Office of Workers' Compensation Programs: Goals and Monitoring	 
Are Needed to Further Improve Customer Communications (03-OCT-00,
GAO-01-72T).							 
								 
This testimony discusses the Department of Labor's Office of	 
Workers' Compensation Programs (OWCP). GAO reviewed how OWCP	 
communicates with injured federal workers, agencies who employ	 
these persons, and medical and other service providers who treat 
them. To evaluate OWCP's system, GAO used criteria suggested by  
the National Partnership for Reinventing Government (NPR). This  
report summarizes GAO's findings on NPR's study of private sector
practices for providing telephone customer service, which	 
included: (1) setting challenging goals for meeting callers'	 
needs for timely and accurate information; (2) collecting	 
credible performance data to measure progress in attaining those 
goals; and (3) improving telephone service by using the 	 
performance data and results to periodic surveys of customers and
stakeholders to determine levels of satisfaction. GAO found that 
OWCP provided consistent customer service regardless of where	 
injured workers live. GAO made 2,400 telephone calls to OWCP's 12
district offices. To compare OWCP's goals and practices for	 
telephone communication with those of model organizations, GAO	 
surveyed three agencies that have won awards for their telephone 
communication practices: the Social Security Administration, the 
Department of Veterans Affairs' Benefits Administration, and	 
Ohio's Bureau of Workers' Compensation. 			 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-01-72T 					        
    ACCNO:   164163						        
    TITLE:   Office of Workers' Compensation Programs: Goals and      
             Monitoring Are Needed to Further Improve Customer
             Communications
     DATE:   10/03/2000 
  SUBJECT:   Claims processing					 
	     Customer service					 
	     Interagency relations				 
	     Program evaluation 				 
	     Telephone						 
	     Workers compensation				 
	     Chicago (IL)					 
	     Dallas (TX)					 
	     District of Columbia				 
	     San Francisco (CA) 				 
	     Seattle (WA)					 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************
GAO-01-72T

United States General Accounting Office
GAO

Testimony

Before the Subcommittee on Workforce Protections
Committee on Education and the Workforce
House of Representatives

For Release on Delivery
10:00 a.m. EDT
Tuesday
October 3, 2000
GAO-01-72T

OFFICE OF WORKERS' COMPENSATION PROGRAMS
Goals and Monitoring Are Needed to Further

Improve Customer Communications

Statement of Michael Brostek, Director
Tax Administration and Justice

Viewing GAO Reports on the Internet
For information on how to access GAO reports on
the INTERNET, send e-mail message with "info" in
the body to:
[email protected]
or visit GAO's World Wide Web Home Page at:
http://www.gao.gov

Reporting Fraud, Waste, and Abuse in Federal
Programs
To contact GAO's Fraud Hotline use:
Web site:
http://www.gao.gov/fraudnet/fraudnet.htm
E-Mail: [email protected]
Telephone: 1-800-424-5454 (automated answering
system)

 (410527)

Statement
Office of Workers' Compensation Programs: Goals
and Monitoring Are Needed to Further Improve
Customer Communications
Page 7                                 GAO-01-72T
Mr. Chairman and Members of the Subcommittee:

Thank you for the opportunity to testify on the
Department of Labor's Office of Workers'
Compensation Programs (OWCP). You asked us to
review how OWCP communicates with injured federal
workers, agencies who employ these individuals,
and medical and other service providers who are
involved in their treatment. In general, our
review focused on how well OWCP's performance
management system was used to respond to
claimants' and other customers' inquiries, both
over the telephone and in written correspondence.
This statement responds to your request by
describing how we conducted our review and what we
found.

To evaluate OWCP's system, we used criteria
suggested by the National Partnership for
Reinventing Government (NPR). 1 NPR performed an
extensive study of high performance customer
service organizations in the private sector and
their best practices for providing telephone
service. NPR stated that the level of service a
customer receives should not vary significantly
across an organization. NPR also found that model
customer service organizations in the private
sector generally follow three consistent
approaches or best practices for telephone
communications. First, they set challenging goals
for meeting callers' needs for timely and accurate
information. Second, they collect credible
performance data to measure progress in attaining
those goals. Third, they continuously improve
telephone service by using the performance data
and results of periodic surveys of customers and
stakeholders to determine levels of satisfaction.

Results in Brief
While OWCP uses these model organizations'
approaches to some extent in communicating with
its customers, its efforts frequently did not
equal those in the private sector that NPR
identified or in the three federal and state
agencies we surveyed that have won awards for
customer service. We found that OWCP

ï¿½    provided widely varying service levels across
its district offices for those attempting to reach
OWCP representatives by phone;
ï¿½    did not set any goals for some important
areas of telephone communications, and the goals
it did set for telephone and written
communications generally  tended to be less
exacting than the goals NRP suggests or that other
customer service operations we surveyed
established for themselves;
ï¿½    did not often collect timely or credible
performance data to gauge progress in attaining
its goals; and
ï¿½    did not adequately survey injured workers,
medical providers, and others to determine levels
of satisfaction, or follow many other practices
that NPR's model organizations use to improve
customer service.

At the conclusion of my testimony, I will discuss
ways that we believe OWCP can improve its
telephone and written communications with
customers.

To get an indication of whether OWCP was providing
consistent customer service regardless of where
injured workers live, we placed a total of 2,400
telephone calls to OWCP's 12 district offices (200
per office). During these calls, we attempted to
obtain information that an injured federal worker
might be calling to request (e.g., status of
compensation payments or medical bill inquiries).
For each call, we recorded how successful the
district office was in providing this information
(e.g., busy signal, no answer after 1 minute,
reached voice mail system, obtained consistent
information about claimant from the automated
system, spoke to an OWCP representative).

