Professors Youngme Moon and John Quelch prepared this case. It was reviewed and approved before publication by a company designate. Funding for the development of this case was provided by Harvard Business School and not by the company. HBS cases are developed solely as the basis for class discussion. Cases are not intended to serve as endorsements, sources of primary data, or illustrations of effective or ineffective management. Copyright © 2003, 2018 President and Fellows of Harvard College. To order copies or request permission to reproduce materials, call 1-800-545- 7685, write Harvard Business School Publishing, Boston, MA 02163, or go to www.hbsp.harvard.edu. This publication may not be digitized, photocopied, or otherwise reproduced, posted, or transmitted, without the permission of Harvard Business School.

Starbucks: Delivering Customer Service

In late 2002, Christine Day, Starbucks’ senior vice president of administration in North America, sat in the seventh-floor conference room of Starbucks’ Seattle headquarters and reached for her second cup of toffee-nut latte. The handcrafted beverage—a buttery, toffee-nut flavored espresso concoction topped with whipped cream and toffee sprinkles—had become a regular afternoon indulgence for Day ever since its introduction earlier that year.

As she waited for her colleagues to join her, Day reflected on the company’s recent performance. While other retailers were still reeling from the post-9/11 recession, Starbucks was enjoying its 11th consecutive year of 5% or higher comparable store sales growth, prompting its founder and chairman, Howard Schultz, to declare: “I think we’ve demonstrated that we are close to a recession-proof product.”1

Day, however, was not feeling nearly as sanguine, in part because Starbucks’ most recent market research had revealed some unexpected findings. “We’ve always taken great pride in our retail service,” said Day, “but according to the data, we’re not always meeting our customers’ expectations in the area of customer satisfaction.”

As a result of these concerns, Day and her associates had come up with a plan to invest an additional $40 million annually in the company’s 4,500 stores, which would allow each store to add the equivalent of 20 hours of labor a week. “The idea is to improve speed-of-service and thereby increase customer satisfaction,” said Day.

In two days, Day was due to make a final recommendation to both Schultz and Orin Smith, Starbucks’ CEO, about whether the company should move forward with the plan. “The investment is the EPS [earnings per share] equivalent of almost seven cents a share,” said Day. In preparation for her meeting with Schultz and Smith, Day had asked one of her associates to help her think through the implications of the plan. Day noted, “The real question is, do we believe what our customers are telling

1 Jake Batsell, “A Grande Decade for Starbucks,” The Seattle Times, June 26, 2002.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

504-016 Starbucks: Delivering Customer Service

2

us about what constitutes ‘excellent’ customer service? And if we deliver it, what will the impact be on our sales and profitability?”

Company Background The story of how Howard Schultz managed to transform a commodity into an upscale cultural

phenomenon has become the stuff of legends. In 1971, three coffee fanatics—Gerald Baldwin, Gordon Bowker, and Ziev Siegl—opened a small coffee shop in Seattle’s Pike Place Market. The shop specialized in selling whole arabica beans to a niche market of coffee purists.

In 1982, Schultz joined the Starbucks marketing team; shortly thereafter, he traveled to Italy, where he became fascinated with Milan’s coffee culture, in particular, the role the neighborhood espresso bars played in Italians’ everyday social lives. Upon his return, the inspired Schultz convinced the company to set up an espresso bar in the corner of its only downtown Seattle shop. As Schultz explained, the bar became the prototype for his long-term vision:

The idea was to create a chain of coffeehouses that would become America’s “third place.” At the time, most Americans had two places in their lives—home and work. But I believed that people needed another place, a place where they could go to relax and enjoy others, or just be by themselves. I envisioned a place that would be separate from home or work, a place that would mean different things to different people.

A few years later, Schultz got his chance when Starbucks’ founders agreed to sell him the company. As soon as Schultz took over, he immediately began opening new stores. The stores sold whole beans and premium-priced coffee beverages by the cup and catered primarily to affluent, well-educated, white-collar patrons (skewed female) between the ages of 25 and 44. By 1992, the company had 140 such stores in the Northwest and Chicago and was successfully competing against other small-scale coffee chains such as Gloria Jean’s Coffee Bean and Barnie’s Coffee & Tea.

That same year, Schultz decided to take the company public. As he recalled, many Wall Street types were dubious about the idea: “They’d say, ‘You mean, you’re going to sell coffee for a dollar in a paper cup, with Italian names that no one in America can say? At a time in America when no one’s drinking coffee? And I can get coffee at the local coffee shop or doughnut shop for 50 cents? Are you kidding me?’”2

Ignoring the skeptics, Schultz forged ahead with the public offering, raising $25 million in the process. The proceeds allowed Starbucks to open more stores across the nation.

By 2002, Schultz had unequivocally established Starbucks as the dominant specialty-coffee brand in North America. Sales had climbed at a compound annual growth rate (CAGR) of 40% since the company had gone public, and net earnings had risen at a CAGR of 50%. The company was now serving 20 million unique customers in well over 5,000 stores around the globe and was opening on average three new stores a day. (See Exhibits 1–3 for company financials and store growth over time.)

What made Starbucks’ success even more impressive was that the company had spent almost nothing on advertising to achieve it. North American marketing primarily consisted of point-of-sale

2 Batsell.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

Starbucks: Delivering Customer Service 504-016

3

materials and local-store marketing and was far less than the industry average. (Most fast-food chains had marketing budgets in the 3%–6% range.)

For his part, Schultz remained as chairman and chief global strategist in control of the company, handing over day-to-day operations in 2002 to CEO Orin Smith, a Harvard MBA (1967) who had joined the company in 1990.

The Starbucks Value Proposition Starbucks’ brand strategy was best captured by its “live coffee” mantra, a phrase that reflected the

importance the company attached to keeping the national coffee culture alive. From a retail perspective, this meant creating an “experience” around the consumption of coffee, an experience that people could weave into the fabric of their everyday lives.

There were three components to this experiential branding strategy. The first component was the coffee itself. Starbucks prided itself on offering what it believed to be the highest-quality coffee in the world, sourced from the Africa, Central and South America, and Asia-Pacific regions. To enforce its exacting coffee standards, Starbucks controlled as much of the supply chain as possible—it worked directly with growers in various countries of origin to purchase green coffee beans, it oversaw the custom-roasting process for the company’s various blends and single-origin coffees, and it controlled distribution to retail stores around the world.

The second brand component was service, or what the company sometimes referred to as “customer intimacy.” “Our goal is to create an uplifting experience every time you walk through our door,” explained Jim Alling, Starbucks’ senior vice president of North American retail. “Our most loyal customers visit us as often as 18 times a month, so it could be something as simple as recognizing you and knowing your drink or customizing your drink just the way you like it.”

The third brand component was atmosphere. “People come for the coffee,” explained Day, “but the ambience is what makes them want to stay.” For that reason, most Starbucks had seating areas to encourage lounging and layouts that were designed to provide an upscale yet inviting environment for those who wanted to linger. “What we have built has universal appeal,” remarked Schultz. “It’s based on the human spirit, it’s based on a sense of community, the need for people to come together.”3

Channels of Distribution

Almost all of Starbucks’ locations in North America were company-operated stores located in high- traffic, high-visibility settings such as retail centers, office buildings, and university campuses.4 In addition to selling whole-bean coffees, these stores sold rich-brewed coffees, Italian-style espresso drinks, cold-blended beverages, and premium teas. Product mixes tended to vary depending on a store’s size and location, but most stores offered a variety of pastries, sodas, and juices, along with coffee-related accessories and equipment, music CDs, games, and seasonal novelty items. (About 500 stores even carried a selection of sandwiches and salads.)

3 Batsell.

4 Starbucks had recently begun experimenting with drive-throughs. Less than 10% of its stores had drive-throughs, but in these stores, the drive-throughs accounted for 50% of all business.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

504-016 Starbucks: Delivering Customer Service

4

Beverages accounted for the largest percentage of sales in these stores (77%); this represented a change from 10 years earlier, when about half of store revenues had come from sales of whole-bean coffees. (See Exhibit 4 for retail sales mix by product type; see Exhibit 5 for a typical menu board and price list.)

Starbucks also sold coffee products through non-company-operated retail channels; these so-called “Specialty Operations” accounted for 15% of net revenues. About 27% of these revenues came from North American food-service accounts, that is, sales of whole-bean and ground coffees to hotels, airlines, restaurants, and the like. Another 18% came from domestic retail store licenses that, in North America, were only granted when there was no other way to achieve access to desirable retail space (e.g., in airports).

The remaining 55% of specialty revenues came from a variety of sources, including international licensed stores, grocery stores and warehouse clubs (Kraft Foods handled marketing and distribution for Starbucks in this channel), and online and mail-order sales. Starbucks also had a joint venture with Pepsi-Cola to distribute bottled Frappuccino beverages in North America, as well as a partnership with Dreyer’s Grand Ice Cream to develop and distribute a line of premium ice creams.

Day explained the company’s broad distribution strategy:

Our philosophy is pretty straightforward—we want to reach customers where they work, travel, shop, and dine. In order to do this, we sometimes have to establish relationships with third parties that share our values and commitment to quality. This is a particularly effective way to reach newcomers with our brand. It’s a lot less intimidating to buy Starbucks at a grocery store than it is to walk into one of our coffeehouses for the first time. In fact, about 40% of our new coffeehouse customers have already tried the Starbucks brand before they walk through our doors. Even something like ice cream has become an important trial vehicle for us.

Starbucks Partners

All Starbucks employees were called “partners.” The company employed 60,000 partners worldwide, about 50,000 in North America. Most were hourly-wage employees (called baristas) who worked in Starbucks retail stores. Alling remarked, “From day one, Howard has made clear his belief that partner satisfaction leads to customer satisfaction. This belief is part of Howard’s DNA, and because it’s been pounded into each and every one of us, it’s become part of our DNA too.”

The company had a generous policy of giving health insurance and stock options to even the most entry-level partners, most of whom were between the ages of 17 and 23. Partly as a result of this, Starbucks’ partner satisfaction rate consistently hovered in the 80% to 90% range, well above the industry norm,5 and the company had recently been ranked 47th in the Fortune magazine list of best places to work, quite an accomplishment for a company with so many hourly-wage workers.

In addition, Starbucks had one of the lowest employee turnover rates in the industry—just 70%, compared with fast-food industry averages as high as 300%. The rate was even lower for managers, and as Alling noted, the company was always looking for ways to bring turnover down further: “Whenever we have a problem store, we almost always find either an inexperienced store manager or inexperienced baristas. Manager stability is key—it not only decreases partner turnover, but it also

5 Industrywide, employee satisfaction rates tended to be in the 50% to 60% range. Source: Starbucks, 2000.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

Starbucks: Delivering Customer Service 504-016

5

enables the store to do a much better job of recognizing regular customers and providing personalized service. So our goal is to make the position a lifetime job.”

