Competing for Customers and Capital

A Mexican Standoff

This is my final article on the exchange wars. If you missed the first two you can check them out at "Exchange Wars: ICE vs. CME" and "Exchange Wars: Play It Again Sam." A series of comments about these posts that I exchanged with a CME trader provide useful background. Bottom line? My analysis leads to a Mexican standoff. Both merger combinations, ICEBOT as well as CMEBOT, have their strengths and weaknesses. But, it's important to note that BOT shareholders win no matter what.

You also will discover an unexpected new approach to stock valuation. In this approach the impact of market growth, maximum earnings and enterprise marketing productivity are just as important in assessing the quality of a merger as are cost and revenue synergies.

A CLEAN CASE
In several ways, the bidding war between CME and ICE for BOT offers an ideal market in which to document the implications of my theory. Why? For one thing, futures exchanges have virtually zero cost of goods sold. What's the marginal cost of an electronic trade made in 100 milliseconds? Virtually zero. Which means overhead costs cannot be loaded onto trading costs.  In fact, ICE reports its sales revenues are equal to its gross profit. Also, there are no third-party resellers sandwiched between the customer and the supplier. Finally, the enterprise marketing costs of futures exchanges are enumerated as line items in their income statements. In short, it’s a clean case.

A TOP DOWN APPROACH
The first thing you'll notice about this new approach to stock valuation is it begins with market demand, not company sales. The tail-wind factor-- growth in strategic group revenues -- creates a favorable environment for company sales growth. In Step 1 you forecast Strategic Group Revenues.

In Step 2 you calculate each company's maximum earnings market share using incremental analysis. This is a sophisticated form of strategic benchmarking.  It equates the incremental cost of a market share point with incremental earnings after enterprise marketing expenses.  This is the point where earnings are maximized. For details see my audio slide show on "The Rule of Maximum Earnings." This Adobe Connect presentation runs about 14 minutes. [Click to enlarge the graphics]

In Step 3 you calculate each company's earnings residual at that share of revenue where they are maximized.  This is important. You don't begin with forecast company sales. Rather you begin with forecast strategic group sales and estimate company sales as a product of its maximum earnings market share.

In Step 4 you calculate each company’s Risk-Adjusted Differential (RAD) and Relative Earnings Productivity (REP) for the most recent period. RAD is the difference between a company's share of market value and share of sales revenue, adjusted for volatility. See "Y'all Buckle That Seatbelt." This slide show runs 18 minutes. REP is the difference between a company's actual and maximum earnings. For details see "Competitive Stock Valuation." It runs 13 minutes.

Step 5 you forecast of the strategic group’s combined shareholder value and each company’s share of that value. Finally, given the number of shares outstanding at the end of the forecast period, the present value of lowball, most likely and highball stock prices are calculated.

ICEBOT's PRICE JUNE 2008
On the last day of trading in March, 2007 the common stocks of BOT, CME and ICE closed at $181.50, $532.46 and $122.21 respectively. Assuming investors built into these prices their expectations on the value of a merger, the price of ICEBOT was $303.71. The market cap of ICEBOT was $18.0 billion. The number of shares outstanding, adjusted for the merger, was 59.32 million.

If ICEBOT maximized earnings and its rejected suitor responded as I hypothesize below, my lowball forecast (minus 2 sigma) of the present value of its stock in June 2008 is $335. The present value of the expected price is $384 and my highball estimate (plus 2 sigma) has a present value of $430 per share ... assuming no changes in the shares outstanding. The next graphic shows the details in the five step competitive stock valuation process.

Step 1. Strategic group revenues were $729 million on March 31, 2007. That's the sum of each company's sales in Q1-07. I applied a rolling same-quarter year-on growth rate to forecast group revenues. Using the historical rate of 38%, the result in Q2-08 is $1,008 million in group revenues. Note, in this graphic the historical numbers are green and the forecast numbers are yellow.

Step 2. Sales revenues of ICEBOT divided by group revenues gives the merged companies a market share of 51.2% in March, 2007. If ICEBOT maximized earnings over the ensuing five quarters, its market share in June, 2008 would be 56.2%. Achieving this share would require doubling enterprise marketing expenses from $121 to $248 million. This would generate ICEBOT revenues of $567 million: a 52% increase over March, 2007 revenues of $373 million.

Built into this forecast is this assumption: if CME lost its bid management would not take it lying down. Rather, they would attempt to minimize their loss of market share by increasing enterprise marketing expenses at an historical rate equal to that of ICEBOT: about 42%. This move significantly reduces ICEBOT's maximum earnings market share by 360 basis points.

