Pricing Analysis for Merrill Lynch

Integrated Choice
Stuart Altschuler
Merrill Lynch Management Science Group, P.O. Box 9065, Princeton, New Jersey 08543–9065
Donna Batavia • Jeff Bennett
Merrill Lynch Strategic Marketing Services, 800 Scudders Mill Road—3F, Plainsboro, New Jersey 08536
Russ Labe • Bonnie Liao • Raj Nigam • Je Oh
Merrill Lynch Management Science Group, P.O. Box 9065, Princeton, New Jersey 08543–9065
stuart_altschuler@ml.com • donna_batavia@ml.com • jeff_bennett@ml.com
• russ_labe@ml.com • bonnie_liao@ml.com • raj_nigam@ml.com • je_oh@ml.com

In late 1998, Merrill Lynch and other full-service financial service firms were under assault.
Electronic trading and the commoditization of trading threatened Merrill Lynch’s value prop-
osition—to provide advice and guidance through a financial advisor. Management decided to
offer investors more choices for doing business with Merrill Lynch. A cross-functional team
evaluated alternative product and service structures and pricing and constructed models to
assess individual client’s behavior. The models showed that revenue at risk to Merrill Lynch
ranged from $200 million to $1 billion. The resulting Integrated Choice strategy enabled Merrill
Lynch to seize the marketplace initiative, changed the financial services landscape, and miti-
gated the revenue risk. As of year-end 2000, client assets reached $83 billion in the new offer,
net new assets to the firm totaled $22 billion, and incremental revenue reached $80 million.
(Financial Institutions: brokerage/trading. Marketing: pricing)

C harles E. Merrill founded the company that bears

his name in 1914 and charged it with the prop-
osition that the financial markets should be accessible
ages nearly $1.4 trillion of this, the rest coming from
international operations.
USPC serves over five million households and
to everyone. Merrill Lynch has delivered on that and small- to mid-sized businesses. It provides brokerage,
is credited with bringing Wall Street to Main Street. banking, and cash services, such as checking, and ATM
One of its major business segments, the Private Client and Visa cards through its Cash Management Account
Group, handles the brokerage and lending services (CMA威). Its asset services include stocks, bonds, mu-
tual funds, transactions of several other complex se-
(Merrill Lynch Annual Report 1999, Harvard Business
curities and products, and research on domestic and
School 2000).
international securities. It serves liability management
Asset gathering has been the Private Client Group’s
needs through diverse offerings of mortgage and home
core business strategy since the 1980s. That coupled
equity loans, securities-based lines of credit, and
with a planning-based, disciplined approach imparts commercial real estate financing. It also offers 401(k)
a long-term perspective rather than the short-term, employee-benefit administration and investment ser-
day-to-day trading approach. This approach has been vices, fixed and variable annuities, variable life-
quite successful. Client assets have increased at a 16 insurance products, and trust- and estate-planning ser-
percent compound annual rate over the past 10 years vices. USPC is a global leader in the full-service
to nearly $1.7 trillion at the end of 2000. US Private financial services arena with more assets under man-
Client (USPC), the domestic side of this group, man- agement than any competitor.

0092-2102/02/3201/0005$05.00 Interfaces, 䉷 2002 INFORMS

1526-551X electronic ISSN Vol. 32, No. 1, January–February 2002, pp. 5–19
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

USPC’s Business Model—Working in January 2000 to offer new services for its affluent
clients. These moves tended to blur the distinctions be-
Through Financial Advisors (FAs)
tween discount brokers and full-service firms.
USPC’s mission is to provide “sophisticated financial
solutions” to its clients to “simplify” their lives. It em-
ploys a highly trained sales force of nearly 16,000 fi- Competitive Trends—Banks and
nancial advisors (FAs) located in approximately 750 Full-Service Consolidations
offices throughout the United States. FAs have strong
In 1977, Merrill Lynch launched its innovative CMA威
working relationships with their clients and give them
accounts, which successfully challenged banks for cash
advice and service to help manage their assets. As part
deposits, and started a radical change. Until then, as a
of the customization, USPC allows its experienced FAs
result of the Securities Exchange Act and the Glass-
some latitude in pricing its products and in serving
Steagall Banking Act enacted in the 1930s, firms pro-
clients. This approach with clients has enabled USPC
viding brokerage and securities services had been
to gather assets and maintain long-term relationships.

By year-end 1999, nearly 31 percent of

Competitive Trends and retail trades were being done online.
Environment
Over the past 25 years, regulatory and technology separate from firms providing banking and insurance.
trends have enabled intense competition in the finan- In the 1980s and 1990s, several banks acquired under-
cial services arena. writing companies, and the operational boundaries be-
tween brokerage and banking activities began to blur.
The late 1990s saw a consolidation trend among ma-
Competitive Trends—Discount jor financial service providers. The merger between
Brokers Dean Witter and Morgan Stanley in 1997 created a bro-
Until 1975, the retail brokerage industry consisted pri- kerage powerhouse. In 1998, the $73 billion merger of
marily of full-service firms that engaged in all aspects Citicorp and Travelers led to the creation of Citigroup,
of the investment process from investment advice to which offered clients credit cards from Citicorp, stock-
execution and follow-up. The deregulation of commis- selling services from Salomon Smith Barney, and in-
sions in May 1975 created the discount brokerage in- surance products from Travelers. One-stop shopping
dustry. Several brokers, notably Charles Schwab, en- for all financial services began to emerge.
tered the industry and reduced commissions to almost
half those charged by full-service firms. Schwab and
other discount brokers focused on providing investors
Competitive Trends—Electronic
with convenient, low-priced access to securities trad- Trading
ing without advice on which securities to buy and sell. Deregulation and technology in the late 1990s led to
They benefited from an increased flow of financial in- the rise of another competitive segment—the electronic
formation in newspapers, financial-related cable tele- brokerage firms. These firms leveraged technology not
vision, and magazines that created informed investors. only in the back office but also in their interface with
Over time, Schwab added research to its product of- customers and thus dramatically reduced costs, even
ferings and arranged to offer its clients initial public beyond what the discount brokers had been able to do.
offerings (IPOs). It added analytic tools for asset allo- They typically did not have branch offices and relied
cation, stock selection, and financial planning to em- on purely electronic order entry.
ulate the sophisticated tools and processes used by the Like discount brokerages in the late 1970s, electronic
full-service firms. To further reposition itself as a “full- trading firms were initially viewed as upstarts attract-
service discount broker,” Schwab purchased US Trust ing low margin, active-trader clients. They built up

