Disturbing Trends at Harley-Davidson: Why It's Time to Sell

Harley-Davidson (HOG) came to my attention a few weeks ago when fellow Reese Fund analyst, Wes Golladay, suggested that HOG may be plagued by inventory issues. As an individual who enjoys a good statistical challenge, I went to work on HOG and found some disturbing trends in one of America’s great brands. These trends lead us to assign a Sell rating to Harley-Davidson and a near-term price target of $50 a share.

Investment Thesis

• Dealer inventory contributing significant EPS: At the end of last year, $1.00 of additional EPS sat on Harley Davidson dealer lots, overhang from too many deliveries over the past two years. Investors are paying an estimated 18x PE TTM versus the current 15.6x TTM.

• Delivery Estimate Indicates Earnings Miss: HOG’s EPS given delivery estimates between 64,000 to 70,000 cycles has a significant statistical probability of being much lower then the current .73 street estimate with our model pegging Q1 between .53 and .62 a share.

• No Work-off in Q1 Dealer Inventory: Even with a strike-reduced shipment count for Q1, no retail inventory will be worked off in this quarter. This will lead to lower shipment needs or pricing pressure for the rest of the year.

• Inventory Balloon: Based on a survey conducted, Harley Davidson dealers are carrying between 2.5x – 3.5x more inventory then they were in Q1 2004.

Given these challenges, Harley likely will use this quarter to take an earnings bath and use its strike in Pennsylvania as cover to do so, putting the stock at significant risk for downgrade and increasing pressure on an already battered stock.

Methodology

Three specific research tools were used to help arrive at this conclusion:

1. An EPS logarithmic regression model that uses seasonality, shares outstanding and deliveries to predict HOG EPS
2. A Holt-Winter’s model to help predict Harley retail sales, trend, and seasonality
3. A survey of 56 dealerships online new inventory conducted between March 28th and March 30th.

Dealer inventory contributing significant EPS

Motorcycle sales are seasonal, but HOG’s revenues are not. In an effort to reduce operational expense, the company delivers cycles consistently throughout the year. Unfortunately, this creates the temptation to beat earnings by shipping more cycles then is necessary. Research began from a base year of Q1 2004 (the first quarter available) where HOG reported both sales and deliveries (deliveries are reported back to Q1 2002). By starting with Q1 2004, you can see the trend of HOG not trying to balance sales and deliveries but over time, delivering more cycles then are needed to the channel leading to dealer inventory build.

These figures are further corroborated by what’s been happening with HOG’s A/R account over the past few years. As you can see in Figure 2, A/R growth has outpaced that of sales as well as the A/R days outstanding measure increasing for the last four years indicating that once again dealers have been receiving more shipments then what is necessary to meet demand.

The conclusion from this information is that HOG dealers now have more cycles on the lot then at any point in the company’s history. Also, HOG has used its channel to achieve its net income estimates. If the additional dealer inventory build was taken out of Harley’s EPS from the previous two years and shipments evenly distributed throughout the quarters then Harley would have reported $3.41 last year and $3.14 in 2005 or a full $1.00 less of EPS over the past two years.

Without dealer build, HOG would have reported net income without inventory build of $891 MM in 2006 and $838 MM in 2005 versus $889 MM in the base year 2004 placing the current P/E of the firm at an estimated 18x versus the current 15.6x.

Delivery Estimate Indicates Earnings Miss

In investigating HOG, it became readily apparent there was a high correlation between deliveries of new cycles and the company’s EPS. By controlling for the 15% of share buyback Harley has engaged in since Q1 2004, a robust EPS model was built that uses deliveries, shares outstanding, and seasonality to determine the company’s EPS from 2002 forward.

The model tested extremely well when predicting future EPS without knowing the current year and has an average miss of -.001 cents, a standard deviation of .039 cents, a mean average percentage error [MAPE] of 5%, and an R2 equal to 94%.

Given the accuracy of the model with just these three parameters and current estimates of 64,000 to 70,000 shipments depending on the source, the model predicts HOG will report excluding extraordinary items between .53 and .62 cents a share; 10 to 20 cents lower then consensus analyst expectations for Q1 2007. Even these shipment numbers may be ambitious; Figure 4 shows the production pace the company will need to achieve to hit certain shipment levels for the strike-shortened quarter.

