Академический Документы
Профессиональный Документы
Культура Документы
in Government
Managing for Performance and Results Series
Thomas H. Davenport
President’s Chair in Information Technology
and Management
Babson College
Sirkka L. Jarvenpaa
James Bayless/Rauscher Pierce Refsnes Chair
in Business Administration
University of Texas at Austin
2008 M a n ag i n g f o r P e r f o r m a n c e a n d R e s u lt s
Thomas H. Davenport
President’s Chair in Information Technology
and Management
Babson College
Sirkka L. Jarvenpaa
James Bayless/Rauscher Pierce Refsnes Chair
in Business Administration
University of Texas at Austin
TA B LE OF CONTENTS
Foreword...............................................................................................4
Executive Summary...............................................................................5
The Strategic Use of Analytics..............................................................6
Strategic Analytics in Government..................................................6
Concepts, Methods, and Tools for the Strategic Use of Analytics.........8
A Model for Assessing Analytical Capability...................................9
Analytics in Government Health Care................................................11
Evidence-Based Medicine.............................................................11
Fraud Prevention..........................................................................12
Disease Management...................................................................13
The DELTA Model in Government Health Care............................14
Supply Chain and Human Resource Analytics in Government............15
The Origins of Supply Chain Analytics.........................................15
Current Approaches to Supply Chain Analytics.............................15
Human Resource Analytics...........................................................16
The DELTA Model for Supply Chain and Human
Resource Analytics.................................................................17
Analytics in Government Revenue Management................................19
Revenue Analysis..........................................................................19
Compliance Systems.....................................................................20
Fraud Detection............................................................................22
Taxpayer Customer Services.........................................................22
The DELTA Model for Revenue Management ..............................23
Intelligence as an Analytical Domain..................................................24
Types of Intelligence.....................................................................24
Need for More Analytics in Human Intelligence...........................25
The DELTA Model for Intelligence Analytics.................................25
Conclusion: Analytics as an Effective Tool for Government................27
Sweden: Strategic Analytics in Government..................................27
Next Steps: Implementing Analytics in Government.....................28
Endnotes.............................................................................................30
About the Authors...............................................................................32
Key Contact Information.....................................................................34
STrategic Use of Analytics in Government
F o r e w o r d
While the opportunities from analytics for improving efficiency and effec-
tiveness in government appear limitless, there is much less clarity about
the readiness of government to embrace analytics. While analytics is often
depicted as a technological innovation, Davenport and Jarvenpaa are careful
to point out that the use of analytics requires managerial innovation. Toni Yowell
E x e c u t i v e S u mm a r y
In recent years, breakthroughs in data-capturing While the opportunities from analytics for improving
technologies, data standards, data storage, and mod- efficiency and effectiveness in government appear
eling and optimization sciences have created oppor- limitless, there is much less clarity about the readi-
tunities for large-scale analytics programs. Several ness of the government sector to embrace the
organizations in the private sector have not only lev- opportunities. Whereas analytics is largely depicted
eraged fact-based decision making, but also created as a technological innovation (often described as
sustained competitive advantage from data-based “business intelligence”), the strategic use of analyt-
analytics. They have built their business strategies— ics in both the private and government sectors also
at least in part—around their analytical capabilities. requires massive managerial innovation. On the
whole, while we found many examples of the suc-
While government organizations and agencies don’t cessful use of analytics in government, we did not
necessarily compete with one another, they use find the elements of leadership, an enterprise orien-
analytics to enable and drive their strategies and tation, and long-term strategic targeting that would
performance in an ever more volatile and turbulent characterize both managerial innovation in general
environment. Analytics and fact-based decision and a strategic focus on analytics in particular. That
making can make just as much or even more of a is, the applications of analytics we discovered were
powerful contribution to the achievement of govern- more tactical than strategic in nature, albeit impor-
mental missions as they can to the accomplishment tant to the successful operations of the organizations
of corporate business objectives. employing them.
www.businessofgovernment.org
STrategic Use of Analytics in Government
government sector to do so. Whereas analytics is framework in the strategic management literature,
largely depicted as a technological innovation specifically the dynamic capabilities literature.
(often described as “business intelligence”), the
strategic use of analytics in both the private and To develop this report, we relied on secondary
government sectors also requires massive manage- literature (on both business intelligence and “the
rial innovation. On the whole, while we found business of government”) to identify agencies or
many examples of the successful use of analytics, external suppliers to government agencies as
we did not find the elements of leadership, an adopters of analytics within the four areas of health
enterprise orientation, and long-term strategic tar- care, supply chain, revenue management, and intelli-
geting that would characterize both managerial gence. We identified a person in charge of either an
innovation in general and a strategic focus on analytical group or a key consultant to that group,
analytics in particular. and conducted a semi-structured telephone interview
with that individual. In several instances, the person
In this report, we explore the successes of analytics invited two or three others from the organization to
in governmental agencies and attempt to develop participate in the conference call in order to provide
an assessment framework for those that are yet to a more accurate and broader description of the ana-
embark on the analytics journey or are still in the lytics activities. In a few cases where analytical
early stages of it. We focus in particular on four activities were well documented in the secondary
governmental areas: health care, logistics, revenue literature, we relied solely on those accounts. We
management, and briefly (because of the paucity primarily focus on analytics in the U.S. government,
of public sources) intelligence. While there are cer- but occasionally address examples and findings in
tainly other domains of government in which analyt- other countries where we could find them.
ics can be applied, these four certainly provide an
overview of the issues involved in their application.
The four sections identify governmental organiza-
tions that are exploiting analytics to meet their stra-
tegic goals. After the description of these activities
and, in some cases, their impact, we discuss key
factors that the agencies have faced in implementing
analytics. We discuss each agency in terms of the
key components necessary for leveraging analytics
in our assessment framework. We ground this
What Is Analytics?
From Competing on Analytics: The New Science of Winning by Thomas H. Davenport and Jeanne G. Harris
(Boston: Harvard Business School Press, 2007)
By analytics we mean the extensive use of data, statistical and quantitative analysis,
explanatory and predictive models, and fact-based management to drive decisions and
actions. The analytics may be input for human decisions or may drive fully automated
decisions. Analytics is a subset of what has come to be called business intelligence:
a set of technologies and processes that use data to understand and analyze business
performance….
In principle, analytics could be performed using paper, pencil, and perhaps a slide
rule, but any sane person using analytics today would employ information technology.
The range of analytical software goes from relatively simple statistical and optimization
tools in spreadsheets (Excel being the primary example, of course), to statistical software
packages (e.g., Minitab), to complex business intelligence suites (SAS, Cognos, Business
Objects), predictive industry applications (Fair Isaac), and the reporting and analytical
modules of major enterprise systems (SAP and Oracle).
www.businessofgovernment.org
STrategic Use of Analytics in Government
The existing literature on strategic orientations to precise risk and asset management; or even hiring,
analytics in the private sector emphasizes several motivating, and managing high-quality human
factors that are almost always present in analytical resources. In the private sector, the implementation
competitors. Having a strong analytical orientation of analytical strategy has required a long, often
would seem to be a function of data and informa- arduous journey. For example, the Barclay’s UK
tion technology (IT), and indeed those resources Consumer Cards and Loans business took more
are critical for analytical success. However, the than five years to implement its “Information-
necessary IT tools, both hardware and software, Based Customer Strategy,” undertaking techno-
are widely available in the marketplace. The requi- logical, process, and organizational tasks to
site data may be more difficult to capture and man- exploit analytics in its credit card and other
age, although private sector firms are increasingly financial businesses.5
able to mobilize it given their investments in enter-
prise software, point-of-sale systems, and electronic Analytical competitors also have strong human
commerce. Providing data for analytical applica- analytical capabilities at the leadership and analyst
tions means that it must be of high quality, sepa- level. They have senior executive teams that are
rated from transaction systems in a data warehouse fully committed to analytical strategies and capa-
or single-purpose “mart,” and consistent through- bilities. They also have a cadre of analytical profes-
out the organization. The most successful analytical sionals who can both perform the needed analyses
competitors also uncover data sources that are new and work closely with decision makers to interpret
for their industry; Progressive Insurance, for exam- and refine the analytical models.
ple, pioneered the use of credit scores for pricing
automobile insurance. What is not widely available in either the public
or private sectors of the economy is the human
A second attribute is that the firm takes an enter- dimension of analytical competition: leadership,
prise-wide approach to managing information and disciplined management, and deep analytical
analytics. An organization may begin by working expertise. It is these human attributes that truly
with particular business problems or functions, but differentiate successful analytical competitors. We
their usage rapidly becomes cross-functional. It is therefore argue that managerial innovation6 is a
usually necessary to take an enterprise perspective better approach to establishing strategic analytical
to pull together the expertise, data, and systems that capabilities than technological innovation.
allow the optimization of organizational relation-
ships and resources. There are now a variety of analytical applications,
or tools, which can be grouped under the term ana-
Analytically focused organizations apply analytics lytics. Some of these applications are used for inter-
to a clear strategic target or intent that they are nal analytics (financial, research and development,
attempting to optimize over time. The target may human resources) and some for external analytics
be based upon strong customer relationships and (customers, suppliers). The box on page 9 describes
loyalty; highly efficient supply chain management; some of the best known analytical applications that
are now either in use by government or could be Figure 1: DELTA Model for Assessing Analytical
used by government in applying analytics to analyz- Capability
ing the activities and programs of government.
