Impact of Community Bank Mergers on Acquiring Shareholder Returns

I. BACKGROUND

Regulatory, technological and competitive changes are encouraging substantial consolidation within the financial services industry, particularly among commercial banks. The Garn - St. Germain Act of 1982 allowed the acquisition of weak financial institutions across state lines.

Technological innovations have revolutionized customer relationship management, cross-selling opportunities, and new product development. To remain technologically competitive, many banks merge to satisfy heightened consumer expectations such as internet banking, debit and credit cards, electronic funds transfer systems, and ATM networks.

Changes in accounting merger rules mandate use of the purchase method of accounting rather than pooling. A Financial Accounting Standards Board (FASB) rule took effect December 31, 2003 that requires banks to separate loans they purchase from loans they originate for accounting purposes. Loss reserves for purchased loans are not eligible to be included in acquired reserves because the purchase is recorded at market value. FASB and the financial regulators have also changed how banks account for loan losses (Davenport 2004. American Banker).

This study examines (1) two market-deepening mergers and one marketexpanding merger by Southern Financial Bank (SFFB) in Virginia and (2) contrasts merging and nonmerging banks in Virginia. The evidence in the literature is that for both types of mergers, returns to acquiring shareholders are not enhanced, but one of the SFBB acquisitions represents a distinct, interesting counter example.

IL THE MERGER PARADOX

Despite the large number of bank mergers over the past twenty-five years, academic studies have failed to produce consistent evidence of value enhancement, cost savings, and economies of scale for acquirers. (see K wan and Wilcox, 1999; Healy et. al., 1997, and De Long, 2001). Bruner (2002, pp. 49 - 50) reviewed 119 studies that used four research designs to investigate the profitability of merger activity: 85 event studies, 15 accounting studies, 13 executive surveys, and 6 clinical or case studies. He concluded (2002, p. 56) that the event studies showed "positive abnormal returns to the seller" but "that in the aggregate, abnormal (or marketadjusted) returns to buyer shareholders from M & A activity are essentially zero."

Becher and Campbell (2005) examined a variety of mergers in the 1990s. For market- penetrating mergers, as well as market-expanding mergers, in the 1997-1999 time frame, they found no excess returns above market returns. Losses resulted where there was significant branch overlap among merger partners.

When an acquired bank serves different geographic areas, different types of borrowers or depositors, or has established market positions in different financial product lines, managers expect mergers to increase returns. Managers expect bank mergers to increase profits by raising rates and fees on retail services or increasing efficiency and reducing costs. Mergers can also reduce bank risk through diversification when the co-variances of the merging organizations' revenue or profit streams are low or negative.

Potential Revenues and Savings

Revenue enhancements can result from expanding financial service offerings, cross-selling services, raising fees and boosting net interest revenue by lowering interest rates on depositors' balances. A broader product-line may enable a bank to cross-sell financial services using customer information gathered in one type of transaction to make an additional sale of a different financial service or product. Reduced average or marginal costs can be achieved by rationalizing back-office activities or spreading costs, such as loan servicing and marketing, over more depositors and borrowers, and applying technology to check clearing and electronic on-line services.

Mergers between banks serving overlapping geographic areas could achieve additional cost savings by closing redundant branches, particularly full-service branches, and expanding electronic services. For withinmarket bank mergers, Houston et al. (2001, pp. 288-89) cite a claim by bank executives that as much as 30 percent of the target bank's noninterest expenses can be eliminated. If the acquired organization is perceived to have lower risk, interest costs on uninsured obligations may be reduced and more uninsured liabilities may be attracted to the bank.

III. SFFB's ACQUISITION OF THREE BANKS

Table 1 provides information on the three Virginia banks that SFFB acquired - Horizon Bank, September 1, 1999; First Savings Bank of Virginia, September 1, 2000; and Metro-Country Bank, August 14, 2002. At the time of the acquisitions, asset sizes were: Horizon Bank - $128 million; First Savings Bank of Virginia - $ 76 million, and Metro-County Bank - $90 million. Each of the three acquired banks, as well as SFFB, emphasized mortgage lending. The data in the top panel of Table 1 are from the complete quarter closest to the effective data of the merger.

Over the four-year period, ending in December 2002, SFFB increased its assets by 275 percent to approximately $970 million. SFFB was an active mortgage lender throughout the period of its rapid growth. At the end of 2002, 74 percent of SFFBs loan portfolio was in mortgages.