Interfaces
6 Vol. 32, No. 1, January–February 2002
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

market share through trading costs as low as $8 per vice chairman of Merrill Lynch, understood the risks
trade (compared to typically $30 for a discount bro- associated with changing a highly successful business
kerage firm and $200 for a full-service firm). In addi- paradigm. He, like many USPC executives, had risen
tion, the upsurge in technology stocks in the late 1990s through the ranks in the sales force and thus was
gave rise to the day-trading phenomenon that fanned acutely aware of the value of the best-trained sales
the growth of this channel. To meet this challenge, dis- force in the industry and the strength of its client re-
counters also began offering online trading. By year- lationships. But he also had a well-deserved reputation
end 1999, nearly 31 percent of all retail trades were in the industry as a visionary, and he could see the
being done online. Research indicated that even afflu- storm clouds gathering.
ent clients were using online services. As more and In late 1998, Steffens formed a task force, led by
more clients traded online at very low costs, pressure Allen Jones, senior vice president of USPC marketing,
on margins increased, particularly for full-service and asked it to recommend a product or service re-
firms. sponse to the marketplace challenge. The task force
By the late 1990s, competition pervaded the
financial-services industry. Regulatory changes had
brought banks into the brokerage business and vice Schwab’s $25.5 billion market
versa. Discounters were well established and had capitalization overtook Merrill
gained significant market share and reputation. Tech- Lynch’s $25.4 billion.
nology—especially the Internet—had placed a lot of
real-time information at the fingertips of investors. was to evaluate all aspects of customer reactions to
Electronic trading firms were using this technology to such an offering. As Jones remarked, “Our challenge
execute trades placed directly by clients effectively and was to make the case that in the emerging new econ-
at very low costs. Trading volume had gone up as omy, price is an integral component of our offer, which
prices went down. we redefined as relationship (based on trust), perfor-
mance (against client’s goals), service (beyond expec-
Merrill Lynch Reexamines Its Value tations), and price (appropriate to client benefit). We
Proposition decided to aggressively compete on each of these com-
ponents, including price” (Harvard Business School
In 1998, amid this swirl of competitive forces, USPC
2000, page 13).
continued to command the largest share of household
Before the task force could get moving, on December
assets and a strong share of affluent households (with
28, 1998, Schwab’s $25.5 billion market capitalization
investable assets of at least $100,000). However, it was
overtook Merrill Lynch’s $25.4 billion for the first time.
becoming increasingly clear that USPC’s pricing struc-
ture was not aligned with the value delivered to its Schwab had increased its assets by 39 percent in 1998,
clients. USPC’s value was in the customized advice while Merrill Lynch’s grew only 18 percent. At that
and guidance its FAs gave to their clients, and yet it moment “Merrill got religion” (Spiro 1999). That was
charged basically for executed trades (through com- an enormously difficult day in the executive suite.
missions). When there was no trade, neither the FAs Merrill Lynch had over a trillion dollars in client assets,
nor USPC got paid. USPC was charging premium compared to less than $500 billion for Schwab, and yet
prices on trade executions (now rendered a commodity their market capitalizations were equal. This added to
by electronic trading) and nothing for its advice and the urgency for a response.
guidance (its value proposition). As Jeff Bennett, first Jones asked the team for help in deciding on fun-
vice president for strategic pricing, remarked, “It was damental product features and their pricing that
like charging for the grocery bag, but not the groceries would be consistent with USPC’s value proposition,
inside” (Harvard Business School 2000, page 13). competitive in the marketplace, attractive to our cli-
John L. “Launny” Steffens, president of USPC and ents, palatable to the FAs, and yet profitable to USPC.

Interfaces
Vol. 32, No. 1, January–February 2002 7
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

He asked Jeff Bennett to lead the effort and the man- an FA could select this option and be charged a fixed
agement science group to participate. percentage of the value of their assets at Merrill Lynch.
They would no longer pay for trades individually; all
Management Science Group trades would be included in the asset-based fee. This
option eventually became known as Merrill Lynch Un-
Merrill Lynch established its management science
limited Advantage (MLUA). It gives clients several
group in 1986, and it has been part of the USPC or-
advantages:
ganization since 1990. The group helped Merrill Lynch
—Clients’ fees to Merrill Lynch would be directly
USPC win the 1997 INFORMS Prize for the effective
aligned with their level of asset appreciation and in-
use and impact of management science on the success
vestment return—clients only pay more if their assets
of the firm. At the time of this analysis, the group had
increase.
nine members.
—FAs and clients could adjust the portfolios, with-
The group’s mission is to aid strategic decision mak-
out the burden of transaction costs.
ing in complex business situations through quantita-
—Clients would know in advance their costs of do-
tive modeling and analysis. It supports executive man-
ing business with Merrill Lynch.
agement and business units by quantifying the impact
The MLUA pricing option has a fee-based, tiered
of different business strategies by analyzing data on
structure with separate rates for different types of eli-
clients, products, services, and the marketplace. Using
gible assets and a minimum account fee (Table 1).
expert-system models and analytical tools, it helps FAs
Trading is unlimited for eligible security types and
deliver value-added solutions and enhance relation-
subject to change. The option also includes, as part of
ships with clients.
the overall fee, free or reduced-price access to a num-
The group provides analytical and business assis-
ber of additional financial services, such as a formal
tance in many of USPC’s core activities. These include financial plan, portfolio reviews, credit lines, and mort-
modeling clients’ investment risk and asset allocation, gage loans. These products have been shown in pre-
optimizing mutual fund portfolios, developing invest- vious internal studies performed by the management
ment strategies, financial planning, developing pros- science group to reduce the likelihood of client defec-
pecting and cross-selling models, analyzing drivers of tion. Any ineligible security trades (for example, unit
client behavior, evaluating promotions, and perform- investment trusts and options) continue to be priced
ing business impact, pricing and compensation on a per-trades basis.
analyses. The second potential new option was a direct online
The management science group focuses on business pricing option for clients wishing to invest online di-
problems to be solved and employs a broad range of rectly. It represented the other end of the spectrum.
operations research and management science tech- This low-cost channel would be targeted to self-
niques, including optimization, simulation, expert sys- directed investors, who do not need the advice and
tems, multivariate statistics, and neural networks. To guidance of an FA, value Merrill Lynch’s brand name,
ensure a focus on implementation and responsiveness, and want to pay a flat fee for each trade. This option
it tackles all aspects of the modeling process, including eventually became known as ML Direct. Under this
collection and validation of data. It has earned the rep-
utation of an objective, analytical entity that has a can- Eligible Assets Equity/Mutual Funds Fixed Income/Cash
do attitude.
First $1 Million 1.00% 0.30%
Problem Overview Next $4 Million 0.75% 0.25%
Over $5 Million 0.50% 0.20%
The task force was asked to recommend new product
Table 1: The final MLUA pricing schedule includes unlimited trades on
structures and pricing options. It focused on two main eligible securities. The minimum fee is $1,500. This fee covers all ac-
options. The first was an asset-based pricing option: count fees, Visa fees, planning services, and reduced fees on credit
clients desiring a relationship founded on advice from products. All numbers are for illustrative purposes only.

