Market Volatility

So I hear this a lot:  "With HFT (High Frequency Trading), hedge funds, ETFs and instant trading via iPad apps (and mobile smart and dumb phones), the volatility in the stock market is intolerable.  I can't take it anymore.  I'm out!"

Or, "with so many darn MBA's, PHD's and rocket scientists trying to outsmart the stock markets, the markets are way too efficient; there's no way to make money in the market anymore!"

Or how about, "with the revolving door of government and Goldman Sachs, the insider connections, people on the outside don't stand a chance".

Actually, I've been hearing these arguments for years.

Market Volatility
So the market is too volatile?   Yes, with HFTs, hedge funds, ETFs and instant trading at any Joe Schmoe's fingertips, the market is prone to sudden, extreme, irrational moves (never mind that that's actually a *plus* for value investors).

So check this out.  This is a chart of the Dow Jones Industrial Average going a long way back, all the way back to 1902, I think.  That's more than 100 years of history.

Long-Term Dow Jones Industrial Average Chart

I look at this chart and scratch my head.  OK, so where is the last five or ten years more volatile than the past 100 years?  Sure, it's more volatile now than, say, the period between 1994 and 2000 when the market never declined more than 10% off the peak.  Remember that period?  I actually think that was the abnormal period.

But what about the period from 1900 - 1950?  That period actually looks a lot more volatile than now!  And guess what?  No hedge funds. No HFTs.  No ETFs.  Nobody can click a button on a cell phone and hit "sell everything NOW!".  Nope.  No futures or options.  No ISDA-based OTC derivatives.   No CMOs, CDOs, or CLOs.  Not even any banks that are "too big to fail".   Nope. None of that. 

In fact, for most of that period, the U.S. was on a gold standard.

I remember back in the 1990s, people really feared a big, supercrash, much bigger than Black Monday that would be driven by the 1-800-GetMeOutNow mutual fund investors.  The automation allowed anyone to dump their stocks/mutual funds with a single phone call.  This scared the industry  to such an extent that I think some mutual fund companies sought credit lines so they can borrow money to meet redemptions (so they won't have to sell into a crash).

And this only got worse in the 1990s with the internet and online brokers.  So in that sense, it's sort of remarkable that we had the worst recession/financial crisis/bear market since the Great Depression and the stock market only went down 50% off it's high, and we didn't even really have a 'crash' day.  (What happened in 2007-2009 was far worse (in terms of the economy, banks etc.)  than what happened in 1987).

Absolutely remarkable.

I remember last fall when the market was up 300 points one day and down 300 the next or some such.  It was very, very volatile.  But the market basically wasn't going anywhere.  People who know I'm involved in the markets asked me, wow, it must be rough dealing with the markets.  I was like, what do you mean?  The market is not going anywhere!  Up 300 one day and down 300 the next is only a problem if your job was to forecast the closing price every single day or you are a daytrader or whatever...  or if you get fed up and dump your stocks on the 300 point down day and change your mind and get back in on the 300 point up day and keep repeating...  Otherwise, it doesn't matter.

Markets Too Efficient

I think someone asked Joel Greenblatt this question some time back when he was speaking at Columbia or somewhere (I wasn't there; I just read an article about it).  Someone asked that if so many smart folks are rushing to Wall Street to work for these proprietary-trading-focused banks and hedge funds, isn't the market getting pretty efficient leaving less for anyone else?

Greenblatt's answer was that this may be true in some areas of investing, but we still had the internet bubble in 1999/2000 (when these supposedly 'smart', highly educated people rushed into internet stocks).  So go figure.

A lot of hedge funds operate with different strategies.  Maybe some spread businesses get tight (merger arbs, convertible arb etc...) with too much money going into it and a lot of funds these days seem to be sort of macro based, event-driven or whatever.

And the ones people complain about are HFT-like; they trade so often they are only interested in mere basis points in 'edge', measured in microseconds or whatever.

