What jobs do people do?

All the recent talk on employment numbers got me curious as to what exactly people are employed in. A quick visit to the U.S. Census Bureau and some Excel magic reveals this graph:

So much for the myth of being a nation of lawyers, baristas and burger flippers!

If you add up the obvious service industries (excluding health), they add up to less than 14%.

If you add up all the professional and white-collar jobs to this, that comes to 45% of the work force. This includes core I.T. jobs, which is only around 2% of the work force! Legal occupations are only around 1%.

Construction and manufacturing workers make up 23%.

Health-related workers come to around 7%.

Education is 6%.

If you are curious as to what each occupation category means, check out The Standard Occupational Classification.

Here's a more interesting way of looking at it: If the U.S. were a deserted island with 10 people on it:

• Gilligan is a salesperson
• Russell is a doctor
• Mary Ann is a secretary, who teaches part time
• Sue is a call center representative
• Ed is an accountant, who dabbles in law and computer programming on weekends
• Marge is a cook
• Jonas is a carpenter
• His friend Al is a metal worker
• Harry is an aspiring actor, who "daylights" as a policeman,firefighter, farmer, and physical therapist.
• And Thurston is the manager of the other 9 :)

AIG and S&P's First Negative Earnings Quarter, Ever

For those of you who don't know, S&P publishes a free and constantly updated earnings forecast here. Essential reading for every investor. In addition to reporting the numbers, you get an 8-quarter forecast of earnings as well as unvarnished commentary from their house analyst.

Here's a snapshot of the current (3/18) report:   

The S&P as a whole lost money in Q4 2008. In their commentary, S&P points out:

• AIGs record setting Q4,'08 As Reported loss of $-61.7B; -$22.95 per AIG share, $-7.10 index impact (first negative quarter for index ever)
• 28%, 138 of the 486 As Reported EPS are negative; index lost more this quarter than it ever made
• 49 issues with mega-$billon losses
• AIG Operating loss of $-28.2B (record); $-10.49 per AIG share, $-3.24 index impact (first negative quarter for index ever)
• 20%, 99 of 494 Operating EPS are negative
• 18 issues with mega-$billon losses
• $-5.2B loss for the quarter, with $-101.3B from the Financials; Non-Financials therefore positive
• Sales (based on reported current membership) are down -8.78%; 42% higher Y/Y (avg +6.64%), 58% lower (avg -18.12%)
• Massive charges warp P/Es (field H33), forward numbers more important - but many investors have a lack of trust in the estimates
• Cash flow now high priority, dividend cuts to 'preserve cash', ride out the storm->companies are worried (so am I)
• No major shift in estimates yet as stimulus/ TARP/ housing/ budget details come out

All I can say is 'Wow".

Corporate Bond Yields Update

We recently looked at the role of corporate bond yields as a price floor. Here's an updated bond yield chart:

Notice that bond yields have started to creep back up, and so has the earnings yield. The good news is that this indicates prices have begun to stabilize back on the bond floor. The bad news is that floor may be falling.

Note: The S&P earnings yield used here is calculated by dividing the current S&P 500 index ($825.88) by 4 times the current Dec. 2008 EPS estimate ($8.73) to yield 4.2%. This is to be consistent with the previous data used in the chart. However, if you use the forecasted 12 month EPS, the earnings yield is actually higher (4.8%) and pretty close to the current AAA bond yield.

S&P Earnings Forecasts

Standard & Poor's regularly publishes the earnings report on the S&P 500 here. This includes a forecast of upcoming quarterly earnings. Here's a chart of the current (1/22/2009) forecasts:

First, note that during the last recession (.com bubble), it took 3 full years for earnings to recover to the previous peak.

Right now, earnings are at 1999 levels, a 55% drop from the peak in mid-2007. The current forecasts go through the end of 2010, and predict a modest gain, but not a recovery.

Forecasts are, of course, largely backwards looking, and are revised constantly. However, this probably captures as best as anything the prevailing thinking on the economy: i.e. that it's going to be a long hard road for a while.

So, When's The Bottom This Time?