On December 31, 2001, SFFB had a capital-asset ratio of 8.2 percent, an increase from 6.5 percent one year earlier. As part of its strategy to maintain its "well-capitalized" rating, (the highest rating recognized by financial bank regulators), SFFB sold 862,500 shares of common stock in October. 2001, raising $16.4 million in Tier I capital. In 2000, SFFB raised $13 million in interest-bearing preferred trust certificates that qualified as regulatory capital without diluting earnings per share.

Acquisitions

In March of 1999, SFFB announced plans to acquire Horizon Bank, which operated four branches. Horizon's loan portfolio was composed of 67 percent mortgage loans, 22 percent small business loans, and 11 percent consumer loans. The definitive merger agreement was signed on May 3, 1999 and the acquisition occurred on September 1, 1999. The transaction was accounted for as a pooling of Horizon's resources with SFFB. The acquisition cost was $21.7 million, which represented 109 percent of Horizon's market capitalization.

First Savings Bank of Virginia (FSBV) had two offices and assets of $76.3 million when it was acquired by SFFB. The transaction was accounted for as a purchase. Seventy-five percent of FSBVs loans were housingrelated. By virtually every standard, FSBV would be judged to have been highly illiquid at the time it was acquired. SFFB's acquisition of FSBV was unique among SFFBs three acquisitions - 44 percent of the FSBV shares were owned by officers and directors and the long-run goal of FSBVs board of directors was to have the bank be acquired. In 1992, the FSBV board had hired a new president and stipulated that his primary goal should be to prepare the bank to be acquired by another bank at an attractive price.

On August 14,2002, SFFB expanded its market area beyond the Northern Virginia suburbs to the Richmond, Virginia area by acquiring MetroCounty Bank of Virginia with $90.3 million in assets. The merger had been announced on April 25, 2002. Nine percent of the acquired bank's assets were in cash, and 78 percent were loans and leases; 45 of the 78 percent were in various types of real estate loans. The acquisition of Metro-County Bank was accounted for as a purchase as required by the revised accounting standards. The purchase price was $20.4 million, 121 percent of Metro County's market capitalization before the announcement.

Southern Financial Bank - the acquiring bank

Southern Financial Bank began operations in 1986 and developed primarily as a real estate lender. Table 2 contrasts SFFB's loan portfolio at the end of 1995 and 2002. While it was acquiring the three banks, SFFB increased its percentage commitment to nonmortgage loans from 12 to 26 percent of its loan portfolio, increasing its business loans from 9 to 24 percent of its loan portfolio.

The combination of internal growth and three acquisitions increased SFFB's total assets from $259 million at the end of 1998 to $970 million in 2002. The book value per share of SFFB stock as of December 31, 1999 was $ 10.11. On March 30, 2000, the last full day of trading before the FSBV merger announcement, SFFB stock traded at $14.19 per share. Using 2,656,196, the number of common shares outstanding on December 31, 1999, the implied market capitalization of SFFB was $37,691,000. Based on the average net income from 1997 - 1999, SFFB stock was trading at approximately 16 times earnings. A three-year average is used because SFFB's 1999 pretax income was reduced by $4 million of merger-related costs associated with the 1999 acquisition of The Horizon Bank of Virginia. SFFB's five-year average ROE from 1997 to 2001 was 11.3 percent. Of the 1,426 non-subschapter-S banks with assets between $100 million and $1 billion surveyed by the American Bankers Association, the top fifty had an average ROE of 20.2 percent; the remaining 1,376 had an average ROE of 11.1 percent (Michael, 2002, p. 32).

The acquisitions of Horizon Bank, FSBV, and Metro-County Bank were SFFB's first three mergers. Hart and Apilado's event study of 10 mergers completed by what they define as inactive acquirers from late 1995 through mid-1997 found no positive abnormal returns to acquirers (Hart and Apilado, 2002, p. 318). The accounting for mergers is significant because post-merger performance evaluations as well as market expectations around the merger announcement date may be affected by an acquirer's accounting policy choices.

IV. ACCOUNTING FOR MERGERS: POOLING VERSUS PURCHASE METHODS

Standards

Until the 2001 adoption of Financial Accounting Standard No. 141, Business Combinations, and 142, Goodwill and Other Intangible Assets, (Financial Accounting Standards Board 2001 a and 200Ib) businesses could choose to account for a merger as either a pooling-of-interests or a purchase. These two methods were supposed to be mutually exclusive. Theoretically, there was not supposed to be a choice of methods. A pooling-of-interests occurs when the owners of two firms become joint owners of a combined firm. Pooling interests offers no basis to revalue assets to current market prices. Practically, SFAS 141 eliminated pooling as a method of accounting for mergers. Requiring purchase accounting instead of pooling has addressed most of the criticisms of one of the weakest accounting practices for business combinations. In their study of acquisitions from 1992 - 1997, Ayers et al. (2000, p. 14) show that when purchase accounting is used instead of pooling, return on equity, earnings per share and market-to-book ratios are all lower.