Step 3. ICEBOT's maximum EBITDA reaches $319 million in June, 2008 – a 65% increase over March, 2007 earnings of $194 million.

Step 4. ICEBOT's risk-adjusted differential (RAD) was -0.45 in Q1-07. To put this in perspective, the long run expectation for RAD was zero in a sample of 347 companies in 29 industries over the ten years form 1991 through 2000. See "Marketing's Impact on Firm Value. " Since its RAD was just slightly below the expectation, the ICEBOT merger seemingly did not excite investors all that much. ICEBOT's relative earnings productivity, however, was extraordinary: REP was -2%. What does this mean? ICEBOT produced actual earnings that were only 2% shy of maximum earnings in March, 2007. The historical RAD and REP numbers were assumed be the same in June, 2008.

Step 5. The combined market cap of all three companies was $36.6 billion in Q1-07. A power function (y = 5988.8*Q#^0.8005) fit the 9 quarters of historical data well (R^2 = 0.97). This produced a forecast of the strategic group's market cap in June, 2008 (Q=14) of $49.5 billion.

At the bottom end of the 95% confidence interval, my theory predicts ICEBOT would create 46% of the group's future market value, leading to lowball stock price forecast of $335. At the upper end of the interval ICEBOT would create 63% of the group's market value, leading to a highball price forecast of $430. Since ICE had an historical Beta of 3.2 compared with BOT's Beta of 0.8, I considered this a more risky combination. So I discounted the forecast June, 2008 stock prices at a rate of 3.3% per quarter.

CMEBOT's PRICE JUNE 2008
On the last day of trading in March, 2007 the common stocks of BOT, CME and ICE closed at $181.50, $532.46 and $122.21 respectively. Assuming investors built into these prices their expectations on the value of a merger, the price of CMEBOT was $713.96. The market cap of CMEBOT was $28.1 billion. The number of shares outstanding, adjusted for the merger, was 39.42 million.

If CMEBOT maximized earnings, and its rejected suitor responded as I hypothesize below, my lowball forecast (minus 2 sigma) of the present value of its stock is $713 in June 2008. The present value of the expected price is $809 and my highball estimate (plus 2 sigma) has present value of $886 per share ... assuming no changes in the shares outstanding. The next graphic shows the details following the five step competitive stock valuation process.

Step 1. Strategic group revenues were $729 million on March 31, 2007. That's the sum of each company's sales in Q1-07. As in the case of ICEBOT, I applied a rolling same-quarter year-on growth rate to forecast group revenues. Using the historical rate of 38%, the result is $1,008 million group revenues in Q2-08. Note, in this graphic the historical numbers are green and the forecast numbers are yellow.

Step 2. Sales revenues of CMEBOT divided by group revenues gave the merged enterprises a market share of 79.4% in March, 2007. If CMEBOT management maximized earnings over the ensuing five quarters, its June, 2008 market share would fall to 69.7%. Theoretically, the merged companies already were over-sized for the expected revenues in this strategic group. As a result, the enterprise marketing expenses required to maximize earnings barely increase by 4%, from $204 to $213 million. This would generate CMEBOT revenues of $703 million: a 21% increase over March, 2007 revenues of $579 million. Built into this forecast is the assumption that if ICE lost its bid for BOT, management would not take it lying down. Rather they would continue their aggressive pursuit of market position by increasing enterprise marketing expenses at their historical rate of 47%. Compared with a more growth neutral increase of 22%, this move by ICE significantly reduces (by nearly 700 basis points) CMEBOT's maximum earning market share in June, 2008.

Step 3. CMEBOT maximum EBITDA reaches $490 million in June, 2008 – a 73% increase over March, 2007 earnings of $283 million.

Step 4. CMEBOT's risk-adjusted differential (RAD) was -0.61. To put this in perspective, the long run expectation for RAD was zero in a sample of 347 companies in 29 industries over the ten years form 1991 through 2000. See "Marketing's Impact on Firm Value." The CMEBOT merger prospect it seems did not excite investors any more than did ICEBOT's, since its RAD also was slightly below the expectation. CMEBOT's relative earnings productivity, however, was quite extraordinary too: REP was just -8%. What does this mean? CMEBOT produced actual earnings that were only 8% less than maximum earnings in March, 2007. These historical RAD and REP numbers were assumed be the same in June, 2008.

Step 5. The combined market cap of all three companies was $36.6 billion in Q1-07. A power function (y = 5988.8*Q#^0.8005) fit the 9 quarters of historical data well (R^2 = 0.97). This produced a forecast of the strategic group's market cap in June, 2008 (Q=14) of $49.5 billion.