To determine what goals OWCP had set to improve
customer service and how the performance data from
these measures are used to improve the program, we
(1) visited OWCP headquarters in Washington, D.C.,
and interviewed officials and obtained
documentation, including strategic and annual
operational plans; (2) visited 5 of the 12 OWCP
district offices (Chicago, Dallas, San Francisco,
Seattle, and Washington, D.C.) and interviewed
district directors and others and obtained
documentation on office practices; and (3)
surveyed the remaining 7 OWCP district offices to
collect some similar information, such as the
extent to which telephone inquiries are recorded.

To compare OWCP's goals and practices for
telephone communication with those of model
organizations, we surveyed three agencies that
have won awards for their telephone communication
practices: the Social Security Administration
(SSA), Department of Veterans Affairs' Benefits
Administration (VBA), and state of Ohio's Bureau
of Workers' Compensation (Ohio's BWC). We asked
OWCP and the three agencies to identify which of
95 telephone "best practices" they use that NPR
identified in its 1995 study.

We did our work between January and September
2000. (See appendix I for additional information
regarding scope and methodology.)

Background
OWCP is responsible for adjudicating and
administering claims of work-related injuries and
illnesses as authorized by the Federal Employees'
Compensation Act (FECA) (5 U.S.C. 8101 et seq., as
amended).2  The FECA program covers nearly 3
million active duty civilian federal employees,
providing benefits to those it determines sustain
an injury or illness in the performance of duty
worldwide.

During fiscal year 1999, FECA's costs totaled
about $1.9 billion in compensation, medical, and
death benefits, and federal employees filed about
167,000 injury notices.  At the end of fiscal year
1999, OWCP was administering about 243,000 ongoing
injury cases, including from previous years, for
partial or total disability.

According to OWCP officials, they receive an
estimated 2.6 million phone calls and 5.5 million
pieces of mail each year from customers--
claimants, medical providers, agencies, and
others. Some mail requires a response-for example,
congressional inquiries on behalf of constituents.
However, district office officials said they
believed that most of the mail does not require a
response. For example, medical reports are used to
assist claims examiners in adjudicating cases but
do not usually require a response. Although OWCP
did not know what proportion of its mail requires
a response, district office officials' estimates
ranged from 1 percent to 7 percent.

The telephone calls and written correspondence are
handled primarily by 12 OWCP district offices
nationwide, which had a total of about 900
employees as of December 1, 1999. These district
offices operate under the authority and guidance
of OWCP headquarters and are responsible for
adjudicating claims from injured workers,
approving wage loss claims, paying medical bills,
and responding to inquiries from customers. Each
district office is responsible for providing
services to claimants living in several states.

OWCP has taken actions to improve customer
communications over the last 5 years. These
actions have included

ï¿½    implementing automated voice response systems
at all 12 district offices to provide information
on the claims, such as the status of bill
payments, 24 hours a day;
ï¿½    expanding automated voice response systems to
allow pharmacy staff to verify claimant's
eligibility and the amounts of drug payments
authorized;
ï¿½    beginning the process of converting incoming
medical bills and other correspondence to a
computerized format to make the information
available to district office representatives via
their computer terminals, enabling them to answer
more queries during initial calls;
ï¿½    giving federal agencies, unions, and
congressional staff direct computer access to
information they need to deal with their
employees' or constituents' cases; and
ï¿½    initiating a communications redesign project
last year--which included  establishing a redesign
team comprised of union members and management to
propose standards, reengineer practices, and make
other improvements in OWCP's communications--and
investigating best practices in public and private
organizations.

OWCP has also taken actions when its monitoring
systems have indicated that district offices have
failed to meet goals for responsiveness to
telephone inquiries. For example, when district
offices failed to meet goals for responding to
telephone inquiries for one or more quarters of a
fiscal year, OWCP's national office counseled the
district directors and required plans for
improvement.

In addition, OWCP's budget request for fiscal year
2001 requested funding for a toll free 800
telephone number for medical authorizations, for
telephone system hardware upgrades, for additional
communication specialists, and for expanded access
to automated information for injured workers. As
of September 22, 2000, the House and Senate
appropriation committees for OWCP had decided not
to fund this request.

Telephone Customer Service Levels Varied Widely
Across District Offices
Although NPR has stated that the level of service
a customer receives should not vary significantly
across an organization, we found that service
levels varied widely for those attempting to reach
OWCP representatives by phone. As figure 1 shows,
the extent to which we were unable to access
district offices' telephone systems on our 2,400
calls-that is, where there was a busy signal, no
answer after 1 minute, or a message erroneously
stating that the phone number was invalid-ranged
from 0 percent in Boston to 54 percent in
Jacksonville.

Figure 1: Percentage of Calls Failing to Access
OWCP Telephone Systems by District Offices

Source: GAO analysis of telephone survey calls to
district offices.

The reasons given for our not being able to access
the phone system varied. For example, in
Jacksonville we frequently experienced busy
signals because district officials there said that
they believe it is better for a customer to
receive a busy signal than to remain on hold for
an extended period of time at the caller's
expense. Conversely, for San Francisco, we
experienced a relatively high frequency of calls
with no answer after 15 rings (about 1 minute).
The San Francisco district director told us this
was caused by a flaw in the phone system that has
existed for years. She said that, although the
customer hears the phone ringing, the system does
not recognize that someone is calling.  She also
said that she had spoken with officials from the
phone company as well as communication officials
in the Department of Labor, but that the problem
remained unresolved. OWCP's Acting Director said
that the problem with the phone system had been
resolved as of September 13, 2000. The Acting
Director also stated that the Washington, D.C.,
office had purchased an additional eight telephone
lines in late July 2000 and that he believes this
will increase the system's accessibility.