SFFB Acquisitions

SFFB's merger with Horizon Bank was a pooling of interests effective September 1,1999. SFFB's merger agreement with FSBV, dated March 31, 2000 committed both parties to qualify the proposed merger for pooling-of-interests treatment; however, when this merger was consummated, on September 1, 2000, it was accounted for as a purchase. In "marking" FSBV assets up to fair market value, SFFB recognized an intangible asset. SFFB also recognized the premium it paid in excess of the fair market value of the net assets acquired as goodwill. SFFB's acquisition of Metro-County Bank in August of 2002 was accounted for as a purchase.SFFB paid a higher premium to acquire FSBV (a purchase) than it paid for its Horizon acquisition (a pooling). Table 1 shows that the premium was calculated as the excess of deal price over the target's market capitalization on the day before the merger announcement for each merger in which SFFB was the acquirer. A higher premium associated with a purchase instead of a pooling contradicts most observations reported in academic studies. One explanation for this anomaly is that the run-up in the target's share price just before the merger announcement was smaller in the case of FSBV.

The net effect of the new purchase standards gives greater latitude to bidders who no longer structure acquisitions to avoid goodwill amortization. SFFB's third acquisition (Metro-County) reflects this new freedom. Unlike SFFB's two earlier acquisitions, the Metro-County shareholders were offered cash and stock B a combination that, prior to SFAS 141 and 142, would have automatically precluded pooling-of-interests and required amortization of any goodwill recognized in the acquisition.

V. PERSPECTIVES ON LOAN LOSS RESERVES

Bank Regulators' Perspective

Loan loss expenses and provisions for losses are major accounting and regulatory concerns because they reflect the quality of a bank's loan portfolio and applications of accounting standards to mergers. This concern is justified because approximately 60 percent of banks' assets consist of loans. As indicated in Table 3 (item 1), SFFB held over 50 percent of its assets in loans, with much in real estate loans, but its peer group (U.S. commercial banks with assets between SlOO million and $1 billion) held a significantly higher percentage of their total assets in loans.

Regulatory perspectives on loans and adequacy of loan loss provisions reflect different philosophies. Federal bank regulators are concerned about the safety and soundness of the banking system and prefer "strong loan loss reserves that are conservatively measured" (Board of Governors of the Federal Reserve System, 1999, p. 2). Regulators view the allowance for loan losses "as a type of capital that should be accumulated during good times to absorb losses during bad times" (Wall and Koch, 2000, p. 2).

The bank regulator is often portrayed as a risk minimizer who will usually want to err on the side of greater protection. Thus, the regulator will generally encourage loan loss accounting such that loan loss allowances will be sure to exceed the actual loan loss expenses by a substantial margin of protection so that a bank's capital is not impaired. A higher level of allowances than actual loan loss expenses will protect a bank's capital since net income after taxes and dividends will not be understated. However, in periods after the provision is recorded, this could result in an overly optimistic message to shareholders and financial markets that determine the market value of the bank.

One of the concerns of the bank regulator, that is hardly shared by the sec or the FASB and the accounting profession, is that substantial loan loss allowances should be accumulated during prosperous times in anticipation of the time when actual loan loss expenses increase, bank profits decline, and the economy is not so healthy. The bank regulators held this perspective for much of the time between 1997 and 2000. By the middle of 2000 many forecasters were projecting the end of the longest economic boom in US history, and by fall of 2001 some analysts were criticizing bank regulators for stringent bank examinations that were resulting in a credit crunch (Rutledge, 2001).

Accounting Regulators' Perspective

Accounting regulators', including the securities and Exchange Commission, the Financial Accounting Standards Board, and the American Institute of Certified Public Accountants, view the provision for loan losses as limited exclusively to events that are occurring in the current period. They insist on excluding the effect of expected future events and consider it a violation of generally accepted accounting principles if a bank increases its loss provision in anticipation of future events. These accounting regulators are more concerned with reflecting current financial results than protection against future circumstances. For a particular year, the bank report of condition and the income and expense statement need to portray the results for the designated dates, without regard for what might occur in future years.

Accounting for loan losses is sometimes misunderstood. Gunter and Moore (2000, p.31) argue that "through provision for loan and lease losses, banks add funds to their [allowance for loan losses]". Beckett (2002, p. Cl) claims that "Investors say they like to see large reserve cushions at banks so that there is already plenty of money set aside should asset quality deteriorate further in the future." From the accounting regulators' perspective, banks' loan loss provisions reduce income and assets but do not set aside or affect cash. From the bank regulators' perspective, banks' loan loss provisions reduce Tier 1 capital by reducing retained earnings and increase Tier 2 capital by increasing the allowance for loan losses.