Interfaces
8 Vol. 32, No. 1, January–February 2002
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

option, most trades on eligible securities (stocks and with asset-specific pricing, stocks and mutual funds
mutual funds) cost $30. could be subject to higher rates than bonds or money
Clearly a critical factor in the ultimate success of ei- funds. The blended fee would be a simpler pricing
ther offering was setting actual prices. For MLUA, this structure to communicate, but the asset-specific fee
was defined by (1) the overall percentage rates clients would make it more palatable for clients to include
their fixed income securities in the account.
Our approach was to simulate client-
choice behavior. Problem Overview—Universe of
Securities
would pay per dollars in assets, and (2) the minimum Clearly, common securities, such as stocks, mutual
fee for the service. The minimum fee would guarantee funds, and bonds, would be included in these pricing
Merrill Lynch a minimum level of revenue from in- options. However, other products, such as unit invest-
vestors whose accounts held smaller than targeted as- ment trusts, options, and futures, could either be in-
set levels. cluded or excluded from the fee. If they were excluded,
However, in addition to the pricing level, several clients would continue to pay for transactions of the
other characteristics of the new offerings had to be excluded securities at traditional rates.
evaluated. The pricing team faced these questions:
—Should Merrill Lynch offer either or both MLUA
Problem Overview—Closed Versus and ML Direct?
Open Architecture —At what pricing level should it offer them?
—How should it structure them in terms of closed
If Merrill Lynch were to offer both MLUA and ML
versus open architecture and blended versus asset-
Direct, clients would have a choice of four service
specific pricing, and for what universe of securities?
channels. In addition to the two proposed services,
they could choose to remain in the traditional
transaction-based pricing structure. They could also Objectives of Our Analysis
choose a wrap account, giving the FA or a designated The questions originally posed to management science
portfolio manager discretionary power over the ac- were rather broad and directional. The group’s first
count, that is, the ability to perform trades according task was to work with the rest of the pricing team and
to the investment objectives established by the client. senior management to define the following objectives:
The firm could offer these service channels through —To determine the total revenue at risk if the only
two formats. Under a closed-architecture format, clients clients choosing the new pricing options were those
would have to select a single channel and have all their who would pay less to Merrill Lynch (we called this
assets covered under the pricing structure for that of- behavior adverse selection);
fering. Under an open-architecture format, clients could —To determine a more realistic revenue impact
freely designate specific assets for each channel. Open based on the likelihood of clients’ adopting one of the
architecture works to the advantage of the clients, al- new services, even if it were not the lowest cost option;
lowing them to put their frequently traded assets in an —To assess the effect on revenue of various pricing
asset-based account and leave their longer-term assets schedules, minimum fee levels, product combinations
in a transaction-based account. and product features; and
—To assess the potential impact on each and every
FA and identify those who would be most affected.
Problem Overview—Blended Versus In formulating this analysis, we had to consider the
Asset-Specific Pricing FAs as well as the clients. Simply providing an attrac-
For MLUA, all assets could be subject to a single, tive and profitable set of offerings to investors was not
blended fee, regardless of asset type. Alternatively, sufficient if the changes in compensation to FAs

Interfaces
Vol. 32, No. 1, January–February 2002 9
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

resulted in high levels of discontent and turnover. Se- had to merge the ledger data with other client details
nior management needed to know how many FAs for our analyses.
(particularly how many top FAs) would be adversely The trades database provides the details of all buy
affected by the new offerings. Consequently, it was es- and sell transactions that took place in 1998. We used
sential that we include all of USPC’s clients in our anal- the data to determine the number of trades eligible for
ysis so that we could aggregate the results at the FA repricing in each account.
level. A random sample of clients would not address We group client relationships by households. What
this objective. we call a household may be composed of multiple ac-
counts and multiple individuals. A household may
Data Collection and Processing
The management science group assembled and eval- We had found the pricing sweet spot.
uated an extensive volume of client-level data. To ac-
curately calculate the revenue impact of the new have multiple accounts with Merrill Lynch for many
pricing options, we assembled a 200-gigabyte compre- reasons: multiple persons (spouses, children, parents,
hensive client database. It was constructed using data and so forth) in a household may have separate ac-
from 1998, which was the most recent full-year data counts, or people may have accounts earmarked for
available and which provided a profile of client assets different purposes (retirement, education, home pur-
and trading activity. We gathered information on five chase, and so forth). Since only assets in individual
million clients, 10 million accounts, 100 million trade retail accounts were eligible for repricing, we elimi-
records, 250 million ledger records, and 16,000 FAs. nated all business accounts from consideration.
For each client, we obtained data for six categories of Among the individual accounts, we did not consider
revenue, four categories of account type, nine asset al- for repricing discretionary managed accounts and cer-
location categories, along with data on number of tain specialty nondiscretionary fee-based accounts.
trades, mutual fund exchanges and redemptions, sales We used the assets in the eligible accounts to deter-
of zero coupon bonds, and purchases of new issues. mine the fees for the MLUA pricing option. Since the
The ledger file contains a record for every source of fees are based on the total household assets and cal-
culated on a tiered pricing structure, we had to identify
revenue for every account at Merrill Lynch. In 1998, 82
all eligible assets in every account. We also had to
different revenue sources were applicable to the pric-
break the assets down into asset-allocation categories
ing analysis, including trade commissions (from trades
(for example, equities, fixed income, cash, and mutual
in various equities, fixed income, and mutual funds),
funds) so that we could model asset-specific pricing.
account fees, asset-based pricing, banking services, in-
The FA database provides profile information on all
surance services, and planning and special services.
FAs, including their length of service and number of
We used this data to calculate the actual revenue de-
clients. We used this data to determine the compen-
rived from each client in 1998, and we used these rev- sation impact of the new pricing options on different
enue calculations as the baseline for comparisons to the groups of FAs. We derived FAs’ actual compensation
new pricing options. in 1998 for the individual accounts from the ledger
For both the MLUA and ML Direct price scenarios, records.
we had to classify each of the 82 revenue sources as We merged, reconciled, filtered, and cleaned data
eligible or ineligible under the new pricing schedules. from all these different production databases for our
For example, at the trade level, certain types of trades, analyses.
such as unit investment trusts (UITs), new-issue buys,
and zero-coupon sells were ineligible for MLUA. Also
certain nontransaction fees and services, such as ac- Model Overview: Client Choice Data
count fees, financial planning, Visa card fees, and elec- Simulation
tronic bill payments, would be part of the bundled of- Our basic modeling approach was to simulate client-
fer and no longer incur separate fees under MLUA. We choice behavior. Using this approach, we used an ini-

