20-page analysis written voluntarily in July 2013 for JPM (and later presented to the CFO) to solve for a problem hidden in company compensation philosophy, sparking a contrarian thinking that later would lead to the founding of HearNotes Inc.
The Branch Vitality Curve - A Sleeping Giant in the Retail Bank
1. – JPMORGAN CHASE –
CONSUMER & BUSINESS BANKING
THE BRANCH VITALITY CURVE:
A SLEEPING GIANT IN THE RETAIL BANK
PATRICK DONOHUE
2. 2
“This company cannot and will not pay the senior people more when the
company does worse.” – Jamie Dimon
OVERVIEW
BRANCH STACK RANKING AND THE HIDDEN FLAW THAT COULD BE
COSTING MILLIONS
The foundation of retail banking, which over the years has been
referred to in company financial reports in formats that includes
‘Regional Banking’ and ‘Consumer & Business Banking,’ ultimately
represents the branch banking business designed to gather deposits
while selling loans, investments, and other financial services to
consumers and small businesses.
Specifically within JPMorgan Chase, branches are viewed as the
epicenter into which virtually every other relevant business can be
funneled. While the desired cross-selling opportunity this creates
could lead to a confusing picture regarding their true performance as
a stand-alone business, historical financial statements dating back to
2000 have been able to accurately identify its isolated contribution to
the bottom line. Thus, it will be fair to assume throughout this
analysis that the underlying research is based on an already drilled-
down portrayal of financial performance in the retail bank.
Traditionally, retail banking at JPMorgan Chase and its heritage
organizations has been almost entirely comprised of seasoned
branches, also called Same Stores. These are branches that have an
established history of foot traffic, transaction counts, and customer
deposit bases. However, in more recent times a significant
investment has been made in the expanding network of new
branches, also called New Builds, which are strategically placed to
capture additional business in previously untapped markets. Since
the 2008 Washington Mutual acquisition, the unlocked footprint in
California and Florida has spurred a major influx of branch
infrastructure in these areas.
This proposal seeks to identify a potentially harmful and yet still
unnoticed drag on financial performance in the retail bank.
3. 3
Specifically, the core of this discovery is embedded in the continued
use of P&L ‘Stack Rank’ and ‘Peer Grouping’ methodology, originally
introduced by Jamie Dimon in 2000 to better understand Bank One
branches and establish a more commensurate incentive plan.
However, the persistence of this forced system of distribution has
painted a distorted picture of profitability, combined with a
compensation structure that allocates a fixed annual sum despite the
current economic reality of stagnant growth and even decline.
Summarized more plainly: historically using the P&L for the purposes
of stack ranking branches has proven to be highly effective in the
short term when competently instituted. But once seasoned, the
profound flaw disguised in its blissful continuance is the inability to
accomplish two vital objectives:
1) Distinguish actual winners from losers due to a progressively
arbitrary system of peer grouping.
2) Account for a potentially uniform decline in revenue,
inadvertently paying out disproportionate bonuses when the
company may in fact be doing worse.
4. 4
THE VITALITY CURVE
A BRIEF HISTORY OF STACK RANKING METHODOLOGY
The Vitality Curve, occasionally referred to as “Rank & Yank” in
pejorative business nomenclature, is a stack ranking performance
method that evaluates an employee’s performance on a bell curve
relative to their peers. Publicly championed by Jack Welch of GE in
the 1980’s, this approach was subsequently adopted by many Fortune
500 companies.
Welch’s specific vitality model called for all managers to rank their
employees on a 20-70-10 scale: the top 20% were considered ‘A’
players, recognized for their proportionately substantial contribution
and thus entitled to greater career and bonus potential. The
subsequent 70% were ‘B’ players, or average employees. The 10%
represented those who were significantly underperforming relative to
their peers and should be actively managed out of the firm.