Do value investors really care if someone picks them off for 1 or 2 bps on a trade?   There may be some unfairness with front-running and things like that, but I think overall, the HFT guys are just doing what the NASDAQ market-makers and NYSE specialists used to do but just doing it better and more efficiently.  (You don't think people on the floor had access to privy information and did some front-running too?  Why would there be any value to being physically on the floor of the exchange if there wasn't some sort of informational advantage?  Is some of the anger against HFTs and ETNs based on the fact that much of the informational advantage has moved away from the floor to these other networks that mere humans can't exploit anymore?)

Flash Crash?

I know people bring up the flash crash as an example of computers run amok, but that to me is not a great example as I tend to think that was just a really simple, stupid mistake; somebody (especially the exchanges) just didn't have a filter that 'flagged' trades that didn't make sense (selling something that's trading at $50/share at $0.01 for example...).   The exchange (and trading firm) should have had a simple screen to reject the trade and demand human intervention/confirmation. 

I remember when trading started to get automated on the street.  There were usually all sorts of fat finger filters to prevent these really stupid errors.  It's not rocket science.

People used the flash crash to point out how the specialist system is better. But wait a second.  What happened on Black Monday?  Didn't people just refuse to answer the phone?  Didn't the specialists just bid 20% below last trade? 

And if the specialist/NASDAQ market-maker system is so good and they provide 'liquidity' to the market for smooth price discovery and whatnot, how come stocks opens down 10% or 20% on some overnight news even when there is a specialist in charge of keeping 'orderly' markets?  Where are the market makers then?  This used to happen way before the HFT's started taking share away from the NYSE floor.   I remember waiting for stocks to open down double digit percentages all the time; the specialist wouldn't open the stock until he had enough buy orders to offset the sell orders and he had to keep moving the price down.  You call that providing liquidity to the market and keeping the market orderly?

I'm not picking on the NYSE and specialists.  I'm just trying to point out that it's really not that different; people complain that HFT's can just leave the market in times of turmoil, but my point is that it's not like the NYSE specialists and NASDAQ market-makers stepped up on Black Monday or other 'event' days to provide liquidity.   

The old system had their problems too just as the current system has problems.  You will never make everyone happy. 

The interesting point, though, is that we haven't had a 20% down day *with* HFTs involved *despite* what we went through in the past few years.

I do have an unproven theory why that might be.

Why Didn't We Crash Big During the Financial Crisis?!

Back in the old days, people had no *information*.  People were blind.  Only the privileged had information  (if you were Gomez Addams, you had a private ticker tape in your house).  When you called your broker and said you wanted to buy 100 shares of IBM, you were at the mercy of your broker and the folks on the floor of the New York Stock Exchange.  And the only information you were going to get was in tommorow's newspaper; the opening price, high price, low price and closing price.

Now you can get Level II, Level III or whatever order book in real time etc... 

When people are empowered with information, that allows people to react.  They can see prices on their screen so when something goes down a lot during the day, they can buy some and not just run in blind fear.  I have no proof, but I tend to think that things are much scarier when you just don't know.  And when you don't know, the natural, reflexive action is to just get out first and think about it later.

But that's just my theory on why we didn't have a huge crash in the last financial crisis.  (Information is power!).   The expectation was of massive, simultaneous liquidation including the pro-cyclical, volatility expanding leveraged ETFs.

Old-Timers Complaining

I hear some old school guys complain about the state of the financial markets including the relationship between brokers and clients and blaming the abolishment of Glass-Steagall, but I tend to point to deregulation of commissions in the 1970s and technology as the culprits (both of which I tend to think are good things, by the way).

Way back in the old days, stock brokerage commissions were fixed so there was no price competition.  It didn't matter where you traded so you looked for a relationship; whoever had the best service got the business because the prices were all the same.  This is sort of like when bank deposit rates were regulated and banks fought for customers by giving away toasters.

This changed in 1975 (May 1, 1975 otherwise known as May Day).   This is the beginning of the propietariazation of Wall Street.  If the brokerage business is not as profitable, then they have to put their own capital to work to make money. 