To this end, the company encouraged promotion from within its own ranks. About 70% of the company’s store managers were ex-baristas, and about 60% of its district managers were ex-store managers. In fact, upon being hired, all senior executives had to train and succeed as baristas before being allowed to assume their positions in corporate headquarters.

Delivering on Service When a partner was hired to work in one of Starbucks’ North American retail stores, he or she had

to undergo two types of training. The first type focused on “hard skills” such as learning how to use the cash register and learning how to mix drinks. Most Starbucks beverages were handcrafted, and to ensure product quality, there was a prespecified process associated with each drink. Making an espresso beverage, for example, required seven specific steps.

The other type of training focused on “soft skills.” Alling explained:

In our training manual, we explicitly teach partners to connect with customers—to enthusiastically welcome them to the store, to establish eye contact, to smile, and to try to remember their names and orders if they’re regulars. We also encourage partners to create conversations with customers using questions that require more than a yes or no answer. So for example, “I noticed you were looking at the menu board—what types of beverages do you typically enjoy?” is a good question for a partner to ask.

Starbucks’ “Just Say Yes” policy empowered partners to provide the best service possible, even if it required going beyond company rules. “This means that if a customer spills a drink and asks for a refill, we’ll give it to him,” said Day. “Or if a customer doesn’t have cash and wants to pay with a check (which we aren’t supposed to accept), then we’ll give her a sample drink for free. The last thing we want to do is win the argument and lose the customer.”

Most barista turnover occurred within the first 90 days of employment; if a barista lasted beyond that, there was a high probability that he or she would stay for three years or more. “Our training ends up being a self-selection process,” Alling said. Indeed, the ability to balance hard and soft skills required a particular type of person, and Alling believed the challenges had only grown over time:

Back in the days when we sold mostly beans, every customer who walked in the door was a coffee connoisseur, and it was easy for baristas to engage in chitchat while ringing up a bag. Those days are long gone. Today, almost every customer orders a handcrafted beverage. If the line is stretching out the door and everyone’s clamoring for their coffee fix, it’s not that easy to strike up a conversation with a customer.

The complexity of the barista’s job had also increased over time; making a venti tazoberry and crème, for instance, required 10 different steps. “It used to be that a barista could make every variation of drink we offered in half a day,” Day observed. “Nowadays, given our product proliferation, it would take 16 days of eight-hour shifts. There are literally hundreds of combinations of drinks in our portfolio.”

This job complexity was compounded by the fact that almost half of Starbucks’ customers customized their drinks. According to Day, this created a tension between product quality and customer focus for Starbucks:

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

504-016 Starbucks: Delivering Customer Service

6

On the one hand, we train baristas to make beverages to our preestablished quality standards—this means enforcing a consistent process that baristas can master. On the other hand, if a customer comes in and wants it their way—extra vanilla, for instance— what should we do? Our heaviest users are always the most demanding. Of course, every time we customize, we slow down the service for everyone else. We also put a lot of strain on our baristas, who are already dealing with an extraordinary number of sophisticated drinks.

One obvious solution to the problem was to hire more baristas to share the workload; however, the company had been extremely reluctant to do this in recent years, particularly given the economic downturn. Labor was already the company’s largest expense item in North America (see Exhibit 3), and Starbucks stores tended to be located in urban areas with high wage rates. Instead, the company had focused on increasing barista efficiency by removing all non-value-added tasks, simplifying the beverage production process, and tinkering with the facility design to eliminate bottlenecks.

In addition, the company had recently begun installing automated espresso machines in its North American cafés. The verismo machines, which decreased the number of steps required to make an espresso beverage, reduced waste, improved consistency, and had generated an overwhelmingly positive customer and barista response.

Measuring Service Performance

Starbucks tracked service performance using a variety of metrics, including monthly status reports and self-reported checklists. The company’s most prominent measurement tool was a mystery shopper program called the “Customer Snapshot.” Under this program, every store was visited by an anonymous mystery shopper three times a quarter. Upon completing the visit, the shopper would rate the store on four “Basic Service” criteria:

• Service—Did the register partner verbally greet the customer? Did the barista and register partner make eye contact with the customer? Say thank you?

• Cleanliness—Was the store clean? The counters? The tables? The restrooms?

• Product quality—Was the order filled accurately? Was the temperature of the drink within range? Was the beverage properly presented?

• Speed of service—How long did the customer have to wait? The company’s goal was to serve a customer within three minutes, from back-of-the-line to drink-in-hand. This benchmark was based on market research which indicated that the three-minute standard was a key component in how current Starbucks customers defined “excellent service.”

In addition to Basic Service, stores were also rated on “Legendary Service,” which was defined as “behavior that created a memorable experience for a customer, that inspired a customer to return often and tell a friend.” Legendary Service scores were based on secret shopper observations of service attributes such as partners initiating conversations with customers, partners recognizing customers by name or drink order, and partners being responsive to service problems.

During 2002, the company’s Customer Snapshot scores had increased across all stores (see Exhibit 7), leading Day to comment, “The Snapshot is not a perfect measurement tool, but we believe it does a good job of measuring trends over the course of a quarter. In order for a store to do well on

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

Starbucks: Delivering Customer Service 504-016

7

the Snapshot, it needs to have sustainable processes in place that create a well-established pattern of doing things right so that it gets ‘caught’ doing things right.”

Competition In the United States, Starbucks competed against a variety of small-scale specialty coffee chains,

most of which were regionally concentrated. Each tried to differentiate itself from Starbucks in a different way. For example, Minneapolis-based Caribou Coffee, which operated more than 200 stores in nine states, differentiated itself on store environment. Rather than offer an upscale, pseudo- European atmosphere, its strategy was to simulate the look and feel of an Alaskan lodge, with knotty- pine cabinetry, fireplaces, and soft seating. Another example was California-based Peet’s Coffee & Tea, which operated about 70 stores in five states. More than 60% of Peet’s revenues came from the sale of whole beans. Peet’s strategy was to build a super-premium brand by offering the freshest coffee on the market. One of the ways it delivered on this promise was by “roasting to order,” that is, by hand roasting small batches of coffee at its California plant and making sure that all of its coffee shipped within 24 hours of roasting.

Starbucks also competed against thousands of independent specialty coffee shops. Some of these independent coffee shops offered a wide range of food and beverages, including beer, wine, and liquor; others offered satellite televisions or Internet-connected computers. Still others differentiated themselves by delivering highly personalized service to an eclectic clientele.

Finally, Starbucks competed against donut and bagel chains such as Dunkin Donuts, which operated over 3,700 stores in 38 states. Dunkin Donuts attributed half of its sales to coffee and in recent years had begun offering flavored coffee and noncoffee alternatives, such as Dunkaccino (a coffee and chocolate combination available with various toppings) and Vanilla Chai (a combination of tea, vanilla, honey, and spices).

Caffeinating the World The company’s overall objective was to establish Starbucks as the “most recognized and respected

brand in the world.”6 This ambitious goal required an aggressive growth strategy, and in 2002, the two biggest drivers of company growth were retail expansion and product innovation.

Retail Expansion

Starbucks already owned close to one-third of America’s coffee bars, more than its next five biggest competitors combined. (By comparison, the U.S.’s second-largest player, Diedrich Coffee, operated fewer than 400 stores.) However, the company had plans to open 525 company-operated and 225 licensed North American stores in 2003, and Schultz believed that there was no reason North America could not eventually expand to at least 10,000 stores. As he put it, “These are still the early days of the company’s growth.”7

The company’s optimistic growth plans were based on a number of considerations:

6 Starbucks 2002 Annual Report.

7 Dina ElBoghdady, “Pouring It On: The Starbucks Strategy? Locations, Locations, Locations,” The Washington Post, August 25, 2002.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

504-016 Starbucks: Delivering Customer Service

8

• First, coffee consumption was on the rise in the United States, following years of decline. More than 109 million people (about half of the U.S. population) now drank coffee every day, and an additional 52 million drank it on occasion. The market’s biggest growth appeared to be among drinkers of specialty coffee,8 and it was estimated that about one-third of all U.S. coffee consumption took place outside of the home, in places such as offices, restaurants, and coffee shops. (See Exhibit 6.)

• Second, there were still eight states in the United States without a single company-operated Starbucks; in fact, the company was only in 150 of the roughly 300 metropolitan statistical areas in the nation.

• Third, the company believed it was far from reaching saturation levels in many existing markets. In the Southeast, for example, there was only one store for every 110,000 people (compared with one store for every 20,000 people in the Pacific Northwest). More generally, only seven states had more than 100 Starbucks locations.

Starbucks’ strategy for expanding its retail business was to open stores in new markets while geographically clustering stores in existing markets. Although the latter often resulted in significant cannibalization, the company believed that this was more than offset by the total incremental sales associated with the increased store concentration. As Schultz readily conceded, “We self-cannibalize at least a third of our stores every day.”9

When it came to selecting new retail sites, the company considered a number of criteria, including the extent to which the demographics of the area matched the profile of the typical Starbucks drinker, the level of coffee consumption in the area, the nature and intensity of competition in the local market, and the availability of attractive real estate. Once a decision was made to move forward with a site, the company was capable of designing, permitting, constructing, and opening a new store within 16 weeks. A new store typically averaged about $610,000 in sales during its first year; same-store sales (comps) were strongest in the first three years and then continued to comp positively, consistent with the company average.

Starbucks’ international expansion plans were equally ambitious. Starbucks already operated over 300 company-owned stores in the United Kingdom, Australia, and Thailand, in addition to about 900 licensed stores in various countries in Asia, Europe, the Middle East, Africa, and Latin America. (Its largest international market was Japan, with close to 400 stores.) The company’s goal was to ultimately have 15,000 international stores.

Product Innovation

The second big driver of company growth was product innovation. Internally, this was considered one of the most significant factors in comparable store sales growth, particularly since Starbucks’ prices had remained relatively stable in recent years. New products were launched on a regular basis; for example, Starbucks introduced at least one new hot beverage every holiday season.

The new product development process generally operated on a 12- to 18-month cycle, during which the internal research and development (R&D) team tinkered with product formulations, ran focus 8 National Coffee Association.

9 ElBoghdady.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

Starbucks: Delivering Customer Service 504-016

9

groups, and conducted in-store experiments and market tests. Aside from consumer acceptance, whether a product made it to market depended on a number of factors, including the extent to which the drink fit into the “ergonomic flow” of operations and the speed with which the beverage could be handcrafted. Most importantly, the success of a new beverage depended on partner acceptance. “We’ve learned that no matter how great a drink it is, if our partners aren’t excited about it, it won’t sell,” said Alling.

