Hsihui Chang

This paper employs Data Envelopment Analysis (DEA) to evaluate the returns to scale patterns of public accounting firms in the post Sarbanes-Oxley Act (SOX) period. Analyzing operations data on top 100 public accounting firms over the period of 2003 similar to 2007, our statistical test results indicate that public accounting firms exhibit decreasing returns to scale in the post SOX period, suggesting that public accounting firms should consider either downsizing and/or divestitures in order to improve their efficiency.

Parallel processing systems are used today in many applications such as in vision, robotics, real-time processes etc. It is therefore important to develop simulators to aid automated design of parallel-processing systems. This Paper discusses development of a Meta-simulator for simulating and analyzing such systems. This could form a part of a general multi-processor system CAD package. The statistical results derived using the simulator can be used to generate Reduced Instruction Set Computing (RISC) elements providing reduced hardware complexity and improved overall system performance. Two methods of implementation - namely using, Instruction Set Processing Specification, ISPS and High Level Language, C are reviewed and compared. Evaluation of candidate architectures requires use of high level and assembly languages for writing benchmark programs for the simulator. We therefore discuss the notion of Meta-Assembler and Meta-Compiler also.