Interfaces
10 Vol. 32, No. 1, January–February 2002
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

tial set of system data and the resulting system-output affect the client’s decision process. In model 2, we used
measures of interest as a baseline. We then introduced a probabilistic simulation feature to capture the com-
change conditions to the data and applied a set of busi- bination of the client’s rational behavior with the cli-
ness rules to the clients to determine their reaction to ent’s FA affinity. In model 3, we used a set of business
the changes and calculated revised output measures. rules based on management judgment to translate the
In our case, the observed system data consists of FA affinity into zones of price indifference. Clients
every revenue-generating component of every account with high (low) affinity to their FAs would have a high
of every client at Merrill Lynch. The system output (low) level of price or economic indifference.
Flexibility was a critical feature of all the models. We
MLUA generated a nearly $50 billion needed to easily study different pricing levels, archi-
tecture of the offerings, eligible assets, and pricing con-
increase in assets and a nearly $80 figurations. With any of the models, we needed to roll
million increase in revenues. up the results from the client level to the firm level to
assess overall revenue impact. In addition, we had to
measures are the resulting revenue at the firm level, determine the impact of changes in pricing policy and
the compensation impact on each FA, and the per- compensation policy by rolling up the data to the FA
centage of clients considered adverse selectors (clients level and incorporating the FAs’ individual payout
who choose one of the new pricing options and as a rates.
result reduce overall revenues). The change conditions All in all, we assessed more than 40 combinations of
consisted of different variations of the two new prod- pricing offerings and architectures. The turnaround
uct structures and pricing options—MLUA and ML time for analyzing new scenarios with a new set of
Direct. offerings was only a few hours.
We also needed to develop the set of business rules
that would determine the client reaction to the new
pricing options. We had several ideas about how to The Rational-Economic-Behavior
model these rules and, as a result, developed three dif- (REB) Model
ferent models: The rational-economic-behavior (REB) model esti-
(1) A rational-economic-behavior model (REB), mates the maximum revenue at risk from adverse se-
(2) A financial-advisor-affinity model based on lection. We calculated the revenue impact for each cli-
Monte Carlo simulation, and ent and then summed to the FA level and the firm level
(3) A financial-advisor-affinity model based on to determine the upper bound of the revenue at risk.
business rules. The model assumes that clients always select the
We designed the REB model to assess the maximum lowest-cost option.
total revenue at risk. In this model, we assumed that a We made the lowest-cost pricing evaluations at the
client would make decisions on a purely rational and household level, usually for multiple accounts. In each
economic basis—close to an extreme scenario. (Choos- household, we assumed people made the decision to go
ing the ML Direct option meant getting no advice and with MLUA or ML Direct pricing on an account-by-
guidance from an FA—a qualitative attribute. Thus, account basis (open architecture). Open-architecture
assuming that everyone who could act on a purely ra- pricing complicates the determination of minimum
tional and economic basis would indeed do so is a little pricing because it permits optimal mixed pricing. That
severe.) Having established an upper bound for reve- is, the lowest cost option may be a combination of tra-
nue at risk using the rational economic model, we ditional pricing and MLUA or ML Direct pricing,
added the second and third models to obtain more re- partly because clients can choose for each account pric-
alistic estimates of revenue impact. We did this by tak- ing decision separately and partly because within each
ing account of the strength of the client’s relationship account, some assets and trades may be eligible but
with his or her FA (FA affinity) and how this might others may not (Figure 1).

Interfaces
Vol. 32, No. 1, January–February 2002 11
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

For the purposes of this analysis, we considered tion, but that does not mean that that option is neces-
three least-cost household pricing outcomes to be sarily a lower-price option. For example, many
feasible: accounts have assets eligible for MLUA but have little
(1) 100 percent traditional (the current baseline), trading activity; by opting for MLUA, the client would
(2) A combination of traditional and MLUA, and pay a higher asset-based fee but receive little benefit
(3) A combination of traditional and ML Direct. from the unlimited trading that this type of account
offers.
Traditionally combinatorial optimization problems
REB—Calculating the Minimum can be solved using standard integer-programming
Household Pricing methods. However, in this case, the problem becomes
We determined the minimum household pricing by intractable because it has to be solved for over 5 million
calculating the cost for each of the three options and households. In addition, it is difficult to formulate as
selecting the lowest-cost option. Option 1 is the actual a traditional optimization because of the complexity of
baseline revenue for 1998 from the ledger records. the MLUA tiered-pricing structure, the minimum fee
requirement, special considerations for the bundled
traditional pricing ⳱ 兺 individual revenue sources nontransaction services, and the presence of both eli-
gible and ineligible assets. Consequently, we used a
∀ household accounts.
heuristic method based on a greedy approach to solve
For options 2 and 3, since the client decides to go to the problem efficiently (see the Appendix for details).
a new pricing on an account-by-account basis, the For ML Direct, because of its simpler pricing struc-
problem becomes a combinatorial optimization prob- ture, a more traditional optimization approach would
lem for which the decision variable is whether or not have worked. But because the greedy approach pro-
an account goes to a new pricing option or remains vided an extremely efficient calculation method for
with the traditional option. An account may have as- evaluating five million households, we also used it for
sets and trades that are eligible for a new pricing op- evaluating ML Direct pricing.