Activity-based costing (ABC): The first step in activity-based management is to allocate costs accurately to
aspects of the business such as customers, processes, or distribution channels; models incorporating activities,
materials, resources, and product-offering components then allow optimization based on cost and prediction of
capacity needs.
Monte Carlo simulation: A computerized technique used to assess the probability of certain outcomes or risks by
mathematically modeling a hypothetical event over multiple trials and comparing the outcome with predefined
probability distributions.
Multiple-regression analysis: A statistical technique whereby the influence of a set of independent variables on a
single dependent variable is determined.
www.businessofgovernment.org
STrategic Use of Analytics in Government
Analytics is increasingly important in health care, their doctors only about half of the time, resulting
and in virtually every society around the globe, in 98,000 annual deaths from medical errors, and
health care is—in part or in whole—a government another 126,000 deaths from physician failures to
responsibility. Even in the United States, where pay- observe evidence-based care protocols for four
ment for health care is largely privatized, govern- common conditions: hypertension, heart attacks,
ment paid 40 percent of the $2 trillion spent on pneumonia, and colorectal cancer.12
health care in 2005. Whether the providers and
payors of health care are public or private, analytics In some institutions, guidelines from evidence-
is key to health care performance across at least based medicine are incorporated into online health
three domains: evidence-based medicine, payment care protocols to be followed by medical practitio-
fraud reduction, and the identification of patients ners in treating patients; in other cases, the practi-
for disease management. tioner is presumed to consult online or print-based
reference sources (though given the above study,
In the United States, the two biggest government this may be a dubious assumption).
health care programs are Medicare (administered by
the federal government) and Medicaid (administered The U.S. federal government has played a role in
by states). The Department of Veterans Affairs (VA) both aggregating and publishing evidence-based
also has a large health care program, the Veterans practice guidelines, and in applying them in its
Health Administration (VHA). All three health care clinical practices. In the Department of Health
programs are increasingly focused on analytics. and Human Services, the Agency for Healthcare
Medicare is perhaps an exemplar of disease man- Research and Quality (AHRQ) attempts to collect
agement, while the states have taken the lead in and distribute evidence-based practices in a vari-
Medicaid fraud reduction. The VHA is one of the ety of medical domains. AHRQ evidence reports
leading health care provider organizations in the synthesize and summarize the published evidence
use of evidence-based medicine. for or against the use of methods of testing, diag-
nosing, treating, managing, or preventing diseases.
The agency commissions research to 14 Evidence-
Evidence-Based Medicine
Based Practice Centers at medical centers based at
Evidence-based medicine (also known as evidence- universities and research institutes in the United
based practice) is simply the use of the best-available States and Canada.
evidence from clinical research studies to guide the
diagnosis and treatment of patients for specific med- The VHA hospital organization is one of the primary
ical problems. This approach may seem obvious, but practitioners of evidence-based medicine in the
physicians do not often consult evidence in making United States, and has pursued it along several dif-
medical decisions; one study found they relied on ferent dimensions, which has led to a high level of
their own memory or intuition in 70 percent of care quality and performance. The VHA’s approach
patient consultations.11 A RAND Corporation study has a number of components:
found that Americans receive appropriate care from
www.businessofgovernment.org 11
STrategic Use of Analytics in Government
that fraud and misuse are contributing to the pilot study in the U.S. called Medicare Health
substantial growth of Medicaid claims. “Anecdotes Support is under way, with over 100,000 partici-
are no longer going to be the norm,” he said. “Now, pants in eight regions. The programs are focused
through statistical-driven reporting … we’re able to particularly on medical conditions that lead to high
back up what we had suspected in ways we were costs. The role of analytics in disease management
not able to do in years past.”16 is to identify the patients most at risk for diseases
and high-cost conditions.
The Department of Veterans Affairs also employs
analytics for fraud reduction through the Veterans Two examples illustrate the problem that disease
Benefits Administration (VBA). The VBA matches management is attempting to address. About 14
income data with the Internal Revenue Service and percent of Medicare beneficiaries have congestive
Social Security to ensure the people who receive heart failure, but these patients account for 43 per-
VA pensions (which are dependent on having low cent of Medicare spending. About 18 percent of
incomes) aren’t getting more income than they are Medicare beneficiaries have diabetes, yet diabetes
reporting. The VBA also analyzes high-value claims patients account for 32 percent of Medicare spend-
checks, which typically go to people who have had ing.18 Disease management programs are designed
long-standing disabilities but have recently filed a to control costs and improve treatment by closely
claim. The organization produces a dashboard that monitoring patients’ conditions, educating them to
the heads of regional offices use to adjust resources, manage more of their own treatment and supporting
and also has a dashboard that managers use to them in bringing about behavior changes that could
evaluate performance and manage their parts of improve their conditions. Patients are also reminded
the business. of tests and treatments needed at particular times in
the course of their diseases.
Of course, fraud reduction can also apply in other
domains of government health and human services. While the programs provide services to all patients,
For example, in 2004, the state of Michigan imple- those at particularly high risk of future medical com-
mented a “reverse wage match” to determine plications are identified through predictive analytics.
whether recipients of day care benefits actually Individuals at risk are identified by disease-specific
received wages from employment; since 2004 the algorithms based on medical coding structures and,
program has identified more than $17 million in in some cases, pharmacy data. The analytical tech-
fraudulent payments. In Food and Cash Assistance niques employed include linear or logistic regres-
programs, Michigan has combined data from recipi- sion analyses, classification/decision trees, and
ent electronic benefits transfers, food assistance neural networks. Identified patients receive addi-
records, participating retailers, and geocode data on tional health care or social/behavioral interventions
store locations to identify sources of fraud. The sys- designed to reduce the risk of avoidable, costly
tem identifies fraud in terms of clients who have left medical interventions in the future.
the state but failed to report their departure, recipi-
ents who travel long distances to patronize stores, Disease management programs have been shown
and excessive reimbursement requests from conve- to foster improved health through prevented or
nience stores and gas stations. While these fraud delayed onset of complications and related diseases,
prevention efforts are important, the dollars spent on resulting in less demand on the system and lower
these programs are substantially smaller than those cost to government payors. Early studies have sug-
on health care; hence, the fraud prevention benefits gested that costs for patients in disease manage-
are much lower as well. ment programs are up to 25 percent less than those
for control groups. However, analysis of initial
outcomes for the 2005 Medicare Health Support
Disease Management
program suggests that cost savings are not yet of
In disease management, Medicare or Medicaid sufficient magnitude to outweigh the management
(or, in some cases, private health insurance) patients fees charged by the private sector organizations
with chronic diseases such as diabetes or heart dis- conducting the pilots.19 Patients and caregivers
ease are enrolled in programs to help manage the have generally expressed high levels of satisfaction
disease and lower the costs of treatment.17 A large
www.businessofgovernment.org 13
STrategic Use of Analytics in Government
with the programs, but greater efficiencies will management” in his critical success factors for the
have to be found if these programs are to prolifer- VA health system, which are critical precursors of an
ate more broadly. analytical orientation. The VA’s targets were primar-
ily around improving health care quality and perfor-
mance (although, as noted earlier, the VA also does
The DELTA Model in Government
analytical work to prevent fraud on the benefits side
Health Care of the organization). Finally, the VA seems to have
Since a major role for government in U.S. health had sufficient analysts to accomplish its analytical
care is as a payor, only some aspects of the strategic objectives (although it also employs outside contrac-
use of analytics would be relevant to government tors). While the VA hospital system could take an
executives. From a payor’s standpoint, the data that even stronger and more strategic analytical focus,
would be of primary interest would most likely the organization seems to have done quite well with
involve whether reimbursements are being conducted the resources available.
effectively and legally. This is, of course, a focus at
both the federal and state levels. But the fact that
government payments are overseen at multiple
levels creates an analytical problem from an enter-
prise perspective. Medicare payments are central-
ized, but Medicaid payments are overseen by states
and even counties. The decentralized nature of the
“enterprise” in this situation often prevents an exten-
sive effort to analyze spending effectiveness and to
reduce fraud. Leadership also suffers from this
fragmented payment approach; there is no visible
national leader to encourage an analytical perspec-
tive on medical payments. The target of analytical
initiatives in health care payment is primarily dis-
ease management for Medicare and fraud reduction
for Medicaid. Finally, there is clearly a shortage of
analysts within government to perform analytical
work on payments. Many of the analysts in this
domain come from either outside disease manage-
ment or consulting firms.