Phew! First the dot-com then the housing bubble. It's been quite a decade, and there is still a whole year left to go!

Take a look at the S&P 500 index since 1990:

Looks like two eerily similar boom and bust periods doesn't it? However, if we dig a little deeper, we'll see that things aren't at all the same, and the market was actually behaving much more rationally for the second bull run compared to during the dot-com bubble. This gives us some useful clues on figuring out when the bottom of this crash will occur.

Let's first consider what happened with earnings over that time. Here is the same index overlaid with aggregate S&P 500 quarterly earnings data:

Notice how price growth in general follows earnings growth. There are distinct periods of earnings growth (1994-2000 and 2003-2007), followed by very abrupt drops (2000-to-2003, and 2007-to-who knows when). Also notice how stock prices grew much faster than earnings from 1997-2000 compared to the 2003-2007 bull run. This shows that investors were much more speculative (or exuberant) in the late 90's than more recently.

We can make the difference easier to see by plotting the earnings data over the extrapolated P/E ratio:

The extrapolated P/E is computed by dividing the closing price for that quarter over 4 times the quarterly EPS. It's a crude and noisy measure, but a simple gauge of what buyer expectations are at that moment.

You'll notice that there are 4 distinct periods of fairly constant P/E ratios (marked by black bars in the chart above). In the earlier dot-com bull run, P/Es hovered around 18 from 1996-1997, then shot up to the 28 range from 1998-2000.

In contrast, P/Es were around 20 in 2003, and went down to below 17 from 2005-2007.

This indicates that buyer sentiment was far more bearish during the most recent bull run. How much more bearish? Well, a P/E of 28 is 65% higher than a P/E of 17, which indicates that investors were expecting at least 65% greater returns from future earnings in the dot-com days.

Earnings Yield

Let's look at this from another perspective. If we invert the P/E, we get earnings/price, or the earnings yield. Think of it this way: If you were to buy an entire company outright, the return in profit for your investment that you would get after tax at the end of the year would be the earnings yield.

Here's the same chart as above, but showing earning yields instead of P/E:

The data is the same, but this view is easier for us to relate to. For example, during the 1998-2000 period, the earnings yield of the S&P 500 was around 3.6% (100/28). To put that number in perspective, you could have instead put your money in a CD for a guaranteed 5.5% return during that time. So on average, investors were choosing to lose money by buying stocks, in anticipation that future earnings growth (and further speculation) would drive prices up.

In contrast, the average (extrapolated) earnings yield of the S&P 500 was between 5% and 6% during 2003-2007. This was also during a time when saving interest rates were at record lows (under 3% until 2005, then under 5.5% through 2007).

This shows that, in contrast to the dot-com boom which was fueled by speculation, the recent bull run was actually largely governed by fundamentals - meaning that stock market investors behaved fairly soberly and rationally (well, it could be that all the irrationally exuberant speculators just shifted their attention to the housing boom during that time!).

The Bond Floor

Looking at corporate bond yields gives us another strong indication that the stock market has largely returned to fundamental valuation.

First some quick background:

There are two main ways large corporations can raise large sums of money: by selling stocks or selling bonds. A bond is pretty much an IOU to pay a certain interest rate over a period of time followed by returning your money for the bond. It's like the company taking a loan from the public market, at a fixed interest rate that's determined by the market.

Now, think about the relationship of earnings yield to bond yields for a moment: If you are the CEO and need to raise money for expansion, you can either sell stocks (which is selling a share of all future profits) or sell bonds (which has a fixed interest rate). Everything else being equal, you would rationally sell whichever has the lower yield.

Moreover, if bond yields drop below earnings yields (or earnings yields rise sufficiently), companies can (and do) borrow money to by back stock - thus lowering the earnings yield until the two equalize. This effect causes bond yields to be a ceiling on earnings yields - and a floor on P/Es.

Just like individuals, companies have credit ratings. Companies such as Moody's and Standard and Poors (S&P) rate the bond issues of companies from AAA (the best, or "prime") all the way down to D ("in default"). A BAA rating is the lowest "investment grade" rating. So the range of yields for companies rated AAA to BAA should generally represent the yields companies in the S&P 500 would have to pay to raise money through bonds.