Figure 1: Eligibility for repricing is determined at the account level and at the assets/trades level within the
accounts. Accounts 1, 2, and 4 are eligible for repricing but Accounts 1 and 2 each have some assets or trades
that are ineligible for repricing. Account 3 is either ineligible for repricing or all of its assets and trades are
ineligible for repricing. In practice, this decision is based on discussions between the FA and client.

Interfaces
12 Vol. 32, No. 1, January–February 2002
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

Once we had completed the minimum-pricing cal- FA Affinity—Monte Carlo

culations for the three pricing combinations, we chose
the lowest for each household, summed over all the
Simulation Model
households, and subtracted the total from the actual For both the REB model and the Monte Carlo model,
1998 revenue to determine USPC’s revenue at risk: the observed system data consist of every revenue-
generating component of every account of every client
revenue at risk ⳱ 兺 (1998 revenue at Merrill Lynch. The output measures are the result-
ⳮ minimum cost option revenue) ∀ households. ing revenue at the firm level, the compensation impact
on each FA, and the percentage of clients considered
Similarly, by summing over households served by
adverse selectors. The difference lies in the business
an FA and subtracting from 1998 values, we found the
rules that capture clients’ reactions to the new pricing
revenue and compensation at risk for that FA (Figure
options.
2). These FA-level assessments gave executive man-
For the Monte Carlo simulation model, we estab-
agement the ability to put faces to all of the aggregate
lished likely price-adoption behaviors by doing the fol-
revenue estimates and to determine the specific inter-
lowing: We segmented clients into 12 segments based
ventions needed to prevent the defections of key FAs.
on the following three attributes:
This added to the risk-mitigation efforts.
—Two levels of FA affinity, high or low;
—Three levels of relative cost of the MLUA option
FA Affinity Model versus actual 1998 cost, higher, same (within 10 per-
We designed the FA affinity model to take account of cent), or lower; and
both rational economic considerations and qualitative —Two levels of relative cost of the ML Direct offer
factors and thus assess the revenue impact more real- versus actual 1998 cost, higher or same (within 10 per-
istically by incorporating the effect of clients’ relation- cent), or lower.
ships with their FAs. We defined FA affinity as the We assessed likely offer-adoption behaviors for each
strength of the client’s relationship with his or her FA segment using managerial judgement, based on mar-
based on client satisfaction (measured by a combina- ket research or experience with clients (Table 2), to es-
tion of client-behavior data and an ongoing, compre- timate the client’s probabilities of choosing from
hensive client-satisfaction survey), asset retention, among the following alternatives:
tenure, and recent completion of a formal financial —Traditional,
plan. We developed two versions, one based on Monte —MLUA,
Carlo simulation and one based on zones of price —ML Direct, or
indifference. —Leave Merrill Lynch.
We then produced a revenue-at-risk figure and re-
ported the impact on FAs, based on the same process
Rank FA Name FA Length 1998 PC Actual Change Change described for the REB model.
of Service Quintile 1998 in PCs in PCs
PCs ($K) ($K) (%)

1 FA Name 1 6 1 1,772 ⳮ529 ⳮ31%

2 FA Name 2 20 1 1,464 ⳮ426 ⳮ29%
FA Affinity—Zones of Price
3 FA Name 3 13 1 1,273 ⳮ51 ⳮ4% Indifference Model
4 FA Name 4 18 1 1,202 ⳮ500 ⳮ42%
In this model, we assumed clients have a zone of price
5 FA Name 5 7 2 1,190 ⳮ11 ⳮ1%
... ... ... ... ... ... ... indifference, that they view all pricing options within
a given distance of the existing traditional costs as hav-
Figure 2: This is a sample FA impact report from the rational-economic-
behavior model. It shows the impact on the top 200 FAs, ranked by their
ing roughly similar costs. For example, clients with
1998 production (PCs), if all of their clients chose the lowest cost pricing high FA affinity may have a large zone of indifference
option. All numbers are for illustrative purposes only. of 30 percent around their current cost, while clients

Interfaces
Vol. 32, No. 1, January–February 2002 13
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

Cost of Offer vs. Client’s Choice

Current Costs (% of clients choosing each channel)

MLUA ML Direct Affinity With FA ML Direct MLUA Traditional Leave ML

Higher Higher or same High 0% 25% 75% 0%

Higher Higher or same Low 15% 0% 75% 10%
Higher Lower High 0% 25% 75% 0%
Higher Lower Low 40% 0% 50% 10%
Same (within Ⳳ10%) Higher or same High 0% 35% 65% 0%
Same (within Ⳳ10%) Higher or same Low 15% 0% 85% 0%
Same (within Ⳳ10%) Lower High 0% 25% 75% 0%
Same (within Ⳳ10%) Lower Low 40% 0% 50% 10%
Lower Higher or same High 0% 50% 50% 0%
Lower Higher or same Low 10% 45% 45% 0%
Lower Lower High 0% 50% 50% 0%
Lower Lower Low 30% 35% 35% 0%
Table 2: This table provides a sample scenario for the FA Affinity Monte Carlo-based model. The probability
of choosing a pricing alternative is displayed for each of the 12 client segments. All numbers are for illustrative
purposes only.