At the bottom end of the 95% confidence interval, my theory predicts CMEBOT would create 59% of the group's future market value, leading to lowball stock price forecasts of $713. At the upper end of the interval CMEBOT would create 75% of the group's market value, leading to a highball price of $886.  Since CME had an historical Beta of 1.4 compared with BOT's Beta of 0.8, I considered this a much less risky combination. So I discounted the June, 2008 stock prices at a rate of 1.1% per quarter.

IT'S THE PROCESS -- NOT THE PRICE
What's the probability the actual prices of the winning bidder will be bracketed by the nominal lowball and highball prices I forecast for June, 2008? It's non-zero. But I wouldn't say I'm 95% confident in these limits. There's a big difference between statistical confidence and practical confidence. The difference is reflected in the actions management of the winning bidder takes over the next four quarters as well as market devlopments. For example, can you be confident management will maximize earnings? How will investors assess their performance? What might happend to demand for futures exchange services in the next year? How will the rejected suitor respond? And, of course, there's always the possibility that the BOT shareholders will reject both bids on July 9, 2007.

Here’s what I am confident of: the utility of this pricing process. Competitive stock valuation is a valuable supplement to the company cash flow analysis used by the M&A guys. For one simple reason: valuing a stock in the context of a company's simultaneous competition for customers and capital provides a richer perspective on the future. What do you think?

~V

Play It Again Sam

In the week since "Exchange Wars" [ICE vs. CME] was published last Sunday, visits to this blog spiked by nearly 500. Given the somewhat esoteric and totally unconventional analysis that appears on my blog, this represented a huge boost in readership. What was responsible? My conclusions were a bit provocative.

First, the analysis of each company plus the object of their affections -- the Chicago Board of Trade (BOT) -- showed that it was in the interest of the Chicago Mercantile Exchange (CME) to pass on this one. I argued they already had 53.8% of the combined revenues of the three companies and acquiring the BOT would give it 84.5% of the market. At that point my analysis said the CME would be over-invested and would, as a result, leave about $100 million in earnings on the table in the first year. This would hurt the merged company's market value.

Second, the analysis documented a good enterprise marketing fit between the IntercontentalExchange (ICE) and the CBOT: the combined companies would maximize earnings at a market share of 52%.

IT TAPPED INTO A CHORD
My post got quite a bit of play the first day from the New York Times DealBook and from SeekingAlpha.com on the second day. Together they generated 109 visitors.  One of these visitors probably sparked the critical reply from Jeff Carter, who I don't know, but who seems to have forgotten more about exchange markets than I'll ever know. My reply to his comments and his reply to mine, spell out all the issues I detail. Suffice it to say, that post tapped into a chord.

THREE DROP-DEAD ISSUES
Since my analysis was static (based on a single period), used "interest expense" as a proxy for the cost of goods sold, and was built on financial accounting data, I was not surprised to read that:

1) In a dynamic, growing environment like futures exchange markets, a single period analysis is incomplete, at best;

2) Reported interest expense is not the "cost of goods sold" in futures exchanges; and

3) Financial accounting numbers aren't real economic data and should not be used in M&A analysis. Instead you must use cash flow numbers.

As you can image these comments motivated me to look more closely into the issue and produce an updated analysis that addresses them. That's the purpose of this post.

FUTURES EXCHANGE REVENUE GROWTH
To create a multi-period analysis the Exchange Wars, I had to use nine quarters of data ending in March, 2007 since all three companies have been public only for that time. During that period the same-quarter, year-on growth rate averaged 38%. This supports Mr. Carter's assertion that a single period analysis is incomplete. Applying this growth rate to the historical data produces the following forecast of group revenues through the 2nd 2008. [click to enlarge charts]

ZERO COST OF GOODS SOLD
Whether or not the "cost of money" reflected in a firm's interest expenses is a reasonable proxy for the "cost of goods sold" in financial institutions is, I believe, debatable. What I was trying to capture is average variable cost of a "round turn" trade. But in this case it really doesn't matter because that cost is so small relative to enterprise marketing expenses it can be ignored. In fact, ICE reports its revenues are equal to its gross profits!

THE COMPETITION FOR CAPITAL IS REAL
Financial analyses by the M&A guys in their proxy documents for ICE and CME reportedly use ten year cash flow analysis of each company to support their merger cases. But, these cash flow projections are made for each company separately without incorporating the plans of the competitive bidder in the projections. In short, the case for ICE ignores the impact of CME its projections and visa versa. Though I have not seen these proxy documents, I expect the M&A guys include some vague language about "over reaching" on the one hand and "under estimating" on the other, to address the fundamental issues involving their competition for capital. This just won't do. The competition for capital is just as real as the competition for customers. M&A analysis should build the dynamic effects of competitive plans into all projections.

