r"O;"~--~~=AAi§'5;;;;:S5""·;;.,sc,,.;:~-,,>,,~
... ---~,,-::...-;,-i.:" ..~:"-'.-.'"'-.- ... -
- ----
--~~--
---
-,-,~--.-.
; ~
EXPLOITING THE SAS SYSTEM - THE EXPERIENCES OF A SMALL AGENCY Introduction The Defence Analytical Services Agency (DASA) is a small Government agency providing statistical and analytical advice and tools to customers in the UK's Ministry of Defence (MoD). In 1991, DASA chose the SAS System as its strategic information package. This paper describes how DASA harnessed the SAS System to its business needs, and gained a competitive edge.
What DASA does DASA is not alone in providing analytical services to the MoD. Statistical and other methods are used by a wide range of specialists employe_d.b.y.the MoD, such as scientists, engineers, psychologists and operational researchers. DASA's efforts are mainly directed at helping the MoD manage its business sensibly. The main services we provide are:in the Manpower area •
current and historic manpower statistics for budget monitoring, manpower resourcing and formulating personnel policy
•
forecasting services for budget planning, manpower planning and formulating personnel policy
•
maintenance of a tri-Service medical events database to monitor the health of the Services, and an associated database for retrieving medical documents
in the Financial and Economic area •
advice on price movements for contracts, provisioning and budget management
•
analysis of MoD expenditure for cash management and central financial policy
•
industrial and economic analyses for financial management and policy making
•
energy use statistics for monitoring the MoD's contribution to the energy efficiency programme
in the Logistics area •
I ~.-
.. -
a project capability to address major logistics problem areas such as inventory management, vehicle fleet management, equipment quality and acceptance schemes, analysis of trials data and through-life costing
285
other services
:4i'l'
\',:
•
statistical surveys, mainly of MoD personnel, to establish data not readily available from administrative sources
• • • •
risk management advice for project and budget managers advice on performance measurement and indicators statistics on military quarters statistics on aircraft accidents
Statistics to Other Government Departments d~ta
to the Central_ ~~istical Office for National Accounts
•
balance of payments
•
estimates of Research and Development expenditure to the Central Statistical Office and for the Cabinet Office
•
manpower numbers to Her Majesty's Treasury, the Office of Population Censuses and Surveys, and the Department of Employment
We employ about 120 people of whom 20% are professional statisticians, 20% IT specialists and the remainder general administrative staff. DASA is part of the UK's Government Statistical Service.
The Creation of DASA Three years ago, around the time DASA chose the SAS System, it was facing a number of challenges arising from general Government policies and from pressures within the MoD. The Government had brought in a number of initiatives to improve the delivery of services in the public sector. •
"The Financial and Management Initiative" which sought to place accountability and responsibility at the same level, so that whoever takes the decision to commit expenditure also pays for it.
•
"The Next Steps" which created government agencies which had clearly defined functions and were given the resources and freedoms within which to operate to meet set targets for performance.
•
"Competing for Quality" which allowed the private sector to compete with the public sector to provide services, through a process called market testing.
We in the public sector had to get better.
286
~~S5S-J.""5S?r-·"'S.~~~~h:'-~-;;::----::::~-:C::'·..:::'_~;'-
j
In the MoD, there was considerable pressure to reduce defence expenditure. The end of the Cold War had raised expectations of a 'peace dividend', and the rundown of the military forces by some 20% was initiated in a study called "Options for Change". A linked study of the MoD's headquarters organisation (of which DASA is a part) also recommended a reduction of 20% in the MoD headquarters numbers. At that time, DASA did not exist. We had been a single organisation previously, but one of the MoD's many re-organisations had split us into three groups each in a different command chain. In fact, we did not meet as an organisation until three levels above DASA's most senior officer. This dispersed organisation was inefficient and ineffectual. The lack of co-ordination resulted, amongst other things, in an unfocused approach to customers who received services from more than one part of the Agency. Few mechanisms existed to meet the changing needs of the MoD. Instead the tendency had been to continue services in areas served historically, often with the main point of service delivery_ being directly to low-level users giving a narrow view of the service required.