In recent years, the company’s most successful innovation had been the 1995 introduction of a coffee and non-coffee-based line of Frappuccino beverages, which had driven same-store sales primarily by boosting traffic during nonpeak hours. The bottled version of the beverage (distributed by PepsiCo) had become a $400 million10 franchise; it had managed to capture 90% of the ready-to-drink coffee category, in large part due to its appeal to non-coffee-drinking 20-somethings.

Service Innovation

In terms of nonproduct innovation, Starbucks’ stored-value card (SVC) had been launched in November 2001. This prepaid, swipeable smart card—which Schultz referred to as “the most significant product introduction since Frappuccino”11—could be used to pay for transactions in any company-operated store in North America. Early indications of the SVC’s appeal were very positive: After less than one year on the market, about 6 million cards had been issued, and initial activations and reloads had already reached $160 million in sales. In surveys, the company had learned that cardholders tended to visit Starbucks twice as often as cash customers and tended to experience reduced transaction times.

Day remarked, “We’ve found that a lot of the cards are being given away as gifts, and many of those gift recipients are being introduced to our brand for the first time. Not to mention the fact that the cards allow us to collect all kinds of customer-transaction data, data that we haven’t even begun to do anything with yet.”

The company’s latest service innovation was its T-Mobile HotSpot wireless Internet service, introduced in August 2002. The service offered high-speed access to the Internet in selected Starbucks stores in the United States and Europe, starting at $49.99 a month.

Starbucks’ Market Research: Trouble Brewing? Interestingly, although Starbucks was considered one of the world’s most effective marketing

organizations, it lacked a strategic marketing group. In fact, the company had no chief marketing officer, and its marketing department functioned as three separate groups—a market research group that gathered and analyzed market data requested by the various business units, a category group that developed new products and managed the menu and margins, and a marketing group that developed the quarterly promotional plans.

This organizational structure forced all of Starbucks’ senior executives to assume marketing-related responsibilities. As Day pointed out, “Marketing is everywhere at Starbucks—it just doesn’t necessarily show up in a line item called ‘marketing.’ Everyone has to get involved in a collaborative marketing effort.” However, the organizational structure also meant that market- and customer-

10 Refers to sales at retail. Actual revenue contribution was much lower due to the joint-venture structure.

11 Stanley Holmes, “Starbucks’ Card Smarts,” BusinessWeek, March 18, 2002.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

504-016 Starbucks: Delivering Customer Service

10

related trends could sometimes be overlooked. “We tend to be great at measuring things, at collecting market data,” Day noted, “but we are not very disciplined when it comes to using this data to drive decision making.” She continued:

This is exactly what started to happen a few years ago. We had evidence coming in from market research that contradicted some of the fundamental assumptions we had about our brand and our customers. The problem was that this evidence was all over the place—no one was really looking at the “big picture.” As a result, it took awhile before we started to take notice.

Starbucks’ Brand Meaning

Once the team did take notice, it discovered several things. First, despite Starbucks’ overwhelming presence and convenience, there was very little image or product differentiation between Starbucks and the smaller coffee chains (other than Starbucks’ ubiquity) in the minds of specialty coffeehouse customers. There was significant differentiation, however, between Starbucks and the independent specialty coffeehouses (see Table A below).

Table A Qualitative Brand Meaning: Independents vs. Starbucks

Independents:

• Social and inclusive • Diverse and intellectual • Artsy and funky • Liberal and free-spirited • Lingering encouraged • Particularly appealing to younger coffeehouse customers • Somewhat intimidating to older, more mainstream coffeehouse customers

Starbucks:

• Everywhere—the trend • Good coffee on the run • Place to meet and move on • Convenience oriented; on the way to work • Accessible and consistent

Source: Starbucks, based on qualitative interviews with specialty-coffeehouse customers.

More generally, the market research team discovered that Starbucks’ brand image had some rough edges. The number of respondents who strongly agreed with the statement “Starbucks cares primarily about making money” was up from 53% in 2000 to 61% in 2001, while the number of respondents who strongly agreed with the statement “Starbucks cares primarily about building more stores” was up from 48% to 55%. Day noted, “It’s become apparent that we need to ask ourselves, ‘Are we focusing on the right things? Are we clearly communicating our value and values to our customers, instead of just our growth plans?’” (see Table B below).

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

Starbucks: Delivering Customer Service 504-016

11

Table B The Top Five Attributes Consumers Associate with the Starbucks Brand

• Known for specialty/gourmet coffee (54% strongly agree) • Widely available (43% strongly agree) • Corporate (42% strongly agree) • Trendy (41% strongly agree) • Always feel welcome at Starbucks (39% strongly agree)

Source: Starbucks, based on 2002 survey.

The Changing Customer

The market research team also discovered that Starbucks’ customer base was evolving. Starbucks’ newer customers tended to be younger, less well-educated, and in a lower income bracket than Starbucks’ more established customers. In addition, they visited the stores less frequently and had very different perceptions of the Starbucks brand compared to more established customers (see Exhibit 8).

In short, the team learned that Starbucks’ historical customer profile—the affluent, well-educated, white-collar female between the ages of 24 and 44—had expanded.

Customer Behavior

With respect to customer behavior, the market research team discovered that, regardless of the market—urban versus rural, new versus established—customers tended to use the stores the same way. The team also learned that, although the company’s most frequent customers averaged 18 visits a month, the typical customer visited just five times a month (see Figure A below).

Figure A Customer Visit Frequency

Source: Starbucks, 2002.

42%

11%

37%

27%

21%

62%

0%

20%

40%

60%

80%

100%

% of Total Starbucks

Customer Base

% of All Starbucks

Transactions

Customer visits 8+ times/month

Customer visits 3-7 times/month

Customer visits 1-2 times/month

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

504-016 Starbucks: Delivering Customer Service

12

Measuring and Driving Customer Satisfaction

Finally, the team discovered that, despite its high Customer Snapshot scores, Starbucks was not meeting expectations in terms of customer satisfaction. The satisfaction scores were considered critical because the team also had evidence of a direct link between satisfaction level and customer loyalty (see Exhibit 9 for customer satisfaction data).

While customer satisfaction was driven by a number of different factors (see Exhibit 10), Day believed that the customer satisfaction gap could primarily be attributed to a service gap between Starbucks scores on key attributes and customer expectations. When Starbucks had polled its customers to determine what it could do to make them feel more like valued customers, “improvements to service”—in particular, speed-of-service—had been mentioned most frequently (see Exhibit 11 for more information).

Rediscovering the Starbucks Customer Responding to the market research findings posed a difficult management challenge. The most

controversial proposal was the one on the table before Day—it involved relaxing the labor-hour controls in the stores to add an additional 20 hours of labor, per week, per store, at a cost of an extra $40 million per year. Not surprisingly, the plan was being met with significant internal resistance. “Our CFO is understandably concerned about the potential impact on our bottom line,” said Day. “Each $6 million in profit contribution translates into a penny a share. But my argument is that if we move away from seeing labor as an expense to seeing it as a customer-oriented investment, we’ll see a positive return.” She continued:

We need to bring service time down to the three-minute level in all of our stores, regardless of the time of day. If we do this, we’ll not only increase customer satisfaction and build stronger long-term relationships with our customers, we’ll also improve our customer throughput. The goal is to move each store closer to the $20,000 level in terms of weekly sales, and I think that this plan will help us get there.

In two days, Day was scheduled to make a final recommendation to Howard Schultz and Orin Smith about whether the company should roll out the $40 million plan. In preparation for this meeting, Day had asked Alling to help her think through the implications of the plan one final time. She mused:

We’ve been operating with the assumption that we do customer service well. But the reality is, we’ve started to lose sight of the consumer. It’s amazing that this could happen to a company like us—after all, we’ve become one of the most prominent consumer brands in the world. For all of our focus on building the brand and introducing new products, we’ve simply stopped talking about the customer. We’ve lost the connection between satisfying our customers and growing the business.

Alling’s response was simple: “We know that both Howard and Orin are totally committed to satisfying our retail customers. Our challenge is to tie customer satisfaction to the bottom line. What evidence do we have?”

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

Starbucks: Delivering Customer Service 504-016

13

Exhibit 1 Starbucks’ Financials, FY 1998 to FY 2002 ($ in millions)

FY 1998 FY 1999 FY 2000 FY 2001 FY 2002

Revenue Co-Owned North American 1,076.8 1,375.0 1,734.9 2,086.4 2,583.8 Co-Owned Int’l (UK, Thailand, Australia) 25.8 48.4 88.7 143.2 209.1 Total Company-Operated Retail 1,102.6 1,423.4 1,823.6 2,229.6 2,792.9 Specialty Operations 206.1 263.4 354.0 419.4 496.0 Net Revenues 1,308.7 1,686.8 2,177.6 2,649.0 3,288.9 Cost of Goods Sold 578.5 747.6 961.9 1,112.8 1,350.0 Gross Profit 730.2 939.2 1,215.7 1,536.2 1,938.9 Joint-Venture Incomea 1.0 3.2 20.3 28.6 35.8 Expenses: Store Operating Expense 418.5 543.6 704.9 875.5 1,121.1 Other Operating Expense 44.5 54.6 78.4 93.3 127.2 Depreciation & Amortization Expense 72.5 97.8 130.2 163.5 205.6 General & Admin Expense 77.6 89.7 110.2 151.4 202.1 Operating Expenses 613.1 785.7 1,023.8 1,283.7 1,656.0 Operating Profit 109.2 156.7 212.3 281.1 310.0 Net Income 68.4 101.7 94.5 181.2 215.1 % Change in Monthly Comparable Store Salesb North America 5% 6% 9% 5% 7% Consolidated 5% 6% 9% 5% 6%

Source: Adapted from company reports and Lehman Brothers, November 5, 2002.

aIncludes income from various joint ventures, including Starbucks’ partnership with the Pepsi-Cola Company to develop and distribute Frappuccino and with Dreyer’s Grand Ice Cream to develop and distribute premium ice creams.

bIncludes only company-operated stores open 13 months or longer.

Exhibit 2 Starbucks’ Store Growth

FY 1998 FY 1999 FY 2000 FY 2001 FY 2002

Total North America 1,755 2,217 2,976 3,780 4,574 Company-Operated 1,622 2,038 2,446 2,971 3,496 Licensed Storesa 133 179 530 809 1,078 Total International 131 281 525 929 1,312 Company-Operated 66 97 173 295 384 Licensed Stores 65 184 352 634 928 Total Stores 1,886 2,498 3,501 4,709 5,886

Source: Company reports.

aIncludes kiosks located in grocery stores, bookstores, hotels, airports, and so on.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

504-016 Starbucks: Delivering Customer Service

14

Exhibit 3 Additional Data, North American Company-Operated Stores (FY2002)

Average

Average hourly rate with shift supervisors and hourly partners $ 9.00 Total labor hours per week, average store 360 Average weekly store volume $15,400 Average ticket $ 3.85 Average daily customer count, per store 570

Source: Company reports.