A CASE FOR AND AGAINST
There is strong evidence to suggest that in the real world, a forced
distribution of employee rankings can lead to a fairly dramatic
improvement in initial performance. Therefore, if the objective is to
launch an ambitious turnaround in a short period of time, this
approach may very well achieve the desired effect. However, it should
also be noted that the net positive effects quickly begin to taper:
Steve Scullen, an associate professor of management at Drake University in Des
Moines, found that forced ranking, including the firing of the bottom 5% or 10%,
results in an impressive 16% productivity improvement -- but only over the first
couple of years. After that, Scullen says, the gains drop off, from 6% climbs in
the third and fourth years to basically zero by year 10. "It's a terrific idea for
companies in trouble, done over one or two years, but to do it as a long-term
solution is not going to work," says Dave Ulrich, a business professor at the
University of Michigan at Ann Arbor. "Over time it gets people focused on
competing with each other rather than collaborating.” 1
As the system survives, a subtle and disturbing risk begins to emerge.
Analogous to a fever that painfully facilitates the destruction of the
1 The Struggle to Measure Performance. Business Week. June 08, 2006
5. 5
host’s unwanted pathogens, failure of the fever to break could
ultimately cripple and destroy the host itself:
First, a very real danger exists that some satisfactory employees will be
misidentified as poor performers…The right thing to do is to strengthen weak
areas where there are a large number of poor performers…Most forced
distribution systems do not produce this result because every area, regardless of
the quality of its employees and its performance, is required to make the same
percentage of cuts. 2
MICROSOFT: A CASE STUDY IN VITALITY CURVE DOGMA
Microsoft, once viewed in business as the pinnacle of innovative
genius, has over the years begun to wane in comparison to its agile
and fluid competitors. This painstaking decline was meticulously
documented by Kurt Eichenwald, two-time winner of the prestigious
George Polk Award, in an expose entitled “Microsoft’s Lost Decade.”
What Eichenwald carefully points out is that while superstars like
Apple and Google were intensely focused on how to best delight
customer with better products, Microsoft’s culture had become
plagued with politics, bureaucracy and self-destruction triggered by a
dogmatic adherence to the Vitality Curve:
At the center of the cultural problems was a management system called “stack
ranking.” Every current and former Microsoft employee I interviewed—every
one—cited stack ranking as the most destructive process inside of Microsoft,
something that drove out untold numbers of employees. 3
Among some of the many insightful excerpts:
“…a lot of Microsoft superstars did everything they could to avoid working
alongside other top-notch developers…those at the top received bonuses and
promotions; those at the bottom usually received no cash or were shown the
door.”
“…One of the most valuable things I learned was to give the appearance of being
courteous while withholding just enough information from colleagues to ensure
they didn’t get ahead of me on the rankings.”
2 The Folly of Forced Ranking. Edward Lawler. July 15, 2002
3 “Microsoft’s Lost Decade.” Kurt Eichenwald. August 2012
6. 6
“…It leads to employees focusing on competing with each other rather than
competing with other companies...”
…In the end, the stack-ranking system crippled the ability to innovate at
Microsoft, executives said.
7. 7
A METAPHOR TO ILLUSTRATE THE BRANCH VITALITY CURVE*
AND WHY IT BECAME NECESSARY
DEAN JD & PROFESSOR SAMSUNG
Imagine a medical school that is faced with a serious dilemma:
lengthy investigations performed by the Board of Medicine revealed
that the majority of students were becoming doctors with a
surprisingly minimal level of effort. The progressively apathetic and
dysfunctional culture had virtually guaranteed the revered ‘MD’ so
long as the students merely showed up to class and occasionally
participated. Aside from the tremendous public embarrassment
suffered by the university, some of those same incompetent doctors
were now being paid generous sums by hospitals to make decisions
regarding a real patient’s life. The ultimatum was clear: produce good
doctors or shut down.
After a necessary impeachment of the former administration, Dean
JD was brought in and tasked to overhaul the medical school’s bogus
grading scale, whip the student body into shape and ultimately
restore prestige to the university.
In his first order of business, Dean JD hired Professor Samsung, a no-
nonsense academic experienced in such matters, to join his team and
institute his proven zero-sum grading system designed to quickly
analyze, identify and remove those medical students who were unfit
to continue their enrollment.
On the first day of the new semester, Professor Samsung assembled a
trepid student body and declared, “From this day forth, whatever the
actual grade you receive in any course is irrelevant. I have rationed
the grades for every 100 students and will only allocate them in the
following proportions:
A+ = 10 Students
B+ = 20 Students
C+ = 20 Students
C– = 20 Students
D = 20 Students
F = 10 Students
* Refers to Same Store compensation philosophy; P&L stack ranking via Performance Levels 1-6
8. 8
“So just be aware that even if you receive a 75% on every exam, if the
average score amongst your classmates is much higher, you may very
well end up failing that course. And unfortunately for you, that could
mean you no longer have a place in this university.”