:~
We also suffered from a lack of visibility at the higher management levels..of the MoD. Although we were performing valuable services, the dispersed organisation obscured the overall value of our services to the MoD. All of this meant that there were questions over our continued survival. One option given serious consideration was to disband us and give the resources to our customers. However, we managed to persuade our leaders that that would be even less efficient. We won the day and were allowed to become an Agency in July 1992. The sting in the tail was that we were told to take a 20% reduction in staff numbers over two years. Meeting such a target, at a time when our customers were actually needing more of our services to help them cope with change, called for radical change. If we did not want to lose whole services we had to make a technological leap and we had to deploy our resources better. The SAS System gave us the technology, and the technology allowed the reorganisation of resources to be made. Manpower Statistics Production The area we concentrated our energies upon was the production of manpower statistics. We had four manpower branches providing manpower statistics and forecasting services to the three Services (Royal Navy, Army and Royal Air Force) and civilian personnel management. We had always operated a policy of collocating manpower branches with their customers to give a knowledgeable service. For each Service, manpower statistics were produced from the computer records of service maintained on the administrative pay computers, but the methods by which statistics were produced varied considerably. The oldest system was the Army system, built in the late 1970's, and run on our behalf by the Army. It consisted of a large number of individually tailored PLI programs, each producing a hard-copy return. Each month the wheel would turn and a roomful of paper
287
_. --- -- -----..-,,.: :: -.::. .:.~..: ~.~ :..~ -~:../:.-.'~'- ~ -::',; -:."~, ,".. " ... -
,:~. -:"'::" ..
.
would be churned out. We would have to extract the information from the mounds of paper to produce our products - analyses and summary information, input to forecasting models and so on. This arrangement suffered from a number of problems - it was not core business to the Army computer centre, so it was given low priority; the system was cumbersome and difficult to maintain; it contained errors and inconsistencies; it was inflexible and unresponsive so that it took months to get a new return commissioned. In fact, the system was obsolete, but because it was going to be expensive to replace, it had been kept soldiering on. However, by mid-1991, it was clear that the system would further deteriorate to an unacceptable level after April 1992, as it would not be providing the information required. The Army planned to disband the Woman's Royal Army Corps and let female personnel join the regiments. It also planned the amalgamation of various regiments and corps. These plans meant that the Army's computer administration systems had to be changed significantly. The Army did not have enough resources to rewrite the statistics system as well as sorting out the administration systems. The Royal Navy and civilian systems we ran ourselves, but they used COBOL instead of PLl. The Royal Air Force system used a more modem approach. Previously, it had run a COBOL-based system like the Army's. However, in the early 1980s, we had moved to a system whereby an extract file was produced which we analysed using the tabulating package TAU (developed by the Office of Population Censuses and Surveys, OPCS, to analyse the national census). The benefits of such a system are immense - we have more control over the system; it is more robust and consistent; it is more flexible and adaptable; we get quicker response; there are lower maintenance costs; and most importantly, it becomes the accepted prime source and people no longer need to keep their own statistics. Even that system was becoming obsolete. OPCS were moving from an ICL-based environment to an IBM-based environment for mainframe computing, and had decided not to port TAU across. Instead they had chosen the SAS System for tabulation. (One of reasons we chose the SAS System was that if OPCS, who had developed TAU, felt confident that the SAS System would meet their needs then we could be confident it would probably meet ours). So faced with these three factors - the need for DASA to make substantial reductions; the collapse of the Army system and the demise of TAU - we set about building our first system in the SAS System. This came to be known as CASSANDRA which stands for Civilian And Service Statistics AND Related Analyses. CASSANDRA CASSANDRA is an integrated system producing statistics on all MoD personnel. CASSANDRA takes data each month from the pay and records of service held on five MoD administrative computers. CASSANDRA is big! Each month it takes data on almost 800,000
MoD personnel - 330,000 regulars (57,000 Royal Navy, 160,000 Army and 110,000 RAF); 356,000 reserves (300,000 Army and 56,000 RAF); and 110,000 civilians. CASSANDRA validates and cleans the data to remove obvious errors, then generates event
288
F~~--=~-£35:.."':::.~~"'-~~~,,"--~'-~""'::C~'...-----'~
;
data, such as intake and outflow, by comparing this month's data with the previous month's data. CASSANDRA produces a large number of regular statistical returns distributed to a wide audience within the MoD. As well as producing paper output, we have used the SAS System to build flexible user systems. The Planning Tables system was developed to replace a large paper return which cross-tabulated· events data profiled by age, such as intake outflow, promotions, with classification data, such as rank, regiment, commission type and so on. These data are used by the manpower branches to prime forecasting models and set assumptions. The ad hoc system was developed to allow our people without knowledge of the SAS systems to answer simple queries, (and relieve pressure on the hard pressed IT staft).
I ,;~
t; ;~
CASSANDRA can also export data to other information systems to make our data more available within the MoD .
.;',
i~
;; "
~}
Although it has taken two years to get CASSANDRA fully operational, the first parts were up and running very quickly -- the basic system the Army was developed in just three months.
i"
t,
-for
~
;,;.
t· (i
The main problems we encountered were through our lack of expertise in the-SAS System we had decided to grow our own experts rather than buy them in. We also had to overcome deficiencies in presentation in the SAS System. Because we were producing high volumes of tables, we used Proc Tabulate rather than Proc Report. Proc Tabulate has problems with some of our basic needs such as sub-totalling, blank lines to group or space out rows, and poor presentation. We got round these by various tricks, but they were and remain a source of irritation.
