1
[Berkshire Hills Bancorp, Inc. Letterhead]
March 1, 2007
VIA EDGAR AND FACSIMILE
- -----------------------
Mr. Donald Walker
Senior Assistant Chief Accountant
U.S. Securities and Exchange Commission
Division of Corporation Finance
100 F. Street, N.E.
Washington, D.C. 20549
RE: Berkshire Hills Bancorp, Inc.
Form 10-K for the Fiscal Year Ended December 31, 2005
Form 10-Q for the Quarterly Periods Ended March 31, 2006,
June 30, 2006 and September 30, 2006
File No. 0-58514
Dear Mr. Walker:
We have received your letter dated January 26, 2007, regarding comments
on the above filings. We appreciate your review and are providing responses to
each of the comments. To facilitate your review, we have repeated each of your
comments followed by our response.
FORM 10-K FOR THE FISCAL YEAR ENDED DECEMBER 31, 2005:
- ------------------------------------------------------
CONSOLIDATED FINANCIAL STATEMENTS
- ---------------------------------
NOTE 1--SUMMARY OF SIGNIFICANT ACCOUNTING POLICIES
- --------------------------------------------------
ACCOUNTING FOR DERIVATIVES, PAGE 44
- -----------------------------------
COMMENT NO. 1:
- --------------
We note your response to comment one of your letter dated December 28,
2006. Please tell us the following with respect to your cash flow hedge of
prime-based revolving home equity loan portfolio:
o how you document with sufficient specificity the hedged cash flows;
o whether the documented hedged risk is the risk of overall changes
in the hedged cash flows or the risk of changes in the hedge cash
flows attributable to changes in prime;
o the specific prime rate index used for the loans and the interest
rate swap;
o how you considered that prime is not a benchmark rate as defined by
SFAS 133;
2
Mr. Donald Walker
March 1 2007
Page 2
o how you ensure that hedged cash flows share the same risk exposure;
o whether there are instances in which the margin on the loans within
the hedged cash flow pool were not zero percent;
o how you ensure that the payment dates on the interest payments
received are the same as the payment dates of the swap;
o whether the timing of rate reset for the loans are the same as the
reset for the swap;
o the specific guidance upon which you relied in determining the
appropriate prospective and retrospective method of assessing
effectiveness;
o the specific guidance upon which you relied in determining the
appropriate method of measuring ineffectiveness; and
o whether you considered any other potential sources of
ineffectiveness aside from the changes in the balance of the loan
portfolio.
COMMENT NO. 2:
- --------------
With respect to your fair value hedges of pools of fixed-rate brokered
certificates of deposit portfolios, please tell us the following:
o which of the hedged risks described in paragraph 21(f) of SFAS 133
is the designated risk being hedged;
o how you determined that the portfolio of deposits satisfies the
requirements of paragraph 21(a)(1) with respect to grouping of
similar assets and liabilities;
o how the documented hedging strategy met the requirements of
paragraph 20(b);
o whether a broker placement fee or upfront fees that take the LIBOR
leg off market is included in the pricing or terms of the swap;
o whether the swap contains an option to mirror death redemptions;
o the specific quantitative and qualitative analysis you performed at
inception to determine that there would not be a material amount of
ineffectiveness, including your assessment of the impact of death
redemptions;
o the specific guidance upon which you relied in determining the
appropriate prospective and retrospective method of assessing
effectiveness;
o whether you considered any other potential sources of
ineffectiveness aside from the changes in the balance of the loan
portfolio, such as differences between the credit risk of the
hedged item and the swap or changes in the swap counterparty's
creditworthiness;
o how you documented your approach to effectiveness testing if there
were changes in the matched terms or changes in counterparty
credit;
o how you determined the amount of ineffectiveness at December 31,
2005 and whether you measure the amount of ineffectiveness on a
periodic basis;
o whether the lack of recording known ineffectiveness was included in
as an audit difference; and
3
Mr. Donald Walker
March 1 2007
Page 3
o the specific guidance upon which you relied determining that
comparing the certificate balances and the notional amounts of the
related swap agreements is an appropriate method of measuring
ineffectiveness.