Combine the above with all the information we now have at our fingertips.  The average high school student with an interest in the stock market today can get on the internet way more information than I was able to get as a professional equity trader working at a major bank in the early 1990s.   Maybe even the late 1990s too.  With all that information, what do you need a broker for?  (I do understand that good brokers are very valuable and clients will gladly pay them fees for their services)

Anyone with an internet connection can get all the SEC filings, annual reports at company websites, price and fundamental data at Yahoo Finance, Morningstar etc.  You can even get whisper numbers and analyst upgrades and downgrades.   And you can get all of this for free or at very low prices. (remember Quotron machines?  I am dating myself here...)

This, to me, is one of the big reasons why the industry has changed.  Yes, there are minus factors to all of this (overtrading by retail investors), but I tend to like price competition and free flow of information.

Revolving Door Between Government and Wall Street

So this is another complaint that I hear about a lot, but frankly, I don't get it.  I know, I know.  People hate banks and bankers.  It's Wall Street's fault for causing the housing bubble and then causing it to collapse.   And yes, I'm just a bank loving, capitalist sympathizer. 

But I tend to think of it another way; the best and the brightest people often go to Wall Street.  Yes, Wall Street causes trouble every now and then.  But still, you tend to have a lot of smart, competent people there.   The top school grads still (well, maybe less today) go there etc.

Now, why would I *not* want those "smart" folks to get involved in the government?  (Well, yes, these very same 'smart' people brought us this mess, I know.   There is good reason to be skeptical about the Bernanke (perfect SAT score?) types too (not to mention the award-winning brilliance at Long-Term Capital Management).  But on the other hand, do we really want Wall Street and government to be run by mediocre people?  I'm not too sure about that.)

OK, so some of the best and the brightest go to Goldman Sachs.  Fine.  And some of the best and the brightest go to Washington.  Fine too.  And then they hook up.  What's wrong with that?  People make it seem so evil if a GS guy becomes Treasury secretary and then hires ex-GS guys.  It's a conspiracy.

But wait a minute.  If I took a government job and I needed to hire people to help me, why wouldn't I hire someone I trust?  Why wouldn't those people be people I actually have experience working with?  Or from an institution that I understand and where I come from?  It's totally natural.  I think it's a good thing that that happens, not bad.  If, for example, a GS guy is not allowed to hire another GS guy, or another financial industry person, *that* would be a problem.  How the heck is he going to find qualified people that he knows well and trusts?! 

So what I see going on is not a conspiracy but just a totally natural thing.   I actually think it's good.  Sometimes, I wish there was more back and forth between the private sector and public sector in Japan, for example.  In Japan, bureacrats are professional bureacrats; they have never run anything in their entire lives and understand nothing about business or the real world.  The only back and forth there is between government and business in Japan is "amakudari", or "descent from heaven" or whatever it was called: government officials get board seats at large corporations after retirement with great benefits. 

You'll never see a banker move to the Ministry of Finance, for example (at least at the minister level).  That makes me go hmmm....   Maybe that's why they are having so much trouble over there.

Anyway, there is a long history of people going back and forth between Wall Street and Washington, but I don't think that's because bankers want to go to Washington to represent their interest (even though that may be so in some cases).   Many people, when they get really wealthy often want to do something else and try to help change the world for the better.  I do believe that many go to Washington with good intent even though not everybody going there is Mr. Smith.  

This country was founded by various interests in the first place.  Forgetting about, for the moment, whether any of this is good or bad, it's sort of always been this way and that's no reason to get in or out of the market. 

You could have said the same thing in 1930, 1940, 1950 or 1960 and I don't know that this revolving door is a reason to be in or out of the stock market.

 I don't mean to say everything is A-OK.   I am just sort of pointing out the sameness of everything over time.  Things just look a little different in every generation, but it sort of seems the same to me.

Tokio Marine Holdings: Japan Fund with Negative Management Fee?

So this is just a quick look at an old trade.  Tokio Marine Holdings has always been known to own a lot of equities and the stock often traded at a discount to it.  A long time ago before insurance companies started to mark their stockholdings to market, it was a classic value play where the real 'value' wasn't represented on the accounting balance sheet.