We also found that our ability to speak to an OWCP
employee varied significantly across districts. Of
the 2,400 calls we made, 1,200 calls were to
either an office phone number designated for
contacting an employee or a central phone number
that gives callers an option for contacting a
representative. As figure 2 shows, the rates at
which we were unable to reach any employee within
5 minutes ranged from 13 percent to 97 percent of
the calls.3 In three offices-Jacksonville, Dallas,
and New York-we were unable to access an employee
on 97, 86, and 80 percent of the calls,
respectively.4

Figure 2:  Percentage of Calls Failing to Access
an OWCP Representative by District Offices

Source: GAO analysis of telephone survey calls to
district offices.

The most frequent reasons why our telephone calls
did not reach an employee were that we were

ï¿½    still receiving a busy signal or no answer
after 15 rings,
ï¿½    transferred to a voice mail box after
selecting the option to speak to a representative,
ï¿½    still on hold 5 minutes after selecting an
option to speak to an employee, or
ï¿½    disconnected after selecting an option to
speak to an employee.

Officials at the five district offices we visited
said that there were too few employees to both
answer the phones and adjudicate claims as well as
perform the other services that they must provide.
On the other hand, an official at OWCP
headquarters said that they did not want to reward
those offices with the lowest telephone access
rates by giving them part of another office's
staff allocation.

When we made our 2,400 telephone calls, we also
attempted on 1,200 of those calls to compare the
information on actual injured workers' claims
provided to us by OWCP headquarters officials with
the same information available on that claimant
through district offices' telephone systems.5 For
example, if OWCP headquarters told us that
claimant Mary Smith was mailed a compensation
check of $550, would the district office where
Mary's claim was handled provide us with this same
information?

We did not include 604 of the 1,200 calls in our
analysis: 138 calls where we could not access the
phone system for various reasons (e.g., busy
signal); 43 calls where we could access the phone
system but not the interactive voice response
system for various reasons (e.g., claim number was
different from that provided by headquarters), and
423 calls where we could not compare the
information because it had been updated after OWCP
headquarters provided it to us.

For the remaining 596 calls, the extent to which
district offices provided us with consistent
claims information ranged from 88 percent to 100
percent. (See appendix II for information on the
accuracy of each district office's interactive
voice system.)  Most of the inconsistent
information involved the dates and amounts of
claimants' compensation checks.

Other communication practices also varied
significantly across district offices:

ï¿½    The Dallas office, unlike most others, used e-
mail for medical authorizations, congressional
contacts, and general inquiries. The four other
district offices we visited did not use e-mail
because of Privacy Act concerns.
ï¿½    The national office and four district offices
have taken steps to provide customers information
through the Internet, while others have not.  The
Internet-linked offices have established a World
Wide Web page to provide information about the
workers' compensation program and the district
offices' procedures and practices.
ï¿½    Most district offices had representatives
available to answer the phone 7 or more hours per
day, but three offices-New York, Philadelphia, and
Boston--were available by phone 6 hours or less
per day, and one of these offices-Boston-had
representatives available for only 4.5 hours.

OWCP Did Not Have Goals for Some Important
Customer Service Areas
OWCP had not set any goals in some important areas
of telephone communications and the goals it did
set for telephone and written communications
allowed OWCP more time to provide responses to
customers than NPR suggests for telephone
communications or that other organizations we
surveyed allowed.  For example, NPR suggests the
following telephone service goals:

ï¿½    99 percent of callers access the telephone
system;
ï¿½    98 percent of callers reach a customer
service representative, and the time waiting on
line be no more than 30 seconds; and
ï¿½    85 percent of callers' inquiries should be
resolved during the first call.

These three basic goals focus on meeting callers'
needs for timely and accurate information. The
other agencies we contacted--SSA, VBA, and Ohio's
BWC--varied in whether they established goals for
these measures.  Three had goals for telephone
access, two had goals for the portion of callers
reaching representatives and the time they have to
wait on line, and one had a goal for resolving
inquiries on the first call.

OWCP had not set goals that conform to any of
these three goals. OWCP did have a goal to return
90 percent of phone calls not related to medical
authorizations within 3 days to persons who leave
messages. That goal could be met by OWCP's calling
the people within 3 days, giving them the status
of their claims, and saying that the answer to
their questions would follow at a later time.
However, OWCP's Acting Director noted that this is
the only response possible for many calls when
OWCP lacks information, such as doctors' reports,
needed to resolve the caller inquiry. OWCP also
had a separate goal established in fiscal year
1999 to return 95 percent of calls related to
medical authorizations within 3 days.

NPR estimates that organizations that answer a
caller's question on the first call will spend
less time and about half the resources as
organizations that take multiple calls to answer
inquiries.  OWCP's Acting Director said that he
recognized the benefits of answering the calls the
first time and that an ongoing OWCP program to
make more claimant information available on
district office computer terminals could help
achieve such a goal. He added, however, that he
believed establishing a goal for answering queries
would be more appropriate for an organization with
a call center whose employees' only responsibility
is answering telephone calls. He explained that
because district offices have many other
responsibilities in addition to answering calls,
some district offices prefer to direct most calls
to voice mail and respond at a later time.
Several district directors told us that, if there
were such a goal, assigning additional employees
to answer calls would take time away from their
adjudication of claims.

The scope of NPR's study did not include
identifying what goals the private sector has for
responding to written inquiries, as it did for
telephone communications.  Thus, we could not
compare the goals that OWCP has established for
the timeliness of written communications with an
NPR suggested standard.