Exhibit 4 Product Mix, North American Company-Operated Stores (FY2002)

Percent of Sales

Retail Product Mix Coffee Beverages 77% Food Items 13% Whole-Bean Coffees 6% Equipment & Accessories 4%

Source: Company reports.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

Starbucks: Delivering Customer Service 504-016

15

Exhibit 5 Typical Menu Board and Price List for North American Company-Owned Store

Espresso Traditions Tall Grande Venti Brewed Coffee Tall Grande Venti Classic Favorites Coffee of the Day 1.40 1.60 1.70 Toffee Nut Latte 2.95 3.50 3.80 Decaf of the Day 1.40 1.60 1.70 Vanilla Latte 2.85 3.40 3.70 Caffe Latte 2.55 3.10 3.40 Cold Beverages Tall Grande Venti Cappuccino 2.55 3.10 3.40 Iced Caffe Latte 2.55 3.10 3.50 Caramel Macchiato 2.80 3.40 3.65 Iced Caramel Macchiato 2.80 3.40 3.80 White Chocolate Mocha 3.20 3.75 4.00 Iced Caffe Americano 1.75 2.05 3.40 Caffe Mocha 2.75 3.30 3.55 Caffe Americano 1.75 2.05 2.40 Coffee Alternatives Tall Grande Venti Toffee Nut Crème 2.45 2.70 2.95 Espresso Solo Doppio Vanilla Crème 2.20 2.45 2.70 Espresso 1.45 1.75 Caramel Apple Cider 2.45 2.70 2.95 Hot Chocolate 2.20 2.45 2.70 Extras Tazo Hot Tea 1.15 1.65 1.65 Additional Espresso Shot .55 Tazo Chai 2.70 3.10 3.35 Add flavored syrup .30

Organic milk & soy available upon request Whole Beans: Bold ½ lb 1 lb Our most intriguing and exotic coffees Gold Coast Blend 5.70 10.95 French Roast 5.20 9.95 Sumatra 5.30 10.15 Decaf Sumatra 5.60 10.65 Ethiopia Sidame 5.20 9.95 Arabian Mocha Sanani 8.30 15.95 Frappuccino Tall Grande Venti Kenya 5.30 10.15 Ice Blended Beverages Italian Roast 5.20 9.95 Coffee 2.65 3.15 3.65 Sulawesi 6.10 11.65 Mocha 2.90 3.40 3.90 Caramel Frappuccino 3.15 3.65 4.15 Whole Beans: Smooth ½ lb 1 lb Mocha Coconut 3.15 3.65 4.15 Richer, more flavorful coffees (limited offering) Espresso Roast 5.20 9.95 Decaf Espresso Roast 5.60 10.65 Crème Frappuccino Tall Grande Venti Yukon Blend 5.20 9.95 Ice Blended Crème Café Verona 5.20 9.95 Toffee Nut Crème 3.15 3.65 4.15 Guatemala Antigua 5.30 10.15 Vanilla Crème 2.65 3.15 3.65 Arabian Mocha Java 6.30 11.95 Coconut Crème 3.15 3.65 4.15 Decaf Mocha Java/SWP 6.50 12.45 Tazo Tea Frappuccino Tall Grande Venti Whole Beans: Mild ½ lb 1 lb Ice Blended Teas The perfect introduction to Starbucks coffees Tazo Citrus 2.90 3.40 3.90 Breakfast Blend 5.20 9.95 Tazoberry 2.90 3.40 3.90 Lightnote Blend 5.20 9.95 Tazo Chai Crème 3.15 3.65 4.15 Decaf Lightnote Blend 5.60 10.65 Colombia Narino 5.50 10.45 House Blend 5.20 9.95 Decaf House Blend 5.60 10.65 Fair Trade Coffee 5.95 11.45

Source: Starbucks location: Harvard Square, Cambridge, Massachusetts, February 2003.

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

504-016 Starbucks: Delivering Customer Service

16

Exhibit 6 Total U.S. Retail Coffee Market (includes both in-home and out-of-home consumption)

Other estimatesa for the U.S. retail coffee market in 2002:

• In the home, specialty coffeeb was estimated to be a $3.2 billion business, of which Starbucks was estimated to have a 4% share.

• In the food-service channel, specialty coffee was estimated to be a $5 billion business, of which Starbucks was estimated to have a 5% share.

• In grocery stores, Starbucks was estimated to have a 7.3% share in the ground-coffee category and a 21.7% share in the whole-beans category.

• It was estimated that over the next several years, the overall retail market would grow less than 1% per annum, but growth in the specialty-coffee category would be strong, with compound annual growth rate (CAGR) of 9% to 10%.

• Starbucks’ U.S. business was projected to grow at a CAGR of approximately 20% top-line revenue growth.

Source: Adapted from company reports and Lehman Brothers, November 5, 2002.

aThe value of the retail coffee market was difficult to estimate given the highly fragmented and loosely monitored nature of the market (i.e., specialty coffeehouses, restaurants, delis, kiosks, street carts, grocery and convenience stores, vending machines, etc.).

bSpecialty coffee includes espresso, cappuccino, latte, café mocha, iced/ice-blended coffee, gourmet coffee (premium whole bean or ground), and blended coffee.

Specialty Coffee 27%

Traditiona l Coffee

73%

2000 – $21 billion

Specialty Coffee 31%

Tradition al Coffee

69%

2002E – $21.5 billion

Specialty Coffee 41%

Tradition al Coffee

59%

2005E – $22 billion

v) Starbuc ks Share

i) Starbuc ks Share

iii) Starbuc ks Share

For the exclusive use of D. Jaramillo-Escobar, 2019.

This document is authorized for use only by Daniela Jaramillo-Escobar in Marketing 300 Integrated Core taught by JOHN CAVALIERE, Montclair State University from Aug 2019 to Feb 2020.

Starbucks: Delivering Customer Service 504-016

Name: Date:

Target Corporation Case

Answers and Analysis

Target Corporation (Target) operates large general merchandise and food discount stores in all of the

United States, with the exception of Alaska Hawaii, and Vermont. The company also has its own credit

card operations and operates a fully integrated online business, target.com. Although the online portion of

target’s business is small relative to the overall size of target, sales are growing at a more rapid pace in the

online business compared to the in-store sales. The company’s philosophy is to offer their customers a

delightful shopping experience and their team members a preferred place to work, and to invest in the

communities in which target conducts business to improve quality of life. Selected information from the

2007 form 10-k of Target Corporation is on pages 228-237.

Required:

1. Analyze the firm’s financial statements and supplementary information. Your analysis should include

the preparation of common-size financial statements, key financial ratios, and an evaluation of

short-term liquidity, operating efficiency, capital structure and long-term solvency, profitability, and

market measures.

2. Identify the strengths and weaknesses of the company.

3. What is your opinion of the investment potential and the creditworthiness of Target Corporation?

Company Overview:

Target Corporation (Target or ‘the company’) operates large format general merchandise and food

discount stores in the US, which include Target and Super Target stores. The company offers both

everyday essentials and fashionable merchandise. Target is headquartered in Minneapolis, Minnesota

and employs 366,000 people. The company recorded revenues of $63,367 million in the fiscal year

ended January 2008, an increase of 6.5% over 2007. The operating profit of the company was $5,272

million in the fiscal year 2008, an increase of 4% over 2007. The net profit was $2,849 million in the

fiscal year 2008, an increase of 2.2% over 2007.

Target Corporation

Consolidated Balance Sheets and common-size Balance Sheets

(In millions, except share and per share date)

Period End Date

2008

02/02/2008

2007

02/03/2007

Assets

Cash and Short Term Investments 2,450.00 12.96% 813 5.53%

Cash & Equivalents 599 3.17% 813 5.53%

Short Term Investments 1,851.00 9.79% 0 0.00%

Total Receivables, Net 8,651.00 45.76% 6,757.00 45.95%

Accounts Receivable – Trade, Net 8,054.00 42.60% 6,194.00 42.12%

Accounts Receivable – Trade, Gross 8,624.00 45.62% 6,711.00 45.63%

Provision for Doubtful Accounts -570 -3.01% -517 -3.52%

Receivables – Other 597 3.16% 563 3.83%

Total Inventory 6,780.00 35.86% 6,254.00 42.53%

Prepaid Expenses 0 0.00% 0 0.00%

Other Current Assets, Total 1,025.00 5.42% 882 6.00%

Total Current Assets 18,906.00 100.00% 14,706.00 100.00%

Property/Plant/Equipment, Total – Net 24,095.00 127.45% 21,431.00 145.73%

Goodwill, Net 60 0.32% 60 0.41%

Intangibles, Net 148 0.78% 152 1.03%

Long Term Investments 0 0.00% 0 0.00%

Note Receivable – Long Term 0 0.00% 0 0.00%

Other Long Term Assets, Total 1,351.00 7.15% 1,000.00 6.80%

Other Assets, Total 0 0.00% 0 0.00%

Total Assets 44,560.00 235.69% 37,349.00 253.97%

Liabilities and Shareholders’ Equity

Accounts Payable 6,721.00 35.55% 6,575.00 44.71%

Payable/Accrued 0 0.00% 0 0.00%

Accrued Expenses 2,109.00 11.16% 2,004.00 13.63%

Notes Payable/Short Term Debt 0 0.00% 0 0.00%

Current Port. of LT Debt/Capital Leases 1,964.00 10.39% 1,362.00 9.26%

Other Current Liabilities, Total 988 5.23% 1,176.00 8.00%

Total Current Liabilities 11,782.00 62.32% 11,117.00 75.59%

Total Long Term Debt 15,126.00 80.01% 8,675.00 58.99%

Long Term Debt 15,126.00 80.01% 8,528.00 57.99%

Capital Lease Obligations 0 0.00% 147 1.00%

Deferred Income Tax 470 2.49% 577 3.92%

Other Liabilities, Total 1,875.00 9.92% 1,347.00 9.16%

Total Liabilities 29,253.00 154.73% 21,716.00 147.67%

Common Stock 68 0.36% 72 0.49%

Additional Paid-In Capital 2,656.00 14.05% 2,387.00 16.23%

Retained Earnings (Accumulated Deficit) 12,761.00 67.50% 13,417.00 91.23%

Other Equity, Total -178 -0.94% -243 -1.65%

Total Equity 15,307.00 80.96% 15,633.00 106.30%

Total Liabilities & Shareholders’ Equity 44,560.00 235.69% 37,349.00 253.97%

Target Corporation

Income statements and common-size Income statements

(In millions, except share and per share date)