with low FA affinity may have a smaller zone of in- even if it is more expensive (benign selection)—only if
difference of only 10 percent. Based on this approach, adverse selection to ML Direct does not occur.
the following outcomes are possible: —If the cost of MLUA is within 30 percent (for high
(1) FA affinity may negate a client’s rational choice FA affinity) or 10 percent (for low FA affinity) of the
of the cheapest offer (adverse selection). cost of the current pricing, we assume the client will
—If MLUA or ML Direct is the least expensive op- adopt the more expensive MLUA offer because of the
tion, and if the client can reduce his or her expenses influence of the FA. For example, the client described
by more than 30 percent (high FA affinity) or 10 per- earlier who had high FA affinity (a 30 percent zone of
cent (low FA affinity) by shifting assets, then the client indifference) would not choose the cheaper ML Direct
will make the shift. but would, however, choose the MLUA option for only
—If the cost of the least expensive option is within $50 more per year (a five percent difference). We as-
30 percent (for high FA affinity) or 10 percent (for low sume that the FA relationship (affinity) would lead the
FA affinity) of the cost of the current traditional pric- client to choose MLUA even though it is more
ing, the client will forsake the cheaper option and re- expensive.
tain traditional pricing. We used these rules, over and above the process de-
For example, a client whose current expense is scribed for the REB model, to determine revenue at risk
at the total firm level and for each FA. Because the
$1,000 might face projected costs for ML Direct of $830
zones of price indifference were based on management
and for MLUA of $1,050. Based on a purely rational
judgment, we also conducted a sensitivity analysis of
response, the client would choose ML Direct and save
revenue at risk, using various values for the indiffer-
$170 (or 17 percent). If the client had high FA affinity
ence zone. We varied the indifference zones to identify
(a 30 percent zone of price indifference), he or she
the breakeven points. This helped us and management
would not switch to ML Direct, since the 17 percent
to become comfortable with the risks.
savings is not large enough to account for the reduc-
tion in service or the hassle of switching. A client with
low FA affinity (a 10 percent zone of price indiffer- Integration of Components
ence), however, would make the switch. Certainly, sound and innovative mathematical and sta-
(2) FA affinity may activate an MLUA relationship, tistical skills were critical to the success of this project.

Interfaces
14 Vol. 32, No. 1, January–February 2002
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

Just as important was teamwork—within the manage- Benefits and Business Impact: ML
ment science group, with the other members of the
Transforms Its Business Strategy
pricing task force, and with executive management.
During this project, members of the team met or Based on the pricing analysis and recommendations
from the team, management decided to move forward
spoke with the leaders of the task force daily and some-
with both MLUA and ML Direct. On June 1, 1999, Mer-
times hourly. Before analyzing a single record of data,
rill Lynch announced the Integrated Choice service, al-
the group clarified the purpose of the analysis, the key
lowing clients to choose a level of advice and manner
assumptions, the important deadlines, and the ulti-
of doing business with Merrill Lynch. Subsequent to
mate objectives. Management science also worked
this announcement, it rolled out MLUA in July 1999
closely with market research and provided support for
and ML Direct in December 1999.
its conjoint-analysis survey that helped us to evaluate
The benefits were significant and fell into four areas:
potential client responses to the features and prices of
seizing the marketplace initiative, finding the pricing
the new product offerings.
sweet spot, improving financial performance, and
We assembled the results from each run of the model
adopting the approach in other strategic initiatives.
into a summary document and distributed it to the Integrated Choice had a profound impact on Merrill
members of the task force. We highlighted key statis- Lynch’s competitive position. It enabled Merrill Lynch
tics and numbers and developed a common display to to seize the marketplace initiative and restored it to a
enable meaningful comparisons across several diverse leadership role in the industry. Speculation on “How
scenarios (Figure 3). Using these results, executive will Merrill Lynch respond to low-cost competitors?”
management specified new pricing alternatives and ar- shifted to “How will the industry respond to Merrill
rangements for testing. Lynch?” We offered a completely new business para-
The worst-case analysis of the revenue at risk clearly digm for investors—a new way of doing business in
had the most profound impact. These estimates financial services. This paradigm shifted the focus
showed that, even in the worst possible (and highly from the commoditization of trading to the value of
unlikely) case, the total revenue at risk did not out- advice, which holds strong appeal for affluent, advice-
weigh the benefits to be gained through superior mar- oriented clients. From a business-strategy perspective,
ket positioning, increased share, and improved it allowed us to more closely align our revenue and
retention. asset growth.

Revised 1998 Revenue ($M) for New Pricing Structure

Client Household Asset Range Traditional FA MLUA ML Direct Discretionary Other Total
Under $50K 151 1 8 2 81 241
$50K to $99K 75 1 15 4 9 94
$100K to $249K 163 5 11 17 17 212
$250K to $499K 163 8 11 31 21 234
$500K to $999K 165 11 11 43 32 262
$1M to $2.49M 155 12 10 47 51 275
$2.5M to $4.9M 66 6 4 20 31 127
$5MⳭ 102 3 2 19 60 186
Revised 1998 Total 1,040 47 60 183 302 1,629
Actual 1998 Total 1,443 61 0 182 302 1,988

Revenue at Risk $358 million

Percent of Clients Shifting Assets to Core 4.2%
Percent of Clients Shifting Assets to Direct 12.2%
Figure 3: This sample output report for executive management shows the shift in revenue for a potential offering
assuming a rational-economic-client response. All numbers are for illustrative purposes only.

Interfaces
Vol. 32, No. 1, January–February 2002 15
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