Nevertheless, for nonpriority mail requiring a
response, OWCP had a goal of responding to 85
percent within 30 days. OWCP also had goals for
responding to priority mail from Congress: 90
percent within 14 days and 98 percent within 30
days. These goals do not compare favorably to
VBA's goal of responding to all written benefit
inquiries within 10 workdays and Ohio's BWC's goal
of responding to written requests the same day, or
within 24 hours of the request's being referred to
another section of the Bureau.6

OWCP also did not have a national goal for
responding to requests for medical authorizations
in the mail, through e-mail, or by fax.
Nonetheless, the five district offices we visited
gave written medical authorization requests
received by mail the same priority that OWCP gave
congressional correspondence. Several of these
five offices have also established their own goals
for medical authorizations received by e-mail or
fax. For example, the Dallas district office
encourages claimants to use e-mail for medical
authorizations. Dallas had a goal of responding to
90 percent of e-mails within 24 hours. The Chicago
district office received 95 percent of its medical
authorization requests by telephone. The district
office has chosen to use the goal for medical
authorizations received by phone-95 percent within
3 days-for authorization requests received by fax.

OWCP Lacked Valid Information to Measure Progress
in Achieving Goals
Often, OWCP did not collect credible performance
data to gauge progress in attaining its goals.
Credible performance information is essential for
accurately assessing agencies' progress toward
meeting existing goals and for setting new goals.7
Decisionmakers must have assurance that the
program and financial data being used will produce
complete, credible, useful, and consistent data in
a timely manner if these data are to inform
decisionmaking.

Progress Measures for Telephone Timeliness Were
Invalid
OWCP's system for measuring its goal of 3 days for
returning phone calls to those who left a message
that required a response did not yield valid
timeliness measurements. It could do this by
either creating a record of all such calls or of a
statistically valid sample. While all calls may
not require a response, district office officials
have stated that most callers have inquiries that
require a response. The national office suggested-
-but did not require--that district employees use
a standardized computer program (CA-110) to make a
recording of all calls requiring a response, as
well as of those in which the content is relevant
to adjudicating decisions.  OWCP did not require
the use of this program for all calls because some
district office employees have complained about
taking time away from their other tasks to record
the information, such as the date and nature of
the call.

We found that all 12 district offices used the CA-
110 system to some extent. However, two of the
five offices we visited-Dallas and Seattle--told
us that they entered only about 15 percent or
fewer of all calls requiring a response and did
not enter calls in the systematic manner that
would be necessary to yield valid results. The
other three offices-Chicago, San Francisco, and
Washington, D.C.--estimated that they entered
about 75, 75, and 95 percent of the calls
requiring a response, respectively. However, our
analysis of the number of calls received and the
number of calls recorded in the CA-110 system
suggests that these estimates are high.8 For
example, San Francisco estimated that it entered
75 percent, but for a 3-month period in fiscal
year 2000, San Francisco received 76,238 calls and
entered 14,502, or 19 percent, in the CA-110
system.

Moreover, four offices-Cleveland, San Francisco,
Seattle, and Washington, D.C--also used
alternative methods to record and track a portion
of their calls. Each of these offices was supposed
to follow a sampling plan approved by OWCP
headquarters when recording information. However,
the OWCP Acting Director said that each of these
four offices had developed a modified version of
the sampling plan and that their plans--while
approved by OWCP--were probably not statistically
valid. We reviewed the national office's sampling
plan for the data to be entered into these systems
and also believe that this plan would not yield
statistically valid results even if implemented as
designed.

The telephone response rates developed using the
CA-110 and other systems indicated that the
district offices were generally meeting their
timeliness goals.  However, because the
performance data were not statistically valid or
collected in enough cases, OWCP could not
determine whether these goals were being met.

Progress Measures for Timeliness of Written
Correspondence Were Also Invalid
We also found problems in the methods that OWCP
used to measure timeliness goals for written
correspondence that undermine the usefulness of
the data. Each district office was required to
take a statistical sample of incoming general,
nonpriority correspondence, and then record and
track the correspondence to determine whether it
was responded to within 30 days. The results of
these samples were to be reported to the national
office on a quarterly basis. Four of the five
district offices we visited each had a different
approach for sampling such correspondence, and
each stated that its approach was not
scientifically developed or developed in a manner
that would produce valid results if projected to
the universe of all responses.9 For example, the
Dallas district office required each claims
examiner to provide his or her supervisor four
pieces of written correspondence requiring a
response per month to determine whether the
response was provided within 30 days. We are
concerned about the validity of this approach
because, among other things, the potential exists
for the claims examiner to give the supervisor
only those letters to which the response was
timely.

The Seattle district office required claims
examiners to log in all general correspondence
received every Wednesday that required a response.
Supervisors were to review the claimants' files
for these letters after 30 days to determine
whether a response had been sent within the 30-day
goal.

We are concerned about the validity of this
approach because, among other things, OWCP
employees may devote extra attention to responding
to letters arriving on Wednesday at the expense of
letters arriving on other days. Although the
performance data showed that this timeliness goal
for fiscal year 1999 was generally met by all
district offices, the results are probably not
statistically reliable.

Conversely, OWCP did seem to have a valid and
reliable system for tracking responses to priority
mail involving congressional requests. For
example, the date of receipt of all congressional
correspondence was to be recorded in the Priority
Correspondence Tracking System. This system
provides reports that track each piece of
correspondence until a response is provided.

OWCP Often Lacked Customer Satisfaction Data to
Know How and Where  Improvements Were Needed
As I said earlier, the third approach that NPR
found model organizations followed was to
continuously improve customer service by using
performance data-including surveys of important
customers and stakeholders-- to identify how and
where improvements are needed.  We found that
OWCP's efforts in this area fell well short of the
best practices that NPR found in the private
sector and in the three agencies we surveyed.

Let me first state the obvious--OWCP did not
measure progress toward  goals that it did not
establish in the first place. OWCP did not have
goals-nor did it have related measures-for three
basic areas of telephone communications:

ï¿½    the percentage of callers able to access the
telephone system,
ï¿½    the percentage of callers who can reach a
customer service representative and their time
waiting on line, and
ï¿½    the percentage of callers' inquiries that are
resolved during the first call.