One of the most important domains for analytics in military approaches were successful, they have
the private sector is in supply chain management, also been accused of a mindlessly quantitative
where companies attempt to optimize resources and approach to decision making (as with “body
distribution channels. More recently, organizations counts” in Vietnam, for example, during McNamara’s
have begun to focus on the “human supply chain,” tenure as secretary of defense in the 1960s, when
or the use of analytics in human resource processes. he recruited a second generation of “whiz kid”
These two areas are also important for governments, analysts to staff his policy offices).
and their most aggressive application has been in
the military. The military remains a leading user of supply chain
analytics today, although private sector firms such as
The Origins of Supply Chain Analytics Wal-Mart typically employ more sophisticated ana-
lytical approaches than the military. In part this is
Supply chain analytics arose out of the analytical due to the greater complexity of the military supply
discipline of “operations research.” These applica- chain, particularly in wartime. In wartime opera-
tions were used early on within branches of the tions, the tempo, volatility, and stakes are dramati-
government—specifically in the military. Because cally higher, and the variables more difficult to
of competitive needs and the drastic consequences model. In wartime, the competition is actively trying
of running out of materiel in wartime, analytics has to (1) destroy your resources and logistics pipeline,
long been employed in forecasting, supply route and (2) to create demand for what you’re already
optimization, and naval and air force operations. short of. There is also no fixed end point to the sup-
Optimization models were employed in World ply chain; it moves as battles unfold and as military
War II by both UK and U.S. researchers—for exam- forces move. Therefore, military analysts have found
ple, they were used to design optimal shipping con- it difficult, if not impossible, to fully model the mili-
voys and bomber flying patterns.21 In England, the tary supply chain during wartime.
Royal Air Force used both radar and operations
analysis to defeat a numerically superior German
foe in the Battle of Britain.22 Sir Arthur Harris said Current Approaches to Supply Chain
his Bomber Command’s Operational Research Analytics
Section “had saved the lives of thousands of aircrew Today, however, some analytical approaches are
and hundreds of aircraft, as well as being an indis-
being employed within the military to manage inven-
pensable aid to military effectiveness.”23 tories and supply lines. The U.S. Army, in particular,
has changed its supply chain model over the past
By the Vietnam War, operations research was widely decade or so from one based on “mass”—moving
accepted and used broadly within military opera- large quantities of heavy goods with a “just-in-
tions, in part because Robert McNamara, a strong case” approach to inventory—to one based on
believer in analytical tools, was secretary of defense. “velocity,” or a more agile, fast-moving supply
At times, the military’s use of analytical techniques chain that operates on a just-in-time inventory
led that of the private sector.24 While many of their basis.25 Rather than detailed analytical modeling,
www.businessofgovernment.org 15
STrategic Use of Analytics in Government
the focus over the past several years has been on commanders to provide insights on how best to
determining and measuring process flows, and achieve their objectives. According to one inventory
converting Army officials to the new supply mind- of field-based U.S. Army operations research ana-
set. Part of the change involved shifting to new lysts in 2005, there were:
metrics, such as the “customer wait time” measure
employed in the private sector. • Six with the Multi-National Forces–Iraq
• Three with the Multi-National Corps–Iraq
Within this velocity-based approach, however, it
is critical—more so than in the private sector—to • Two with Combined Forces Command–
avoid shortages of critical supply items. To improve Afghanistan
the likelihood that needed parts are available, the
• One with Combined Joint Task Force-76 in
Army employs an analytical approach called “dollar
Afghanistan
cost banding.”26 This algorithm for inventory man-
agement adjusts “the criteria for determining whether • One with the Headquarters, 3rd Infantry
an item should be added or retained according to Division (UEx)
the item’s criticality, mobility impact, end item den-
sity, and dollar value.” Items that don’t cost much According to Lieutenant General David Melcher, the
to store as inventory, but are mission critical, are deputy chief of staff, their presence was at the request
more likely to be kept in stock than larger, heavier, of field commanders, and the organizational designs
and more expensive items. The dollar cost banding for Army transformation call for specified numbers of
algorithm has proven to be more effective than the analysts at particular points in the field hierarchy.27
previous “days of supply” algorithm used by the
Army to prevent stock-outs while still lowering the Human Resource Analytics
cost of carrying inventory. Most importantly, repair
times for key military equipment (such as tanks) The military has also increasingly employed ana-
have been reduced by as much as 29 percent. The lytical approaches to the human resources “supply
decision rules for the new algorithm also can be chain.” Particularly in wartime with an all-volunteer
automated, which reduces the burden on Army military, the U.S. armed forces are turning to
supply managers. analytical decisions related to recruitment. The
analytical domains include forecasting, recruiting
Logistical analyses may happen both at some remove segmentation and pipeline models, attrition models,
from battlefield activities and in close proximity to and force reduction strategies. Forecasting analytics
them. They may be done in isolation or as part of an generally takes place at the level of large military
overall set of battlefield course-of-action alternatives populations (for example, specific services such as
presented to battlefield commanders. Battlefield the Army or the civilian force within the Army) and
operations analysts may also collaborate closely with includes modeling and testing of alternative policies
and resource (for example, enlisted vs. civilian vs.
Capacity planning: Finding the capacity of a supply chain or its elements; identifying and eliminating bottlenecks;
typically employs iterative analysis of alternative plans.
Demand-supply matching: Determining the intersections of demand and supply curves to optimize inventory
and minimize overstocks and stockouts. Typically involves such issues as arrival processes, waiting times, and
throughput losses.
Modeling: Creating models to stimulate, explore contingencies, and optimize supply chains. Many of these
approaches employ some form of linear programming software and solvers, which allow programs to seek
particular goals, given a set of variables and constraints.
contractor) mixes, and the impact of realignments, programs, and a “Manpower and Training” initiative
incentives, retirements, and so forth. that “applies sophisticated economics and social sci-
ence methodologies to Army personnel and training
Of course, military staffing can become a political issues. It stresses quantitative analysis and testing of
issue that may overshadow analytical planning for alternative policies and resource mixes, all oriented
optimized resource levels. For example, RAND toward choosing appropriate strategies for manning,
analysts studied the ratio of occupying forces to training, and structuring the Army for the future.”32
population levels in several “nation-building”
exercises after World War II, and concluded that Consulting firms also play a role in analytical
in order to keep the peace in Iraq, over 500,000 research and practice development. Booz Allen
troops would be required.28 The U.S. administration Hamilton, for example, has worked with the U.S.
found that number politically unpalatable, however, and other governments on analytical projects since
and deployed less than a third of that number. the late 1940s. Today large consulting firms around
Analysts may have done their jobs (and hindsight the world work on complex projects assisting gov-
has largely suggested that they were correct), but ernments to manage operations and supply chains.
decision makers relied on other criteria.
Researchers at several military-oriented universities
Recruiting for the armed forces is increasingly viewed and colleges assist in analytical supply chain
as similar to a marketing and sales exercise, and projects. The Industrial College of the Armed Forces,
many of the same techniques employed in commer- Defense Acquisition University, the Naval Postgrad-
cial firms to attract customers are being employed uate School, and some faculty at the U.S. service
to attract recruits. Segmentation models employ academies have done substantial work on behalf
geodemographic data to determine which types of of the military in the realm of supply chain and
potential recruits should receive which promotional human resources.
materials.29 Pipeline analyses model the number of
Inside the military, the primary sources of supply
recruits at every stage in the recruiting process (simi-
chain analytical expertise are within individual
lar to a sales lead pipeline) and alert recruiters when
services, and in cross-service organizations such
the pipeline falls below desired levels. The U.S.
as the Defense Logistics Agency (DLA). Demand
Army Recruiting Command, for example, uses geo-
planners at the DLA, for example, work with mili-
spatial data in its Graphical Accessions Mapping
tary customers to develop demand plans based on
Analysis Tool (GAMAT), which tracks the progress
statistical forecasts. Individual services also have
of recruits from first contact through boot camp and
logistical units, such as the Army Materiel Command
presents the information to recruiters in geographic
organization, that create demand plans for service-
form.30 Recruiters also have access to online market
specific supply items. Unfortunately, the prevalence
penetration analyses and their own performance
of different organizations doing supply chain work
reports.31 Attrition models are used to identify recruits
for the military often causes problems. For example,
that have had some contact with the Army, but who
there is no unified approach to how to deal with
have not yet enlisted and who are in danger of fall-
supply containers in war zones.
ing out of the process.
