United (UAL) and overbooking


Overbooking itself is an airline industry staple.  It’s also a sensible practice.

That’s partly because there’s always someone who doesn’t show up for a flight.  For example, business people who travel often may book seats on three or four different flights to their next destination on a given day.  No matter what time their morning client meeting ends they’ll be able to get away easily.  They just cancel the ones they won’t use as the day develops.

operating leverage = large profit for the last seats sold

A plane that takes off with unused seats is a lost revenue opportunity.  Like unused hotel rooms, there’s no way to sell them later on.  And both hotels and airlines have a ton of operating leverage–there’s maybe $10 in marginal cost associated with an incremental seat/room.  So the profit contribution left on the table from non-use is immense.

UAL’s overbooking algorithm

I’ve noticed with UAL, which I’ve used regularly for years, that all of the last half-dozen or so flights I’ve been on recently have been overbooked.  That’s after seeing maybe one overbooked flight in the prior year.

My (unscientific) guess is that UAL tweaked its overbooking algorithm a few months ago in an effort to squeeze a bit more profit out of its flights.

My experience is that when companies begin to operate by trying to figure out where the the line is that marks the minimum a customer will accept–and then tries to get as close to that line as possible without crossing it, disaster inevitably happens.

I first encountered this phenomenon with a former management at Marriott that later went on to wreak havoc at Disney and at Northwest Air.  Their idea was that no one would notice if they lowered the room ceilings on newly-constructed hotels by an inch or two and shrank the room square footage by, say, 5%.  Yes, construction costs were a bit lower.  But I remember those rooms as being vaguely unsettling when I entered.  Business customers, the heart of a hotel chain’s profits, fled in droves.

If I’m correct, the remedy for UAL is simple.  Just go back to the former algorithm.  If, however, the corporate culture at UAL is to provide customers with the minimum acceptable service–and, to be clear, I don’t know that it is–more trouble is likely brewing.


big day for Amazon (AMZN)–why?

AMZN reported 2Q15 results after the close last night.  They were very good.

Sales were up 20% year-on-year; expenses rose by 17%, three percentage points less.  As a result, the company reported an operating profit of $464 million vs. a loss in the second period of 2014.

More than that, AMZN’s cloud services division, AWS, had revenue growth of 81% yoy and a quintupling of segment profits (basically operating profits less stock option expense) to $391 million.  AWS, broken out as a separate segment for the first time after 1Q15, remained a bit more than a third of the AMZN total.


AMZN posted an overall profit of $.19 a share for the quarter, vs. analysts’ expectations of a loss of $.13 a share and a deficit of $.27 per share in the year-ago quarter.

On the announcement, the stock immediately rose by 15% in aftermarket trading.

AMZN opened up by 20% this morning, before drifting down steadily during the day to close +9.8% in a market that was down just more than 1%.


Why the strong advance?

I have no good explanation, although I do have some ideas.

1. The obvious factor that changed overnight was the earnings announcement.

It contained a significant positive earnings surprise, one that makes it more likely that the company will earn, say, $1- a share in the current year. It makes the analyst consensus of $2.78 a share for 2016 more believable.   On the other hand, the stock was trading at $482 before the earnings report, or 173x the 2016 consensus.  Looking at the stock price another way, let’s say that at maturity for its businesses (whenever that may be), AMZN shares will be trading at 20x earnings.  To sustain the pre-earnings report price, that would imply a burst of rapid growth that shoots earnings up to around $24 a share.  That would be something like a doubling of earnings each year for the next five or six.

That’s already baked in the cake.  A buyer of the stock at this level must believe that $24 a share in eventual earnings is way too low.

I find it hard to believe that a $.32  per share earnings surprise during one quarter–when expectations were already sky-high=-would be enough to add 20%, or even 10%, to AMZN’s perceived market value.

2.  A second hypothesis…

What if investors are beginning to separate AMZN into two parts, AWS and everything else, and are doing a sum-of-the-parts evaluation.  To me, this sounds a little more plausible.  What would the numbers look like?

Let’s say that in 2016 AWS will comprise half of AMZN’s earnings and AMZN Retail the remainder.  To make the figures easier, let’s say each half earns $1.50 a share next year.

Let’s assume AMZN retail can grow in earnings at 20% a year for a long time, and that we’d be willing to pay 50x current results–a big number for a retail stock–for that future profit stream.  If so, AMZN Retail is worth $75.  To reach a sum-of-the-parts value of $482, AWS must therefore be worth about $400, or close to 270x its 2016 eps.  Ok, while I personally wouldn’t be willing to pay that much for AWS, I can see how someone else might.  However, I still don’t understand why confirmation that a holder at 270x earnings isn’t insane would cause the multiple to expand.  (Also, before I’d be comfortable valuing AWS as a separate company, I’d want to know more about how AMZN apportions revenues and costs among segments to ensure the published numbers don’t flatter AWS.  I’d also think long and hard about the possible effect of stock options.)