Officials from SSA and VBA told us that setting
goals for those areas (e.g., telephone access) for
their agencies and measuring the results has
proven useful in identifying areas where customer
satisfaction levels needed improving. For example,
data from SSA's customer satisfaction surveys in
1993 showed that access was the single biggest
factor affecting customer satisfaction. According
to an SSA official, SSA began collecting access
data and found that callers attempting to reach
SSA at the busiest times were getting busy signals
50 percent of the time. SSA established a
telephone access goal and continued to collect
access data, explore new technologies, and acquire
additional telephone capacity. By 1996, SSA said,
it was able to set and achieve a goal of 95
percent of the callers reaching the system within
5 minutes, and customer satisfaction scores
improved accordingly.

My point in using this example is that without
goals or measures in these three basic high
priority areas, OWCP was not in a position to know
what levels of customer service it was providing
and where and how telephone services needed to be
improved.

For the goals that OWCP did establish, we found
that it often did not collect credible performance
data from surveys of (1) injured workers, (2)
medical providers, or (3) employees that was
sufficiently reliable or done in a timely manner
to measure progress or to set goals for improving
customer service.

OWCP Did Little to Gauge Customer Satisfaction
Levels
NPR found that model telephone service
organizations in the private sector survey (1)
their customers frequently to determine how
satisfied they are with the services provided and
(2) their employees who are answering the phones
for job satisfaction levels and ideas to improve
their services. In addition, Executive Order
12862, issued September 11, 1993, directs
departments and agencies to survey customers to
determine the kind and quality of services they
want and their level of satisfaction with existing
services.

Every organization must decide how frequently it
can survey its customers in a cost-effective
manner. However, we found that OWCP (1) did not
obtain information on injured workers'
satisfaction with services as frequently and by
utilizing as many techniques as other model
organizations do; (2) did not survey other
important stakeholders, such as medical providers
and federal agencies; and (3) did not survey the
OWCP employees who are answering the phones.

OWCP did do customer satisfaction surveys of
injured workers. Since 1996, OWCP has hired a
contractor to conduct customer satisfaction
surveys about once each year to determine
claimants' perceptions of several aspects of the
FECA program, including overall service, the
timeliness of responses to telephone inquiries,
and the timeliness, thoroughness, and accuracy of
written responses to claimants' inquiries.10  The
claimants are selected on a random sample basis.
During these surveys, injured workers are asked to
recall situations that occurred up to 1 year in
the past.

NPR recommends as a best practice that
organizations survey their customers constantly to
determine satisfaction levels for existing
services and to gather requests for new services.
Ohio's BWC surveys approximately 600 injured
workers weekly by mail.

The OWCP Acting Director said that OWCP does not
conduct a customer satisfaction survey of medical
providers because it had difficulty doing so in
the past. That is, OWCP officials said they had
previously attempted to survey medical providers;
however, when they called the representative in
the medical provider's office often could not
identify the individual who had previously called
OWCP.

Some district office directors and other officials
cited outreach programs that, while not
necessarily systematic, were an attempt to gain
more input from a broader selection of customers.
For example, the Seattle district office, in May
2000, initiated a program to begin calling a
sample of all telephone callers within the same
week they called. The district director said that
the office asked the callers whether the response
was appropriate and if the caller was satisfied
with the representative's response. OWCP national
office officials said that these efforts did not
ensure that consistent questions were used or that
a random sample of all district office customers
was surveyed.

OWCP also has not surveyed its own employees
regarding customer service or employee
satisfaction within the last 5 years. According to
NPR, employee satisfaction is measured as
routinely as customer satisfaction in model
customer service organizations. NPR cited as
benefits of such employee surveys: obtaining
information on how to improve the work processes
that lead to improved customer service, as well as
identifying employee morale issues that could lead
to customer service problems. In addition,
Executive Order 12862 directed agencies to survey
front-line employees on barriers to, and ideas
for, matching the best in business. Officials from
Ohio's BWC told us that they have systems in
place, such as an annual employee survey that
captures information about employee morale.

OWCP Followed Fewer NPR Best Practices Than the
Other Three Agencies We Visited
     NPR identified 95 best telephone service
practices of customer service organizations in the
private sector. Of these 95, we categorized 20
practices as being related to performance
measurement and 68 as related to improving
customer service through use of performance data.11
Each of these "practices" also contained several
major activities. For example, the practice of
managing customer dissatisfaction with telephone
service included related activities, such as
establishing a dedicated customer relations team
to receive complex customer issues and complaints
and recording caller complaints for use in
identifying the root causes of problems. While no
organization would be expected to apply all of
these best practices, we wanted to determine the
extent to which OWCP was applying these practices
in comparison to SSA, VBA, and Ohio's BWC.
Consequently, we asked these organizations to
characterize to what extent-"all," "some," or
"none"--they followed the activities within each
of these practices.

     Of the 20 best practices related to measuring
performance, OWCP stated that two best practices
were not applicable because they applied to call
center operations and OWCP does not have any call
centers. Of the 18 that OWCP said were applicable,
it reported that it performed "all" or "some" of 9
of the 18 practices, or 50 percent. This compares
to SSA, VBA, and Ohio's BWC, who reported that
they performed 20, 17, and 20, or 100, 85, and 100
percent, of all 20 NPR best practices,
respectively. Of the 9 practices OWCP indicated it
did not perform, SSA, VBA, and Ohio's BWC
responded that they applied 8 of them to all or
some extent. Examples of these 8 performance
measuring practices follow.

ï¿½    Call monitoring: Senior managers regularly
listen in on live calls in order to stay in touch
with the customer. Team leaders participate in
group monitoring sessions to ensure consistency of
measurement.
ï¿½    Accessible to customers: A customer feedback
loop is built into every phase of the customer
service delivery process. It is convenient and
easy for customers to contact world-class
organizations.