If we plot the S&P 500 earnings yield chart over a chart of corporate bond yields, this is what it looks like:

Notice that the earnings yield has been below bond yields all the way up to 2003 (in fact, they have been below AAA bond yields since the early 1980s). This means that during that period, other factors (such as speculation) were the primary factors that drove stock prices. However, starting in 2003 and before the crash in 2007, the average S&P 500 earning yield has hovered around the comparative bond yields.

Let's zoom in on that 2004-2008 period to make it more clear:

In effect, bond yields provided a solid floor for stock prices during that time, and were what pushed prices up (or kept them from dropping).

We can also look directly at the stock buyback levels during that time to see the magnitude of their effect. Between 2004 and mid-2007, buybacks increased four-fold, only dropping off after earning yields declined:

Between April 2003 and September 2007 (which was before the market peak), S&P 500 companies bought back a total of $1.5 trillion of stock. Over that same period, the total market cap of the S&P 500 grew by $5.6 trillion, so the buy-backs contributed (very) roughly 28% to the total gain in price over that time. Without them, prices would have been roughly 28% lower (and earnings yields 28% higher) than they were.

Here's a visual of what the index performance would have looked like if we were to remove the investments due to buybacks:

This great article from S&P goes into more depth on buybacks. Of course, there were many other factors at play during this time, but this clearly shows the powerful effect of bond yields on the market.

Money Exit Stage Left: Enter The Credit Crisis

With this in mind, it's easier to understand what's happening during the credit crunch and what the Fed is fundamentally worried about. Look at what happened to bond yields since September 2008:

When the credit crisis hit, there was a sudden shortage of money available to lend and so borrowers, particularly the relatively less credit-worthy, were forced to pay higher rates. The effect on the stock market was that the floor was suddenly dropped from under it, causing prices to crash.

This is primarily why the Fed took such drastic action, including dropping short-term treasury rates down to zero (yes, they are essentially lending money for free) in order to lower downstream lending rates. This is what caused the subsequent drop in bond yields, although you will note that they are still 1-1.5% higher than where they were at the beginning of the year.

So What Does This All Mean?

While there are plenty of other factors at play (a changing U.S. political administration, a continuing global recession, and ever-present global uncertainty), this is what I interpret this data to mean:

1. A lot of speculation had been "wrung" out of the market from the dot.com crash, which shows that overall investor sentiment has been neutral-to-negative on stocks since 2003, which unlikely to improve in the near term.
2. The Fed is aggressively doing everything they can to pull down interest rates, and eventually they will, causing bond yields to fall back to where they were.
3. When that happens, stock earning yields will equalize with bond yields again, and will rise as earnings rise out of the recession.

If you use S&P's current estimate for Q4 2009 earnings, the average earnings yield at today's prices will be 4.4%. This is already close to the AAA bond yield of 4.75%, but still a long way from the BAA yield of 8.12%. Assuming the Fed's actions manage to bring the BAA yield back down to 6.5% where it was before the recession, by S&P's own estimates, prices could still have a ways to drop (to get the earnings yield to go from 4.4% to 6.5% with constant earnings will require a price drop of 32%).

The thing to keep in mind though is that we have been talking about averages - using a single number to represent the entire market. In reality of course, there is a very wide band of earnings yields across all the companies in the S&P 500. For strong (AAA rated) companies with earnings yields above 4.75% (below a P/E of 21), they should already be at their floor price now. Likewise, BAA rated companies below a P/E of 12.3 should be close to their floor.

Ok, So What Should A Savvy Investor Do?

Keep a close eye on corporate bond yields over the next few months. If they keep their current levels or continue to go down, then it means the Fed is doing their job and credit is flowing back into the markets. That's a sign that the bottom is near (assuming earnings stay on track or better).

Even easier, keep an eye on this blog for updates on how all this plays out.

Happy New Year 2009!