3.  The explanation for AMZN’s rocket ship ride that I’m leaning toward, however, is more technical.  Two factors may be involved.  At what Google Finance reports as 21+ million shares, today’s trading volume in AMZN was 7x normal.  The sharp opening spike suggests to me that algorithmic trading computers were at work reacting to the earnings report, not humans.  Humans, I think (?!?), would have a better sense of valuation.  I also suspect that the report and immediate upward move triggered a lot of short covering.

I’m partial to #3 because I think the whole reaction is a little  crazy.

Why is any of this important?  AMZN is a high-profile, large-cap stock with almost two decades of operating history.  There’s got to be a way to make money from the possibility that something like AMZN’s big move will occur with other similar names.



the inventory problem: holding costs vs. stockout risks

The domestic auto industry reported November vehicle sales yesterday.  The numbers were very good.  But most of the (negative) media attention centered on the elevated level of inventories–about three months worth of sales–on dealer lots. Yes, that may eventually be a worry, but I don’t think it’s the right way to look at the current situation.

The auto news also gives me the occasion to write about the balancing act every manufacturer and retailer faces in deciding how much inventory to have.

the simplified story

There’s an often convoluted dance between supplier and distributor/end user about return policy, payment terms, co-op advertising…in negotiating over how much of a product to buy and at what price.  Nevertheless, the decision about how much inventory to hold ultimately comes down to weighing two opposing risks:

stockout costs.  This is when your brilliant national advertising campaign, your sterling reputation for high quality and service–or sometimes just random factors–prompt a potential customer to either go online or enter a physical store with the intention of buying an item.

You’re out of stock.  You try to interest him in a substitute, or promise to have the item tomorrow.  He says thanks, leaves and buys the item somewhere else.

You’ve lost a sale.  And the person you’ve disappointed is at least marginally less likely to have you first on his list next time he’s shopping.

That’s stockout costs.

inventory holding costs are much more straightforwardly quantifiable.

There are three main factors:

-financing costs, which in today’s world are negligible;

-liquidity risk of having your capital tied up in inventory rather than in cash during the time it tales you to make a sale; and

-the possibility that the items either become obsolete, go out of style, or–like fresh food–exceed their shelf life before they can be sold.  Then your asset has become a wirtedown.


(In the stock market, there are always complications.)

In good times, companies want to hold more inventory (because they see stockout as a greater risk than holding costs); in bad times sentiment reverses and everyone wants to hold as little as possible.

If prices are rising, procurement managers see the chance to make windfall profits and order more than they need; if prices are falling–as is chronically the case in industries like consumer electronics–inventories are kept trimmed to the bone (except in really good times, when everyone throws caution to the winds).

In industries with low fixed, high variable costs, manufacturers see no percentage in upping production volumes.  In industries like autos, with high fixed costs and therefore tons of potential operating leverage, there’s a tremendous incentive to make extra units once a firm reaches breakeven.

The competitive structure of an industry doesn’t change the nature of inventory risk, but it can change who it is who’s assuming them.  This may not always be obvious from even a detailed study of the working capital sections of the balance sheet.  If a manufacturer were to have a policy of unlimited returns (that would be crazy, but let’s just suppose), then it–not anyone farther down the distribution chain–would ultimately be responsible for any unsold goods.


More tomorrow.






calculating operating leverage

I’ve written a number of posts on operating leverage.  You can use the search function on the blog to get them.

The basic idea is that a company has both fixed costs, which it must pay whether it sells anything or not, and variable costs, which are a function of the number of things a company sells.  Once a company covers its fixed costs through sales, the operating profit on additional sales can be very high.  This is a key source of positive–and negative–earnings surprise.

As a practical matter, though, how can we calculate how operating leverage works in a given company?

For some firms, it’s impossible.   Take 3M (MMM).  It makes a gazillion different items, many of them sold in massive quantities. For an investor, there’s no way to see very deeply inside the company.

We also have to realize that the data we get from any company’s financials is going to be imperfect, at the very least during our initial look.  If we take the time and energy to compare our projections to the actuals the company publishes, listen very carefully during management conference calls for clues, and call the company every once in a while, we may be able to refine the numbers we come up with in a surprisingly significant way.

Nevertheless, for smaller companies that sell only one or two main products, there’s a very simple way to get an idea of whether a firm has significant operating leverage or not:

–take two most recent consecutive quarters

–subtract the revenue reported for Q1 from the revenue reported from Q2

–subtract the operating income of Q1 from that of Q2

–calculate an incremental operating margin by dividing the operating income change by the sales change

–compare that with the operating margin achieved during either quarter.