Period End Date

2008

02/02/2008

2007

02/03/2007

2006

01/28/2006

Revenue 63,367.00 100.00% 59,490.00 100.00% 52,620.00 100.00%

Total Revenue 63,367.00 100.00% 59,490.00 100.00% 52,620.00 100.00%

Cost of Revenue, Total 43,766.00 69.07% 41,073.00 69.04% 35,703.00 67.85%

Gross Profit 19,601.00 30.93% 18,417.00 30.96% 16,917.00 32.15%

Selling/General/Administrative

Expenses, Total

12,670.00 19.99% 11,852.00 19.92% 11,185.00 21.26%

Research & Development 0 0.00% 0 0.00% 0 0.00%

Depreciation/Amortization 1,659.00 2.62% 1,496.00 2.51% 1,409.00 2.68%

Interest Expense (Income), Net

Operating

0 0.00% 0 0.00% 0 0.00%

Unusual Expense (Income) 0 0.00% 0 0.00% 0 0.00%

Other Operating Expenses, Total 0 0.00% 0 0.00% 0 0.00%

Operating Income 4,625.00 7.30% 4,497.00 7.56% 3,860.00 7.34%

Interest Income (Expense), Net

Non-Operating

0 0.00% 0 0.00% 0 0.00%

Gain (Loss) on Sale of Assets 0 0.00% 0 0.00% 0 0.00%

Other, Net 0 0.00% 0 0.00% 0 0.00%

Income Before Tax 4,625.00 7.30% 4,497.00 7.56% 3,860.00 7.34%

Income Tax – Total 1,776.00 2.80% 1,710.00 2.87% 1,452.00 2.76%

Income After Tax 2,849.00 4.50% 2,787.00 4.68% 2,408.00 4.58%

Minority Interest 0 0.00% 0 0.00% 0 0.00%

Equity In Affiliates 0 0.00% 0 0.00% 0 0.00%

U.S. GAAP Adjustment 0 0.00% 0 0.00% 0 0.00%

Net Income Before Extra. Items 2,849.00 4.50% 2,787.00 4.68% 2,408.00 4.58%

Total Extraordinary Items 0 0.00% 0 0.00% 0 0.00%

Discontinued Operations 0 0.00% 0 0.00% 0 0.00%

Net Income 2,849.00 4.50% 2,787.00 4.68% 2,408.00 4.58%

Target Corporation

Statements of cash flows

(In millions, except share and per share date)

Period End Date

2008

02/02/2008

2007

02/03/2007

2006

01/28/2006

Net Income/Starting Line 2,849.00 2,787.00 2,408.00

Depreciation/Depletion 1,659.00 1,496.00 1,409.00

Amortization 0 0 0

Deferred Taxes -70 -201 -122

Non-Cash Items 634 497 579

Discontinued Operations 0 0 0

Unusual Items 28 53 70

Other Non-Cash Items 606 444 509

Changes in Working Capital -947 283 177

Accounts Receivable -602 -226 -244

Inventories -525 -431 -454

Other Assets -38 -25 -52

Accounts Payable 111 435 489

Accrued Expenses 62 389 351

Taxes Payable 0 41 70

Other Liabilities 124 100 0

Other Operating Cash Flow -79 0 17

Cash from Operating Activities 4,125.00 4,862.00 4,451.00

Capital Expenditures -4,369.00 -3,928.00 -3,388.00

Purchase of Fixed Assets -4,369.00 -3,928.00 -3,388.00

Other Investing Cash Flow Items, Total -1,826.00 -765 -761

Sale of Fixed Assets 95 62 58

Other Investing Cash Flow -1,921.00 -827 -819

Cash from Investing Activities -6,195.00 -4,693.00 -4,149.00

Financing Cash Flow Items -375 -5 58

Other Financing Cash Flow -375 -5 58

Total Cash Dividends Paid -442 -380 -318

Issuance (Retirement) of Stock, Net -2,267.00 -720 -1,025.00

Issuance (Retirement) of Debt, Net 6,791.00 101 386

Cash from Financing Activities 3,707.00 -1,004.00 -899

Foreign Exchange Effects 0 0 0

Net Change in Cash 1,637.00 -835 -597

Net Cash – Beginning Balance 813 1,648.00 2,245.00

Net Cash – Ending Balance 2,450.00 813 1,648.00

Target Corporation

Key financial ratios

RMBS Trading Desk Strategy

Understanding Mortgage Dollar Rolls October 2, 2006

Sharad Chaudhary 212.583.8199 sharad.chaudhary@bankofamerica.com

RMBS Trading Desk Strategy

Ohmsatya Ravi 212.933.2006 ohmsatya.p.ravi@bankofamerica.com

Qumber Hassan 212.933.3308 qumber.hassan@bankofamerica.com

Sunil Yadav 212.847.6817 sunil.s.yadav@bankofamerica.com

Ankur Mehta 212.933.2950 ankur.mehta@bankofamerica.com

RMBS Trading Desk Modeling

ChunNip Lee 212.583.8040 chunnip.lee@bankofamerica.com

Marat Rvachev 212.847.6632 marat.rvachev@bankofamerica.com

Vipul Jain 212.933.3309 vipul.p.jain@bankofamerica.com

Dollar rolls serve as the primary channel for mortgage securities borrowing and lending in the MBS TBA market, and as such they play a critical role in providing liquidity to

the market. The borrowing rate associated with a dollar roll provides significant

information on technicals in the MBS market. A background in the conventions and

calculations associated with dollar rolls is thus critical to understanding relative value

in the MBS market. The primary goal of this primer is to provide this background.

Functionally speaking, dollar roll transactions are a form of securities lending and are closest in spirit to repurchase agreements (repos). In a typical repo, a lender agrees to

sell securities to a buyer in return for cash. At the termination of the transaction, the

securities are resold at a predetermined price plus an interest payment. A dollar roll is

analogous to a repurchase transaction except that the party borrowing the securities has

the flexibility of returning “substantially similar” securities, instead of the same ones.

In addition to providing financing opportunities for mortgage pass-through positions and enhancing liquidity in the TBA market, dollar rolls also serve other needs. Dollar

rolls are used to obtain collateral for CMO deals and TBA transactions, to avoid

operational issues associated with taking delivery of mortgage pools, to hedge specified

pool positions, and to express a view on prepayment speeds.

As financing transactions, dollar rolls can offer attractive borrowing rates and a considerable portion of this primer is devoted to explaining how to calculate the

implied financing rate associated with a particular dollar roll. The sensitivity of roll

pricing to different factors such as prepayment speeds and reinvestment rates is also

explored.

Historical data suggest that dollar rolls can offer financing advantages of as much as 15 bps to 100 bps versus 1-month LIBOR. This advantage should not be construed as a

“free lunch”, however, since it comes with risks attached. These risks include being

redelivered collateral with inferior prepayment characteristics relative to the original

securities that were delivered and prepayment risk. Credit risk and liquidity risk play a

relatively minor part in dollar rolls.

Our primer concludes with a “real life” example of how dollar rolls trade in practice. We review the history of roll levels on FNMA 6s and comment on the various factors

that led to the roll trading in a 2 to 15 tick range between 2003 and 2004.

This document is NOT a research report under U.S. law and is NOT a product of a fixed income research department. This document has been prepared for Qualified Institutional Buyers, sophisticated investors and market professionals only. To our U.K. clients: this communication has been produced by and for the primary benefit of a trading desk. As such, we do not hold out this piece of investment research (as defined by U.K. law) as being impartial in relation to the activities of this trading desk. Please see the important conflict disclosures that appear at the end of this report for information concerning the role of trading desk strategists.

RMBS Trading Desk Strategy

I. INTRODUCTION

As one of the most liquid fixed income sectors in the world, mortgage-backed securities

(MBS), particularly those backed by agency guarantees, constitute an active subsector of

global securities lending markets because of their high credit quality and liquidity. Securities

lending transactions in MBS are usually structured in one of two ways: as repurchase agreements (repos), or as dollar rolls. Repurchase agreements are securities lending transactions in which one party (the lender) agrees to sell securities to another party (the

borrower) in return for cash (or securities), with a simultaneous agreement to repurchase the

same securities at a specific price at a later date.1 At the termination of the transaction, the

securities are resold at the predetermined price plus an interest payment (calculated based on

a previously determined interest rate). A dollar roll is analogous to a repurchase transaction,

except that it provides the party borrowing the securities with additional flexibility of

returning “substantially similar” securities, instead of the same ones.2 In addition, unlike a

repo, the party borrowing the security owns the principal and interest payments generated

during the roll period.

The popularity of mortgage repos and dollar rolls can be gauged from the fact that parties to

these agreements span the entire range of institutional participants in the mortgage market:

broker-dealers, GSEs, banks, pension funds, hedge funds, insurance companies, mutual

funds and overseas investors. The needs and motivations of these participants vary from

transaction to transaction, depending upon whether they are acting as a borrower or lender of

securities (or sometimes even both). For example, a broker-dealer might borrow securities to

cover a short position, or simultaneously borrow and lend securities to earn a high rate on

the securities loaned versus the securities borrowed. To give a specific example relevant to

the mortgage sector, during times of heavy Agency CMO issuance activity the demand for

mortgage pools by broker-dealers can lead to very attractive financing rates for MBS holders

who choose to “roll” their securities to these dealers. The ability of MBS investors to “roll”

the pass-through securities they have in position lowers funding costs and can significantly

enhance returns on these positions versus other fixed-income alternatives.

As our discussion and examples suggest, dollar roll transactions play an important role in

enhancing liquidity in the MBS market by facilitating market making and the creation of

structured securities. Unfortunately, it is not easy to quantify this liquidity because no

official statistics exist for the volume of dollar roll transactions since they take place “over-

the-counter.” In other words, dollar rolls are privately negotiated transactions between two

parties that take place through a trading desk or an electronic trading system. However, data

from TradeWeb and the Bond Market Association suggest that dollar rolls constitute

1 Securities lending has been defined as the “temporary exchange of securities, generally for cash or other securities of at least an equivalent value, with an obligation to redeliver a like quantity of the same securities on a future date.” An excellent introduction to this subject can be found in Securities Lending Transactions: Market Developments and Implications, Technical Committee of the International Organization of Securities Commissions (IOSCO), Bank of International Settlements (BIS) Committee on Payment and Settlement Systems (CPSS), July 1999. 2 “Substantially similar” is defined in the American Institute of Certified Public Accountants’ Statement of Position 90-3 as meaning that the original and returned security should be of the same agency/program, original maturity, and coupon and both should satisfy “Good Delivery” requirements. Some of the requirements associated with “Good Delivery” are spelled out in The TBA Market section.