 Of the 68 best practices for improving customer
service, OWCP said that 6 practices were not
applicable because they applied to call centers.
Of the 62 practices OWCP said were applicable,
OWCP reported that they generally followed to all
or some extent 31 of the 62 practices, or 50
percent. This compares to SSA, VBA, and Ohio's
BWC, who reported that they followed 65, 51, and
66, or 96, 75, and 97 percent, of the 68 NPR best
practices to all or some extent, respectively. Of
the 31 practices OWCP indicated it did not
perform, SSA, VBA, and Ohio's BWC each responded
that they applied 18 of them to all or some
extent. Examples of these 18 practices follow.

ï¿½    Information queuing: Callers waiting in the
queue are provided with information as to the
expected length of delay, allowing them to choose
whether to stay in queue or hang up.
ï¿½    Resource allocation strategies: Continuous
evaluations of key performance indicators help to
ensure the appropriate alignment of resource
allocations with planning objectives. Well-
established benchmarking programs help identify
improvement opportunities for cross-functional
teams.12

Conclusions
OWCP has been concerned about the level of
services that it provides to its customers-injured
workers, medical service providers, and agencies.
The annual surveys of injured workers that OWCP
has contracted for have surfaced issues, like
access to service representatives, that OWCP has
taken actions to address.  For instance, the
installation of the automated response systems in
district offices have given customers an
alternative means of getting answers to common
questions.  OWCP has also begun to set goals and
use data about goal achievement to manage the
customer service aspect of district office
operations.

However, OWCP's customer service efforts fell
short of practices used in three organizations we
surveyed and in model organizations in the private
sector that NPR studied. According to our
telephone survey, customers' access to OWCP's
telephone systems and to customer service
representatives varied widely by district office.
OWCP had not established goals for several
important aspects of telephone service. Where OWCP
had set telephone or written communication goals,
they generally tended to be less exacting than the
goals NPR suggests or that other customer service
operations we surveyed established for themselves.
Further, for those goals OWCP had established, the
systems for collecting performance information
were not yielding credible information that OWCP
could use to make properly informed decisions
about its customer service operations. In
addition, OWCP officials reported using
proportionately about one half as many best
practices related to measuring performance and
proportionately from one half to two thirds as
many best practices related to improving customer
service through the use of performance and other
data as did the three award winning customer
service operations we surveyed. By developing
goals for important areas of telephone
communications, determining if the timeliness
goals for telephone and written communications can
be made more exacting, and beginning to reliably
measure both customer satisfaction and goal
achievement, OWCP can lay the foundation for
better serving its customers' needs for timely and
accurate information.

Recommendations for Executive Action
We recommend that the Secretary of Labor require
the Director of OWCP to

ï¿½    establish goals for all important areas of
OWCP's telephone and written communications with
injured workers and other customers and revise as
appropriate existing goals to better ensure that
customers' needs for accurate and timely
information are met;
ï¿½    collect credible performance data on progress
toward these goals, including timely periodic
surveys of injured workers', medical providers',
and agencies' satisfaction with OWCP's services
and surveys of OWCP employees to gauge their job
satisfaction and to gather ideas on how to improve
services; and
ï¿½    use these performance data and survey results
to identify areas needing improvement and to
develop strategies for achieving those
improvements, including new and revised goals
where appropriate.

OWCP's Acting Director said that he agreed with
our recommendations and would continue to explore
ways in which to improve customer communications.

That concludes my statement, Mr. Chairman. I would
be happy to answer any questions that you or other
Members of the Subcommittee may have.

Contacts and Acknowledgements
For further contacts regarding this testimony,
please contact Michael Brostek at (202) 512-9039
or Alan Stapleton at (202) 512-3418. Individuals
making key contributions to this testimony
included Jeanne Barger, Thomas Davies Jr., James
Turkett, Michael Valle, and Cleofas Zapata Jr.

_______________________________
1 The National Performance Review, now entitled
the National Partnership for Reinventing
Government, was begun in 1993 under the direction
of Vice President Gore. NPR directed federal
agencies to build a customer focus into their
operations to eliminate unnecessary bureaucracy,
streamline processes, and serve the public more
cost effectively.
2 OWCP also adjudicates and administers claims
authorized by the Longshore and Harbor Workers'
Compensation Act, which covers employees engaged
in maritime employment, and for recipients of
Black Lung compensation.
3 We chose to wait no longer than 5 minutes for a
representative from the time we made a selection
requesting a representative because we believed
this was a conservative approach, given VBA's goal
of reaching a representative within 3 minutes and
NPR's guidance and Ohio BWC's goal of 30 seconds.
4 We also placed 1,200 calls initially to the
district offices' automated telephone voice
response systems, followed by an attempt to
contact a representative. The proportion of
representatives we were unable to contact using
this approach was even higher than when using our
other approach. However, we made a conservative
assumption that a caller whose primary reason for
calling was to speak to a representative, would
call the specific number for accessing a
representative at those district offices with such
a number, or immediately attempt to access a
representative at an office with only one number.
5 For the 1,200 calls we placed attempting to
obtain automated voice information, we were able
to access most of the district offices' telephone
systems on a higher proportion of calls than we
did for the 1,200 calls attempting to access a
representative. For calls attempting to access the
automated system, we were unable to access
district offices' telephone systems on from 0 to
37 percent of the calls to each office.
6 Prior GAO testimony (Veterans Benefits
Administration: Problems and Challenges Facing
Disability Claims Processing, GAO/T-HEHS/AIMD-00-
146) stated that VBA-with an average time for
processing claims of 205 days--was far from
reaching its strategic goal of 74 days. While VBA
did not meet its goal for processing claims, VBA
officials provided reports stating that it did
meet the 10-day goal for responding to nonpriority
correspondence 97 percent of time in fiscal year
1999. A VBA official said that just telling
inquirers that they would be contacted at a later
date with an answer did not constitute a valid
response for purposes of calculating the rate. We
did not independently verify VBA's response rates.
SSA does not have goals for responding to written
correspondence.
7 Managing for Results: Challenges Agencies Face
in Producing Credible Performance Information (GGD-
00-52, Feb. 4, 2000).
8 The five district offices provided us with
information on the number of calls received that
was based on periods from 4 weeks to 3 months.
These offices also provided us with the number of
calls recorded in their CA-110 systems over
various periods of time.
9 Chicago district office officials stated that
they track all of their nonpriority written
correspondence.
10 The surveys were conducted in 1996 and 1998 by
the same contractor. In 1997, OWCP contracted with
another contractor to perform a similar survey.
11 We determined that the other seven best practice
categories were not related to either improving
customer service through the use of performance
and other data or to measuring performance.
12 In a second response to our request after
reviewing our draft testimony, OWCP revised its
responses to 15 of the best practices, including
increasing from 8 to 14 the number of practices it
considered not applicable because the activities
applied to call centers. Based on OWCP's revised
response, OWCP considers 3 of the 20 best
practices for performance measurement not to be
applicable and reported that it performs all or
some of 11 of the remaining 17 practices, or 65
percent. Of the 68 best practices for improving
customer service, OWCP said that 11 were not
applicable. OWCP reported that they followed to
all or some extent 39 of the remaining 57
practices, or 68 percent. We reported OWCP's
original responses in comparison to those of the
other organizations' responses because the other
organizations did not have an opportunity to
revise their responses to our survey. In addition,
we did not have an opportunity to analyze OWCP's
revisions, including its position on the NPR
practices that it viewed as applicable only to
call center operations.