An example:

Harley-Davidson (HOG–I own shares, despite the fact the ticker symbol spells a word) sells motorcycles, spare parts and branded merchandise.

During 2Q13, the company posted motorcycle-related revenues of $1.631 billion and operating profit of $362.9 million.  The operating margin was 22.2%.

During 1Q13, HOG had motorcycle revenues of $1.414 billion and operating profit of $279.0 million.  The operating margin was 19.8%.

The quarter-on-quarter revenue difference was $217.9 million, and the q-on-q operating profit difference was $83.1 million.  The operating margin on the extra production was 38.1%.

In HOG’s case, we can go on to make a number of refinements.  We can try to separate out the profits from sales of merchandise and spare parts, which are relatively small in revenue in comparison with motorcycles but which carry higher margins.  And we can examine whether the much higher margin on incremental sales comes from manufacturing efficiency or from leveraging SG&A (it’s the latter).

But the main point is clear.  HOG makes almost twice as much on incremental sales as it does on average sales.  And we found this out just by making some simple subtractions.


the Intel (INTC) 3Q12 preannouncement: studying operating leverage

the preannouncement

As I wrote about yesterday, INTC preannounced weaker than expected 3Q12 earnings.  The main culprit?  …worldwide general economic slowdown.

The company said it now expects revenues of $13.2 billion for the quarter, down by 7.7% from the $14.3 billion it guided to when it announced 2Q12 earnings two months ago.  The gross margin will come in at 62% instead of 63%.  Virtually all other cost items will remain the same.

looking at leverage

This isn’t much data.  But it’s enough for us to see two things about the company, manufacturing leverage and leverage on SG&A (Selling, General and Administrative) expenses.

manufacturing leverage

two kinds of costs

In the simplest terms, in every accounting period employees get paid and the accountants apportion costs for the use of the factory and the machinery in it, whether or not anything gets build.  So, in a sense these are indirect costs of manufacturing.  In the short run, they’re relatively fixed.

In addition, there’s the cost of the materials–electricity, gas, silicon, who knows what else–that get used up in making INTC chips.  These are direct costs.   Their total in any period is variable, depending on how many chips get made.

Accountants assign each chip a total cost that depends on two factors:  the out-of-pocket cash (variable cost) spent making it plus its share of indirect costs, a figure that depends on how busy the factories are.

gross margin

Total cost ÷ sales price = gross margin.

separating the two

Is there a way to find out how much of the total cost is variable and how much depends on how well sales are going in a given quarter?  In INTC’s case, yes.

Management has just told us that sales will be $1.1 billion less than anticipated and that this fact will lower the gross margin by a percentage point.  That’s not because the variable cost of making a chip has changed; it’s because the indirect (or fixed, or overhead) costs of running the factories are being distributed over a smaller number of chips.  (It’s a little more complicated than that, but not a worry in this case.)

Another way of saying this is that in order to get to the new, lower, sales and operating profit estimates, INTC has subtracted the sales price of the extra chips it won’t sell and only the variable cost of making those chips.  If we calculate the change in estimated gross profit and divide by the change in sales, we’ll get a variable cost margin for those “extra” chips.

Here we go:

$13.2 billion x .62  =  $8.18 billion in gross profit

$14.3 billion x .63  =  $9.01 billion in gross profit

The difference is $.83 billion, the gross profit lost from lower sales.   This gross profit   ÷ $1.1 billion in lost sales   =  75.5%.

Therefore, 75.5% is the profit margin from producing/selling an extra chip during the quarter.  That’s the manufacturing leverage INTC gets at current production levels for getting/losing additional sales.

Note, too, that the new operating profit is 9.1% less than the original estimate.  That compares with a 7.7% drop in sales.  So, while there is operating leverage in the manufacturing operation, but at current production levels it’s not huge.

SG&A leverage

INTC has two types of SG&A.  One is R&D.  The other is the typical SG&A that any industrial company has. The two items are roughly equal in size.  This quarter they’ll amount to $4.6 billion.

Let’s subtract that from both the original gross profit estimate and the new guidance.

$8.18 billion  –  $4.6 billion  =  $3.58 billion in operating income

$9.01 billion  –  $4.6 billion  =  $4.41 billion in operating income

Now calculate the percentage drop in operating income that our 7.7% decline in sales produces.

It’s 18.8%!

To recap, the 7.7% fall in sales produces a 9.1% drop in gross profits and an 18.8% contraction in operating profits.  Of the 11.1 percentage point differential, 1.4 comes from the manufacturing process, 9.7 from SG&A leverage.

In other words, the operating leverage at INTC is coming from SG&A, not manufacturing.  If INTC wanted to reduce costs in a way that would affect current reported profits the most, it would attack either R&D or “normal” SG&A.