Supply chain and human resource analytics for the The DELTA Model for Supply Chain
U.S. military are most actively pursued outside of
and Human Resource Analytics
government, in research centers, consulting firms,
and universities. The RAND Corporation has been The DELTA model for supply chain and the “human
particularly active in both supply chain and human supply chain” has some similarities to those of the
resource activities on behalf of the U.S. Army. other analytical domains in government. The
RAND’s Arroyo Center is the Army’s only federally amount of data for this area is increasingly not a
funded research and development center. In addition problem, as the various branches of the armed
to traditional military research programs on strategy services and the Defense Logistics Agency install
and technology, Arroyo has a “Military Logistics” enterprise resource planning (ERP) systems that
program that carries out analysis on supply chain track movements of goods and people. The pri-
mary problem from the data perspective is that
www.businessofgovernment.org 17
STrategic Use of Analytics in Government
www.businessofgovernment.org 19
STrategic Use of Analytics in Government
the states stay within their budgets (for example, right people, for the right action at the right time.36
Florida and California). Although revenue analysis The applications address one or more of the follow-
applications are widespread in the U.S. as well as ing areas:
other countries at the sectoral (macro) level, they
are still evolving at the micro level. Desktop hard- • Discovery of nonfiling businesses and individu-
ware and graphical interfaces have eased the model als with potential tax liabilities
permutations and enabled real time, iterative runs
• Selection of those taxpayers for audit who are
during legislative deliberations.
likely to underreport
The Congressional Budget Office of the U.S. federal • Application of risk-based collection strategies
government has developed an individual income tax for each collection case
model to project individual and aggregate tax liabil-
ity for future tax years. Such projections are neces- Whereas in the U.S. and Canada, many of the
sary to ensure that sufficient tax revenues will be applications focus on identifying nonfilers and/or
forthcoming to fund new governmental programs. optimizing tax collection strategies, the emphasis
Several states, including Iowa, have developed fore- in European tax authorities has been the application
casting models that predict individual income. Such of analytics to the auditing of tax returns.
forecasting helps to understand the shifts in taxable
revenue as the baby-boom generation retires; Iowa’s
Discovery of Nonfilers
tax code includes several preferential provisions for
By the 1980s, the U.S. Internal Revenue Service
the elderly.34
(IRS) had developed scoring systems to identify
those most likely to be nonfilers or those underre-
In 1994, the state of California passed a law that
porting their taxes. Called the Taxpayer Compliance
required the California Department of Finance to
Measurement Program, the scoring system was vali-
develop a dynamic revenue analysis model for tax
dated through detailed audits of selected taxpayers.
bills with significant fiscal effects. The law was
The use of scoring has been demonstrated to result
enacted to deal with the highly volatile revenue
in higher tax assessments.37 More recently, the IRS
system. The state was experiencing rapid increases
has improved its capabilities to detect tax cheaters
in personal income tax as a percentage of the total
among corporations with offshore operations.
tax revenue, with increased concentration of income
at the high end. High-income taxpayers have volatile
Many states have built data warehouses and engaged
income sources (stock options and capital gains).
in matching records to identify nonfilers of taxes.
California’s dynamic revenue model was to take into
These include states like Virginia and Massachusetts.
account the probable behavioral responses of tax-
The state of Massachusetts uses an off-the-shelf tax
paying individuals and businesses. The size of the
recovery program to perform record matching with
model had grown to some 1,100 equations at the
nearly 50 databases such as Internal Revenue
time the law was sunset in 2000 because California’s
Service, U.S. Customs, state licensing boards, business
policy makers did not extend the law. Although the
filing of unemployment insurance, customer records,
quantitative benefits of the dynamic revenue estima-
and so on. Such matching programs can identity a
tion faced many challenges (particularly because of
company reporting a certain number of employees
data and specification problems), the qualitative ben-
on its payroll to state tax agencies, while providing
efits were much less under debate. Dynamic revenue
another figure when reporting unemployment com-
forecasting provided new and useful insights and
pensation benefits. The system identifies taxpayers
opened new lines of discussion regarding the ramifi-
who owe taxes but have not filed returns, as well as
cations of tax policy changes.35
people who are underpaying or overpaying. Discovery
of nonfilers cost the state $6 million while bringing
Compliance Systems in $325 million in additional revenues.
Data analytics can help reduce tax gap (lost reve-
nue) by improving tax collection with the same or Massachusetts is also using Clearinghouse, a col-
even reduced resource levels. Compliance applica- laboration among eight state tax commissioners,
tions focus on getting the right cases assigned to the to search tax cheaters who are exploiting state
boundaries. Massachusetts has spent about help to increase this knowledge by developing the
$500,000 to launch the operation. The goal is to profiles of those taxpayers which were found to be
make this system operate in real time (verifying the highly compliant during prior audits. These models
accuracy of the claim when the taxpayer claims a help to identify those taxpayers that are most likely
credit from the state of Massachusetts). to make errors and predict how large of an error
they have made. The predictive models are used to
One of the first states to use predictive modeling identify compliance problems at the time a tax
analytics to identify nonfilers was Texas. The state return is filed. The solutions are capable of data
enacted enabling legislation in 1997 that allowed mining thousands of tax returns in seconds, and can
it to build a data warehouse and a year later to have potentially help save years of tracking, investigation,
a system in production for the tax discovery (nonfil- and collection costs. Audit selection is most effec-
ers) group. The enabling legislation was needed to tive when there can be an early detection and rapid
accommodate benefit-based funding of the system. follow-up. In most cases, the audit selection solu-
The system matches data from U.S. Customs and tions provide only decision or informational support
Border Protection, employment records, state motor to the auditors; it is up to the human auditors to act
vehicle and private airplane registrations, and so on, upon the information.
with franchise tax (business tax) filers and sales tax
records. It has found many nonfiling businesses for Analytics-based audit systems have been imple-
franchise and sales tax liabilities. For example, one mented in South Carolina, Iowa, Texas, Virginia,
business had bought three airplanes in one year but Kansas, Massachusetts, and Connecticut. The
was reporting no sales revenue. As the system was California Board of Equalization (which manages
benefit based, Texas spent on the vendor some $4.7 sales tax) is also making progress toward the full
million that helped return $48 million. The state has deployment of scoring models for audit selection.
also considered tapping into commercial sources Texas located $400 million in unpaid state revenues
such as Dun and Bradstreet databases to find non- between 1998 and 2004, and expects to collect
registrant businesses and nonfiling taxpayers, but about $70 million a year over time. Virginia reported
has yet to do so. to have located $73 million. South Carolina
expected to collect $100 million in the first five
The state of California has employed a variety of years from its compliance solutions.
data sources—including IRS data, wage and sales
tax information, property sales, and mortgage inter-
Collection Strategy
est—to identify nonfilers. The California Franchise
The earliest applications of analytics in tax agencies
Tax Board identified 800,000 nonfilers, recovering
were in collections. Currently, about 20 percent of
$350 million in taxes. Costs were recouped within
the states in the U.S. have implemented risk-based
one year. The agency estimates $36 million annually
collections including California, South Carolina,
in new tax revenue.