RESPONSE TO COMMENT NOS. 1 AND 2:
After further review, management has determined that it does not have
the necessary documentation to meet the criteria for hedge accounting under
Financial Accounting Standard No. 133 ("FAS 133"). The subject instruments were
entered into by Woronoco Bancorp, Inc. before its acquisition by Berkshire Hills
Bancorp, Inc. (the "Company") in June 2005. The subject instruments are the only
hedging instruments owned by the Company. Although management continued to
perform the quarterly analysis of hedge effectiveness subsequent to the
acquisition, the Company is not in possession of a formal re-designation of the
hedges, as required under DIG Issue E15. Furthermore, although management
continued to apply the "Dollar Offset" method to measure retrospective hedge
effectiveness quarterly, these analyses did not adequately consider the
prospective hedge effectiveness in accordance with FAS 133 as it relates to the
impact of the death maturity calls embedded in the brokered certificate of
deposit portfolio. Based on the facts presented above, management has
discontinued use of hedge accounting for all subject instruments.
In accordance with Staff Accounting Bulletin 99, management evaluated
the impact of discontinuation of hedge accounting on income of the unadjusted
differences in all periods from June 1, 2005 (the date of acquisition of
Woronoco Bancorp) to December 2006 under the rollover method. Under this
approach, there were two quarters (Q2 2006 and Q3 2006) in which the unadjusted
differences were at or moderately greater than 5% of pre-tax and post-tax
income. The cumulative adjustment for the period June 1, 2005 to December 31,
2006 results in an unadjusted difference of (1.1%) and (1.2%) of pre and post
tax net income, respectively.
- ----------------------------------------------------------------------------------------------------------------
% of Income Q2 2005 Q3 2005 Q4 2005 Fiscal 2005
- ----------------------------------------------------------------------------------------------------------------
Pre-tax impact (0.4%) (2.9%) (3.7%) (2.9%)
- ----------------------------------------------------------------------------------------------------------------
Post-tax impact (0.2)% (3.0%) (3.6%) (3.8%)
- ----------------------------------------------------------------------------------------------------------------
- -------------------------------------------------------------------------------------------------------------------
% of Income Q1 2006 Q2 2006 Q3 2006 Q4 2006 Fiscal 2006
- -------------------------------------------------------------------------------------------------------------------
Pre-tax impact (2.2%) (5.5%) 7.3% (0.2%) 0.1%
- -------------------------------------------------------------------------------------------------------------------
Post-tax impact (2.2%) (5.0%) 7.7% (0.2%) 0.1%
- -------------------------------------------------------------------------------------------------------------------
Based on Staff Accounting Bulletin 108, management also evaluated the unadjusted
differences under the iron curtain method in Q4 2006 and for the year 2006. The
pre-tax and post-tax impacts in Q4 2006 were (6.9%) and (6.8%), respectively.
For the year 2006, the difference was 1.6% for both pre-tax and post-tax income.
Additionally, management considered the following factors outlined in
SAB 99 to determine whether any of the periods were impacted materially by this
unadjusted difference:
o The impact on the amount of expense recognized was reviewed and
considered immaterial.
4
Mr. Donald Walker
March 1 2007
Page 4
o The unadjusted difference does not mask a change in earnings or
other tends.
o The unadjusted difference does not hide a failure to meet analysts'
consensus expectations for the enterprise during any period.
o The unadjusted difference does not change a loss into income during
any period.
o The unadjusted difference does not affect the registrant's
compliance with loan covenants or other contractual requirements.
o The unadjusted difference does not conceal an unlawful transaction.
o The unadjusted difference would not cause a significant market
reaction.
o There are no other significant unadjusted differences and there are
no other qualitative or quantitative factors that management
believes would impact the assessment of the materiality of this
unadjusted difference.
After carefully considering all the factors outlined above, management
concluded that this unadjusted difference was not material to the financial
statements in any of the periods shown above. Management intends to record the
cumulative change in fair value in earnings in the first quarter of 2007. The
Company sold the swaps hedging the time deposits on February 28, 2007. The
Company is scheduled to remit approximately $466,000, which will result in a
loss on sale of approximately $141,000 pre-tax and $83,000 post-tax in Q1 2007.
