estimated the model every quarter based on a rolling data sample
of the last four quarters. We then evaluated the accuracy of the model's
forecasts by comparing them to the actual BOPEC ratings assigned.
2 presents our analysis of the forecast accuracy of our model relative
to actual BOPEC changes (i.e., upgrade, no change, or downgrade). We transform
our BOPEC forecasts into BOPEC change forecasts by examining how far they
are from the median forecast for their BOPEC peer-group. If our transformed
forecast is one full rating grade below its peer-group median, then the
BHC is forecast to improve. If the transformed forecast is one full rating
grade above its peer-group median, then the BHC is forecast to worsen.
Otherwise, the transformed BOPEC forecast indicates no change in BOPEC
"No changes" to BOPEC make up the largest category and are well forecasted
at all horizons between four quarters and one quarter prior to their assignment.
They are forecasted correctly about 67% of the time. BOPEC downgrades
are forecasted correctly about 30% of the time at four quarters prior,
and that percentage improves to 56% at one quarter prior. Upgrades are
forecasted correctly about 50% of the time at four quarters prior and
60% of the time at one quarter prior. The change forecast should be compared
to the "naive" forecast where, given no information about a specific BHC's
condition, the probability that the BHC is either upgraded or downgraded
is about 25%. Overall, the transformed BOPEC forecasts are accurate about
55% to 65% of the time. These results strongly indicate that the BOM model's
forecasts are capable of detecting actual BOPEC assignments up to four
quarters prior and could thus be useful for supervisory monitoring purposes.
This research suggests that the BOPEC off-site monitoring model can summarize
supervisory information on BHCs in a simple and practical way. As with
the off-site monitoring model currently used for supervisory bank ratings,
this model could give supervisors a tool for detecting potentially significant
changes in BOPEC ratings up to four quarters ahead of time. Therefore,
it offers practical value for banking supervisors, who would be interested
in having accurate early warnings of changes in BHC conditions.
|Jose A. Lopez
Krainer, J., and J.A. Lopez. 2001. "Incorporating Equity Market
Information into Supervisory Monitoring Models." Federal Reserve Bank
of San Francisco Working Paper 01-14. http://www.frbsf.org/publications/economics/papers/2001/wp01-14bk.pdf