The inevitable collision between the accounting and bank regulators concerning accounting for loan losses occurred in the fall of 1998. sec Chairman Arthur Levitt stated that some banks were using loan loss provisions as "cookie jar reserves" to "stash accruals . . . during the good times and reach into them when needed in the bad times" (Levitt, 1998, p. 8). Less than six weeks later, The Wall Street Journal (Brooks, 1998, p. A8) reported analysts' speculation that the sec had "decided to make an example of SunTrust [Banks]... [a] company ... known in the banking industry for its conservative lending practices and pristine asset quality." In response to an sec review, SunTrust reduced its loan loss provisions by a total of $100 million and increased after-tax income by $61.1 million (SunTrust Banks, Inc., 1998, 3).

The controversy among the banking community and financial regulators continues to be a current issue as shown by the American Banker article of April 21, 2003 (Davenport, 2003b). The new regulation does not allow acquiring banks to acquire loan loss reserves and the result may force banks to make considerable changes to their accounting procedures. The goal of the FASB task force was

"to determine how generally accepted accounting principles should be applied to banks' loan-loss reserves" (Davenport, 2003a, American Banker).

Item 2 of Table 3 shows that, except for 2001, SFFB's allowance for loan losses in relation to its total loans receivable has been virtually the same as its peer group. Item 3 of Table 3 shows that SFFB's annual provision for loan losses has been much higher than its peer group, but this is a function of the rapid growth of its loan portfolio (item 4, Table 3). SFFB's growth in loans is consistent with its merger-driven growth in assets (item 5, Table 3).

Vl. FINANCIAL PERFORMANCE DURING EXPANSION

Operating Ratios

During this rapid expansion period, the four operating measures that portray much of SFFB's financial performance are: market capitalization, efficiency, return on assets (ROA) and return on equity (ROE). Calomiris and Karceski (1998) examined similar performance and efficiency ratios for nine banking cases during mergers.

From January 4, 1999 to August 30, 2002, SFFB's market capitalization increased from $34 million to $132 million (a 288 percent increase in 44 months). Table 4 shows that SFFB's efficiency ratio has improved both in absolute terms and, relative to its FDIC-insured bank peer group; in absolute terms, SFFB's expense ratio declined from 64 to 53.5 percent of revenues; SFFB's ratio moved from 3 percentage points below its peers to 9 percentage points above its peers between 1998 and 2001. Table 4 also shows that between 1998 and 2001 SFFB's ROA and ROE improved in absolute terms and relative to its bank size peer group. By 2001, SFFB's ROA increased to equal that of its peer group, and its ROE moved from 2 percentage points below its peers (11.53 vs. 13.57) to almost 4 percentage points above its peers (16.13 vs. 12.24).

In a previous study, the authors (Mayer-Sommer, Sweeney, and Walker, 2005) apply a market model to test whether returns for SFFB are different from the returns on the NASDAQ banking Index before and after each acquisition. After acquiring FSBV, SFFB's returns significantly exceeded the excess returns on the NASDAQ Banking Index for banks in the same size class ($100 million - $1 billion in assets). The returns following SFFB's acquisition of Horizon Bank as well as the acquisition of Metro-County Bank were not significant for SFFB shareholders, which is consistent with findings of no excess returns from mergers by Bruner (2002) and Becher and Campbell (2005).

Impact of Bank Acquisitions on Shareholder Returns

SFFB's acquisitions invite a contrast of how the financial impact of its acquisitions by merger compare to others in the State of Virginia during the same time period. For Virginia banks comparable in size to SFFB, shareholder returns have been examined for the period January 1999 through December 2002.

Virginia is the headquarters for 33 bank holding companies with total assets between $100 million and $1 billion. Fifteen of these banks merged with other banks between 1999 and 2002 and 18 did not. SFFB had total assets of $259 million at the beginning of 1999, but grew to $970 million in assets by the end of 2002.

A t-test can be employed to examine whether there was a difference between average returns to shareholders of the 15 banks that merged, AR(M), versus average returns to shareholders of the 18 banks that did not merge, AR(NM). The null hypothesis is that shareholders of banks that did and did not merge earned equal returns; the alternative hypothesis is that the average returns were not the same.

