This is a recommendation to short Bankrate (RATE). Acquisitions and aggressive accounting have clouded RATE’s underlying weak business trends. The stock has rallied 50% from its low earlier this year, buoyed by the market upturn and hope, providing an attractive short entry point.

RATE’s organic page view growth has slowed to about zero and the heralded pricing power of the business is overstated. Because of RATE’s aggressive accounting, Street estimates for EBITDA margin expansion in 2007 are likely too aggressive. I discuss these issues more fully below. RATE has several attractive features for a short:

 

Bankrate sells banner add space (graphic ads) on their websites and charges advertisers a per click fee for hyperlinks. RATE combines graphic ad and hyperlink revenue and reports it as on-line publishing revenue. Graphic ads are about 60% of on-line publishing revenue; hyperlinks are the balance. RATE’s primary websites are bankrate.com, interest.com and mortgage-calc.com. On-line publishing represents about 80% of revenue, with the rest from a no growth/declining business that sells print advertising.

 

 

The odds are high that RATE misses 2007 EBITDA estimates. Revenue estimates assume the best of all worlds in 2007 despite weak trends recently. Analyst estimates fail to account for the highly likely reversal in 2007 of RATE’s 2006 aggressive accounting for expenses and other items. This reversal in 2007 will pressure EBITDA margins.

 

RATE missed 3Q06 revenue and EBITDA estimates. Based on data from compete.com (a website that tracks unique users and page views) it appears that RATE will struggle to make 4Q06 estimates as well. This report is really a call on 2007, so I won’t go over the 4Q detail here. Compete.com will release data on the month of December in January that should provide more color as well.

 

 

Street estimates call for a significant acceleration in revenue growth next year to 24%, up from 13% organic growth in 2006 (see pro forma disclosure in 10Q). This amount of acceleration is unlikely given poor fundamentals in the mortgage origination market, Bankrate.com’s non-existent core traffic growth and signs that Bankrate is already struggling to maintain its current low teens growth rate (rocketing DSO, revenue missing 3Q estimates, RATE increasingly paying for traffic since 2Q06).

 

On the surface, the Street is projecting a deceleration of revenue growth from 62% in 2006 to 24% in 2007, but that comparison is confounded by acquisitions. September 2006 YTD pro forma growth was only 13% and will likely clock in around 13% for the full year 2006. And growth of 13% is of poor quality, with DSO up 20 days from 46 at the end of 2005 to 66 days at the end of 3Q06. Growth in 2004 on-line publishing revenue, untainted by acquisitions, was only 8%. The historical record and likely future trends do not support analysts’ estimate of 24% revenue growth next year.

 

RATE’s revenues are a function of the price they charge (for graphic ads and hyperlinks) and the number of page views for their websites. Consensus is $98.9m in 2007 revenue; my estimate is $92m. Even if RATE meets consensus revenue in 2007 through paid search (more on this below), it is likely that EBITDA margins will be lower than analysts’ estimates due to aggressive accounting this year.

 

Following are my model assumptions for page views in 2007:

 

Growth in RATE’s annual page views and growth in mortgage origination are highly correlated with an r-squared of 92% (t-stats highly significant). Given trends in mortgage origination it seems unlikely that RATE will see significant page view growth in 2007.

 

 

Other non-mortgage segments of the business, such as ads for CDs and money markets, may help offset mortgage related weakness, but note that RATE experienced a significant shift away from mortgage related business to other products in 2006 yet page views (ex acquisitions) were still flat. If you believe that mortgage originations will fall significantly next year, then RATE’s organic page views will likely be down. I don’t believe this is necessary for the short to work.

 

Because of Bankrate’s flagging traffic, management is experimenting with paid search traffic. Paid search is what it sounds like: RATE pays Google, etc. to boost traffic to Bankrate.com.

 

Management lumps together direct visits by users to Bankrate.com and partner-generated traffic; this accounts for ninety percent of traffic. Lumping direct traffic and partner-generated traffic together is a bit misleading. RATE shares revenues with partners, so in terms of cost to RATE, partner-generated traffic is more like paid search traffic. Direct traffic is free.