A Look at Unemployment

Here's a chart of the historical unemployment rate from FRED (the Federal Reserve Economic Data service):

Two observations:

First, the absolute best time to be entering the labor force in the last 40 years was right before 2000, when the unemployment rate hit a low of 3.8% in April of that year.

Second, the unemployment rate increased between 2 to 4 percent during each of the recession periods in the graph, with the current increase at already 2.3% since March of 2007.

FDIC Data

The FDIC website is a treasure-trove of information on how the banking industry, and by extension the economy, is doing.

Here are some charts from right before the October market meltdown.

This one shows the credit-card loss and personal bankruptcy rates:

Notice the huge spike leading up to and subsequent fall off from the more stringent bankruptcy laws passed in 2005.

This chart shows the non-current (i.e. late-payment) rate on real-estate development loans since 1990:

Yes, that rate is climbing, but wow was it bad in 1990!

Here's a similar chart on the rise of non-current loans on private residence mortgages since 2005:

Yikes, over 5% of home mortgages are late!

Finally, here's a chart showing a corresponding fall-off in new mortgages:

 

These charts only show data up to September 2008. The next set due out next month will show what happened in the next 3 months.

A Tale of 3 Computer Makers

Numbers can tell you a lot.

These are the recent annual earning statements for Dell and HP, plotted as a chart. Each value is shown as a percentage of revenues for that year:

       

Dell's COGS are a higher percentage of revenue (81% in 2008) compared to HP (76%). This is a big deal when net margin is near 5%. Dell's direct model may be more efficient than traditional retail channels, but that doesn't translate to higher margins for Dell. HP's higher overall margins means that they can leverage earnings from their server and printer products to subsidize their battle for greater PC and notebook market share.

Two other things are striking: First, HP has been pretty aggressive in getting their SG&A spending down in line with Dell's (11% vs. Dell's 12% in 2008). However, HP spends 3 times as much as Dell on R&D percentage-wise. That's 6 times as much in absolute numbers ($3.8B vs $610M in 2008). Their efficiency, coupled with a greater investment in R&D, positions HP to do better against Dell in the future.

Now, compare both of them to Apple's numbers:

SG&A spend is at similar levels (around 12% of revenue in 2008), so is R&D (3%), but there's a big difference in their COGS ratios. Apple started the decade looking very similar to Dell & HP, with COGS at 77% in 2001. However, by 2007 it is only 66%, translating to an over 10% increase in net margin, which was 14.9% in 2008, over double HP's. This was a result of Apple changing the game by entering the higher-margin music-player and smart phone business, and that has put Apple in a whole different league.

What's more astounding is that Apple accomplished this transformation with very similar R&D budgets to Dell and HP. However, there was a significant jump only recently in 2008 where Apple increased R&D spending by 40% to $1.1B at a time when HP trimmed their's by 2%. One can only imagine what that $300M extra is working towards after the development of the iPhone.

(Note: Dell's Fiscal Year 08 ended in February 08, while Apple and HP's fiscal years ended in Sept and Oct respectively).

Computer Science Graduate Statistics

I really like the idea of tracking key stats as part of our investment club activities. Following our discussion last Monday, I took a quick look at graduating CS degree numbers. They tell a vivid, if unsurprising, story.

Here's the big picture from NSF and CRA (Computing Research Association) numbers:

(From http://www.cra.org/info/education/us/index.html)

There are huge peaks around 1986 and 2003, coinciding with 4 years after the birth of the PC and the .com boom respectively. Presumably, entering freshman decide their majors based on current economic and technological events. Unfortunately, the picture may be different 4 years later.

The CRA also surveys the number of incoming CS majors, which of course is a leading indicator for how many graduates there will be 4 years later:

(From http://www.cra.org/CRN/articles/may08/taulbee.html)

These are absolute totals from a subset of schools, but they do show a corresponding huge 2x drop in enrolments in 2005-2007. This means that we can expect a low in the number of graduating CS engineers starting in 2009.

Since the swings are so huge (2x), it's tempting to make broad generalizations, such as the need to relax immigration limits or face increased outsourcing of engineering work. Presumably, it will also lead to increased price premiums for domestic talent, which is good for those already in the profession. :)