This represents the cumulative mark to market impact since acquisition date. The
swaps hedging the loan portfolio mature in May 2007 and had an unrealized loss
of approximately $7,000 as of February 28, 2007.
FORM 10-Q FOR THE QUARTERLY PERIOD ENDED SEPTEMBER 30, 2006:
- ------------------------------------------------------------
MANAGEMENT'S DISCUSSION AND ANALYSIS
- ------------------------------------
COMPARISON OF FINANCIAL CONDITION AT SEPTEMBER 30, 2006 AND DECEMBER 31, 2005,
- ------------------------------------------------------------------------------
PAGE 20
- -------
COMMENT NO. 3:
- --------------
We note your response to [paragraph] four of our letter dated December
28, 2006. Your response does not appear to clearly explain how using the
probable rate of loan losses over the expected average life of each loan pool
helped you derive the loss factors which would result in your estimate of losses
incurred as of the period-end. Please explain to us how your current methodology
estimates the losses incurred as of the period-end and does not represent an
estimate of the expected losses over the life of the related loan pools. In your
response please clearly bridge the gap between deriving the average annual
expected loss rate and the portfolio estimated loss factor that is applied to
the pool of loans as of the period-end. Aside from demonstrating that you
multiplied the estimated annual loss rate by the estimated average life of loan
category to arrive at the estimated loss factors, it does not appear as though
your response clearly explains the interrelationship between the average annual
expected loss rate and the estimated loss factors.
5
Mr. Donald Walker
March 1 2007
Page 5
RESPONSE TO COMMENT NO. 3:
The Company's loan loss allowance methodology for general pool reserves
reflects estimated loan losses incurred as of the financial statement date in
accordance with Statement of Financial Accounting Standards No. 5 ("SFAS 5").
The methodology is based on an assessment of losses inherent in the loan pools
at each balance sheet date, and does not anticipate the occurrence of specific
future events. It does anticipate the passage of time for those losses incurred
to emerge in problem loan performance and a further passage of time for
collection and workout processes to be conducted, and losses to be confirmed and
charged off.
THIRD QUARTER LOSS EVENTS
The Company determined that certain events had occurred in the third
quarter, resulting in a higher assessment of probable incurred losses at
September 30, 2006, even though such losses were not yet identifiable in terms
of specific loans or borrowers.
The events assessed by the Company in the third quarter included: (1)
the sharp decline in residential real estate activity; (2) the Federal Reserve
Board's cessation of interest rate increases after 17 consecutive increases over
the prior two years; and (3) the combined impact of higher interest rates and
energy prices including the impact of energy price highs in the middle of the
year. Management determined that these events were signaling that changes had
occurred in the environment which had reduced borrower debt service ability and
that it was probable that more borrowers were no longer servicing their debt out
of cash flows as they had done in the past. Additionally, the majority of the
Company's loans have interest rates which are not fixed for the entire term of
the loan. Based on rate increases over the prior two years, additional future
increases in debt service requirements were contracted and predictable for the
majority of loans, contributing to the losses which were inherent in the loan
portfolio at the financial statement date. Management further determined that
borrower liquidity reserves were also impacted by these economic events, which
meant that they would not be replenished in the ordinary course of events.
Additionally, management believed that the value and liquidity of collateral was
negatively impacted by these events, which would contribute to higher inherent
losses in the loan portfolio. While the Company did not have information with
which to specifically identify the loans with inherent losses at the financial
statement date, management determined that the higher probable and estimable
losses on specific loan pools should be identified and recorded in the loan loss
allowance.
METHODOLOGY
Recognizing that SFAS 5 requires accrual of losses that are probable
and estimable as of the balance sheet date, management utilized historical data
in comparable economic periods to develop expected annual loss rates for its
primary loan pools. Management then developed time horizons over which losses
inherent in the portfolio were expected to be recorded, making a reasonable
estimate based on average expected pool lives and expected loan behaviors. The
reserve factor for each pool was then determined as the product of the loss
rates and time horizons.