THE COMPETITION FOR CUSTOMERS IS REAL TOO
For all its virtues in financial analysis, there is one serious failing in cash flow analysis – you can't identify the flow of cash from its source. In other terms, there is no such thing as "activity based cash flow analysis."  Despite its limitations, the enterprise marketing expense that firms report in their income statements are as close as a mere mortal can get to building the competition for customers into M&A analyses.

CME's MAXIMUM EARNINGS MARKET SHARE
The results of a maximum earnings market share analysis of the CME over time do not change much from the single period analysis. This chart tells the story.

Actual share (white) has fallen ten share points from 59.3% in March, 2005 to 49.1% in March 2007. At the same time, maximum earnings market share fell from about 60 to 55 share points. Notice how close the CME came to maximum earnings market share in every period until June 2006. Then the gap widened significantly from 100 basis points in March 2006 to 620 basis points in March 2007. What's going on here? Over the nine quarters ICE's revenues increased 280%. This aggressive market expansion has increased the opportunity both for CME and BOT. ... all ships rise with the tide.

CME's MAXIMUM AND ACTUAL EARNINGS
The results are reflected in CME's maximum earnings after enterprise marketing expenses. They achieved almost the greatest relative earnings productivity possible -- the biggest difference between actual and maximum earnings was $8 million in December, 2006.

BOT'S ACTUAL AND MAXIMUM EARNINGS OVER TIME
Group revenues are forecast to increase from $729 to $1,008 million by June 2008. The CBOT's maximum earnings market share increases from 30.73% to 51.23% over the period. This drives its revenues up from $223.94 to $516.30 million.

Since its "cost of goods sold" is barely measurable, these revenues become its gross profits. In order achieve maximum earnings, CBOT's enterprise marketing expenses increase from $86.72 to $251.81 million. This drives EBITDA to increase over two fold, dropping enough earnings to the bottom line to increase EBITDA per share from $1.91 to over $5.01 by the 2nd quarter of 2008. Both EBITDA and enterprise marketing to revenues increase dramatically.

PLAY IT AGAIN SAM
How can the Chicago Board of Trade possibly achieve a market share of 51% in just five quarters? By merging with the IntercontinentalExchange. It this a good idea? Yes, from an enterprise marketing perspective.

THE SCALING FACTOR
But there's another serious concern raised in my discussions with Jeff Carter: can the ICE trading platform be expanded quickly and reliably enough to take on the additional volume of the CBOT? It's no small task. I calculate from CFTC reports that the CBOT had 513 million Open Interest contracts in 2006 on a wide range of commodities. Some of these are not traded on the NYBOT. That exchange had just 160 million Open Interest contracts in 2006. The scaling factor is over 3.5 times. Can ICE deliver on its promise?

~V

Exchange Wars

DUKING IT OUT
On June 14, 2007 the New York Times DealBook reported the latest development in the Exchange War between the "ICE" and the "CME" over a property they both want:

As two of the Chicago Board of Trade’s rivals continue duking it out over the exchange, its board of directors spoke up on Thursday: it favors the Chicago Mercantile Exchange.

But the IntercontinentalExchange Inc. (ICE) seems to want the Chicago Board of Trade (BOT) badly. Dow Jones MarketWatch reported late Tuesday June 12 that the company had enhanced its offer. It now represents a 10.2% premium over Chicago Mercantile Exchange (CME) bid. And

ICE also said it plans to file a preliminary proxy statement to oppose the proposed acquisition of CBOT Holdings by the Chicago Mercantile Exchange (CME) and solicit votes against the proposed CBOT/CME merger at the CBOT shareholders meeting scheduled for July 9.

Given the reported challenges faced by ICE in shifting CBOT's action to its New York Board of Trade platform, as well as the premium it has already offered to pay for the property, one wonders what's behind this ambitious strategy. CEO Jeffrey Sprecher says "It's a growth play." Is he right?

THE THREE PARTS OF A DOLLAR
To understand the motives behind an audacious strategic move like Mr. Sprecher's, it's useful to start with the enterprise marketing concept of "the three parts of a dollar." Why? Because it divides the sales dollar up into three mutually exclusive and exhaustive parts that define, with the use of financial accounting data, the company's underlying business model.