2

RMBS Trading Desk Strategy

anywhere from one-third to two-thirds of all “To Be Announced” (TBA) transactions.3

These numbers and information on Agency MBS trading volumes released by the Federal

Reserve Bank of New York lead us to an estimated daily volume of $20 billion in dollar roll

transactions in 2006. However, we should point out that a daily number is somewhat

misleading in this context since most roll activity takes place two to five days before TBA

settlement dates. Turning to the composition of participants in these transactions, based on

trading desk flows, we estimate that money managers and U.S. banks account for nearly

50% of dollar roll activity, with the other half attributable to the GSEs, mortgage servicers,

insurance companies and overseas investors.

The primary goal of this primer is to describe and discuss the conventions and

calculations associated with dollar roll transactions. Consequently, we begin by

sketching some of the key conventions associated with the MBS TBA market, which is

the market where these transactions take place. The next section explains some of the

basic aspects of dollar roll transactions and provides practical examples of how and

why these agreements are executed by market participants. From there, we delve into

the nuts and bolts of calculating the all-important financing rate associated with these

transactions. This is followed by a sensitivity analysis that demonstrates how this

financing rate depends upon prepayment assumptions and the reinvestment rate. The

sensitivity analysis provides a natural segue into a more detailed discussion of the risks

associated with the dollar roll—with redelivery and prepayment risks being the crucial

ones. The paper concludes with a review of recent trading activity in the rolls associated

with a specific coupon to provide readers with some “real-life” background for the

dollar roll market.

3 TradeWeb is a multi-dealer electronic auction system that links fixed-income securities dealers with buy-side institutions. Besides TBA-MBS, other products traded on the platform include US Treasuries, agencies and several other types of debt securities. The Bond Market Association is a private trade association that represents security firms and banks that underwrite, distribute, and trade debt securities. The Association’s membership includes all major dealers.

3

RMBS Trading Desk Strategy

II. THE TBA MARKET

As a preface to an explanation of the dollar roll market, it is useful to review some

features of the “To Be Announced” (TBA) market for mortgage pass-throughs that play

a particularly important role in dollar roll transactions. The TBA nomenclature captures

two important features of how this market operates:

The market is a forward delivery market with participants entering into forward contracts to buy or sell MBS on a monthly settlement schedule set by the Bond Market

Association. A settlement date is typically 30 days into the future but can be as much

as two to six months in the future. The current settlement month is known as the front month and the settlement months following the front month are typically known as back months. The actual numbers of pools and pool numbers are not known at the initiation of the forward contract but are “announced” on 48-hour day: two business

days before the agreed settlement date, prior to 3pm EST. Other general trade

parameters such as the agency type, price, coupon and par amount are known at the

time the trade is executed.

The pools delivered by the seller on the settlement date have to satisfy Good Delivery Guidelines. These are also established by the Bond Market Association and detail the specifics associated with confirming and settling MBS. For example, the guidelines

govern the maximum number of pools per lot and maximum allowable variance

between the face amount of the pools delivered and the agreed-upon face amount.

Current standards for Good Delivery include a maximum of three pools per lot and a

maximum variance of 0.01% per lot. In other words, for a $1 million lot, the sum of

current face amounts of the pools should be within 0.01% of $1 million, or between

$999,900 and $1,000,100.

Settlement fails on TBA transactions turn out to play an important role in the economics of dollar rolls. In a TBA transaction, the seller does have the option to fail to

deliver the securities to the buyer. In this situation, the buyer does not have to pay the

seller until the securities are delivered. The price of the security, including the accrued

interest that is to be paid, does not change. In essence, then, the buyer is compensated

for the fail by reinvestment income on the money that was to be paid for the pools.

4

RMBS Trading Desk Strategy

III. DOLLAR ROLLS The close tie between the dollar roll and TBA markets arises from the fact that the process of

trading rolls is indistinguishable from that of trading TBAs from the perspective of a pass-

through trading desk. To understand why this is the case, recall the definition of a dollar roll

again: these are essentially repurchase agreements in which the seller of securities is

obligated to repurchase securities that are substantially similar to, but not identical with, the

securities originally sold. It is useful to further unpack this definition a little and think of a

dollar roll as a combination of two simultaneous transactions: a buy and a sell order for a

mortgage pass-through security for two different settlement dates. By convention, the Seller of roll is the party who sells mortgage securities for a settlement month (the front month) and agrees to buy back “substantially similar” securities for a future settlement month (the

back month). On the other side of the transaction, the Buyer of roll is the party who buys securities for the front month and agrees to sell “substantially similar” securities for the back

month. The equivalence of the dollar roll and TBA markets becomes clear when we realize

that the roll buyer’s commitment to return securities in the second leg of the dollar roll can

be hedged by getting long TBAs for the future settlement month.

Pricing on a dollar roll transaction can be requested from a pass-through desk by specifying

the size of the MBS position to be rolled (based on original face amount), long/short,

agency, maturity, coupon, the month the position is expected to settle, and the future

settlement month that the position should be rolled to. The roll price (also known as the drop) is equal to the difference between the purchase and sale prices of the mortgage security on the two settlement dates and is expressed in ticks. For example, a trading desk

might tell a customer: I can offer you $500 million of the FNMA 6 Sep/Oct roll at 11+.

As an intuition building exercise, it is useful to review some concrete examples of roll

transactions.

Investor A is long FNMA 6s for October delivery. At some time prior to October settlement day, investor A decides it does not want to have a forward position on

FNMA 6s for October settle and rolls the position forward to November. The decision

to roll could be based on an attractive drop between the October and November prices,

or operational issues such as not being willing or able to take delivery of the security

in October. By selling the roll, Investor A remains invested in mortgages and can

potentially also lock in an attractive financing rate on the funds obtained during the

roll period.

Mechanically, Investor A sells FNMA 6s for October settlement and purchases FNMA

6s for November settlement. The only transfer that takes place on the October

settlement date is a cash transfer that reflects the difference between the price of

FNMA 6s that the investor agreed to pay when they purchased them and the front

month price of FNMA 6s when they rolled these bonds (both prices are for October

settlement).

Investor B holds a collection of FNMA 6 pools and decides to lend these pools to the pass-through trading desk of broker-dealer X for October delivery in return for

“substantially similar” FNMA 6s pools in November. Investor B is motivated to do

this transaction by the fact that it can obtain attractive financing rates over the

October-November roll period. Based on recent history, the comparative advantage of

5

RMBS Trading Desk Strategy

financing rates obtained through dollar rolls can be anywhere from 15 bps to 100 bps

over 1-month LIBOR.

Mechanically, investor B sells FNMA 6s from its inventory to pass-through desk X for

October settle and buys FNMA 6s for November settle. The FNMA 6s held in Investor

B’s inventory are transferred from its inventory to the inventory of the pass-through

trading desk.

Investor B delivers FNMA 6s to pass-through desk X for October settle. At the same time, it enters into a forward commitment to purchase FNMA 6s for November settle.

Before November settlement, Investor B and desk X agree to roll the transaction to

December. Desk X sells the FNMA 6s for October settle to investor C while

simultaneously agreeing to buy FNMA 6s for November settle with investor D. On

being notified of Investor B’s intention to roll their 6s from November to December,

Desk X sells the November/December FNMA 6 roll to Desk Y.

This sequence of transactions demonstrates how the dollar roll market adds liquidity to

the pass-through market by providing an outlet for market making pass-through trading

desks to maintain a neutral position in pass-throughs or go long or short.

The Agency CMO trading desk of broker-dealer X can use the dollar roll market both for taking delivery of the collateral required for issuing a CMO and for hedging the

collateral that they have in inventory for soon-to-be-issued CMOs. Normally, CMO

deals are collateralized by current coupon pools, but these pools can be hard to find

because of the lag between current interest rates and the time when the corresponding

pass-through pools are actually created.4 A supply crunch in CMO deal collateral

impacts the roll market though increased demand for collateral in the front month. This

bids up prices for front-month settlement TBAs relative to back-month TBAs and

increases the drop in the roll market. In addition, CMO deals are frequently done using

“story collateral” i.e., collateral with desirable prepayment characteristics. In this case,

the roll may be bid up to induce the owners of the bonds of these desirable

characteristics to either deliver them in to TBA or to sell them outright.

This list of transactions does not exhaust all the possible uses dollar rolls can be put to. For

example, rolls can be used for hedging a specified pool position: when rolls are trading rich, the attractive characteristics of specified pools are worth less since the cost of hedging

specified pools with TBAs increases (or the carry advantage of specified pools is reduced).5

Finally, sophisticated investors can use the roll market to express a view on prepayment

rates on different coupons. As discussed later in this paper, it is sometimes beneficial to

maintain ownership of pass-throughs rather than roll them and vice versa, depending on the

prepayment rate realized in the front month of the roll, and whether the pass-through is a

discount or a premium security.

4 It typically takes four to eight weeks to close a mortgage. 5 Specified pool trades take place outside the TBA market and are executed by specifically identifying certain pools that an investor wishes to acquire. In general, investors “pay up” for these specified pools—for example, for the prepayment protection offered by a premium pool with a low average loan size or for the extension protection offered by slightly seasoned discount pools relative to TBA discount pools.

6

RMBS Trading Desk Strategy

Comparing Mortgage Dollar Rolls and Repos

To round up our introductory review of dollar roll transactions, it is instructive to compare a

dollar roll to a mortgage repurchase transaction (also known as a repo):

All the cash flows generated during the roll period belong to the buyer of the dollar roll. Thus, for example, if a record date falls during the term of the roll, the principal and

interest for the front month are paid to the buyer of the dollar roll.6 In a repo

transaction, principal and interest go to the original owner.

Securities “substantially similar” to the original ones borrowed can be returned in a dollar roll. In a repo, the ownership of the security is not transferred and the same

security needs to be returned.

There is considerable flexibility in terms of setting the length of repo transaction, with typical maturities ranging from one to 30 days. The settlement dates for dollar rolls

typically match TBA settlement dates.

The collateral backing a repo transaction can consist of small collections of pools, agency CMOs and non-agency CMOs, while dollar rolls trade in the TBA market. The

perceived fungibility or interchangeability of mortgage pools plays a crucial role in

terms of being able to perform roll transactions.

The flexibility to deliver substantially similar collateral in a dollar roll makes dollar roll

transactions far more common than mortgage repos. In general, mortgage repos are

collateralized by structured securities or specified pools. From both an operational and a

financing advantage perspective, it usually makes more sense to roll pass-throughs than to

repo them.

6 The record date is the date for determining the registered owner of the next scheduled payment of principal and interest for the mortgage security. The record date for MBS is typically the last business day of the month.