To better understand how the medical school fared with this
approach, let’s take a look at the decade-long journey of stack ranking
that began in 2000 at Bank One.
9. 9
THE BANK ONE CASE STUDY
INTRODUCTION OF THE STACK RANK
Upon the arrival of Jamie Dimon to Bank One in 2000, the company
was facing an array of integration, risk management and leadership
headaches. In the midst of overhauling a company that would post a
$511 million loss that year, Dimon spearheaded a major headcount
reduction, conservative balance sheet management, the severance of
unprofitable corporate banking relationships, and the introduction of
the branch P&L with its corresponding incentive structure:
Previously, all branch managers had received bonuses ranging from 5 percent to
12 percent of their salary. Henceforth, the top 10 percent of branch managers
were to receive a bonus equal to 100 percent of their salary; the next 10 percent
received 50 percent; the next 10 percent received 30 percent, and the bottom 30
percent would receive no bonus at all. 4
While this was a necessary approach to compensation and
performance management at the time, it was certainly not Dimon’s
invention. It was actually a tactic learned while working alongside his
legendary mentor Sandy Weill during their now-famous takeover of
Consumer Credit in 1986. At the time, Weill summed up the new
compensation philosophy in his customary bluntness:
“The top ten percent of managers will receive a bonus equal to their salary,
doubling their compensation. The bottom ten percent will be out of the job.” 5
THE CYCLES
Historically speaking, if Bank One in 2000 commenced the first cycle
of Vitality Curve management, future economic events would create
two additional opportunities to restart the process. And while the
firm as a whole continued to dramatically improve operations, the
following analysis is specifically isolated to retail banking’s
performance over the course of 12 years and three distinct cycles: 6
4 Last Man Standing. Duff McDonald. 2009
5 Tearing Down The Walls. Monica Langley. 2003.
6 Analysis excludes the first year of each cycle to better reflect normalized performance.
10. 10
Cycle I: Bank One (2000-2003)
Cycle II: Bank One + JPMorgan Chase (2004-2007)
Cycle III: JPMorgan Chase + Washington Mutual (2008-?)
CYCLE I
Financially speaking, the results of Cycle I were the most dramatic of
all three. Retail banking revenue increased a modest 4% from 2001
to 2002, but net income increased an astonishing 39% during the
same period. The following year, from 2002 to 2003, revenue and
net income increases were 4% and 7%, respectively. And while the
last year of the cycle proved the most optimal on a per unit basis, the
most profound macroeconomic improvement was realized in the first.
In other words, after establishing a birds-eye view of branch
performance, it was considerably easier to remove the clear losers
that were a drag on profitability:
2001
Profit Per Branch: $580,000
Bankers Per Branch: 1.27
2003
Profit Per Branch: $850,000
Bankers Per Branch: 1.96
Best Period of Net Income Increase: 2001-2002
Net Income Increase: 39%
Best Year of Per Unit Performance: 2003
Ending # of Branches: 1,841
CYCLE II
By 2004, Bank One and JPMorgan Chase had merged, combining the
two firms’ assortment of branches to create a logical cost savings and
economies of scale. And as one would expect, the overhauling of
branch performance management was again necessary for the
business to thrive. Said in another way: it was time to reboot the
stack rank.
This time around, primary responsibility was delegated to Charlie
Scharf, CEO of Retail Financial Services, who noticed a familiar
cultural shortcoming on the opposite side of this new alliance:
11. 11
Before they came, half of branch managers earned a bonus of $8,000 to $18,000.
In the new system, the top generators of revenues and profits could earn as much
as $65,000 in bonuses, whereas the poorest would earn nothing and quite
possibly lose their jobs. 4
2005
Profit Per Branch: $1,050,000
Bankers Per Branch: 2.7
2007
Profit Per Branch: $730,000
Bankers Per Branch: 3.06
Best Period of Net Income Increase: 2005-2006
Net Income Increase: 4%
Best Year of Per Unit Performance: 2005
Ending # of Branches: 3,152
CYCLE III
In 2008, JPMorgan Chase embarked upon the nation’s largest FDIC-
aided bank acquisition, Washington Mutual, expanding the branch
network from 3,152 to 5,474. While it was an obvious financial
growing pain of expanding its branch network overnight by 74%,
there was also a significant challenge in that many of those branches
were blatantly unprofitable. Thus, the time had come to yet again
reboot the stack rank.
