AUDIT ANALYTICS AUDIT

ESSAY 1: CONTINUOUS AUDITING—A NEW VIEW
4.2 Time Focus
Auditing has been retroactive since its inception. Auditors examined past
accounts for accuracy and reported perceived discrepancies. Figure 1-7
illustrates that auditing can both be reactive and predictive
(Kuenkaikaew, 2013). When predictive, the auditor (Vasarhelyi and
Halper, 1991; Vasarhelyi, Alles & Williams, 2010) will rely on models
(standards) to predict results (performance) in an account (transaction)
(Kogan, et al., 2014). This prediction is compared with actuals in near-real
time to detect substantive variances in monitored processes. Much of the
recent research on CA has recently focused on developing better models
for actual comparisons (Chiu, Liu, and Vasarhelyi, 2014; Brown et al.,
2007). These variances, from improved models, are treated either as an
alert to the management and audit function or, if the system has reliable
filters, to prevent faults from progressing toward execution. Modern
systems combine management action and assurance. Much conceptual
work is needed in the re-definition of concepts such as auditor
skepticism, independence, materiality, auditor role, audit objectives, and
so on. Many of these needs are motivated by the ever-increasing level of
automation in corporate business systems and the correspondingly
automated nature of tools used by individuals. The advent of a
progressively bring-your-own-device (BYOD) environment (Loras et al.,
2014) is affecting the locus of the control and assurance. Some of the
BYOD tools like smart phones already incorporate predictive algorithms
to perform a set of integrative functions for the user. These functions
associate typical behavior with data integration to decrease key-strokes
by the user. For example if the device detects a request for contact and a
telephone number or an address, it may immediately prompt a call or a
map to the location.
Auditors will eventually have predictive procedures to drive them to
data level prediction (Kuenkaikaew, 2013), procedural prediction (based
on the experience of other auditors using the tools and maybe the
guidance of the audit plan), and control prediction (where weaknesses in
controls or process changes drive the activation or re-parameterization of
controls.)
Intelligent preventive controls are progressively permeating the corporate
IT ecosystem and personal devices. The relationships between processes
that have always existed may now be explored analytically and visually
for management and assurance purposes. If the predictive ability of
models is high and processes modularized and discrete, it may be
possible to prevent an error, automatically correct an error, or correct a
control deficiency prior to its occurrence. For example an insurance
company develops a forensic model to determine if a particular claim
29