Discussion of impact of information technology on public accounting firm productivity.

I. INTRODUCTION

I am grateful for the opportunity to discuss Banker et al. (2002) (hereafter "the paper" or BCK). This paper undertakes an important research question; what is the benefit to public accounting firms of implementing audit software and information and knowledge-sharing

applications (ASIKSA)? This paper has an excellent research design in that using a single firm over time controls for many possible confounding effects from using cross-sectional data. In addition, having substantial access to the research site allows for a triangulation approach, using qualitative and quantitative data to address the same research question.

An important contribution of this paper is its exposure of Data Envelopment Analysis (DEA) to a wide audience. DEA provides a method for analyzing multiple inputs and multiple outputs to determine efficiency. Most information technology (IT) research questions involve multiple inputs such as technical IT skills, IT infrastructure, other IT resources and capabilities, and other factors of production. New IT often results in multiple outputs such as improved decision making and timeliness of reporting, increased and broader information availability, increased productivity, and increased profitability. Traditional techniques restrict researchers to a single output and only a few inputs. DEA can expand the value of IT research by adding other relevant inputs and outputs. Despite the paper's numerous contributions, I do have some questions regarding the research question, methodology, and conclusions, which I address in the following section.

II. DISCUSSION

Research Question

Papers published in the mid-1990s show that IT can result in increased efficiency (e.g., Brynjolfsson and Hitt 1995, 1996; Dewan and Min 1997; Hitt and Brynjolfsson 1996; Lichtenberg 1995). What the literature has yet to conclusively demonstrate is how to turn IT spending into profitable, competitive-advantage-granting investments.

The strategic management literature tells us that for strategic actions to lead to a sustained competitive advantage, they must be value-creating, different from what competitors are doing, and prohibitive to copy (Mata et al. 1995; Bharadwaj 2000). BCK state that their analysis demonstrates "the value impact of IT investment in public accounting firms." However, it might have been useful for readers if the magnitude of the IT expenditures in 1998 was provided to gauge the value-creating potential of the ASIKSA. Without knowing IT expenditures, ROI and increase in value are difficult for the readers to discern. It is difficult to judge the return on investment without knowing the amount of the investment. Similarly, it is difficult to discern the value impact of ASIKSA without considering the cost of the new systems.

The second criterion for strategic implementations to lead to a sustainable competitive advantage is that the selected strategy is different from what competitors are doing. It might have been beneficial to have information on the state of the accounting industry in the FIRM's country to know if the FIRM is a leader or follower in the use of ASIKSA.

The third criterion for strategic implementations to lead to a sustainable competitive advantage is that the selected strategy must be prohibitive to copy by competitors. It does not appear as if the FIRM did anything that other accounting firms would not have a relatively easy time replicating. Any competitive advantage gained through the implementation of ASIKSA will likely be short-lived. IT investments that usually increase value the most are those that are not imitable by competitors. Although there might be a small increase in firm value from implementing ASIKSA, more value could be generated from an IT investment that leads to a sustainable competitive advantage.

Methodology

Measures Used

The primary benefits from ASIKSA are better decision making and reduced time required to complete audit tasks. There may be better measures of these benefits than revenue and spending measures aggregated across audit engagements. Although the study controls for the effects of inflation by using constant dollar measures, it does not control for price index changes in audit fees and labor costs. It would have been helpful to know if audit fees and labor costs changed at the same rate from 1997 to 1999. If the audit fee price index changes at a different rate than the labor price index, then the discrepancy will show up in efficiency measures, but is unrelated to the new ASIKSA. Thus, other measures may be called for. Whereas dollars are confounded by other factors, hours and number of personnel are not.

Compelling arguments about the efficiency of the audit software could be made by examining the number of hours worked, or size of the audit staff assigned to the same client in 1997 and again in 1999. One of the junior auditors in the FIRM claims, "With a computer, my working hours on an engagement can be reduced by more than half of that without the computer." An audit senior at the FIRM noted that due to the implementation of a client database, "work hours for a client can be reduced significantly." An alternative unit of measure for efficiency in this analysis might be hours.

BCK note the impact of ASIKSA on upper-level audit staff: "Since the FIRM's audit software organizes all audit evidence collected by juniors and seniors in an electronic format, audit managers are likely to be more effective when reviewing such data." Measures such as time spent by managers and partners on a particular engagement could proxy for these positive effects of ASIKSA. Upper-level audit personnel should spend less time on engagements in 1997 compared to the same engagement in 1999. Likewise, if ASIKSA increases the audit staff's access to a wide variety of experience-based knowledge, then upper-level audit personnel should spend less time answering questions and guiding audit staff on a particular engagement. The number of hours worked by managers and partners on engagements in 1997 compared to the same engagements in 1999 could capture this benefit of ASIKSA.

Another measure of the benefits from ASIKSA would be the total number of days that pass from the start of an audit to its finish. This is different from the number of hours worked on the audit, because it also counts days when the audit team is delayed and not working. These delays could be due to reviews by managers or partners, time waiting for further instructions or answers to questions, etc. ASIKSA might pay off in reducing these delays by speeding up the review process and reducing the delays due to waiting for answers to technical or procedural questions. The total number of days that pass from the start of an audit to the finish in 1999 compared to 1997 would capture all of these benefits.

One possible measure of improved audit quality and decision making is to compare the number of additional tests required by managers and partners on a particular engagement in 1997 compared to the same engagement in 1999. The number of decisions made by lower-level staff that was overturned by higher-level staff in 1997 compared to 1999 could proxy for changes in the quality of decisions due to ASIKSA.