Missouri, Michigan, Massachusetts, Virginia,
Arizona, and Iowa. Texas has also recently imple-
Matching applications of different data sources often
mented its collection models, although the models
produces too many leads. Given the limited staff
are under further development. The collection appli-
available to follow up, predictive modeling is used
cations enable agencies to maximize overall collec-
to prioritize the leads and predict the tax dollars
tions while minimizing the amount of resources to
owed by the organization or individual taxpayer.
handle the collection of unpaid tax. Risk-based col-
Data mining is also used to analyze if existing tax-
lections work on the assumption that taxpayers have
payers have liabilities for other tax types than what
different levels of risk of nonpayment. The applica-
they have filed.
tions assign taxpayers to different classes of collec-
tion strategies/treatments based on their likelihood
Audit Selection of paying. Different strategies treat taxpayers differ-
Predictive models are also used to select tax returns ently from the beginning to the end of the process.
for audits. The tax administrators have much knowl- Those considered low risk are sent a “softer” letter
edge of what types of errors taxpayers make and and allowed more time to respond than those con-
why based on historical data. Predictive models sidered high risk. The agencies report that many
www.businessofgovernment.org 21
STrategic Use of Analytics in Government
accounts close before a phone call is needed. High- season had to proceed without any electronic
risk taxpayers are moved to enforced collections as detection system in place. The IRS estimated that
quickly as possible because timely field enforcement it paid over $300 million in bogus tax refunds. The
increases the probability of collection. For example, treasury inspector general for tax administration
in Iowa, the increased revenues from an improved released a report on an investigation into the trou-
collection strategy were estimated to be $4 to $5 bled system in spring of 2007.40 Work is under way
million in less than one year. Annual projections are to restore operations of the client/server-based EFDS
in the $4.5 to $9 million range with future opportu- system for the tax year of 2007.
nities identified worth $12 to $29 million per year.38
In addition to refund fraud, preparer fraud has
In some states (for example, South Carolina and grown more significant. For example, preparers
Massachusetts), the models not only segment tax- convince uninformed people that they qualify for
payers, but also in some cases determine what a “refund program” and then file on their behalf.
actions to take and drive interactions with the cli- New York State has developed a predictive modeling
ents. These compliance solutions go beyond predic- application that evaluates returns in semi-real time
tion to optimization. Optimization extends analytics to look for preparation patterns (for example, high
into the area of treatment strategy design, taking into deductions), audit questionable returns, and then
account the taxpayer’s history and circumstances.39 potentially prosecute for preparer fraud. New York
Optimization yields higher returns but also requires State has also developed a very successful analyti-
major organizational change in the agencies. cal system to identify fraud and abuse of the Earned
Income Tax Credit (a welfare benefit for low-
Canada Revenue Agency has also embarked on an income individuals).
ambitious plan to build sophisticated models for
auditing and collections, along with the United Businesses, particularly self-employed businesses,
Kingdom and Australia. Of the continental European sometimes falsify income to qualify for tax credit
countries, in Belgium and Sweden the tax authori- programs. Iowa has experienced a decline in
ties have major programs under way to exploit data corporate income tax collections for about two
analytics in audit selection of tax returns. decades. The drain in tax collections points to
corporate income tax credits. In response to state
legislation, the Iowa Department of Revenue devel-
Fraud Detection
oped a Tax Credits Tracking and Analysis System to
As tax refund fraud schemes have become more monitor and consolidate tax credits by taxpayers.
sophisticated, tax agencies have had to deploy more The agency is also expanding the system to include
powerful methods to stop fraudulent return checks analysis of all credits awarded by agencies to certain
before they go out, while maintaining expected self-employed business classes (for example, con-
refund times and avoiding inconvenience for com- tractors and consultants).
pliant taxpayers.
At the U.S. Internal Revenue Service, the Electronic Taxpayer Customer Services
Fraud Detection System (EFDS) evaluates every The emerging area for analytics in revenue manage-
tax return requesting a refund to rate potentially ment is taxpayer customer services. The applica-
fraudulent individual tax returns. About 80 percent tions are similar in some ways to customer services
of returns requested a refund in 2005. The system analytics systems that have long been in commercial
includes only personal information (no private use in the private sector. These systems are typically
sector data and no other agency data). The proto- based in customer relationship management (CRM)
type was first introduced in 1995 and deployed applications, and identify customers who are likely
nationwide for the tax year of 1996. candidates for some sort of intervention. Taxpayer
services provide proactive interventions to reduce
At the IRS, a consulting firm was hired to deliver an nonfiling and underreporting. The applications
updated web-based version of the existing client/ exploit models that evaluate inbound telephone
server-based system by January 2006. However, the calls and assess the likelihood that the taxpayer has
implementation was unsuccessful and the 2006 tax not registered for all necessary tax types, does not
understand the tax law change, will become a non- Revenue management also lacks an enterprise focus,
filer, or will likely have certain types of questions. although there is an increasing awareness of the
Such systems also often involve an outreach educa- need to integrate analytical activities with broader
tional component. The Massachusetts Department of work flow processes at the departmental level and
Revenue is an agency that has worked on taxpayer even at the agency level. But tax and revenue man-
analytics services. Future versions of the system may agement is spread through a variety of different tax
leverage data sources such as auditors’ notes. agencies with widely different missions. We found
little collaboration among the different state agen-
Some of the taxpayer services will be built on the cies charged with revenue responsibilities regarding
case management systems that are already in use their analytics. Even within the same tax and reve-
in many states. These systems are designed to man- nue agencies, departmental boundaries hindered
age all interactions with customers across various collaboration on analytics. There was, however,
channels, including mail, call centers, and websites. considerable collaboration across state agencies
For example, the California Franchise Tax Board uses (visits, presentations, conferences) with similar
a case management system with an integrated web responsibilities. Much of this collaboration occurred
and interactive voice response front end for through professional and federal groups as well as
improved customer service. The system can be skilled and experienced consultants. Many of these
used by nonfilers to request additional time to reply consultants had built their expertise in the private
to their notices and grant extensions online via a sector and are now leveraging it in the public sector.
web-based interface. However, consultancies tend to “contribute more to
the innovative process, than to the new sustainable
capabilities” in daily work practices.41 Those agen-
The DELTA Model for Revenue
cies that have had sustained success with analytics
Management (such as Massachusetts and Texas) have engaged
About 20 percent of the states in the U.S. deploy themselves in long-term relationships with the same
analytics as part of their compliance systems. As consultancies. Yet, as one of these consultants with
one consultant remarked, “Analytics represent a a long-term agency relationship noted, “Consultants
niche market, with few states taking it seriously.” cannot lead these projects….”
Many of these are in relatively early stages of imple-
mentation. Although the access to data is increas-
ingly less of a problem (some note that the problem
is too much data), data quality can be an issue,
although usually a manageable one. Many states
have implemented data warehousing strategies.
However, some states have built up analytics infra-
structure but because of the lack of analysts and
the necessary leadership are unable to exploit their
infrastructure. In some states, leadership remains
skeptical whether data analytics yields better out-
comes than traditional methods.
www.businessofgovernment.org 23
STrategic Use of Analytics in Government
In addition to health care, supply chain manage- to take electro-optical, radar, or infrared images.
ment, and revenue management, there are, of The National Geospatial-Intelligence Agency
course, other analytical domains within government. coordinates imagery collection and processing.
One of the most important is intelligence. Some
• Measurement and Signature Intelligence, or
areas within intelligence are highly analytical—for
MASINT, is the collection of technically derived
example, the perusal of global telecommunications
data that describes distinctive characteristics of a
traffic. Traditional “spying” or intelligence agent
specific event such as a nuclear explosion. The
activities, however, are much more difficult environ-
Defense Intelligence Agency and the military
ments in which to gather data, quantify observa-
services are the primary MASINT collectors.42
tions, and perform quantitative analyses.
There is little doubt that there is considerable analyt-
Types of Intelligence ical processing applied to signal intelligence, or
If one breaks down intelligence into four different SIGINT. The National Security Agency is one of the
types of information gathered, as did a U.S. congres- world’s largest consumers of high-powered comput-
sional commission in 1996, the latter three types are ers, and numerous vendors assist intelligence agen-
heavily analytical and quantitative: cies with specialized software and hardware. For
example, Nice Systems, an Israel-based vendor of
• Human source intelligence, or HUMINT, is the analytical software for the analysis of communica-
operational use of individuals who know or tions content, notes that it works with governments
have access to sensitive information that the as well as corporations. The company offers intelli-
Intelligence Community deems important to its gence agencies “interception, mediation, collection
mission. The Central Intelligence Agency and and analysis of telecommunication interactions for
the Defense HUMINT Service, an element of both telephony and Internet data.”43 However, neither
the Defense Intelligence Agency, are the primary Nice nor other sources reveal the details of how
collectors of HUMINT. analytical approaches are applied in this context.