 

Paid search traffic will help page growth in the short term, but I am skeptical that the economic return on paying search engines to direct traffic to RATE will be as positive as expected (expectations are high for this initiative). RATE’s advertising customers also pay to advertise on search engines, so RATE will increasingly be competing with its customers as it increases paid search.

 

Direct traffic to RATE’s customers’ websites is far more valuable to RATE’s customers than traffic to RATE’s website. Washington Mutual, for example, would much rather potential customers click on its Google Washington Mutual add than a Bankrate add. When a WaMu customer clicks on a WaMu ad on Google the customer goes directly to the WaMu site and sees nothing but WaMu product. When the customer clicks on the RATE ad on Google, the customer is bombarded with ads from dozens of WaMu’s competitors. If RATE does start diverting search engine traffic away from its customers’ websites, RATE’s customers are likely to respond negatively.

 

One effect of paid search will be to boost revenue, but marketing expenses will also rise. RATE will be paying GOOG, et. al for the traffic and RATE expenses this payment in marketing. However, because of the issues outlined above, it seems unlikely that paid search will deliver economic returns anywhere near the high expectations for this initiative.

 

My 2007 model assumptions:

 

There are three issues with prices that suggest the much heralded pricing power of the business may disappoint in the future. First, revenue and thus prices in 2006 have been artificially inflated by aggressive revenue recognition. Second, RATE’s share of internet users is declining, suggesting that advertisers may be unwilling to accept price increases as large as the market expects. Finally, RATE’s graphic ad prices are declining sequentially this year. The CEO seemed rather defensive about price declines on the 3Q06 call.

 

Analysts model pricing by dividing page views into revenue. Poor quality revenue has served to inflate these price calculations, in turn rendering unreliable analysts’ extrapolation of current pricing trends. Pricing is overstated because RATE recognized revenue in 2006 that it could not and will not collect. See the accounting section below for more detail.

 

Bankrate.com’s declining reach suggests that advertisers should be increasingly less accepting of price increases. Indeed, graphic ad revenue per page view in 3Q06 was below both 1Q and 2Q rates and was flat with 4Q05. Graphic ads were 58% of online publishing revenue (the total of graphic ads and hyperlink revenue).

 

Alexa.com provides data on reach. According to Alexa.com, Bankrate.com’s reach declined significantly this year and is back to levels of 2002-2003. Reach measures share of Internet users. The data below shows that for every one million internet users, 650 went to Bankrate.com’s website in 2006. All else equal, advertisers will pay less per ad through time for websites that are losing share of users. RATE’s declining reach suggests that growth of its ad space prices may not meet the market’s high expectations.

 

 

Hyperlink (42% of online revenue) pricing has been strong. The company implemented a 20% price increase in October 2005 and another 20% in October 2006. I have assumed that the October 2006 price increase holds and that the company implements another 5% in January, in line with analysts’ expectations.

 

On line publishing (graphic add plus hyperlink) prices dropped in 3Q versus 2Q, prompting several questions by analysts on the 3Q06 conference call. The CEO was evasive regarding the price decline, a red flag. On the conference call the CEO said an analyst’s assertion that pricing declined was “inaccurate.” However, the analyst was correct; revenue divided by page views declined for both the graphic ad and hyperlink business q/q (by 9% and 3%, respectively). Upon further questioning by another analyst later on the call, the CEO backpedaled, saying that he understood the question now and:

 

“One of the things that was impacted, and I better understand the question now. Some of the page views that were generated were from Interest.com which doesn’t monetize at the rate that Bankrate does. Some of the pages, generated page views were from the new mortgage calc [acquired August 2006], which the page views were in there, but not much revenue because we really hadn’t started networking mortgage calc and selling that. So you are working off of a larger base, but we weren’t really selling that so that would probably have some impact on skewing the numbers a little bit.”

 

Turns out this assertion by the CEO isn’t true either. Pricing declined q/q adjusting for mortgage-calc and Interest.com has been part of the company for a year. Back out page view contributions from mortgage-calc (5.5-6.0M in 3Q; mortgage-calc acquired August 4, 2006) in 3Q06 and assuming that mortgage-calc contributed no revenue (it did, but assuming it didn’t biases the results in favor of the CEOs assertion) and the result is that on-line publishing prices were down 2% q/q with graphic ad prices down 5% and hyperlink prices up 1%.