This is an old Marty Whitman (Third Avenue Value) favorite, and I noticed that the Third Avenue International Fund still owns it (but not much compared to other non-Japanese insurance companies).

Anyway, Japan is taking a beating for various reasons and even though I am itching to get bullish Japan I'm not quite there yet.  As I keep saying, so many things have sort of lined up for Japan to be a great opportunity; bear market for 20+ years, universally unloved, time for a turn etc.  But even if the idea sounds good, the bottoms up view looks horrible (low ROEs across the board, shareholder unfriendliness etc.).

But I thought I'd take a look at this as it would be a nice way to play Japan.

Anyway, here is the 10-year chart of the Nikkei 225 index.  It's basically heading back to the lows of 2009.

Nikkei 225 Index Past 10 Years

and here's Tokio Marine Holdings (TMH; TSE ticker 8766):

Tokio Marine Holdings Past 10 Years

TMH is the oldest and largest property/casualty insurance group in Japan and as far as I can recall, it has been well-regarded in terms of management.  At one time they renamed themselves Millea Holdings and filed 20-F with the SEC and had a U.S. listing.  They have delisted from the U.S.

Anyway, here is the U.S. GAAP combined ratios (for the P/C business) for the period when they were listed here:

                Combined
                ratio (U.S. GAAP)
1999         95.0%
2000         98.8%
2001         99.8%
2002         94.3%
2003         91.7%
2004         92.6%
2005       103.2%
2006         97.2%

This is just to show that it was a decent business under U.S. GAAP; no funky accounting (not that U.S. GAAP is not funky).

TMH was actually formed in 2002 after the merger between Tokio Marine and Nichido Fire (and renamed at the time Millea Holdings).

Here are some basic figures for TMH:

This is all Japanese GAAP based.  Net assets per share is basically what we would call shareholders' equity.  In Japanese GAAP, the shareholders' equity is adjusted by what we would call accumulated other comprehensive income (securities marked to fair or market value) and the adjusted shareholders' equity is called net assets.

I put the ROE figure in there, but that's sort of meaningless as it includes gain on sales of securities but not changes in values of securities holdings.

So it seems that TMH trades pretty close to net assets much of the time since 2003, averaging about 1.06x.

TMH BPS vs. Nikkei 225 Index

The above chart is excluding dividends as I couldn't find good dividend data on Japanese stock indices.   I just looked at the EWJ (MSCI Japan Index ETF) net investment income to average assets as a proxy for dividend yields.  Here is the comparison between the Japanese market dividend yield to TMH dividends to net assets:

                   EWJ              TMH

2007           0.68%            1.5%

2008           1.11%            2.3%

2009           1.46%            1.8%
2010           1.20%            2.0%
2011           1.53%            2.0%
Average     1.19%            1.9%

(investment income to average assets for EWJ as of August of each year; TMH dividend to net assets is as of March of the following year)

In the chart above, between 2002-2012 (March-end), the Nikkei rose 2.64%/year while TMH net asset per share grew 2.32%/year.  The Nikkei did 0.32% better. 

For the past five years, the Nikkei declined 10.2%/year and TMH declined 10.3%/year for around a 7 bps difference.   However, the dividends to net assets at TMH was 0.73% higher than for the EWJ (Japan proxy) as seen in the above table.

That's really not much, but it is positive tracking error, or positive slippage.  

Nothing to get excited about, though.

TMH Stock Holdings
So this is where we get to the fun stuff.  Check this out:



The net assets is, as I said above, what we call shareholders' equity.  As of the end of March 2012, the net worth of TMH was around 1.9 trillion yen.  TMH has a large equity portfolio and this, at market value, was also 1.9 trillion yen.  So basically, if you buy TMH stock at book value, you are getting a Japanese equity portfolio; the net assets will move up and down with the stock market plus or minus whatever the insurance business does.