Given that the company will have to meet its all-time production pace just to make .67 in EPS, we believe HOG will use the quarter to take an earnings bath and use the strike as cover to do so.

Note: If indeed this does come to pass then watch for HOG to take additional loss reserves against HDFS.

No work-off in Q1 Dealer Inventory

Originally, the Holt-Winter retail sales model created to predict Q1 sells had dealers selling ~74,000 cycles in Q1. However, with UBS analyst Robin Farley’s note that sales were down and Craig Kennison of Baird reporting sales increase between 0% and 2%, we adjusted the model with these expectations and the seasonal loss over the two months plus a prediction for March. These led to the conclusion that dealer inventory will be flat to slightly increasing for the quarter but still leave dealers with ~7,000 cycles more then last year at this time heading into the peak selling season. Without significant discounts towards the end of the year to increase retail sells or a reduction in shipments, YOY dealer inventory will likely continue to increase once again.

Inventory Balloon

Since HOG does not report dealer inventories, current dealer inventory had to be estimated. By conducting a survey of 56 online dealers new inventory, an average of 63 cycles per dealer was arrived at and with it a 95% statistical confidence that HOG dealers carry between 62,000 and 101,000 cycles. However, this number can be refined further by overlaying the cumulative change in cycles per quarter. With these together we feel comfortable in estimating that HOG dealers currently carry between 80,000 and 100,000 cycles or between 2.5x and 3.5x more inventory then they carried in Q1 2004. This additional inventory is most likely affecting HDFS because of its floor-plan financing arrangements, creating even more significant pressure on the unit.

Conclusion

All companies aspire to beat the market and its near-term estimates. In this case, it looks as though HOG did so by using its dealer network as an unwitting accomplice, allowing it to increase EPS and hoodwink $7.50/share from shareholders based on the current 15 P/E multiple. But as the adage goes, all such acts must come to an end. The Reese Fund analysts believe HOG may ultimately be forced to rob today’s shareholders in order to pay for yesterday’s. Given that net income without dealer build would have been flat for 2004 versus 2006, the apparent slowing in retail sales, and the impending Q1 earnings miss, we assign a sell to HOG stock and a near-term price target of $50 a share

Disclosure: Grant Case is a 2nd Year MBA student at the Kelley School of Business and a Reese Fund analyst. Neither he nor the Reese Fund has a position in HOG.

Comments

[Nick Mulcahy]

"Much has been made of Frank H. Knight's (1921: p.20, Ch.7) famous distinction between "risk" and "uncertainty"... In Knight's interpretation, "risk" refers to situations where the decision-maker can assign mathematical probabilities to the randomness which he is faced with. In contrast, Knight's "uncertainty"... refers to situations when this randomness "cannot" be expressed in terms of specific mathematical probabilities. As John Maynard Keynes was later to express it:

"By `uncertain' knowledge, let me explain, I do not mean merely to distinguish what is known for certain from what is only probable. The game of roulette is not subject, in this sense, to uncertainty...The sense in which I am using the term is that in which the prospect of a European war is uncertain, or the price of copper and the rate of interest twenty years hence...About these matters there is no scientific basis on which to form any calculable probability whatever. We simply do not know." (J.M. Keynes, 1937)"

cepa.newschool.edu/het...

[Dav Mix]

Thanx Grant. This is a reasonable approach and an excellent analysis. Feel free to present commentary on additional companies, I for one would appreciate the info as it is difficult to find an un-opioniated voice today. I'm sure others will not agree with your findings, but historical numbers are difficult to bury. I like HOG and have owned them in the past and will probably own them again. Obviously, the actual stock COST has to be considered to determine the stock's VALUE, and each invester must consider the potential for a given time frame, so what sounds bleak in your analysis may be the gem to another! Thanx again, Dave

[Nick Mulcahy]

www.marketwatch.com/ne...;siteid=yhoof

[PeterWellington]

This is some great analysis Reese. However, I think you may have made an important mistake in the table, "Dealer Inventory Contributing Signficant EPS." In column 2, Deliveries, you have included Buell-brand sales in the total (roughly 10,000 per year). However, in the Reported Retail Sales line you are counting only Harley-brand bikes--Buell is not included. About 80% of the "inventory-build&... you calculate from 2004-2006 is just the sale of Buell bikes--not inventory-build at all. A careful reading of the 10-Q shows this difference.