 

As noted, page views, a key revenue driver, are highly correlated with mortgage origination. Also, about 60% of RATE’s on-line business is tied to mortgage financing; the rest to financial institution advertising for CDs, money markets, etc. If you believe as I do that the housing downturn will last several more years and that we are at a cyclical peak in financial services profits and therefore advertising by financial institutions, then this short may be especially appealing to you. The principal foundation for my belief that this current housing downturn will last several more years is an October 2005 Fed paper:

http://www.federalreserve.gov/pubs/ifdp/2005/841/ifdp841.pdf.

 

The authors of this paper examine housing markets in 18 developed countries over 35 years and conclude that housing downturns last an average of five years and that the entire prior run up in real house prices in the upturn is subsequently reversed in the downturn.

 

 

Because of RATE’s aggressive accounting, TTM EBITDA appears overstated by at least 20% or $4.5m. A year ago (3Q05, 4Q05) and two years ago (4Q04), RATE required between zero and $1.5M of non-cash, non-tax related working capital. Because of the issues outlined below, that amount has ballooned to $6M. Had RATE kept its DSO the same through higher bad debt expense (or not recognized revenue they subsequently admitted was uncollectible), accrued for expenses in line with sales growth and not postponed the recognition of expenses (other assets are ballooning), EBITDA would have been about $5-6M lower than RATE reported. The evidence is clear that aggressive purchase accounting for the 4Q05 acquisitions also played a role in boosting 2006 EBITDA.

 

RATE acquired three businesses in 4Q05 for $40m. At the time of the acquisitions, RATE’s total assets were $54M, so the acquisitions were large relative to the size of the business, giving management plenty of accounting maneuvering room. Given RATE’s balance sheet changes around the acquisitions and subsequent poor performance of the acquisitions, I suspect a motivation for the acquisitions was to prop up flagging growth in revenue and earnings. Six months after the acquisitions RATE raised $90m in a secondary and management sold over $30m of shares concurrently.

 

The addition of a negative working capital balance from the acquisitions and subsequent increase in working capital after the acquisitions had the effect of inflating RATE’s earnings. The purchase price allocation for the acquisitions (see 2005 10K) attributes ($1.8M) to working capital and ($1.8M) for purchase liabilities (an accrued expense). For purposes of discussion, I combine these two amounts for a net addition to RATE of ($3.6M) of working capital. Negative $3.6m of working capital for the acquired companies is out of line (too low) for the type of acquired businesses.

 

Management does not disclose the exact breakdown of the ($1.8M) of working capital. We can see that RATE’s bad debt allowance jumped from 6% of gross accounts receivable at 3Q05 to 16% of gross accounts receivable after the acquisitions in 4Q05, reducing working capital by $1M. Since the 4Q05 acquisitions management reduced the allowance to 13% of gross AR as of 3Q06.

 

We can see by the cash flow statement that the $1M increase in RATE’s bad debt allowance from 2Q to 3Q was through the acquisitions and not reported by RATE as bad debt expense on its income statement. This large bad debt reserve attributed to the acquired companies allowed RATE to subsequently recognize less bad debt expense through SG&A than otherwise would have been the case.

 

The allocation of ($1.8M) of the purchase price to purchase liabilities has allowed RATE to subsequently recognize less expense than it otherwise would have. Purchase liabilities are defined in the 10K as contract amounts for Internet hosting, co-location content distribution and other infrastructure costs. Management has reduced this account by $1.1M during 2006 and as of 9/30/06 it stood at $0.7M. Since this $1.1m of expense was already accrued with the acquisition, it did not reduce earnings this year, but has of course reduced cash flow as the amounts were paid. Management will have to start expensing/accruing for this expense soon as the accrual is now down to a low level.

 

Another area that may or may not be part of these acquisition adjustments is deferred revenue. Looking back at the filings you will notice that the 2005 10k (which includes the acquisitions) deferred revenue balance was $414k at 12/31/05. But in the 3/31/06 10Q deferred revenue is shown as $1,176k at the same date, 12/31/05. I could find no explanation for the change. In any case, deferred revenue has declined to almost nothing, falling 76% q/q from 2Q06 to 3Q06. The decline was not seasonal. The 2Q to 3Q drop in deferred revenue represented 5% of 3Q06 revenue.