• Signals intelligence, or SIGINT, consists of infor- There is also considerable interest and activity in the
mation obtained from intercepted communica- area of analytics for image intelligence, or IMINT.
tions, radars, or data transmissions. The National A walk down Main Street in virtually any city will
Security Agency is the primary collector. Within confirm that video image capture is a growth indus-
the SIGINT discipline, there are subcategories of try for both the private and public sectors. Because
communications intelligence, electronic intelli- there are not enough humans to economically
gence (essentially emanations from radars), and analyze all the video footage for terrorism or crime
foreign instrumentation signals intelligence threats, companies are embracing analytical
(such as automated data from space vehicles). approaches to identifying threats. This capability is
• Imagery intelligence, or IMINT, is the use of called “video analytics” or “intelligent video surveil-
space-based, aerial, and ground-based systems lance.” The Central Intelligence Agency’s (CIA)
venture capital arm, In-Q-Tel, has invested in one
company that provides this type of solution.44 It is implies that the methods and techniques of
obvious why that organization would be interested analysis are informal, idiosyncratic, unverifi-
in the technology, but of course no details of its able, and perhaps even unexplainable …
interest are publicly available. there is no formal system for measuring and
tracking the validity or reliability of analytic
Quantitative analysis can also be applied to other methods, because they are both perceived
intelligence contexts, such as identifying state and and employed within the context of idiosyn-
regional instability. A Military Operations Research cratic tradecraft.46
Society Working Group in 2004 reviewed instability
forecasting approaches and identified two different Quotations from Johnston’s 489 interviews illustrate
approaches:45 the issue:
• Strategic forecasting models: forecast several • “What we do is more art and experience than
years ahead the likelihood that states will fail or anything else.”
will become unstable based on quantitative
• “Science is too formal. We can’t actually run
analysis of social, political, demographic, and
experiments here.”
economic factors. Examples include the CIA’s
Political Instability Task Force and the Center for • “How would you actually test a hypothesis in
Army Analysis’ ACTOR (Analyzing Complex intelligence?”
Threats for Operations and Readiness) model.
• “Science is what you do in a lab.”
• Operational forecasting models: monitor,
• “We’re not scientists; we’re analysts. We don’t
assess, and forecast trends in behavioral interac-
generate the data.”
tions between people, organizations, and states,
and predict changes at the event level. Examples
The absence of a more rigorous scientific approach
include Psynapse Technologies’ ABC Terrorism
to methods means that the intelligence domain will
Prediction Model, which mines textual reports
find it difficult to ever measure or improve itself in a
to forecast terrorist attacks, and the Center for
disciplined way. Johnston’s findings are a reminder
Army Analysis’ FORECITE Monitor, which col-
that simply talking about “analysis” does not neces-
lects data for indices of the “character and
sarily mean that a rigorous, scientific analysis of
intensity of interactions between individuals,
data is being undertaken. His anthropological
organizations, and states.”
research approach is also a reminder that the culture
of an organization (or, in this case, a larger commu-
Need for More Analytics in Human nity) can be an important determinant of the analyti-
Intelligence cal approach it employs.
www.businessofgovernment.org 25
STrategic Use of Analytics in Government
Conclusion: Analytics as
an Effective Tool for Government
The transformation to analytical competition in organization. The agency has an aggressive goal of
private sector firms is a long-term, broadly focused reducing the lost revenue from the “tax gap” by half
organizational transformation. In order to truly com- by 2012. The agency has invested in analytics infra-
pete on their analytical capabilities, organizations structure as part of its knowledge-based strategy.
must transform not only their technology and data, This strategy calls for increasing the agency’s
but also their cultures, their business processes, knowledge of taxpayer behavior; the knowledge
and the day-to-day behaviors of their employees. is leveraged to take actions that improve citizens’
Historically, information technology applications confidence in the tax authority; improved confi-
that challenge the prevailing institutional logic are dence increases tax compliance. The analytics
short-lived and ultimately unsuccessful. Sustained directly supports the knowledge-based strategy and
long-term change in the public sector will require transformation of the underlying institutional logic.
the same types of organizational transformation and
managerial innovation as seen in the private sector. The agency set a broad goal of improving the confi-
dence of taxpayers in the tax authority. Studies helped
Few agencies we studied were examples of such it to determine that confidence in the tax authority is
transformations. In these agencies, the interviewees the most effective and efficient way to improve com-
described mutual adaptations to policy or strategy pliance. The analytics activities brought knowledge
from leveraging analytics. For example, in one about taxpayer behaviors and their common errors.
state tax agency, analytics has begun to have an This knowledge was used to change the tax forms so
impact on the strategy design of the agency. Analytics that taxes are done right from the start. Because the
is used to improve customer service in addition to analytical activities are linked to a larger agency “tar-
compliance. This is also accompanied by structural get” of taxpayer confidence building, the analytics
changes to leverage the knowledge more efficiently activities are visible and of interest to the highest lev-
between the areas of auditing, collections, and els of the agency leadership.
customer services. Other agencies noted changes
in tax legislation because of the knowledge gained The knowledge from analytics is not only used to
from analytics. change the agency managerial processes and inter-
actions with taxpayers, but also to shape the broader
tax system. One of the agency’s studies revealed that
Sweden: Strategic Analytics in
about 35 percent of the sales of international share
Government funds were declared incorrectly by private persons
We did find an example of a highly strategic appli- in their tax returns. The errors were largely due to
cation of analytics in government. In Sweden, the oversight or lack of information about how to
leadership of the national tax agency has risen to declare the sales. The agency engaged in a commu-
the occasion of transforming its managerial pro- nications campaign to educate the taxpayers. The
cesses. The transformation has come on the coattails errors continued and even increased even after the
of a new broadened agency mission and centraliza- agency’s communications campaign. A follow-up
tion of what was formerly a highly decentralized study was undertaken to understand more deeply
www.businessofgovernment.org 27
STrategic Use of Analytics in Government
www.businessofgovernment.org 29
STrategic Use of Analytics in Government
Endnotes
Her Majesty’s Stationery Office, 1963, pp. xvii–xviii. Program Analysis Section, Iowa Department of Revenue,
2. M. W. Kirby, “Operations Research in World War January 2007.
II: RAF Bomber Command and the Defeat of Germany,” 5. J. D. Vasche, “Whatever Happened to Dynamic
Lancaster University, December 3, 1999, p. 27. Revenue Analysis in California? Annual Revenue
24. McNamara and his World War II “whiz kid” Estimation & Tax Research Conference, Portland, Oregon,
colleagues (10 young veterans of the Army Air Force September 17–20, 2006.
Statistical Command) were brought into Ford Motor 6. T. Heinz, “Tax Compliance: Yesterday and Today,
Company and other companies to bring some of the FTA Technology Conference, August 2006.
scientific, analytical approaches to management that 7. J. Alm, “Tax Compliance and Administration,”
they had employed in the military. For further discussion in Handbook on Taxation, eds. W. Hildreth and J. A.
of McNamara’s Ford Motor Whiz Kids, see J. A. Byrne, The Richardson, pp. 741–768, Marcel Dekker, Inc., 1999.
Whiz Kids: The Founding Fathers of American Business— 8. T. London, “Combining Analytics and Business
and the Legacy They Left Us, Currency, 1993. Process Reengineering to Improve Tax Administration,”
25. J. Dumond, et al., Velocity Management: The CGI, August 8, 2007.
Business Paradigm that Has Transformed U.S. Army 9. J. Taylor, “Analytics for Tax Collection: From
Logistics, Santa Monica, CA: RAND Corporation, MR- Predictive Models to Optimization,” Fair Isaac
1108-A, 2001. Corporation, Tax Volume 1, June 7, 2005.
26. K. J. Girardini, et al., “Dollar Cost Banding: A 40. The Treasury Inspector General for Tax
New Algorithm for Computing Inventory Levels for Army Administration, “Oversight of the Electronic Fraud
Supply Support Activities,” RAND Corporation, 2004; Detection System Restoration Activities has Improved, but
online at http://www.rand.org/pubs/monographs/2004/ Risks Remain, March 29, 2007, Reference Number: 2007-
RAND_MG128.pdf. 20-052, http://www.tigta.gov.
27. Military Operations Research Society, Phalanx: 41. B. Swanson, Workshop on How Emerging
The Bulletin of Military Operations Research 38:3, 2005, Technologies Are Shaping and Are Shaped by Strategy,
online at: sponsored by the JSIS, Montreal, December 9, 2007.
http://www.mors.org/publications/phalanx/v38n3.pdf. 42. From “An Overview of the Intelligence
28. J. Dobbins, et al., “America’s Role in Nation- Community,” in “Preparing for the 21st Century: An
Building: From Germany to Iraq,” RAND Corporation, Appraisal of U.S. Intelligence,” Commission on the
2003, online at: Roles and Capabilities of the United States Intelligence
http://www.rand.org/pubs/monograph_reports/MR1753/. Community, March 1, 1996; online at http://www.fas.org/
29. Major J. Baird and L. Clingan, U.S. Army, irp/offdocs/int023.html.
“Innovations in U.S. Army Recruiting: Using Custom 4. Nice Systems website at www.nice.com.