6
Mr. Donald Walker
March 1 2007
Page 6
EXPECTED ANNUAL LOSS RATES
The preponderance of the increase in the Company's loan loss allowance
at September 30, 2006 was due to higher inherent losses in the commercial loan
portfolio. This portfolio was analyzed in three pools: construction and
development; commercial mortgages; and commercial business loans. For each of
these pools, the Company reviewed historic data extending back into the 1990s,
as the data was available, both for the Company's loss experience and industry
loss experience. The Company focused on periods which it deemed most similar to
the current period, and the Company considered various subjective factors which
were also relevant to assessing probable loan losses inherent in the portfolio.
The annual loss rates were estimated based on this methodology.
For the commercial loan pools, management considered the following
factors for the expected annual loss rates that were used in its methodology:
o Construction and development loans. These loans were evaluated for the
----------------------------------
historic period of 1991 - 1997 and loss patterns were reviewed both for
Berkshire Bank and for the industry. The analysis considered the impact
of real estate slowdowns and higher interest rates in determining the
loss rate inherent in the current loan pool.
o Commercial mortgages. A number of periods were evaluated over an
--------------------
historic period of more than fifteen years and the period of 1996 -
1998 was selected as most similar to the current environment. Interest
rates peaked in 1995 and economic activity slowed, and the loss
recognition that followed this was viewed as the best indicator of
losses inherent in the current loan pool. The loss rate estimate
included an evaluation of the historic experience of Berkshire Bank and
of the industry for this period.
o Commercial business loans. Based on a review of almost a decade of
-------------------------
historic data, the period of 2002 - 2004 was selected as most relevant
to the estimate of commercial business loan losses. The estimate of
inherent losses was developed from the historic loss recognition data
for both Berkshire Bank and the industry for this period.
LOSS TIME HORIZONS
Management estimated the loss time horizons as equivalent to the
average lives of the loan pools, which in most cases were equal to half of the
estimated average time to maturity for the loan pools. Accordingly, the average
lives used are less than the total future period in which losses are probable
and estimable based on the average contractual future terms of the loan pools.
Additionally, it should be noted that contractual loan terms in many cases do
not reflect the probable future lives of the outstanding loans. A significant
number of relationships in the commercial loan portfolio regularly obtain
extensions and modifications in the ordinary course of business, lengthening the
probable loan life. Further, loans that develop credit problems will probably
have longer lives than indicated by the existing contractual maturities due to
workout arrangements that may be agreed to. Management believes that the time
horizons established for
7
Mr. Donald Walker
March 1 2007
Page 7
each loan pool were reasonable for the recognition of losses inherent in the
pools. It is implicit in the Company's methodology that there will be additional
future losses that are probable and that are not reserved, either because they
cannot be accurately estimated or because the loan loss events will occur in
future periods beyond those that the Company estimated as likely for the
emergence of existing inherent losses. The estimated loss factors determined
based on average annual expected loss rates do not include these additional
likely future loan losses.
Although management considered the rate of loan loss migration, the
Company does not have sufficient data to identify specific loss migration
histories for these loan pools. Alternatively, management evaluated the nature
of the loss events identified in the third quarter and how such events will
likely impact each loan pool.
Management believes that it is reasonable to anticipate that loan
impairments will require a period of one to two years before the negative cash
flows of the borrowers appear as current period loan performance problems and
there can be another period of one to two years after loan performance problems
appear during which collection and workout actions are taken. Additionally, a
portion of the Company's loan portfolio is comprised of loans outstanding under
contractual credit commitments. These commitments have an effect of prolonging
the loss recognition period for losses inherent in loans outstanding as of the
financial statement date. Since most of the Company's loans are secured, and
real estate is the primary form of collateral, the liquidation time alone can be
in the area of 6 - 12 months. In many cases this is extended by borrower
defenses, including bankruptcy proceedings. Additionally, most of the Company's
commercial loans are guaranteed by the owners of the borrower. The impact of
these guarantees is to reduce the amount of loan loss, but also to extend the
time for loss recognition due to guarantor actions and defenses raised during
the workout process. While management anticipates that FAS 114 reserves would
emerge as appropriate for specific collateral dependent loans, the actual loss
confirmation would often be made at the time of the final collateral
liquidation, which in most cases is expected to be within a period of two to
four years from the time that the loss event occurred. Since the time horizons
used in estimating the pool reserve factors were within this general timeframe,
management determined that its estimates of the probable and estimable losses
were consistent with the general expected loss migration patterns for the loan
pools. Consequently, it was management's determination that its loan loss
allowance reflected the losses that were inherent in the loan portfolio as of
the financial statement date.