By 2009 retail banking was already approaching the cycle’s sweet
spot of branch synergy in terms of per unit performance and net
income increase. However, the retail bank has since struggled with
an array of economic and regulatory challenges that over time have
begun to erode collective and individual branch profitability. In
addition, the sheer number of bankers has risen to a staggering level
of 4.22 per branch, which is in all probability adding to the current
financial drag:
2009
Profit Per Branch: $760,000
Banker Per Branch: 3.49
2012
Profit Per Branch: $580,000
Bankers Per Branch: 4.22
Best Period of Net Income Increase: 2010-2011
Net Income Increase: 5%
Best Year of Per Unit Performance: 2009
Ending # of Branches: 5,614
12. 12
WHAT HAPPENS IF THERE IS NO CYCLE IV?
By year-end 2012, average branch profitability had been reduced to
$580,000, which essentially mirrors the business performance in
2001 when the Bank One journey had just begun. This is because the
Vitality Curve is a highly effective solve for rapid improvement, but
not one for sustainable performance measurement or compensation
planning. And without any economic catalyst to spark its necessary
reboot, the outcome may be inevitable self-cannibalization.
13. 13
A METAPHOR TO ILLUSTRATE BRANCH PEER GROUPING*
DEAN JD’S EPIPHANY
After four tireless years, Dean JD and the administration became
widely celebrated for besting a culture that invariably created
underachieving doctors and a lackluster university. The graduation
rates of their students improved to those of higher performing
medical schools, and it seemed that every student left standing had a
rightful claim to continue their rigorous pursuit of becoming a doctor.
In fact, it was so effective that a world-renown hospital became
interested and asked Dean JD to specifically identify his absolute best
students. Based on his recommendation, those talented and
intellectually superior students would be granted the opportunity to
train at the prestigious hospital during their residency. And in three
years when they passed their boards, would go on the highly coveted
payroll.
Dean JD, hardly being able to contain his enthusiasm, pledged to
oblige the hospital and recommend only his best and brightest to join
their ranks. After all, Professor Samsung’s tried-and-true approach
would clearly delineate the winners from the losers. However, when
he began perusing students’ detailed transcripts, Dean JD
experienced a startling revelation: he had absolutely no idea who the
top performers actually were.
THE PEER GROUP DILEMMA
Every semester, a student was required to complete five courses, each
comprised with a varying number and combination of their peers,
based on projected aptitude in that particular subject. The student
body, now composed of over 1,000 pupils, had taken dozens of
courses over the years.
This is where the puzzle begins: a small population of students
happened to stay in the precise composition of classmates during
* P&L Knowledge Center refers to this grouping methodology based on “deposit size, transaction
levels, growth rates, and branch type to achieve the fairest and most equitable groupings among
branches.”
14. 14
their progressive study of one particular subject. But by the 4th year,
the difference between the top 10% and the bottom 10% was no
longer substantial. For example, if a student took Anatomy I, II, III,
and IV with the exact same classmates every year, by the 4th year the
A+ student had an average exam score of 85%, while the F student
had an average of 75%. Dean JD was disturbed. It did not seem at all
reasonable to fail a student who very well may have been a good
doctor because they trailed the top student by a mere 10%. On the
inverse, he could not in good conscience recommend the A+ student
as a clear intellectual superior. To better understand the distinction,
he dug further into the data.
The interesting detail he uncovered was that most students did not
have the same combination of classmates in progressive courses. In
fact, while they tried their best to create a fair system by grouping
classmates based on perceived subject aptitude, it had now become
more of a random and arbitrary assortment that was reshuffled
every semester, thus creating inherently inequitable groupings.
Testing his concern, he identified a student named Ryan he had come
to know well over the years that was very capable, but unique in that
he had actually studied economics instead of pre-med prior to his
enrollment. When Ryan was recently grouped in an organic
chemistry course with peers who had extensive science backgrounds,
he barely passed with a D. But surprisingly, his D ultimately equated
to an average exam score of 80%. Dean JD could not believe it. Ryan
had clearly been introduced into an unfair peer group, but still
managed to achieve a respectable grade that under Samsung’s system
had nearly derailed his dreams of becoming a doctor.