H^sub 0^ : AR(M) = AR(NM)

H^sub A^ : AR(M) ? AR(NM)

The returns are monthly data for each bank over 48 months (January 1999-December 2002). There are not 48 observations for each bank because some began public trading during the period or previously traded under a different ticker symbol. For the merger group, 13 of the 15 banks traded for all 48 months, another bank traded for 47 months, and one traded for 36 months. Among the group of 18 banks that that did not merge, 5 banks' shares traded throughout the 48 months, 4 banks traded for 47 months, 3 banks traded for 44 or 45 months, 3 banks traded between 35 and 37 months, and a single bank traded for each of 23, 12, and 10 months. To avoid a potential bias in the results, none of the banks were eliminated from the analysis. However, differences in numbers of trading months requires adjustments to calculations of means and standard deviations for the number of months that shares of each bank were traded.

The hypothesis test requires an aggregate mean and aggregate standard deviation for each of the groups of merging and nonmerging banks. Aggregating returns across banks and the 48 months within each group requires an assumption about the variation of returns across the 15 banks within the merger group and across the 18 banks within the nonmerger group. It is assumed that among the 15 merging banks the standard deviations of returns are equal. Among the 18 banks that did not merge, it is assumed that the standard deviations of these banks' returns are equal.

These assumptions do not imply equal standard deviations of returns between the 15 banks that merged and the 18 banks that did not merge. The two groups made different acquisition decisions. It is assumed that the aggregate standard deviation for the 15 merging banks is unequal to the aggregate standard deviation of returns for the 18 banks that did not merge. The different assumptions about equal or unequal variances across returns dictates how the aggregate standard deviation is computed for each group of banks and then for the aggregate standard deviation for the t-test for the null hypothesis: AR(M) = AR(NM).

Table 5 provides a summary of the process for the hypothesis test of equal versus unequal shareholder returns for the Virginia merging and nonmerging banks. Shares of the 15 merging banks traded across a total of 707 months, while shares of the 18 nonmerging banks traded during 715 months. The aggregate mean monthly returns for the merging banks were 0.9773 percent with a standard deviation of 11.4838 percent. The mean monthly returns for thel8 nonmerging banks were 0.9801 with a standard deviation of 8.6159. The relatively high standard deviations were not unexpected because the various regions of Virginia are economically diverse and the size range is rather wide ($100 million to $1 billion).

The test for whether the percentage changes in mean returns of the merging and nonmerging banks are different is the difference in means, AR(M) - AR(NM) = -.0028, divided by the standard deviation for the difference; where there is no basis to assume the standard deviations of the two groups are equal. The standard deviation for the difference in monthly returns between the two groups is 0.5388. The test statistic is -0.0052 = -.0028/.5388.

This t-statistic is not statistically significantly different from zero at any meaningful probability level. The alternative hypothesis AR(M) ? AR(NM) is rejected. For Virginia banks in the size class of $100 million - $1 billion, between 1999 and 2002, there is no basis to believe that the percentage change in shareholder returns for the acquiring banks is different from the percentage change in shareholder returns for banks that did not acquire.

This confirms the finding of other researchers who have asked the question, "Why pursue an acquisition strategy if there is no evidence of gains to acquiring shareholders?" This study examines a group of firms that rarely have been examined with respect to this question - firms that are often too small to be included in standard data bases - CRSP and Compustat. Further research needs to be conducted on the nature of private gains to management of this sample of firms to see if the beneficiaries of pursuing such a strategy are managers who are also, especially in the case of community banks, significant shareholders.

VII. SUMMARY AND CONCLUSIONS

This study expands the evidence that bank mergers rarely enhance acquiring shareholder returns, regardless of whether the merger deepens market penetration or is market-expanding. Only when a merger has unusual characteristics, as was the case for the acquisition of FSBV by SFFB, did shareholders earn excess returns. The directors of FSBV had established the goal for the bank to be acquired eight years before SFFB acquired it and officers and directors held a large percentage of the outstanding shares. SFFB's strategy to combine internal growth with acquisitions by merger led to expanded market penetration and substantial increases in shareholder value. Unlike SFFB, the banks in Virginia that merged and those that did not have similar average returns over the period that SFFB completed its three acquisitions.

SFFB shifted its lending focus from being almost entirely a real estate (88 percent of total assets in 1995) to a more diverse portfolio with 24 percent of its loans to business and 74 percent of its loans in mortgages. This made SFFB an attractive merger partner; subsequent to SFFB's acquisitions, it was acquired, by Provident Bankshares of Baltimore in April 2004.

The revised U.S. accounting rules that mandate using purchase accounting for bank mergers, instead of pooling, and the different perspectives for loan loss provisions by bank regulators and the accounting profession did not seem to have a negative impact on SFFB performance.