For the commercial loan pools, management considered the following
factors for the expected loss recognition periods that were used in its
methodology:
o Construction and development loans. These loans normally have two-year
----------------------------------
terms, during which projects are expected to be built and either sold
or converted to permanent financing. These loans typically involve
periodic inspections and disbursements of new monies. They are a form
of asset-based lending. Normally, projects have to be completed and
recapitalized before loss recognition can be made. The Company views
its two-year estimated time horizon as a reasonable and low-end
estimate for the loss recognition period for these loans.
8
Mr. Donald Walker
March 1 2007
Page 8
o Commercial mortgages. These loans normally have long maturities, are
--------------------
normally serviced from cash flows and often there are limited other
sources of liquidity. These loans can take the longest time before
inherent losses become evident in loan nonperformance and losses are
recognized. Because of the comparative stability of the collateral,
workout processes and bankruptcy courts tend to allow the most time
before collateral liquidation for this type of loan. The Company
believes the 3.5-year time horizon estimated for loss recognition is
reasonable based on the characteristics of these loans.
o Commercial business loans. These loans include a variety of loan types,
-------------------------
including lines of credit, working capital loans and equipment loans.
The performance of these loans is often related to seasonal or annual
processes. The Company views the loss recognition period for these
loans as being between the two periods discussed above. A time horizon
of 2.5 years was used, which is reasonable based on the variety of
circumstances that are involved with the emergence and management of
the losses inherent in this portfolio.
OTHER MANAGEMENT OBSERVATIONS
Management believes that it identified the most relevant historical
loss periods as a basis for determining pool loss factors in the current
environment. The most relevant historical loss periods do not include the
Company's highest loss periods, as management does not believe that such periods
are reflective of the current environment. The Company has observed that loss
emergence in its markets can be a protracted process, particularly for
commercial loans with longer effective lives and extended workout times. The
Company has observed that its markets are slower to overheat and slower to
evidence losses than some larger or more urban markets. The Company believes
that the annual loss rates, time horizons and pool loss factors are consistent
with its experience in recognizing loan losses in prior times when credit
conditions contributed to higher losses. In reviewing its history, the Company
noted that there were several years in which its annual charge-off rate on
commercial loans exceeded the entire total pool reserve factors as of September
30, 2006.
CURRENT CONDITIONS AND INDUSTRY FACTORS
The Company evaluates its estimates of inherent loan losses in
comparison to industry averages for such estimates and the Company also
communicates with informed industry associates to evaluate indicators of losses
inherent in similar loan pools in the industry. Management believes that its
estimates of inherent loan losses at September 30, 2006 were within a reasonable
range for its major loan pools based on these comparisons. The Company also
evaluates industry trends for problem loans and it believes that the increase it
has recently witnessed is consistent with its own risk estimates as of September
30, 2006. The Federal Deposit Insurance Corporation's most recent quarterly
profile states that in the third quarter, total nonperforming loans increased at
the highest quarterly rate in five years. Management believes that this
information corroborates its assessment that probable loan losses inherent in
its portfolio increased during the third quarter. Management monitored trends in
nonperforming loans and loss recognition in the fourth quarter. Nonperforming
loans increased, charge-offs increased
9
Mr. Donald Walker
March 1 2007
Page 9
and a higher level of impaired loan reserves was identified in the fourth
quarter. These changes were consistent with the increase in the allowance in the
previous quarter.
* * * * *
If you have any questions about our responses or require any additional
information, please do not hesitate to call me at (413) 236-3194.
Very truly yours,
BERKSHIRE HILLS BANCORP, INC.
/s/ Michael P. Daly
Michael P. Daly
President and Chief Executive Officer
cc: Joyce Sweeney, Securities and Exchange Commission
John S. Millet, Chief Financial Officer, Berkshire Hills Bancorp, Inc.
Jean M. Joy, Wolf & Company, P.C.
Lawrence M. F. Spaccasi, Muldoon Murphy & Aguggia LLP
Scott A. Brown, Muldoon Murphy & Aguggia LLP