However, when Ryan took an elective business course that allowed
him to leverage his relatively superior economics foundation, he
ranked #1 in the class. And much like JD discovered before, the
failing student, who previously had limited schooling of economics,
was ousted with a respectable average exam score of 75%.
The evidence was clear: even though the medical school had achieved
an optimal level of students through Samsung’s approach, its
continuance was unnecessarily weeding out great students, and even
worse: the now random peer grouping of students could actually be
15. 15
what dictated their ultimate fate in the rankings. It was the academic
equivalent of an amateur golfer unknowingly competing with
professionals on the PGA tour and never being given a handicap or an
explanation that adequately described his predictable yet highly
demoralizing defeat.*
A VITALITY CURVE LEFT UNCHECKED
There was one more analysis Dean JD decided to undertake: the
progression of overall course averages between when they originally
instituted the stack rank and the present time. He was shocked at his
findings.
Four years ago, the course average was an abysmal 58%. The
following year that rose dramatically to 76%, validating Samsung’s
approach and rendering further observation a supposedly futile
exercise. However, what he saw now was that the next year it actually
lowered to 69%, and then most recently back down to 58%. He
scratched his head in frustration, wondering how their broad
performance could possibly be declining, when he finally realized
what had happened in each respective year:
1) They instituted Samsung’s vitality model, which spurred the
dramatic increase in productive behaviors while at the same
time removing the poorest performers.
2) Collectively, the resulting student body was a superior group of
individuals who competed fiercely to achieve the highest grade
possible relative to their peers.
3) The course load became gradually more difficult in the third
year, and they brought in more demanding professors, while the
line between true winners and losers began to obscure.
4) After four consecutive years of removing ‘bottom’ performers in
a progressively more challenging academic environment, the
student body’s overall performance had begun to go backwards
in self-destruction, all while the university continued allocating
an unchanged proportion of grades.
* P&L Knowledge Center’s actual response to the Peer Group Dilemma in section
Ranking/Growth Score Dropped vs. Last Year – What’s Driving Performance? – “It
is important not to get too discouraged about your current performance. But it IS good
motivation to sit with your District Manager to understand the impact of business rule changes
for all branches.”
16. 16
RETHINKING THE METHODOLOGY
Upon this revelation, Dean JD realized he had much bigger concerns
than who to recommend to the prestigious hospital. He now only had
three vital goals until equity was restored:
1) Eliminate Professor Samsung’s stack ranking system
immediately. He was convinced it was not only unnecessary
now but becoming disturbingly destructive.
2) Cease the granting of any grade until it could be determined
that it was properly warranted for a given course.
3) Figure out how to account for a knowingly more difficult
academic environment in a fair way while still giving the
students the grades they rightfully deserved.
17. 17
A METAPHOR TO ILLUSTRATE THE REALIGNMENT OF BRANCH
COMPENSATION WITH ACTUAL PERFORMANCE
DEAN JD AND PROFESSOR NEWTON
Dean JD decided to recruit Professor Newton, a noted and respected
academician, to bring justice back to the system and implement
reasonable grading metrics.
When Newton arrived, he visually demonstrated his plan. He pulled
out a scratch paper and inscribed the following scale:
A+ = 115% and above
B+ = 107% to 114.9%
C+ = 98% to 106.9%
C– = 91% to 97.9%
D = 85% to 90.9%
F = less than 85%
Dean JD looked incredulous. “What are these figures?”
“Those are called Growth Scores*,” replied Newton. “We use this
score to determine how well a student has performed in a particular
course based on our expectation, or what we call Commitments**.”
“How do you come up with the commitment?”
“By looking at the historical average over a certain period of time, say
five years, and then deciding where to set the bar for this year’s
students.” Newton again went to the other side of his scrap paper and
identified the three step process:
1) Start with the course’s long-term historical average.
2) Use that figure to establish the student’s commitment, and
assign a corresponding grade based on performance relative to
that commitment with this formula:
* In the New Build Incentive Plan, Growth Scores refer to the weighted average of pretax profit
and revenue growth relative to commitments to determine commensurate bonus potential.
