【正文】
ess model—personalized service and customized financial services (., small business loans) in the case of small banks and efficient distribution of relatively uniform types of financial services (., credit cards and home equity loans) in the case of large banks. The business model of the small bank requires relatively high cost, while larger banks can keep cost low. Under this line of reasoning, both types of banks should have a role to play in the future financial services marketplace. Noheless, differences in PROFEFF are important because ultimately small and large banks pete for capital. For example, the decision of a smaller bank to join or not to join a large banking organization through a merger is ultimately a subjective decision about how its capital can be best employed. Given these considerations, two important questions raised by Berger and Mester (1997) must be considered before we proceed. The first is the appropriate variable—assets or equity—to use in normalizing profits in puting the PROFEFF measure. The second is the use of one frontier or several frontiers in paring banks of different sizes. Because PROFEFF, when normalized by equity, measures how well a bank utilizes its financial capital, we choose to use this measure. Some earlier studies paring large and small banks, such as Akhigbe and McNulty (2020), use assets and find small banks have higher PROFEFF. Use of equity can be expected to produce the opposite result since large banks use more leverage than small banks. In other words, the PROFEFF measure that we use is closer to return on equity, which should show greater PROFEFF for large banks. Normalizing by assets is likely to produce the opposite result. Since we want to consider the sources of the differences in PROFEFF, we use three different frontiers for small, medium, and large banks. This is consistent with the assumption that their focus, and their basic business model, is different. This procedure allows the PROFEFF measures to have maximum flexibility—small bank PROFEFF and its frontier are not constrained or affected in any way by the activities and balancesheet structure of large banks, and vice versa. Thus, when we look at the determinants of PROFEFF for the three groups, if they are different, this will reflect real differences, and if they are the same, it will not be because the same frontier was imposed on all banks. We recognize the alternative argument that, in paring the performance of different banks, one normally wants to use the same test, not two or three different tests.(We made this argument ourselves in an earlier paper.) Profit Efficiency Trends for Various Bank Size Groups PROFEFF has declined sharply in recent years for small banks, from in 1995 to in 2020. We consider the hypothesis that this decline may reflect an increasing number of de novo banks in the small bank category. FDIC data indicate that between 1992 and 1994 only 74 new banks per year were chartered, which no doubt reflects the depressed state of the banking industry at that time. In contrast, in the six year period from 1995 to 2020, there were an average of 175 new bank charters per year. Many of these banks remain small for a number of years after being chartered. DeYoung and Hasan (1998) show that de novo banks are much less profit efficient than older, similarly sized banks. In Table 1 the percent of banks in the under $100 million dollar category that are de novos (age under 10 years) has increased from perce