COST OF GOODS SOLD
The "cost of goods sold" at the Chicago Board of Trade in 2006 was 0.24%. Does this mean the company's gross margin is over 99%? Yes. Think of these three companies as retailers of risky financial products to highly select clients. Here's how the CBOT defined its business in its 2006 annual report:

CBOT Holdings was formed in April 2005 as a Delaware corporation for the purpose of becoming the holding company of the CBOT which was founded in 1848. Today, the CBOT is the world’s leading marketplace for trading agriculture, grains and U.S. Treasury futures as well as options on futures. In 2006, 13% of the global listed futures and options on futures contracts traded at the CBOT.

It's clear the company does not have a "cost of goods sold" like a product manufacturer or retailer. So what is cost of goods sold at such companies? Simple. The interest paid on borrowed money. The CME was the least efficient with an interest cost of goods sold equal to 8.45%. ICE was the most efficient with a cost of just 0.07%.

ENTERPRISE MARKETING EXPENSES
Calculating the enterprise marketing expenses of these three companies is a little more complicated. And you can't take the SG&A expenses reported in online services. Often the result make no sense. For example, Yahoo! Finance reported that the Chicago Mercantile Exchange had SG&A expenses of +$24.615 million in the 2nd quarter of 2006!

You have several options in assembling enterprise marketing expenses from the income statements of these exchange services. The least attractive option is to comb through each company's annual report and pick out all the line items that fit this definition:

Enterprise marketing expenses are the costs of everyone (and everything) that affects a company's performance in its competition for customers and capital.

If you're interested in the details of which expenses fit this definition see my audio slide show on "Enterprise Marketing Expenses." It runs about 12 minutes.

The most attractive option in my experience is to subscribe to EDGAR Online I*Metrix services:

I-Metrix is a suite of interactive data and analytical tools from EDGAR Online that provides quick and accurate, XBRL-tagged financial statement data via Microsoft Excel and an easy to use web interface or via a direct data feed.

Here's a list of all the line items in the income statements reported by I*Metrix that include at least a single entry for one (or more) of these three companies in their 2006 income statements:

Each of these enterprise marketing expenses influence the company's performance in competing for customers and capital. Here are the enterprise marketing expenses of each company.

There are two big advantages in using I*Metrix to assemble enterprise marketing expenses. First, the line items are standardized across all companies (whether or not they report the expense separately). Second, with a single mouse click you can drill down into the original SEC document to discover which items, if any, were combined.

COULD IT BE MARKET SHARE?
Revenues for the group of three firms totaled just over $2 billion at the end of 2006. The CBOT captured 30.7%, ICE got 15.5%, and CME led the group with 53.8%.

Could it be that ICE covets BOT because the combined firms would achieve maximum earnings market share and leapfrog the Merc?

I applied The Rule of Maximum Earnings (this audio slide show runs 12 minutes) to the data for the CBOT.

CBOT's actual operating income was $354 million on a market share of 30.7%. Maximum OIBD of $418 billion occurs at 45.5%. How could the Chichao Board of Trade grow that much additional share? If BOT were acquired by ICE their combined market share would be 46.2%! That's only 52 basis points from BOT's maximum earnings share.

WHY LEAVE $100 MILLION ON THE TABLE?
Okay, but what about the alternative? The Merc already owns 53.8% of this market. If it wins the exchange war, the company will own 84.5% of the market. This will maximize its earnings, right? Wrong. This chart shows what happens to market share and earnings if CME wins the bet for the CBOT.

If the Merc wins the bidding, combined earnings of the merged companies would be $968 million at a market share of 84.5% on sales revenues of $1,711 million (in red). While maximum earnings of the combined firms are $1,078 million at a 75% market share on revenues of $1,520 million (in green).

Even if the Merc bid doesn't run into anti-trust issues, the market eventually will drive it to downsize to a point closer to its maximum earnings market share. Why? Because the merged companies will leave over $100 million in earnings on the table.

A BIG ROLE FOR ENTERPRISE MARKETING
Where does traditional marketing fit in this seemingly complicated story? Look back at the list of line items from the income statements that contribute to enterprise marketing efforts. The first one is Marketing and Advertising expenses. Only the Chicago Mercantile Exchange reported any of these traditional expenses. And the total was just $16.74 million in 2006. In a strategic group with revenues of $2.0 billion. Traditional marketing's slice of this pie isn't even rounding error.

Does this mean traditional marketing has only a small role to play in financial institutions that serve a relatively small set of highly select customers? Yes, but enterprise marketing can step in and offer CEOs and CFO's unexpected insights into the challenges they face in the competition for customers and capital --- even in esoteric markets. In this case, it appears that the CBOT should think about switching its bet to the ICE before the window closes on July 9, 2007. What do you think?

~V