 

Interestingly, management did not provide pro forma disclosure for the recent acquisition of mortgage-calc, even though it represented almost 5% of page views and would have been 7% of page views had mortgage-calc been acquired at the beginning of 3Q instead of in August. 

 

RATE’s rocketing DSOs call into question the quality of revenues. DSO increased sequentially every quarter form 43 days in 3Q05 to 66 days in 3Q06, up an eye popping 50% y/y. Management acknowledged that it had problems collecting from some customers and has quit selling to these customers. DSOs (calculated net of the bad debt reserve) indicate that the problem is worsening, has not been fully addressed and that further large bad debt expenses are likely.

 

Excluding the $3m accrual for a legal settlement in 3Q (which analysts exclude from EBITDA), RATE’s accounts payable and accrued expenses fell by $2.5m from 4Q05 to 3Q06. This does not fit with revenue growth of $5.6M.  During the same period a year ago (i.e. 4Q04 to 3Q05) accounts payable and accrued expenses increased by $1.3M. Part of the decrease this year is explained by the large amount of accrued expenses that came onto the balance sheet with the 4Q05 acquisitions.

 

Prepaid assets have doubled in the last nine months, possibly indicating over capitalization of expenses (this account must eventually reverse, decreasing earnings).

 

Poor capital allocation. RATE spent $40m to acquire three businesses in 4Q05: MMIS, FastFind and Interest.com. It is safe to say that so far, MMIS and FastFind have destroyed economic value. We have no financial statement data on Interest.com, but its page views declined significantly this year.

 

RATE paid $10m for FastFind. The revenue from FastFind declined to $0.9M in 3Q06, down a whopping 40% from 2Q06, even though management had expected revenue to increase Q/Q. Management expects FastFind revenue to decline further to $0.6M in 4Q06. Rate paid $30m for MMIS and Interest.com. MMIS represents around 70% of print publishing revenue now. The print business operating loss has significantly worsened since the addition of MMIS. For the first nine months of 2006 (which includes MMIS) the print publishing operating loss worsened to ($2.8M) from ($0.7M) in the same period a year ago (which excludes MMIS).

The stock is richly valued at 31x trailing TTM EBITDA. Since I believe EBITDA is significantly overstated, operating cash flow is probably a better metric, although operating cash flow may well be over stated as a result of the acquisitions. In any case, RATE is trading at 60x TTM cash from operations and 77x if we exclude the tax benefit from options and assume a full cash tax rate.

 

With companies whose EBITDA I believe suspect, I like to use cash flow from operations to arrive at EBITDA.

 

 

EBITDA from CFO of $11.6M compares to management’s adjusted EBITDA of $24M for the TTM period ended 9/30/06. Management’s guidance for EBITDA for 2006 is about $28M.

 

The difference between the EBITDA from CFO and management’s adjusted EBITDA is primarily a use of $8m for a build up of working capital (see accounting discussion). More than half of this can not be explained by growth in the business.

 

I believe that the return on the short will be a combination of a reduction in EBITDA expectations of at least 20% and multiple compression. Multiple compression is likely when the market realizes that EBITDA is overstated and EBITDA growth expectations are reduced.

 

Several members of management sold significant portions or all of their holdings with the May 2006 secondary at $48.25, with directors and officers selling a total of over $30m of stock on the secondary. Form 4s and the insider selling page on Yahoo! Finance indicate that executives sold twice what is indicated in the S-3. The ex head of sales, Stalzer, sold all of his holdings after leaving the company in November. Other insider sales in November included the VP- Publisher Varsell ($645k) and CTO Hoogterp ($545k).

 

Head of sales (chief revenue officer) Stalzer left the company last fall.

 

Through September of this year RATE has already granted options equivalent to 4% of shares outstanding. In 2005 RATE granted 4% of shares outstanding. In 2004 RATE granted options that represented a whopping 12% of shares outstanding.