7

RMBS Trading Desk Strategy

III. PRICING DOLLAR ROLLS Dollar Roll Calculations

Let’s review the definition of a dollar roll once again: a dollar roll is a combination of two

simultaneous transactions: a buy and a sell order for a mortgage pass-through security for

two different settlement dates. Thus, a natural entry-point for thinking about the value of

dollar rolls is to analyze the economics of holding onto pass-through securities versus

“rolling” them. To this end, Figure 1 itemizes the cash flows associated with the two

alternatives. It would be more advantageous to roll securities versus holding them if the cash

flows in the “Financing” box of the figure added up to more than the cash flows in the

“Holding” box.

The forward repurchase price plays a crucial role in determining which of the two

alternatives presented in the figure is more advantageous. If this were a repo transaction, the

repurchase price would equal the initial sale price, plus an interest payment at a previously

determined interest rate (the repo rate). The repo rate is typically lower than prevailing short-

term rates since the repo transaction is a form of collateralized borrowing. Thus, the lender

of securities (the borrower of money) benefits because even though they are paying more to

repurchase their securities, they are (typically) able to invest the funds received from the

initial sale of securities at a higher rate than their borrowing rate (the repo rate).

A dollar roll introduces one crucial twist to the above situation. Since the principal and interest earned on the security over the roll period belong to the lender of cash (the roll

buyer), in an upwardly sloping yield curve environment, the forward price is usually

lower than the initial price to compensate the seller of the roll for losing one month of

positive carry on the position.7 Other than that, the situation is exactly analogous to the repo agreement discussed above. After taking the lost carry into account, the forward

price is set at a level that defines a financing rate for the transaction. The security owner

then compares this financing rate to their other alternatives to determine whether to

hold or “roll” their securities.

Figure 1. Cash Flows Associated with Rolling versus Holding Pass-throughs

Cash Flows of the Roll (Financing the Mortgage Securities) + Sale of Security at initial price (plus accrued interest) + Reinvestment income on sale proceeds – Repurchase of security at the forward price (plus accrued interest)

Cash Flows From Holding Mortgage Securities + Coupon Payment + Scheduled payment of principal + Prepaid principal + Reinvestment income on interest and principal received during roll period – Discount on interest and principal received after the roll period

Source: Banc of America Securities

7 The carry on a fixed income investment is usually defined as the interest income on the position less the cost of financing it. In case of dollar rolls, an investor needs to also consider the premium/discount received on pay-downs. Note also that the forward

price (back-month settlement price) need not be lower than the initial price (front-month settlement price) even when the yield

curve is upwardly sloping if prepayment speeds on premium coupons are very fast.

8

RMBS Trading Desk Strategy

A worked example illuminates several of these issues. Recall that the roll price, which is referred to as the drop in market parlance, is equal to the difference between the purchase and sale prices of the mortgage security on the two settlement dates. An investor who owns

30-year FNMA 6s enters into a dollar roll transaction by selling $1 million of the

September/October 2004 roll on 30-year FNMA 6s. As of August 27th 2004, FNMA TBA 6s

for September settlement were trading at 103-14 and the drop for the September/October roll

was 11.8 ticks. Thus, the investor is committing to simultaneously selling $1 million FNMA

6s for September settlement at 103-14 and buying $1 million FNMA 6s for October

settlement at a price of 103-2.2.

In addition, we make the following assumptions:8

The 1-month reinvestment rate over the roll period for the investor is equal to 1.77%; TBA FNMA 6s have a WAC of 6.50% and a WAM of 356 months; The expected prepayment speed for this collateral in September is 28.2% CPR; The payment date for September principal and interest payments is October 25, 2004

(following FNMA’s stated delay of 54 days).

Based on these assumptions, cash flows from the roll transaction are shown in Figure 2. The

seller of the roll delivers $1,000,000 face value of FNMA 6s and receives $1,036,708.33 on

the front- month settlement date (September 15). The seller then reinvests these proceeds at

1.77% until October 14th (the back month settlement date) and obtains $1,038,186.51. To

facilitate the comparison to the buy-and-hold situation, we assume that what the seller buys

back is the amount left after a total principal pay-down of $28,131.44.9

Figure 2. Cash Flows from Rolling FNMA 6s

Date Transaction Cash Flow 15-Sep Sell $ 1,000,000 FNCL 6s @ 103-14 $1,034,375.00

Plus 14 days accrued interest (6% * 14/360 * 1,000,000) $2,333.33

$1,036,708.33

Reinvest proceeds @ 1.77%

29 actual days/360 * 1.77% * 1,036,708.33 $1,478.17

$1,038,186.51

14-Oct Purchase $ 971,868.55 FNCL 6s @ 103-2.2 -$1,001,692.77

Plus 13 days accrued interest (6% * 13/360 * 971,868.55) -$2,105.72

-$1,003,798.49

Net Proceeds from the Roll $34,388.02

Source: Banc of America Securities

Now, consider the situation in Figure 3 where the investor holds on to the $1 million FNMA

8 Readers with access to Bloomberg can verify the results obtained in Figures 2 through 4 by using the assumptions listed below and a roll calculator such as Bloomberg’s Roll Analysis function (FNCL 6 <Mtge> RA <Go>). 9 The total principal pay-down includes both the scheduled and unscheduled principal payments. The unscheduled principal payment assumes a prepayment rate of 28.2% CPR in September.

9

RMBS Trading Desk Strategy

6s instead of rolling these bonds. In this case, they would own $971,868.55 face of FNMA

6s on October 14 because of principal pay-downs. Also, because the investor held the MBS

as of the September record date, they are entitled to the September principal and interest

payments on the pools. These payments will be received on the payment date of October 25.

Specifically, as Figure 3 shows, a total cash flow of $33,131.44 would be received on this

date consisting of an interest payment of $5,000 and a principal payment of $28,131.44. To

make the situation comparable to Figure 2, we need to discount these cash flows back to

October 14. The principal and interest payments received on October 25 are worth

$33,113.53 on October 14 when discounted at an annualized reinvestment rate of 1.77%.

Figure 3. Cash Flows from Holding FNMA 6s

Date Transaction Cash Flow 25-Oct Receive Principal and Interest on FNMA 6s

Receive Interest (6% * 30/360 * 1,000,000) $5,000.00

Receive Principal:

Scheduled Principal $927.14

Prepaid Principal (@28.2% CPR) $27,204.30

$33,131.44

Net Proceeds from Holding FNMA 6s (as of Oct 14) $33,113.53 Present value of Principal and Interest Cash Flow Rcvd on Oct 25

33,131.44/(1 + (1.77% * 11/360))

Cash Value of Rolling vs Holding FNMA 6s $1,274.48

Source: Banc of America Securities

Thus, the owner of the 6s earns an additional $1,274.48 ($34,388.02 – $33,113.53) during

the roll period by rolling their securities instead of holding onto them. In this specific

example then, it is more advantageous to roll.

Calculating the Implied Financing Rate of a Roll

The drop of a dollar roll along with the principal and interest cash flows received over the

roll period together imply a certain financing rate for the seller of the roll. Figure 4 shows

how to calculate this rate. The logic of the calculation goes as follows: the situation in Figure

2 is preferable to Figure 3 because the financing rate being offered to the seller of the roll is

less than the prevailing reinvestment rate. For a given drop, the advantage in Figure 2

decreases as the reinvestment rate goes down and the “breakeven” point occurs when it is

equal to the implied financing rate. At this point, the drop simply reflects the lost carry at

prevailing reinvestment rates. Thus, to find the implied financing rate, we need to solve for

the reinvestment rate in Figure 2, which equates the cash flows in this situation to Figure 3.

10

RMBS Trading Desk Strategy

Figure 4. Calculating the Implied Financing Rate for Rolling FNMA 6s

Date Transaction Cash Flow 15-Sep Sell $ 1,000,000 FNCL 6s @ 103-14 $1,034,375.00

Plus 14 days accrued interest (6% * 14/360 * 1,000,000) $2,333.33

$1,036,708.33

r = implied financing rate Reinvest proceeds @ r%

29 actual days/360 * r% * 1,036,708.33 $1,036,708.33 * 29/360 * r%

$1,036,708.33 * (1 + 29/360 * r%)

14-Oct Purchase $ 971,868.55 FNCL 6s @ 103-2.2 $1,001,692.77

Plus 13 days accrued interest (6% * 13/360 * 971,868.55) $2,105.72

$1,003,798.49

Present value of Principal and Interest Cash Flow Rcvd on Oct 25 $33,113.53

$1,036,912.02

To calculate the implied financing rate, solve for r in the equation: $1,036,708.33 * (1 + 29/360 * r%) = $1,036,912.02

r = 0.24% Source: Banc of America Securities

Calculating the Breakeven Drop of a Roll

The breakeven drop is the difference between the front and back month TBA settlement prices such that the implied financing rate is equal to prevailing short-term reinvestment

rates. We typically take the short-term rate to be 1-month LIBOR. Figure 5 lists the

computations involved in calculating the breakeven drop for the roll on FNMA 6s. The

figure shows that the breakeven drop is equal to 7.6 ticks, which is 4.2 ticks less than the roll

price of 11.8 ticks. Thus, the investor earns 4.2 ticks of additional carry by rolling their

mortgage pools instead of simply holding onto them. The net advantage of rolling versus

holding tends to zero as the difference between the quoted drop and the breakeven drop

decreases.

Figure 5. Calculating the Breakeven Drop for Rolling FNMA 6s

Date Transaction Cash Flow 15-Sep Sell $ 1,000,000 FNCL 6s @ 103-14 $1,034,375.00

Plus 14 days accrued interest (6% * 14/360 * 1,000,000) $2,333.33

$1,036,708.33

Reinvest proceeds @ 1.77%

29 actual days/360 * 1.77% * 1,036,708.33 $1,478.17

$1,038,186.51

14-Oct X = breakeven drop Purchase $ 971,868.55 FNCL 6s @ 103-14 – X $971,868.55 * (103-14 – X)%

Plus 13 days accrued interest (6% * 13/360 * 971,868.55) $2,105.72

$2,105.72 + $971,868.55 * (103-14 – X)%

Present value of Principal and Interest Cash Flow Rcvd on Oct 25 $33,113.53

$35,219.25 + $971,868.55 * (103-14 – X)%

To calculate the breakeven drop, solve for X in the equation: $1,038,186.51 = $35,219.25 + $971,868.55 * (103-14 – X)%

X = 7.6 ticks

Source: Banc of America Securities

11

RMBS Trading Desk Strategy

Roll “Specialness”

We say a roll is special, when the roll price (drop) is higher than the breakeven drop. This is same as saying that the implied financing rate is less than alternative money market rates. As

mentioned before, “special” rolls are created in times of heavy refinancing and CMO

production activity, and can contribute significantly to the carry on a mortgage position.

