【正文】 d then populate this framework using a bination of surveys and semistructured interviews. We then pare the results of this exercise with a synthesis of the literature on IT decision making to understanding how practices vary across firms and the extent to which this is consistent with best practices as described in previous literature. Finally, we will pare these processes to internal and external performance metrics to better understand which sets of practices appear to be most effective.　　To make these parisons concrete, we examine both the general decision process as well as the specific processes used for two recent IT investment decisions :the adoption of puterbased home banking (PC banking), and the development of the corporate web site. These decisions were chosen because they were recent and are related but provide some contrast。 in particular, PC banking is a fairly well defined product innovation, while the corporate web presence is more of an infrastructure investment which is less welldefined in terms of objectives and business ownership.　　Overall, we find that while some aspects of the decision process are fairly similar across institutions and often conform to best practice as defined by previous literature, there are several areas where there is large variation in practice among the banks and between actual and theoretical best practice. Most banks have a strong and standardized project management for ongoing systems projects, and formal structures for insuring that linemanagers and systems people are in contact at the initiation of technology projects. At the same time, many banks have relatively weak processes(both formal and informal) for identifying new IT investment opportunities, allocating resources across organizational lines, and funding exploratory or infrastructure projects with long term or uncertain payoffs.　　The reminder of this paper is organized as follows. Section 2 describes the previous literature on performance of financial institutions and the effects of IT on performance. Section 3 describes the methods and data. Section 4 describes the current academic thinking on various ponents of the decision process and pares that to actual practices at the banks we visited. Section 5 describes the results of our indepth study of PC banking projects and the summary, Section 6 contains a similar analysis for the Corporate Web Site and discussion and conclusion appear in Section 7. How Financial Firms Decide on Technology(Part Three) 　　Previous Literature　　Performance of Financial Institutions　　There have been a number of studies that have examined the efficiency of the banking industry andthe role of various factors such as corporate control structure (type of board, directors, insider stock holdings, etc.), economies of scale (size), economies of scope (product breadth), and branching strategy。 see Berger, Kashyup and Scalise (1995) and Harker and Zenios (forthing) for a review of the banking efficiency literature. While there is substantial debate as to the role of these various factors, there is one unambiguous result: that most of the (in) efficiency of banks is not explained by the factors that have been considered in prior work. For example, Berger and Mester (1997) estimate that as much as 6590% of the xinefficiency remains unexplained after controlling for known drivers of performance. A similar story also appears in insurance where xefficiency varies substantially across firms when size, scope, product mix, distribution strategy and other strategic variables are considered. It has been argued that one must get inside the black box of the bank ot consider the role of organizational, strategic and technological factors that may be missed in studies that rely heavily on public financial data.　　 Information Technology and Business Value　　Early studies of the relationship between IT and productivity or other measures of performance were generally unable to determine the value of IT conclusively. Loveman (1994) and Strassmann (1990) ,using different data and analytical methods both found that the performance effects of puters were not statistically significant. Barus, Kriebel and Mukadopadhyay (1995), using the same data as Loveman, found evidence that IT improved some internal performance metrics such as inventory trunover, but could not tie these benefits to improvements in bottom line productivity. Although these studies had a number of disadvantages (small samples, noisy data ) which yielded imprecise measures of IT effects, this lack of evidence bined with equally equivocal macroeconomic ananlyses by Steven Roach (1987) implicitly formed the basis for the productivity paradox. As Robert Solow (1987) once remarked, you can see teh puter age everywhere except in the productivity statistics.　　 More recent work has found that IT investment is a substantial contributor to firm productivity, productivity growth and stock market valuation in a sample that contains a wide range of industries. Brynjolfsson and Hitt (1994,1996) and Lichtenberg (1995) found that IT investment had a positive and statistically significant contribution to firm output . Brynjolfsson and Yang (1997) found that the market valuation of IT capital was several times that of ordinary capi