Appendix I
Scope and Methodology
Page 23                                GAO-01-72T
     To determine how OWCP communicates with
injured federal workers, agencies who employed
these workers, and medical and other service
providers who are involved in their treatment, we
performed the following audit steps.

We placed 2,400 telephone calls to OWCP's 12
district offices (200 per office), to assess the
accessibility of telephone representatives at each
of the 12 OWCP district offices and whether each
office's automated integrated voice response (IVR)
system's data was consistent with the national
office's claim data. We generated listings for our
callers which provided the dates, times, and phone
numbers to be called, and when applicable, the
claim information to be accessed. OWCP provided
information about the breakdown of IVR call types
for four of the twelve OWCP district offices
(Chicago, Philadelphia, Kansas City, and Seattle)
during a 2-month period.  We used this information
to build a nongeneralizable profile of the
distribution of IVR calls across the following six
types of claim information for each district
office (number of calls per district office in
parentheses):

ï¿½    claimant calling about a bill payment ( 47),
ï¿½    medical provider calling about a bill payment
  (8),
ï¿½    medical provider calling about a periodic
  roll payment  (24),
ï¿½    claimant calling about a compensation check
  (10),
ï¿½    claimant calling about physical therapy
  authorization  (2), and
ï¿½    medical provider calling about physical
  therapy authorization  (9).

For the IVR calls, we were provided with
identifiers, such as claim numbers and employer
identification numbers, that enabled us to enter
the IVR and identify specific transactions on each
office's automated database.  We worked with OWCP
to identify a time period of earlier transactions
from the national database that would still exist
in each district office's IVR system during our
test period in June and July 2000.  OWCP sampled
records from their national database for this time
period until they obtained the required number of
records of each type, for each district office.
OWCP's selection of IVR cases did not strictly
constitute a random sample of cases for each
office from the specified time period, since the
cases were selected in case number order, which
generally reflected a chronological order, until
the required number of cases was obtained.
However, since we had no reason to believe that
time ordering of the cases was associated with
whether a district office's IVR data would match
the information in the national database, we
accepted OWCP's selections for use in our test.

All of our test calls attempted to reach an OWCP
representative, either directly or after accessing
the IVR system.  We therefore randomly scheduled
all our test calls only during the hours that each
office told us that a "live" representative should
be available to respond to customer inquiries.
Saturdays, Sundays, and Tuesday, July 4, 2000,
were excluded from our test days.

For district offices that had the same telephone
number for the automated voice response system and
for a representative, we called that number 200
times and attempted to reach a representative
immediately upon accessing the system on 100 of
those calls and attempted to access claim
information on the automated system and then
access a representative on the other 100 calls.
For offices that had separate numbers for the
automated voice response system and the
representative, we called each number 100 times.
As before, upon completion of the automated voice
response system, we attempted to reach a
representative. All representatives contacted were
informed that they had participated in a GAO
telephone survey. When attempting to access an
OWCP office's telephone system, we let the
telephone ring 15 times (over 1 minute) before
making the determination that the telephone system
did not answer.  And, in waiting for a
representative to answer the line, we waited at
least 5 minutes after making the selection to
speak to a representative before determining that
the call was not answered. We used the 5 minute
time period because we wanted to be more
conservative than VBA's goal of a caller's
accessing a representative within 3 minutes and
NPR's guidance and Ohio's BWC's goal of 30
seconds. For each call, we recorded how successful
the district office was in providing services an
injured worker or medical provider might desire.
The information recorded included busy signals, no
answer after 1 minute, whether we reached an
office representative, whether we reached the
automated voice response system, and whether
consistent information was provided about the
claimant by that system, such as the amount of a
medical payment or the status of medical
authorization for physical therapy. We conducted
this telephone survey over a 6-week period in June
and July 2000, and attempted the calls throughout
the business hours listed for each district
office.