Dollar rolls trading special can have a significant impact on pay-ups for specified pools and

on the attractiveness of the coupon with the special roll versus other coupons in the coupon

stack. The attractiveness of “specified” characteristics decreases as the “specialness” of the

roll increases because the carry advantage of TBAs from special rolls reduces the value of

specified pool characteristics. The same logic applies to the relative attractiveness of

different coupons across the coupon stack – in general, TBA pass-through coupons with rich

rolls tend to trade at rich price levels relative to other coupons because of their carry

advantage in the roll market.

Roll “Trading-at-Fail”

When the implied financing rate is zero, the roll is said to be trading at fail. Why would

someone offer to lend money at a 0% interest rate? This can occur when a mortgage trading

desk or some investors are short a coupon and cannot find the collateral necessary for

satisfying TBA delivery requirements or if the CMO desk urgently needs mortgage

collateral to complete a planned issuance.

If a trading desk fails to meet TBA delivery requirements, the TBA buyer effectively

finances their purchase at a 0% financing rate. As we pointed out in the TBA section of the

paper, this is because the TBA buyer does not have to pay for the purchase until the

mortgage pools are delivered. Theoretically, the maximum value for a dollar roll drop should

not exceed the cost of failing to deliver the underlying security for the roll period. However,

strong demand for collateral from CMO desks or stringent requirements for failing to meet

TBA delivery requirements occasionally lead to rolls trading through fail levels close to

TBA settlement dates.

An interesting point about dollar rolls trading through the fail levels should be noted here.

There will be some scenarios in which the collateral that is likely to be delivered for back-

month TBAs will be substantially worse than the collateral delivered for the front-month.

This usually occurs when new production in a coupon changes the characteristics of TBA

delivered collateral by so much that an investor is better off by not rolling their bonds even

when the roll is trading through fail levels. It is worth pointing out that standard roll

calculations assume that the collateral delivered for back-month is same as the collateral

delivered for front month plus one month of additional aging.

12

RMBS Trading Desk Strategy

IV. SENSITIVITY ANALYSIS OF DOLLAR ROLLS

Our analysis in the previous section can be distilled into a few simple observations. First, the

value of a dollar roll is the difference between the actual drop and the breakeven drop on the

underlying security. Whenever the actual roll price is greater than the breakeven drop, or

conversely, whenever the available reinvestment rate exceeds the implied financing rate,

rolling a security may be preferable to holding it.10 The calculations presented in Figures 2 through 5 should also make it clear that in addition to the actual drop, the value of the roll

also depends upon the prevailing reinvestment rate and the prepayment speed over the front

month of the roll. In what follows, we explore how the implied financing rate for a roll

transaction changes as a function of the roll price and prepayment speeds. We also explore

the relationship between the breakeven drop, prepayment speeds and the reinvestment rate.

Sensitivity Analysis of the Implied Financing Rate and the Breakeven Drop

Figure 6 shows the implied financing rate for the roll transaction we have been discussing

above as a function of the roll price and prepayment speeds. Moving along a row of the

figure gives us a feeling for how the breakeven financing rate changes as a function of the

drop assuming a fixed prepayment speed. Similarly, moving along a column tells us how the

breakeven rate changes as a function of prepayment speeds assuming a fixed drop.

Not surprisingly in light of our analysis, for a given prepayment rate, the breakeven

financing rate increases as the drop decreases and vice-versa. For a fixed drop, the figure

also shows that the implied financing rate decreases as prepayment speeds increase. The

crucial point here is that the 6s are a premium security and as such faster prepayment speeds

decrease the value of the captured carry for the buyer of the roll. By keeping the drop fixed,

they are lending money to the seller of the roll at more attractive rates.

Figure 6. Breakeven Finance Rate Sensitivity on FNMA 6s

CPR (%) -14.8 -13.8 -12.8 -11.8 -10.8 -9.8 -8.8 Range 13.2 -0.23 0.14 0.51 0.88 1.25 1.62 1.99 2.22

18.2 -0.42 -0.05 0.32 0.69 1.05 1.42 1.79 2.21

23.2 -0.62 -0.25 0.12 0.48 0.85 1.21 1.58 2.20

28.2 -0.83 -0.47 -0.10 0.26 0.63 0.99 1.35 2.18

33.2 -1.06 -0.69 -0.33 0.03 0.39 0.75 1.11 2.17

38.2 -1.30 -0.94 -0.58 -0.22 0.14 0.50 0.86 2.16

43.2 -1.56 -1.20 -0.85 -0.49 -0.13 0.22 0.58 2.14

Range 1.33 1.34 1.36 1.37 1.38 1.40 1.41

Drop (32nds)

Source: Banc of America Securities

Conversely, as shown in Figure 7, the breakeven drop decreases as prepayment speeds and

reinvestment rates increase. As before, faster prepayment speeds on premium securities

render them more unattractive for holding. This increases the value of a roll at a given

reinvestment rate by lowering the breakeven drop.

10 The point is that there are some risks associated with dollar rolls. These risks are detailed in the next section.

13

RMBS Trading Desk Strategy

Figure 7. Breakeven Drop Sensitivity on FNMA 6s

CPR (%) 0.27% 0.77% 1.27% 1.77% 2.27% 2.77% 3.27% Range 13.2 13.3 12.0 10.6 9.2 7.9 6.5 5.2 -8.2

18.2 12.8 11.5 10.1 8.7 7.4 6.0 4.6 -8.2

23.2 12.3 10.9 9.6 8.2 6.8 5.4 4.0 -8.3

28.2 11.8 10.4 9.0 7.6 6.2 4.8 3.4 -8.3

33.2 11.2 9.8 8.4 7.0 5.6 4.2 2.8 -8.4

38.2 10.5 9.1 7.7 6.3 4.9 3.5 2.0 -8.5

43.2 9.8 8.4 7.0 5.6 4.1 2.7 1.3 -8.6

Range 3.5 3.6 3.6 3.7 3.8 3.8 3.9

Reinvestment Rate

Source: Banc of America Securities

How do these results change when we look at discount securities? In this situation, an

increase in the prepayment rate at a given roll price increases the implied financing rate. In addition, the breakeven drop increases with an increase in prepayment speeds. The point is that faster prepayment speeds on discounts render them more attractive for holding and this

decreases the value of a roll at a given drop. This is in contrast with the roll example on

premium 6s discussed above – there a rise in prepayment speeds at a given drop decreased

implied financing rates. However, the relationship between the drop and the implied

financing rate remains unchanged – the implied financing rate decreases with an increase in

the drop for both premium and discount pass-throughs.

14

RMBS Trading Desk Strategy

V. RISKS ASSOCIATED WITH DOLLAR ROLLS

The dollar roll market has historically offered as much as 15 bps to 100 bps of a financing

advantage versus the 1-month LIBOR rate. However, as Chicago economists have taught us

over the past several decades, there is no such thing as a “free lunch” in financial markets, or

at least the cafeteria that offers one is not open for any length of time. We can categorize

some of the risks assumed by the seller of a dollar roll transaction as follows.

Redelivery Risk. While the mortgage pass-through pools returned in a dollar roll transaction will have to meet good delivery requirements, these requirements do not say anything about

the risk characteristics of these pools. In particular, these pools can and frequently do have

less desirable prepayment characteristics than the originally delivered pools – the seller

receives faster prepaying pools for rolls on premium passthroughs and slower prepaying

pools on discounts. This type of adverse selection, therefore, makes dollar rolls an unattractive financing option for the holders of vintage and specified pools unless the

characteristics of the pools to be returned are agreed on beforehand or the roll is rich enough

to compensate for the value of the collateral characteristics lost in the transaction.

Prepayment Risk. At the time of entering into the roll transaction, neither party to the transaction knows what the actual prepayment speed during the front month of the roll is

going to be. The value of rolling versus holding a premium security is increased by a faster

prepayment speeds and reduced by a slower prepayment speeds in the front month of the roll

(see the discussion accompanying Figures 6 and 7). For a discount security, the relationship

between the roll’s value and realized prepayment speeds is opposite.

Credit Risk. The risk of one of the parties to a dollar roll defaulting during the roll period is called credit risk. The risk is mitigated to a large extent by the fact that either party could sell

or buy the security in the market in case of a default. However, the risk of adverse market

movements during this period could make this transaction a little painful. The lack of a

haircut in a dollar roll further aggravates the credit risk problem.

Liquidity Risk. The inability of a one party to deliver securities to the other on the settlement date is defined as liquidity risk. This may arise in case of a market disruption such

as a squeeze. This is different from credit risk since the delinquent party will typically be

able to settle the transaction at a later date.

Our catalog of risks thus suggests that some due diligence is required before we can

conclude a particular dollar roll will “enhance” the carry on an MBS position. In particular,

the redelivery risk by itself suggests that roll prices should trade somewhat above carry. The

expected prepayment characteristics of the redelivered pool need to be considered in

conducting the analysis of whether it is better to hold or roll.

15

RMBS Trading Desk Strategy

VI. A BRIEF HISTORY OF THE FNMA 6s ROLL

We conclude the primer by taking a look at how rolls actually trade in practice. Figure 8

illustrates roll prices on 30-year FNMA 6s from the beginning of 2003 to October, 2004.

The convention we use for the rolls associated with these prices is most easily understood by

looking at a specific month. In particular, suppose we look at roll values in the month of

August. Starting from the first day of the month, these values denote levels for the

August/September roll up until the pool notification day for August settlement, and refer to

levels for the September/October roll from the next day onward.

Turning our attention back to Figure 8, notice the dramatic changes in roll levels over the

observation period: the roll traded between a range of 2 to 15 ticks over the period in

question, reaching its lowest point in mid-June 2003 and its highest point in mid-May 2004.

What is behind these changes in roll prices? The figure clearly shows that there is a strong

correlation between interest rates as represented by the MBS current coupon rates and the

FNMA 6s roll. On average, roll prices were higher when interest rates were high and lower

when interest rates were low.11 Over the first half of 2003, the 6s roll gradually declined and

reached its lowest point of two ticks in mid-June (this reflects roll prices for the July/August

roll). This decrease on the premium 6s was largely due to the increase in refinancing activity

that resulted from rates decreasing, culminating in multi-year lows in mortgage rates in June

2003. At this juncture, 30-year 6s were prepaying at 60%–80% CPR. With the expectation

of suffering substantial losses from prepayment of principal on a premium security at par,

roll buyers were willing to pay very little to own 6s collateral from one settlement date to the

next settlement date. Notice how the 6s roll reached another local minimum in early March

of 2004 as interest rates rallied and renewed concerns of heightened prepayment rates.