Missing Variables, Confounding Effects

There are numerous possible omitted variables that might be correlated with the dummy variable, POSTITINV in 1998 and 1999. In order to establish causality between ASIKSA and changes in efficiency, possible omitted variables must be ruled out as alternative explanations. For example, changes in the legal environment in the FIRM's country such as new securities laws, laws that affect audit risk, tort reform, or new labor laws could result in changes in efficiency as measured in the paper, but are not due to the ASIKSA. Other possible omitted variables include changes in the FIRM's budgeting process, business environment, and organizational learning.

How new engagements in 1998 and 1999 compare to the FIRM's engagements in 1997 is another possible confounding effect. For example, if new audit business consists of relatively "easy" audits compared to audits in place in 1997, then the average efficiency of the FIRM would increase, even if ASIKSA had no effect on the efficiency of the FIRM. The FIRM also might have taken on new clients in 1998 and 1999 that were considered "high risk" and therefore paid higher audit fees relative to the amount of work necessary to complete the audit. These types of engagements would increase the efficiency measures used in the paper, but are not due to ASIKSA. One way to control for many of these issues is to compare efficiency only in the subset of engagements that were in place in 1997 and in 1999.

It is impossible to list all of the possible correlated omitted variables here. Any economy- or industry-wide structural changes and any changes in the FIRM not due to ASIKSA will all show up in measures of the FIRM's efficiency, yet are unrelated to ASIKSA. A research design such as matching with a second audit firm, controlling for changes in the FIRM from 1997 to 1999, and matching engagements in 1999 with engagements in 1997 could mitigate many of these problems.

III. RESULTS

Although we cannot definitively answer if the implementation of ASIKSA resulted in increased efficiency, it does not appear so. As shown in Table 1, vertical analysis of the FIRM's income statements shows the annual labor costs for accounting professionals within the FIRM increased from 54.5 percent of the FIRM's revenues to 58.9 percent from 1997 to 1999. Annual operating costs decreased only slightly over the same period, from 20.1 percent in 1997 to 19.4 percent in 1999. Profit margin decreased from 25.4 percent in 1997 to 21.7 percent in 1999. Particularly striking is that revenues added since implementation of the ASIKSA have only an 11.3 percent profit margin. Given that, except for the very largest public accounting firms, the FIRM operates in an industry characterized by economies of scale in auditing (Cheng et al. 2000), ASIKSA did not appreciably improve efficiency.

Any benefits in the year after implementation of ASIKSA are most likely more than offset by initial setup, adjustment, learning costs, and delays due to unfamiliarity with the new system. The FIRM started and completed implementing ASIKSA in 1998. I would expect 1999 to be a year of adjustment/acclimation, not a year of improvement. I do not doubt that the ASIKSA will eventually increase efficiency, and wonder if the results would be different if 2000 was used in the analysis instead of 1999.

IV. CONCLUSIONS

BCK do an admirable job in trying to address an important topic, and I hope that theirs is the first of several papers in this area that will assist audit firms in making strategic IT decisions. The paper makes a contribution by using qualitative and quantitative methods to provide insight into the implementation process and benefits from ASIKSA. In addition, BCK show the benefits of using DEA analysis in IT efficiency research.

TABLE 1

Vertical Analysis

                      1997        1999      Change    1997   1999

Revenue         $3,564,006  $4,817,692  $1,253,686  100.0%  100.0%
Profit Cost      1,943,725   2,836,482     892,757   54.5    58.9
Operating Cost     716,306     935,357     219,051   20.1    19.4
Profit Margin      903,975   1,045,853     141,878   25.4    21.7

                Change

Revenue         100.0%
Profit Cost      71.2
Operating Cost   17.5
Profit Margin    11.3

All numbers are in 1996 dollars, and disguised by multiplying by a
scalar.

Revenue = annual revenue of the FIRM;

Profit Cost = annual labor costs for accounting professionals in the
FIRM

Operating Cost = annual operating costs for the FIRM, including
administrative staff, rent, utilities, office supplies, travel, postage
and telephone, advertising, training, depreciation and amortizatin, but
excluding all items related to accounting professional labor and IT
costs; and

Profit Margin = Revenue -- (Profit Cost + Operating Cost)

REFERENCES

Banker, R. D., H. Chang, and Y. Kao. 2002. Impact of information technology on public accounting firm productivity. Journal of Information Systems 16 (Fall): 209-222.

Bharadwaj, A. 2000. A resource-based perspective on information technology capability and firm performance: An empirical investigation. MIS Quarterly 24 (March): 169-196.

Brynjolfsson, E., and L. M. Hitt. 1995. Computers as a factor of production: The role of differences among firms. Journal of Economic Innovation and New Technologies (3) May: 183-199.

-----. and -----. 1996. Paradox lost? Firm-level evidence on the returns to information systems spending. Management Science 42 (4): 541-560.

Cheng, T. W., K. L. Wang, and C. C. Weng. 2000. Economies of scale and scope in Taiwan's CPA service industry. Applied Economics Letters 7 (6): 409-414.

Dewan, S., and C. Min. 1997. The substitution of information technology for other factors of production: A firm level analysis. Management Science 43 (12): 1660-1675.

Hitt, L., and E. Brynjolfsson. 1996. Productivity, profit and consumer welfare: Three different measures of information technology. MIS Quarterly 20 (2): 121-142.

Lichtenberg, F. R. 1995. The output contribution of computer equipment and personnel: A firm level analysis. Journal of Economic Innovation and New Technologies (3): 201-217.

Mata, F., W. Fuerst, and J. Barney. 1995. Information technology and sustainable competitive advantage: A resource-based analysis. MIS Quarterly 19 (4): 487-505.

I acknowledge the financial support provided by the Virginia Paul Dee Professorship at the University of New Hampshire.