If the test calls we made were considered random
samples of customers' telephone experiences during
the test period, the following statements could be
made about the precision of the estimates:

ï¿½    Estimates of the proportion of calls in each
  district in which customers were unable to access
  the telephone systems, and the proportion of calls
  in each district in which customers were unable to
  reach an employee within five minutes, have
  sampling errors of no more than 10 percentage
  points.

ï¿½    Estimates of the proportion of IVR calls with
  consistent data (among IVR calls for which
  transactions could be tested) have sampling errors
  of no more than 10 percentage points unless
  otherwise noted in table II.1.

The data we collected, however, were from test
calls rather than "actual" customer calls.
Characteristics of the test calls that might
affect the outcomes we measured, such as time of
day, day of week, or subject matter, might not
have mirrored the profile of these characteristics
among "actual" customers' phone calls in any
district during the test period. Therefore, the
results displayed for individual districts might
differ from the ones we might have obtained by
sampling "actual" customer calls, by amounts
larger than the stated sampling errors.

ï¿½    At OWCP's headquarters in Washington, D.C.,
we interviewed knowledgeable officials, reviewed
strategic and operational plans for fiscal years
1995 through 2000 to identify goals and measures
related to responding to customer inquiries, and
obtained communication reports showing data that
would indicate OWCP's performance for the same
period.  We also discussed with these officials
methods for testing the communications at OWCP
district offices, and obtained claimant and
medical provider information so that we could
perform a telephone survey of the 12 district
offices.

ï¿½    We visited 5 of the 12 OWCP district
offices-Chicago, Dallas, San Francisco, Seattle,
and Washington, D.C. At these five offices, we
conducted an in-depth review; we interviewed
regional directors, district directors, claims
managers, claims examiners, and workers
compensation assistants to gain an understanding
of the communications at OWCP from various
perspectives. We obtained available communication
reports, accountability review reports, reports
measuring written and telephone communication
performance, and other applicable communication
data. We selected the five offices based on
several factors to obtain a mixture of offices of
differing sizes and levels of performance. The
primary factors were (1) the number of employees
and the number of cases managed at these offices,
which ranged from among the lowest to among the
highest of all the district offices; (2) OWCP's
telephone and written responsiveness measures,
which indicated that some offices had and some had
not met national performance goals; and (3) the
proximity of two of the offices to our Washington,
D.C., and Dallas Regional Office staff.

ï¿½    We surveyed the other seven district offices-
-Boston, Cleveland, Denver, Jacksonville, Kansas
City, New York, and Philadelphia--via a
questionnaire and obtained general information
about their communication practices, such as the
number of employees each assigned to respond to
telephone calls.

ï¿½    To compare OWCP's goals and practices for
telephone communication with those of leading
organizations, we also surveyed three agencies
that have won awards for their telephone
communication practices: the Social Security
Administration, Department of Veterans Affairs'
Bureau of Benefits Administration, and state of
Ohio's Bureau of Workers' Compensation (BWC). We
asked OWCP and the three agencies to identify
which of 95 telephone "best practices" they used
that NPR identified in a 1995 study. We did not
attempt to verify or validate their responses. In
addition, we interviewed and obtained documents on
communication performance goals and practices from
officials at the three organizations and compared
OWCP's performance goals and measures with those
of private sector organizations identified in the
NPR study and those of the three organizations we
surveyed.

We did our work between January and September 2000
in accordance with generally accepted government
auditing standards.

Appendix II
Access Rates For And Accuracy Of OWCP District
Offices' Automated Interactive Voice Response
Systems
Page 24                                GAO-01-72T

Table II.1: Number of GAO Survey Telephone Calls
Able to Access and Complete OWCP Interactive Voice
Response (IVR) Systems, Number of Transactions
Tested, and Number of Transactions With Consistent
Claims Data by District Office
District        Number of   Number of    Number of    Number of      Number
office         calls made  calls able   calls able transactions (percentage)
               to attempt   to access  to complete      testeda          of
                access to         IVR          IVR              transactions
                      IVR                                              with
                                                                 consistent
                                                                      datab
Bostonc               100         100           98           59      55(93)
Chicagoc              100          79           77           51      48(94)
Clevelandc            100          93           91           54     49(91)d
Dallasc               100          95           95           51      48(94)
Denver                100          96           96           60      56(93)
Jacksonville          100          84           63           32     28(88)e
c
Kansas City           100          93           93           59      58(98)
New Yorkc             100          90           90           52     47(90)d
Philadelphia          100          95           81           46      44(96)
San                   100          63           63           31    31(100)d
Francisco
Seattle               100          99           97           59      55(93)
Washington            100          75           75           42     42(100)
DC
Totalsf              1200        1062         1019          596     561(94)
aAlthough we made 100 calls to each district, the
number of transactions tested at each district are
less than 100 because either we could not access
the automated voice response system (e.g., busy
signals, no answer, etc.), could not complete the
transaction (e.g., lost connection, system
rejected claim input, etc.), or the system had
been updated by more recent data, (i.e.,
information was updated after OWCP provided the
sample claims).
b We compared data obtained from automated voice
response system with claim data provided by the
national office and determined whether each
district office's system's data were consistent
with the national office's claim data. We did not
independently verify the accuracy of the national
office's data.
c The district office had separate telephone
numbers for accessing the automated interactive
voice response system and accessing an OWCP
representative. The other district offices had
only one main number.
d If the test calls were considered as a random
sample, the lower bound of the sampling error
associated with this estimate is 11 percentage
points less than the value of the estimate.
 e If the test calls were considered as a random
sample, the lower bound of the sampling error
associated with this estimate is 16 percentage
points less than the value of the estimate.
f An average for all OWCP district offices cannot
be provided, because OWCP could not provide the
total number of telephone calls each office
received, and thus we could not weight the sample
to accurately reflect the impact of each district
office's performance on a national average.
Source: GAO analysis of telephone survey.
*** End of document ***