** Financial targets based on projections of a New Build’s annual pretax profit and revenue.
18. 18
[Final Grade – Commitment] / [Commitment] + 1 x 100% = Growth
Score*
3) Reward with higher grades those students that can adequately
exceed the commitment (>107%) and penalize with lower
grades those students who are unable to reasonably emulate
historical performance (<91%).
Dean JD, still wrapping his mind around the concept, started to
frightfully imagine returning to a culture where every student was
sliding by and the competitive atmosphere they fostered would be
undone. He even considered what would happen if every student
ended up achieving an A+, and how that could be an expensive lesson
for the administration.
Newton, however, was insistent. “Empirically speaking, it is highly
unlikely that everyone will get an A+. For example, consider a course
with a five-year average of 75%. We decide to set the Commitment at
76% knowing that the overall student quality this year has improved
and is better equipped to outperform the average.”
Newton explained that if a student received a final grade of 85%, this
is how we would calculate their Growth Score:
[85%-76%] / [76%] + 1 x 100% = 111.84%
“The student achieves a Growth Score of 111.84%, which as we know
is a respectable B+. What his peers do is irrelevant, because we
control the commitment. But even if they all collectively perform
better, then that’s a good thing, isn’t it? And regardless, the
commitment will be adjusted the following year to reflect that
improvement, making it a constantly moving target based on real
world conditions.”
Dean JD thought for a moment, finally nodding his head in
agreement. Newton’s approach would indeed restore fairness while
still allowing the administration to identify and recognize the best
performers. The puzzle had finally been solved.
* Actual formula cited in the New Build Incentive Plan minus the weighted average.
19. 19
“You cannot lift the wage earner up by pulling the wage payer down.” –
Abraham Lincoln
CONCLUSION
PROPOSED SOLUTION FOR REVISED BRANCH PERFORMANCE &
COMPENSATION PHILOSOPHY
Today, the internal approach for evaluating Same Store performance
within JPMorgan Chase begins with the annual process of reassessing
how to adequately peer group thousands of branches that are
comprised of an endless sum of individual differences, so as to fix
compensation expenses while reserving the bulk of it for “top
performers.” This was first introduced in 2000 to Bank One, and
since then two distinct economic catalysts has facilitated the
necessary recycling of this methodology.
However, banking in today’s environment is faced with an
immeasurable quantity of legal, regulatory and interest rate
challenges. Therefore, it does not make good economic sense to
perpetuate this arbitrary peer group and Vitality Curve performance
method when the business’ overarching financial performance is
suffering. Instead of rewarding the best of the worst, branches should
adopt a hybrid performance model borrowed from the New Build
Incentive Plan and revised for Same Stores.
New Builds, it should be noted, tend to be a much more profitable
undertaking. This is partly because extensive analysis is performed in
advance of its construction to determine the necessary business
justification while projecting its probable financial performance.
Then the branch team is tasked with achieving its corresponding
revenue and pretax profit commitments, rewarding only those who
significantly surpass economic expectations. If the banking center
fails to deliver on its targets, the staff (and its entire senior
management hierarchy) is compensated with an appropriately
reduced sum. And most importantly, their performance is justifiably
independent of their ‘peers,’ who may be doing better or worse for a
great multitude of reasons.
20. 20
Same Store branches should be analyzed for their realistic economic
growth potential to determine probable performance. Most likely the
place to start is with a calculated historical average of its five-year
Growth Score, properly adjusted and restated to reflect prior results.
And in the end, the branch should be given an appropriate, fair and
thoughtful commitment, leaving it entirely up to the team to justify
their ultimate compensation.
Consumer & Business Banking has performed extremely well in the
achievement of heightened customer satisfaction, unparalleled
balance growth and the fundamental notion of a performance-based
culture. But the fact is that performance today should be tied to
reality, not to each other.
As true ‘business owners,’ it should be inherently understood that if
the store generates less revenue every year, siphoning out the same
amount of expenses will invariably wipe out the enterprise. In the
end, it will not matter if its failure was attributed to higher costs,
perpetual lawsuits or a devastating reduction in profit margins. This
is because as our firm’s leadership most assuredly understands, what
drives America’s great economic engine forward lies not in the
complacent acceptance of the world that once was, but in the resolve
for a determined adaptation, innovation and response to the world
that could be.