In the wake of our June 1999 announcement, many walk in the door opening up Unlimited Advantage ac-
competitors scrambled to develop their own versions counts” (Merrill Lynch Direct TV Broadcast (DBS), July
of MLUA. For example, Morgan Stanley Dean Witter 27, 2000).
announced the debut of iChoice in October 1999, de- Bill Hill: “MLUA overall makes your job a lot easier
signed to compete with MLUA by combining online to do and frees up everyone within the team to do the
services with full-service advice and guidance for a most important things, which are to service your cli-
single, asset-based fee. Salomon Smith Barney (the As- ents” (Merrill Lynch Direct TV Broadcast (DBS), June
set One account), Paine Webber (Insight One), A. G. 26, 2000). “One of the biggest advantages with MLUA
Edwards (Client Choice), and others made similar is the resistance level goes down enormously. You
moves. But none of them came close to gathering the don’t get asked the question everyone hates to get
assets accomplished by Merrill Lynch. asked—how much is it going to cost me?” (Merrill
This was because we had found the pricing sweet Lynch Direct TV Broadcast (DBS), July 27, 2000).
spot. When the management team first started think- Corby May: “I think it’s such a professional way to
ing about pricing MLUA, it was thinking of a fee do our business and . . . this is probably the best deal
higher than one percent on equities. Without the mod- on the street” (Merrill Lynch DBS, June 26, 2000).
els and analysis, we would not have found the right Merrill Lynch has realized a number of substantial
price. The one percent fee was the sweet spot because bottom-line financial gains since the launch of Inte-
it balanced several factors: the client’s price elasticity, grated Choice. At an aggregate level, through year-end
our revenue at risk and profitability, competitive of- 2000 it made the following gains:
ferings, and possible defections among our top FAs. —It gathered $83 billion in assets in MLUA, $22 bil-
As a result of the pricing analysis, management bet- lion of them new assets to the firm.
ter understood the risks and was able to mitigate them. —The number of MLUA accounts increased 80 per-
We determined that the revenue at risk ranged be- cent between 1999 and 2000.
tween $200 million and $1 billion. Moreover, the top —Merrill Lynch gathered $3 billion in assets in ML
FAs most adversely affected by the new offer would Direct. While this is smaller than the MLUA benefits,
have defected at significant levels. Our analysis iden- ML Direct was intended as an asset-retention strategy
tified specific FAs who were most at risk, and we for clients who preferred to manage some of their as-
worked with them to reduce the risk that they might sets themselves.
defect. We conducted specific and meaningful discus- One concern the firm had in introducing these new
sions with them to gain understanding of their issues products was cannibalization. While some of the fi-
and concerns. Changes were incorporated into the final nancial benefits came from existing clients who mi-
offering to address their concerns. grated to the new offerings, Merrill Lynch achieved a
In fact, the introduction of MLUA proved advanta- significant amount of growth. This came from three
geous to our FAs. In the months following the rollout, sources: new clients, retention of existing clients, and
discussions with many FAs revealed that MLUA had additional assets from both new and existing clients.
a positive impact on their business. Here are just a few Both MLUA and ML Direct attracted new clients to
samples of what they had to say. Merrill Lynch. In the case of MLUA, 48 percent of en-
Eric Bartok: “When I sit down with clients who have rollees during 2000 were new to the firm. In the case
been at other firms for the last 10, 15, 20 years, I’m of ML Direct, 60 percent were new to the firm.
really not having a problem bringing those assets, The second source of growth was increased client
transferring them over into MLUA, because of the way retention. It is clear from our market research and the
it’s price structured, and the benefits that they’re get- marketplace that a growing segment of clients are
ting” (Merrill Lynch Direct TV Broadcast (DBS), June looking for these alternative styles of business, and
26, 2000). “We are adding extreme amounts of value now they can get them without leaving Merrill Lynch.
and, ultimately, having extremely happy clients mov- Third, we found that MLUA induced clients to bring
ing forward, who continue to refer more people to new business to the firm in addition to the $22 billion

Interfaces
16 Vol. 32, No. 1, January–February 2002
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

in new assets in MLUA accounts mentioned earlier. In important decision we as a firm have made since we
a study completed in September 2000, we evaluated introduced the first cash-management account in the
the business lift for clients enrolled in MLUA com- 1970s.”
pared to a control group of non-MLUA clients with Acknowledgments
similar characteristics. The lifts represent incremental Several people have contributed to the success of this effort. We
benefits that Merrill Lynch derived from MLUA. Ap- extend our sincere thanks to Launny Steffens, president of USPC,
plying these results to the total MLUA client base, we and Allen Jones, senior vice president of marketing, for their lead-
ership and support throughout the pricing analysis. They played a
found that MLUA generated
critical role in helping us understand, analyze, and then communi-
—A nearly $50 billion increase in assets and cate the strategic importance of this work. We thank Bill Henkel,
—A nearly $80 million increase in revenues. first vice president of strategic marketing, for his on-going support
The success of the pricing analysis led Merrill Lynch of our analytical work. We extend special thanks to Tom Spencer, of
to adopt the same modeling approach in other projects. AT&T Laboratories, who acted as our coach during the Edelman
process. Last but not least, we thank the rest of our team in man-
Most notably, in January 2001, USPC embarked on an
agement science—Cigdem Gurgur, Manos Hatzakis, Jukti Kalita,
effort to redesign FA compensation, which affects sev- Gretchen Marsh-Ferino, Raj Patil, Charlie Pearlman, Shweta Shafi,
eral billion dollars of Merrill Lynch expenses. To sup- Steve Spence, Zhaoping Wang, and Lihua Yang—for their hard work
port this project, Management Science is leveraging the and insistence on excellence on a daily basis.
work from the Integrated Choice pricing analysis to
Appendix. Rational Economic Behavior Model:
evaluate alternative compensation plans and their im- Solution to Combinatorial Optimization
pact on individual FAs. A literature search did not reveal any heuristics that could directly
address a problem of this complexity, but the basic pricing problem
could be formulated as a variation of the traditional asset-allocation
Conclusion problem, in which case a greedy approach could be used. The prob-
lem in using a greedy algorithm was determining what measure to
A lot has changed since December 1998 when Schwab use as the basis for the allocation. For MLUA, the final decision was
overtook Merrill Lynch in market capitalization. Merrill to use the revenue velocity of the assets eligible for repricing. Rev-
Lynch made a bold decision to shift its business para- enue velocity is a traditional measure in the financial industry used
digm, and the results have been very positive. As noted to determine the profitability of an account. In this case, we calcu-
in an article in the New York Times in January 2001, “To- lated it as the 1998 revenue for fees eligible for repricing divided by
the eligible assets, that is, the current profitability of the assets eli-
day Schwab is worth $40 billion, about 30 percent less
gible for repricing. For ML Direct, we replaced revenue velocity by
than Merrill, and a growing number of analysts are pre- the number of eligible trades, which is a more appropriate measure
dicting that the gap will widen” (McGeehan 2001). In- for this pricing option.
deed, as of May 2001, that gap has widened to over 50 In the following MLUA pricing example (Table 3), we have a
percent. Integrated Choice was an important factor in client household consisting of five separate accounts. A household
this reversal of fortune. may have multiple accounts for many reasons, such as multiple per-
sons in the household with separate accounts (spouses, children,
The importance of Integrated Choice to USPC can-
parents, and so forth) or accounts ear-marked for different purposes
not be overstated. Launny Steffens, president of USPC, (education, retirement, home purchase, and so forth). Account 1 is
commented, “The decision to implement Integrated a fixed-income and cash account and Accounts 2 through 5 are eq-
Choice was an unprecedented change in strategy for uity and mutual fund accounts.
us. Management Science and Strategic Pricing pro- 1998 Revenue
vided the modeling and analyses that enabled me and Revenue Assets Velocity
my executive management team to better understand Account 1 (FI/Cash) $150 $200,000 0.0008
the revenue risks. This is the kind of thing that kept Account 2 (Equity) $1,500 $55,000 0.0273
me up at nights! . . . We have moved forward like a Account 3 (Equity) $3,400 $125,000 0.0272
Account 4 (Equity) $800 $25,000 0.0320
bullet train and it is our competitors that are scram-
Account 5 (Equity) $400 $90,000 0.0044
bling not to get run over.” Table 3: In this example, there are five accounts with varying levels of
David Komansky, CEO of Merrill Lynch and Chair- revenues and assets. The revenue velocity of each account is calculated
man of the Board, called Integrated Choice “The most as a proportion of revenues to assets.

