The Problem

The Government of Canada had been using a legacy payroll system for over $40$ years. It wanted

to integrate its systems that was used to pay about $290,000$ employees from $101$ agencies and

departments with wages of about $$22$ billion per year.

Payroll was processed in many different locations, often using different forms and processes.

There were over $1,300$ people who were considered to be payroll specialists. Their jobs included

the recording and processing of payroll payments and adjustments, and correcting payroll errors.

Many processes, such as calculating partial pay period amounts, were done manually.

Sometimes, the same employee $($particularly contract employees$)$ were paid twice by different

agencies or departments.

The IT Solution

Public Services and Procurement Canada $($PSPC$)$ wanted to save money on payroll processing

by adopting one common payroll system that was more comprehensive and versatile and

accessible by all departments and agencies: a modern database approach. It also intended to save

money by physically consolidating the payroll processing for close to half of the departments and

agencies at a new pay centre in Miramichi, New Brunswick.

The traditional systems development life cycle takes a long time. The initial proposal to replace

the existing system was developed by PSPC in spring $2009.$ Extensive consultation took place,

and a public request for proposals was issued, and the results carefully analyzed during the

period June to October $2010.$ Over six months later, in June $2011,$ IBM was selected to

customize the Peoplesoft commercial payroll software into a new system that PSPC called

Phoenix.

It took over a year and a half $($to December $2012)$ for the budget allocated to the Phoenix project

to be approved, and another two years for the customization to be completed, which was pilot

tested in June $2015.$ During the first two months of $2016,$ independent advisors $($S$.$ J$.$ Systems

and Gartner$)$ were asked to review whether the software was ready to implement and to assess

conversion plans and contingency plans at individual departments and agencies.

The PSPC oversight group, called the Public Service Management Advisory Committee, was

informed and was part of the decision$-$making process. On February $24,2016,34$ departments

and agencies went live with Phoenix, followed by the remaining $67$ on April $21,2016.$

The results were chaos.

COMM $213$: Management Information Systems $($Fall $2023)$

$3$

The Result

In spring $2018,$ the Auditor General of Canada reported in the Building and Implementing the

Phoenix Pay System report that $$the Phoenix pay system is less efficient and less cost$-$effective

than the old system, and thousands of employees have not been accurately paid or paid on time.$$

In $2017,$ the Auditor General had reported that a year and a half after Phoenix launched, there

were $150,000$ federal employees with errors in pay that had to be corrected. Vulnerable

employees, like students working on a summer contract, did not get paid for months. Some lost

their year at school because they could not afford to pay tuition. Some contract employees were

paid even after their contract expired and wondered about the impact on their income taxes and

when they would be asked to repay the funds.

Over a thousand payroll employees were re$-$hired after they had originally been laid off, to help

correct the errors. Over a year after the system had been implemented, the estimated amount of

payroll that was outstanding and still to be corrected stood at over $$520$ million.

A separate satellite payroll office was established simply to handle the backlog of errors, so that

the $$regular$$ payroll could be effectively processed.

Information systems and functional departments usually set targets on quality control and

accuracy. PSPC$$s target for payroll accuracy was $95$ percent, which still would have resulted in $5$

percent of employees not being paid accurately, a rather poor record. Sadly, as of August $23,$

$2017,$ only $49$ percent of payroll processing met the accuracy standards of the PSPC$.$ That meant

that $51$ percent of payroll payments were going out wrong, creating more errors. Fifty$-$one

percent of $290,000$ is an awful lot of errors.

As of June $2018,$ PSPC stated that it could take another five years and an estimated additional

$$500$ million per year $($which calculates out to a total of $$2\mathrm{.}5$ billion$)$ in programming,

maintenance, and other costs to stabilize the system.

Questions

$1.$ What are some actions that PSPC could have taken to more effectively assess the

readiness of the system? $(2)$

$2.$ Why was parallel processing not used and could it have prevented the mess? $(2)$

$3.$ What SDLC implementation method was used? Discuss the advantages and

disadvantages of using it in this application? $(2\mathrm{.}5)$

$4.$ What actions could PSPC have taken to mitigate the effects of hardship that arose due to

employees not being paid for several months? $(1\mathrm{.}5)$