Segmentation to Reach America’s Youth,” Claritas 44. N. Schachtman, “The New Security: Cameras that
Precision Marketing Conference, April 23, 2007. Never Forget Your Face,” The New York Times, January 25,
0. “Map Intelligence Gives U.S. Army Recruiting 2006, online at http://www.nytimes.com/2006/01/25/tech-
Clear View of Talent Pool,” May 18, 2007, online at: nology/techspecial2/25video.html?_r=1&oref=slogin.
http://intellibriefs.blogspot.com/2007/05/map-intelligence- 45. Military Operations Research Society Workshop:
gives-us-army.html. The Global War on Terrorism: Analytic Support, Tools and
1. Oracle Corporation, “Information Streamlines: Metrics of Assessment, 30 November–2 December 2004,
Optimizing U.S. Federal Government Agency Naval War College, Newport, Rhode Island, described in
Performance,” 2007. Online at: Phalanx 38:1, 2005, http://www.mors.org/publications/
http://www.oracle.com/industries/government/ phalanx/v38n1.pdf.
oracle-solutions-data-integrity-br.pdf. 46. R. Johnston, Analytic Culture in the U.S.
2. RAND Arroyo Center website at Intelligence Community: An Ethnographic Study, CIA
http://www.rand.org/ard/about.html. Center for the Study of Intelligence, 2005.
. J. R. Layman, “Individual Income Tax
Nonwithholding in Virginia: Data Analysis and Revenue
Forecasting,” September 18, 2007, 2007 FTA Revenue
Estimating Conference. September 18, 2007.
4. A. R. Harris, “Forecasting Taxable Income for the
Individual Income Tax Micro Model, Tax Research and
www.businessofgovernment.org 31
STrategic Use of Analytics in Government
Ab o u t t h e a u t h o r s
Davenport’s most recent book (with Jeanne Harris), Competing on Analytics: The New Science of Winning,
became a best-seller. His previous book, Thinking for a Living: Getting Better Results from Knowledge
Workers, was named one of the best business books of 2005 by the Financial Times. Prior to this, he wrote,
co-authored, or edited 10 other books, including the first books on business process reengineering and
achieving value from enterprise systems. He has written over 100 articles for such publications as the
Harvard Business Review, Sloan Management Review, California Management Review, and the Financial
Times. He has also been a columnist for CIO, InformationWeek, and Darwin magazines. In 2003,
Davenport was named one of the world’s “Top 25 Consultants” by Consulting magazine. In 2005, he was
named one of the top three “Business/Technology Analysts” in the world by Optimize magazine. In 2007,
he was named one of the 100 most influential people in the information technology industry by Ziff-Davis
magazines and was the highest-ranked business academic.
In 1997, she was a distinguished visiting scholar at the University of Melbourne. Her other visiting scholar
appointments include the Harvard Business School; MIT’s Sloan School of Management; Queens University,
Canada; Jyvaskyla University, Finland; Hong Kong Science and Technology, China; City University of Hong
Kong, China; and Helsinki University of Technology, Finland.
Jarvenpaa received her B.S. from Bowling Green State University and her M.B.A. and Ph.D. from the
University of Minnesota.
www.businessofgovernment.org 33
STrategic Use of Analytics in Government
K e y c o n t a c t I n f o r m a t i o n
Sirkka L. Jarvenpaa
Center for Business, Technology, and Law
McCombs School of Business
University of Texas at Austin
Austin, TX 78703
(512) 471-1751
e-mail: sirkka.jarvenpaa@mccombs.utexas.edu
website: www.btl.mccombs.utexas.edu
To download or order a copy of a report, visit the IBM Center for The Business of Government website at: www.businessofgovernment.org 35
CENTER R E P O RT S AVA I L A B L E
36 To download or order a copy of a report, visit the IBM Center for The Business of Government website at: www.businessofgovernment.org
Growing Leaders for Public Service San Diego County’s Innovation The Challenge of Developing Cross-
(2004, 2nd ed.) Program: Using Competition and a Agency Measures: A Case Study of
Ray Blunt Whole Lot More to Improve Public the Office of National Drug Control
Services (2000) Policy (2001)
Pay for Performance: A Guide for William B. Eimicke Patrick J. Murphy and John Carnevale
Federal Managers (2004)
Howard Risher Innovation in the Administration of The Potential of the Government
Public Airports (2000) Performance and Results Act
The Blended Workforce: Maximizing Scott E. Tarry as a Tool to Manage Third-Party
Agility Through Nonstandard Work Government (2001)
Arrangements (2005) Entrepreneurial Government: David G. Frederickson
James R. Thompson and Sharon H. Bureaucrats as Businesspeople (2000)
Mastracci Anne Laurent Using Performance Data for
Accountability: The New York City
The Transformation of the Rethinking U.S. Environmental Police Department’s CompStat
Government Accountability Office: Protection Policy: Management Model of Police Management (2001)
Using Human Capital to Drive Change Challenges for a New Administration Paul E. O’Connell
(2005) (2000)
Jonathan Walters and Charles Dennis A. Rondinelli Moving Toward More Capable
Thompson Government: A Guide to
Creating a Culture of Innovation: Organizational Design (2002)
Designing and Implementing 10 Lessons from America’s Best Run Thomas H. Stanton
Performance-Oriented Payband City (2001)
Systems (2007) Janet Vinzant Denhardt and Robert The Baltimore CitiStat Program:
James R. Thompson B. Denhardt Performance and Accountability
(2003)
Managing for Better Performance: Understanding Innovation: Lenneal J. Henderson
Enhancing Federal Performance What Inspires It? What Makes It
Management Practices (2007) Successful? (2001) Strategies for Using State Information:
Howard Risher and Charles H. Fay Jonathan Walters Measuring and Improving Program
Performance (2003)
Seven Steps of Effective Workforce Government Management of Shelley H. Metzenbaum
Planning (2007) Information Mega-Technology:
Ann Cotten Lessons from the Internal Revenue Linking Performance and Budgeting:
Service’s Tax Systems Modernization Opportunities in the Federal Budget
(2002) Process (2004, 2nd ed.)
Barry Bozeman Philip G. Joyce
Innovation
Advancing High End Computing: How Federal Programs Use Outcome
Managing Workfare: The Case of the Linking to National Goals (2003) Information: Opportunities for
Work Experience Program in the New Juan D. Rogers and Barry Bozeman Federal Managers (2004, 2nd ed.)
York City Parks Department (1999) Harry P. Hatry, Elaine Morley, Shelli B.
Steven Cohen The Challenge of Innovating in Rossman, and Joseph S. Wholey
Government (2006, 2nd ed.)
New Tools for Improving Sandford Borins Performance Management for
Government Regulation: An Career Executives: A “Start Where
Assessment of Emissions Trading A Model for Increasing Innovation You Are, Use What You Have” Guide
and Other Market-Based Regulatory Adoption: Lessons Learned from the (2004, 2nd ed.)
Tools (1999) IRS e-file Program (2006) Chris Wye
Gary C. Bryner Stephen H. Holden
Staying the Course: The Use of
Religious Organizations, Anti-Poverty Transforming Government Through Performance Measurement in State
Relief, and Charitable Choice: A Collaborative Innovation (2008) Governments (2004)
Feasibility Study of Faith-Based Satish Nambisan Julia Melkers and Katherine
Welfare Reform in Mississippi (1999) Willoughby
John P. Bartkowski and Helen A. Regis
Moving from Outputs to Outcomes:
Business Improvement Districts and Managing for Practical Advice from Governments
Innovative Service Delivery (1999) Performance and Around the World (2006)
Jerry Mitchell Results Burt Perrin
An Assessment of Brownfield Using the Balanced Scorecard:
Redevelopment Policies: The Using Evaluation to Support Lessons Learned from the U.S. Postal
Michigan Experience (1999) Performance Management: A Guide Service and the Defense Finance
Richard C. Hula for Federal Executives (2001) and Accounting Service (2006)
Kathryn Newcomer and Mary Ann Nicholas J. Mathys and Kenneth R.