~
;.'.
r
"i
;~~
~ ;.
.L1"' $:
.( ( 'i
~
MEDICS
.l~
,ff, '1
h i!
Meanwhile, we set about converting another COBOL system to the SAS System.
~1"
,~
One of jobs we perform concerns medical information on service personnel. We take a microfilm copy of certain medical documents, each of which is given a unique camera number, and maintain a computer index which allows the records for an individual to be easily retrieved by listing all the necessary camera numbers.
~
~
~f;. J"
,f.
t~: J
We also code certain medical events to the Manual of International Statistical Classification of Diseases, Injuries and Cause of Death (ICD) and the NATO Cause of Injury codes in the Forces Supplement to the ICD Manual. This coded information is used to monitor the health of the Armed Forces and as a research tool.
:£
~: .W J:
Ii
1.~;
I "
l,,'
We get the data on forms from medical centres and hospitals. To make sure we attribute medical documents or events to the correct individuals, we check the service identity number and name shown on the forms. The service identity number is checked against a master list and if valid, and the name corresponds, then the document or event is accepted .
{
~,:
;J.: J'
~'
."" ~
$. ,
~ '.
I "
' Ii
\
I.
MEDICS is linked to CASSANDRA for personnel data, and for aggregate data for analyses .
\.. MEDICS too is big! Some 8 million documents are currently logged in the microfilm index,
tko,
i ~
¥
-
..
289
with about 8000 new documents being added each week. On the coded database, there are some 3 million events logged, with about 3000 new events per week.'· MEDICS is an on-line system which allows our people to enter new events to the database. It validates data on entry and highlights problems, such as overlapping dates. It allows correction of existing data. It produces regular reports and has a sub-system for answering ad hoes enquiries.
ETHOS Another linked system is ETHOS - the Ethnic Origin Survey. The MoD wishes to ensure equal opportunities for all Service personnel and is collecting information on their ethnic origin, which can be analysed to see if there is any underlying difference between ethnic groups in retention or promotion rates. The data are being collected on forms through a census being conducted by DASA. ETHOS will hold the ethnic origin code and use service identity number as the key to link through to CASSANDRA for personal information. As well as sub-systems for analysis, ETHOS will have a sUb-system to monitor response to the census and produce lists of
personnel who have not responded for subsequent follow-up action: It will also have a subsystem to respond to enquiries under Data Protection legislation. ji
Future Developments Future developments using the SAS System include:•
the conversion of a generalised manpower forecasting model developed by DASA in the early 1980's and written in COBOL and FORTRAN. Although the model is extremely powerful and one of our key tools, it needs to be brought up to date.
•
the rewrite of the BASICI spreadsheet system used to manipulate defence price indices data.
•
more small user-systems for particular needs - even EIS
We also need to overhaul all the systems developed so far. Although these are less than three years old, the SAS System has moved on considerably and we need to take advantage of the new features such as object orientation, to make our systems as efficient and robust as possible.
Lessons Learnt So what have we achieved? First we made a technology leap - though it was not a leap into the unknown as our experience of package software made us aware of what was achievable, and we had the advantage of having a firm handle on our data. The technology leap did what we required it to do - it allowed us to re-organise and· take the
290
20% cut in staff without cutting outputs. By building systems quickly '!Ie gained a competitive edge. We now hold a source of easily accessible corporate data which we can exploit. We have a responsive system which we can use to help our customers over this next difficult period when once again there is severe pressure to reduce defence expenditure without reducing defence capability. This has helped us to increase our value to and our visibility within the MoD.
~
~ ~
h
l
Hopefully this will stand us in good stead when we are subject to market testing.
!,
f
The lessons to be learnt from our experience are:
'~.
r;
~
0
" b
•
harness Information Technology to you business needs and use IT to change the way you do business - Business Re-engineering
•
do it quickly to get a
~
t:
f
.f:
~ompeti~ive
ecige :
~~pid
Application Development
~~.
,i~
;!.'
Our opinion of the SAS System? Despite the limitation on presentation, it is very, very good. We made the right choice for our strategic information package. .
i~
f
}, ~1
~: 'r !" ~1 ~l,
rr
~,
-:;-:;
Ii,
i'
Il
~
.' "~.:
Ii
~~ ;1
~f.. ~
.
~
l.}:
J~X
J;"
ji
8,;
f; W,
~ ~
! t
~'
f Ii'
~
g ~
r i
I .' j
\ )
.
\