Interfaces
Vol. 32, No. 1, January–February 2002 17
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

The first step in evaluating which accounts should go to the new MLUA
pricing option is to rank the accounts in reverse order of revenue 1998 Revenue MLUA Incremental
velocity (Table 4). This allows us to evaluate the most profitable Revenue Assets Velocity Revenue Revenue
accounts first since they are the ones most likely to benefit from the
Account 1 $150 $200,000 0.0008 $3,550 $600
new pricing options.
Account 5 $400 $90,000 0.0044 $2,950 $900
Account 3 $3,400 $125,000 0.0272 $2,050 $550
1998 Revenue Assets Revenue Velocity
Account 2 $1,500 $55,000 0.0273 $1,500 $0
Account 4 $800 $25,000 0.0320 Account 4 $800 $25,000 0.0320 $1,500 $1,500
Account 2 $1,500 $55,000 0.0273 Table 7: The accounts are resorted in order of increasing revenue ve-
Account 3 $3,400 $125,000 0.0272 locity and a search is conducted until the first account with an MLUA
Account 5 $400 $90,000 0.0044 incremental revenue lower than the 1998 revenue is identified. This ac-
Account 1 $150 $200,000 0.0008 count (Account 3) and all the accounts that follow (Accounts 2 and 4) will
Table 4: The accounts are sorted by decreasing revenue velocity to eval- benefit from the MLUA pricing option.
uate the accounts in order of likelihood that they will benefit from the
In this example, Accounts 1 and 5 are better off staying with the
MLUA pricing option.
traditional pricing whereas Accounts 3, 2, and 4 are better off under
the MLUA pricing for a total cost of $2,600 compared to $6,250 under
We then calculate the new revenue cumulatively for each account
the all-traditional pricing. By evaluating the incremental costs rather
(Table 5). That is, we add the eligible assets of each succeeding ac-
than the cumulative costs, we ensure that low-velocity accounts like
count to the eligible assets of all of the accounts with higher revenue
1 and 5 are left in the lower traditional pricing schedule.
velocities to calculate the new MLUA pricing since MLUA tiered
price breaks are based on all eligible assets in a household.
References
Harvard Business School. 2000. Merrill Lynch: Integrated choice.
1998 Revenue MLUA
Case N9–500–090, March 20, Cambridge, MA.
Revenue Assets Velocity Revenue
McGeehan, Patrick. 2001. Schwab downgrade is latest in a series of
Account 4 $800 $25,000 0.0320 $1,500 blows to online brokers. The New York Times (January 10) C1.
Account 2 $1,500 $55,000 0.0273 $1,500 Merrill Lynch Annual Report. 1999.
Account 3 $3,400 $125,000 0.0272 $2,050 Merrill Lynch Direct TV Broadcast (DBS). 2000. Managing the in-
Account 5 $400 $90,000 0.0044 $2,950 vestment process with unlimited advantage. June 26.
Account 1 $150 $200,000 0.0008 $3,550 ——. 2000. Positioning unlimited advantage. July 27.
Table 5: The eligible assets for each account are added to the eligible Spiro, Leah Nathans. 1999. Merrill’s battle. Business Week (November
assets of all the accounts with higher revenue velocities and repriced 15) 259.
using the MLUA pricing option.
Launny Steffens, President, US Private Client, Vice-
We then calculate incremental revenue associated with adding Chairman, Merrill Lynch and Company, 4 World Fi-
each account’s eligible assets to the total MLUA eligible household nancial Center FL 32, New York, New York 10080,
assets (Table 6).
made the following comments during the Edelman
competition and in a written memo: “The decision to
1998 Revenue MLUA Incremental
Revenue Assets Velocity Revenue Revenue implement Integrated Choice was an unprecedented
change in strategy for us. Management science and
Account 4 $800 $25,000 0.0320 $1,500 $1,500
Account 2 $1,500 $55,000 0.0273 $1,500 $0
strategic pricing provided the modeling and analyses
Account 3 $3,400 $125,000 0.0272 $2,050 $550 that enabled me and my executive management team
Account 5 $400 $90,000 0.0044 $2,950 $900 to better understand the revenue risks. The overall risk
Account 1 $150 $200,000 0.0008 $3,550 $600 ranged from $200 million to $1 billion in revenues. This
Table 6: The incremental revenue associated with each account is cal- is the kind of thing that kept me up nights! The risks
culated by subtracting the MLUA revenue from the account immediately
were also very critical at the individual financial ad-
preceding it.
visor levels.
We rerank the accounts in the reverse order and compare the “Analysis of many scenarios, with different pricing
incremental revenue for each account to the 1998 revenue (Table 7). points and how the offering should be structured in
The first account for which the incremental revenue is less than the terms of open vs. closed architecture, blended vs. asset-
1998 revenue keeps the new MLUA pricing as do all of the following
specific pricing, and universe of securities and services,
accounts.

Interfaces
18 Vol. 32, No. 1, January–February 2002
 ALTSCHULER ET AL.
Pricing Analysis for Merrill Lynch Integrated Choice

helped us to mitigate these risks. This helped me and relationship for one simple fee. For us, it puts renewed
the board of directors to make the final decision to pro- power into our asset gathering engine. MLUA had
ceed with both Unlimited Advantage and Direct offer- leaped to $83 billion under management by the end of
ings. It was also used to help convince the sales force 2000, and accounted for $22 billion of net new money.
that Integrated Choice was the right strategy for the And it allowed us to seize the initiative in the market-
long term. place. We have moved forward like a bullet train and
“ML Unlimited Advantage has been an unqualified it is our competitors that are scrambling not to get run
success. To our clients, it delivers the total financial over.”

Interfaces
Vol. 32, No. 1, January–February 2002 19