Scheirer Thompson
To download or order a copy of a report, visit the IBM Center for The Business of Government website at: www.businessofgovernment.org 37
CENTER R E P O RT S AVA I L A B L E
Performance Leadership: 11 Better Managing “Big Science”: A Case Public Deliberation: A Manager’s
Practices That Can Ratchet Up Study of the Human Genome Project Guide to Citizen Engagement (2006)
Performance (2006, 2nd ed.) (2002) Carolyn J. Lukensmeyer and Lars
Robert D. Behn W. Henry Lambright Hasselblad Torres
Performance Accountability: Managing the New Multipurpose, A Manager’s Guide to Choosing
The Five Building Blocks and Six Multidiscipline University Research and Using Collaborative Networks
Essential Practices (2006) Centers: Institutional Innovation in (2006)
Shelley H. Metzenbaum the Academic Community (2003) H. Brinton Milward and Keith G.
Barry Bozeman and P. Craig Boardman Provan
Implementing OMB’s Program
Assessment Rating Tool (PART): Assessing Partnerships: New Forms The E-Government Collaboration
Meeting the Challenges of Integrating of Collaboration (2003) Challenge: Lessons from Five Case
Budget and Performance (2006) Robert Klitgaard and Gregory F. Studies (2006)
John B. Gilmour Treverton Jane Fedorowicz, Janis L. Gogan,
and Christine B. Williams
The Philadelphia SchoolStat Model Leveraging Networks: A Guide for
(2007) Public Managers Working across From Forest Fires to Hurricane
Christopher Patusky, Leigh Botwinik, Organizations (2003) Katrina: Case Studies of Incident
and Mary Shelley Robert Agranoff Command Systems (2007)
Donald P. Moynihan
What All Mayors Would Like to Extraordinary Results on National
Know About Baltimore’s CitiStat Goals: Networks and Partnerships in A Manager’s Guide to Resolving
Performance Strategy (2007) the Bureau of Primary Health Care’s Conflicts in Collaborative Networks
Robert D. Behn 100%/0 Campaign (2003) (2007)
John Scanlon Rosemary O’Leary and
Engaging Citizens in Measuring and Lisa Blomgren Bingham
Reporting Community Conditions: Public-Private Strategic Partnerships:
A Manager’s Guide (2007) The U.S. Postal Service-Federal
Alfred T. Ho Express Alliance (2003)
Oded Shenkar presidential
Strategic Use of Analytics in
Government (2008) The Challenge of Coordinating “Big transitions
Thomas H. Davenport and Science” (2003)
Sirkka L. Jarvenpaa W. Henry Lambright The President’s Management
Council: An Important Management
Communities of Practice: A New Innovation (2000)
Tool for Government Managers (2003) Margaret L. Yao
networks, William M. Snyder and Xavier de
Souza Briggs Government Reorganization:
Collaboration, Strategies and Tools to Get It Done
and partnerships Collaboration and Performance (2004)
Management in Network Settings: Hannah Sistare
The Challenge of Managing Across Lessons from Three Watershed
Boundaries: The Case of the Governance Efforts (2004) Performance Management for
Office of the Secretary in the U.S. Mark T. Imperial Political Executives: A “Start Where
Department of Health and Human You Are, Use What You Have” Guide
The Quest to Become “One”: An (2004)
Services (2000)
Approach to Internal Collaboration Chris Wye
Beryl A. Radin
(2005)
Leveraging Networks to Meet Russ Linden Becoming an Effective Political
National Goals: FEMA and the Safe Executive: 7 Lessons from
Cooperation Between Social Security Experienced Appointees (2005,
Construction Networks (2002)
and Tax Agencies in Europe (2005) 2nd ed.)
William L. Waugh, Jr.
Bernhard Zaglmayer, Paul Judith E. Michaels
Applying 21st-Century Government Schoukens, and Danny Pieters
to the Challenge of Homeland Getting to Know You: Rules of
Leveraging Collaborative Networks Engagement for Political Appointees
Security (2002)
in Infrequent Emergency Situations and Career Executives (2005)
Elaine C. Kamarck
(2005) Joseph A. Ferrara and Lynn C. Ross
Managing Across Boundaries: A Donald P. Moynihan
Case Study of Dr. Helene Gayle and Six Trends Transforming Government
the AIDS Epidemic (2002) (2006)
Norma M. Riccucci Mark A. Abramson, Jonathan D.
Breul, and John M. Kamensky
38 To download or order a copy of a report, visit the IBM Center for The Business of Government website at: www.businessofgovernment.org
Reflections on 21st Century Trans-Atlantic Experiences in Health Reforming the Federal Aviation
Government Management (2007) Reform: The United Kingdom’s Administration: Lessons from
Donald F. Kettl and Steven Kelman National Health Service and the Canada and the United Kingdom
United States Veterans Health (2006)
The Management of Regulation Administration (2000) Clinton V. Oster, Jr.
Development: Out of the Shadows Marilyn A. DeLuca
(2007) Improving Service Delivery in
Cornelius M. Kerwin Transforming Government: The Government with Lean Six Sigma
Revitalization of the Veterans Health (2007)
Strengthening Homeland Security: Administration (2000) John Maleyeff
Reforming Planning and Resource Gary J. Young
Allocation (2008)
Cindy Williams Transforming Government: Dan
Goldin and the Remaking of NASA SUPPLY CHAIN
(2001)
W. Henry Lambright MANAGEMENT
Social Services
The Power of Frontline Workers Digitally Integrating the
Implementing State Contracts for in Transforming Government: Government Supply Chain:
Social Services: An Assessment of The Upstate New York Veterans E-Procurement, E-Finance, and
the Kansas Experience (2000) Healthcare Network (2003) E-Logistics (2003)
Jocelyn M. Johnston and Barbara S. Timothy J. Hoff Jacques S. Gansler, William
Romzek Lucyshyn, and Kimberly M. Ross
Making Public Sector Mergers
Delivery of Benefits in an Emergency: Work: Lessons Learned (2003) Enhancing Security Throughout
Lessons from Hurricane Katrina (2007) Peter Frumkin the Supply Chain (2004)
Thomas H. Stanton David J. Closs and Edmund F.
Efficiency Counts: Developing the McGarrell
Capacity to Manage Costs at Air
Force Materiel Command (2003) Investing in Supply Chain Security:
Michael Barzelay and Fred Thompson Collateral Benefits (2005, 2nd ed.)
Strategy and James B. Rice, Jr., and Philip W. Spayd
Transformation The Transformation of the
Government Accountability Office: RFID: The Right Frequency for
The Importance of Leadership: The Using Human Capital to Drive Government (2005)
Role of School Principals (1999) Change (2005) David C. Wyld
Paul Teske and Mark Schneider Jonathan Walters and Charles
Benchmarking Procurement
Thompson
Leadership for Change: Case Studies Practices in Higher Education
in American Local Government (1999) Transforming the Intelligence (2007)
Robert B. Denhardt and Janet Community: Improving the Richard R. Young, Kusumal
Vinzant Denhardt Collection and Management of Ruamsook, and Susan B. Purdum
Information (2005)
Managing Decentralized Elaine C. Kamarck
Departments: The Case of the U.S.
Department of Health and Human Executive Response to Changing
Services (1999) Fortune: Sean O’Keefe as NASA
Beryl A. Radin Administrator (2005)
W. Henry Lambright
Corporate Strategic Planning in
Government: Lessons from the Ramping Up Large, Non-Routine
United States Air Force (2000) Projects: Lessons for Federal
Colin Campbell Managers from the Successful 2000
Census (2005)
Transforming Government: The Nancy A. Potok and William G.
Renewal and Revitalization of the Barron, Jr.
Federal Emergency Management
Agency (2000) The Next Government of the United
R. Steven Daniels and Carolyn L. States: Challenges for Performance
Clark-Daniels in the 21st Century (2005)
Donald F. Kettl
Transforming Government: Creating
the New Defense Procurement
System (2000)
Kimberly A. Harokopus
To download or order a copy of a report, visit the IBM Center for The Business of Government website at: www.businessofgovernment.org 39
BOOKS
40
About the IBM Center for The Business of Government
The IBM Center for The Business of Government connects
public management research with practice. Since 1998, we
have helped public sector executives improve the effectiveness
of government with practical ideas and original thinking. We
sponsor independent research by top minds in academe and the
nonprofit sector, and we create opportunities for dialogue on a
broad range of public management topics.
e-mail: businessofgovernment@us.ibm.com
website: www.businessofgovernment.org