Needless to say, if you are bullish the Japanese stock market, you can see why this would be an interesting idea.  You get one-for-one Japanese stock market return plus something extra.  That's sort of like buying an equity fund but getting paid a management fee instead of paying it, isn't it?

This is not a static situation, though, as TMH is in the process of selling down their equity portfolio.  Here is the history of the amount of sales since 2002 (in billions of yen):
         

Since 2002, TMH has sold 1.3 trillion yen worth of stock out of their portfolio of equities (held for business relationships).   Naturally, the question is what they do with that cash.  From the above, it seems that they have spent 305 billion yen on dividends and 538 billion yen on share repurchases, so a little more than 60% of sales proceeds was returned to shareholders.

It's also remarkable that even though they sold 1.3 trillion yen in stocks out of their portfolio since 2002, stocks at market value are still 1.9 trillion yen versus 2.0 trillion yen back in 2002.

What's in the Portfolio?
With the entire net worth of the company invested in equities and it being the primary driver of value and share price, obviously we would want to know what's in the portfolio.  In the Japanese filings, there is a list of names of the largest holdings.  The annual filing for the year just ended hasn't been filed yet (similar to 10-K in the U.S.), but here is the top ten names from last year's filing:

Total value of the equity portfolio was 2.1 trillion yen.

Positions as of March 2011:

Name                               #shares                  Value (yen at market)
Toyota Motor Corp          67 mn                    225 billion
Mitsubishi Corp               84 mn                    195 billion
Honda Motor Co.             56 mn                    176 billion
Mitsubishi Estate             37 mn                      52 billion
Nissan Motor Co.            65 mn                      48 billion
Asahi Glass                      43 mn                     45 billion
Suzuki Motor Corp          20 mn                     37 billion
Terumo Corp                     8 mn                      36 billion
Samsung Fire & Marine    1 mn                      27 billion
Itochu Corp                      31 mn                     27 billion
Total:                                                              868 billion

Toyota itself was more than 10% of stock holdings (and therefore also 10% of net asset value of TMH and the market cap).  Mitsubishi Corp and Honda are also close to 10%.  From there the concentration tapers off as the fourth largest holding is less than 3% of the equity portfolio.



Negative Management Fee?
So let's see.  How do we know that this so-called management fee is negative and not positive?  Does TMH make money as an insurance company?

Since TMH has a combination of businesses, from domestic property and casualty, domestic life and international, we can't just do a simple combined ratio analysis.

So I did something simple.  It may not be clean, but it sort of makes sense to me.

If you look at the table above, you will see the equity portfolio at market value and at cost.  So you notice that the equity portfolio at cost was 1.3 trillion yen back in 2003 but is down to 828 billion yen in 2012.  That's because they are selling down their equity holdings.

Most of this portfolio is part of the Japanese cross-holding thing;  they own the shares to enhance business relations with partners (clients, basically).

I was going to use the above ROE to show that the business makes money even without equity gains, but since ROE includes "realized" gains, that's no good.  How do we know that TMH hasn't simply sold just enough stock every year to show a profit?

Ordinary Profits Less Gains on Sale of Securities
So to check that, I decided to just take the ordinary profits of TMH and then deduct gain on sale of securities less loss on sale of securities (to deduct "net" gain on sale of securities) and then deduct impairment loss on securities.

This way, I would get a sort of non-market related profit of TMH.  It does include investment returns as far as interest income and dividends go.  I also left in their trading gains and losses as those aren't part of the liquidation of the large equity portfolio.   The impairment losses are not necessarily equity portfolio losses, but I decided to take that out too as that is not really an ordinary, business profit/loss.

Profits Excluding Gains on Sale of Securities

So judging from this table, it looks like TMH is profitable even excluding the gain on sale of securities, much of which is the equity portfolio.

I think it's safe to say that the "management fee" is negative.  There may be other extraordinary gains to offset operating losses here and there, but I think the securities gains are the bulk of the lumpy gains.

The annual reports include something called the "adjusted earnings" and "adjusted ROE" and also "adjusted BPS", but I don't have the comfort with those figures yet but they can be proxies too of profitability excluding market factors.  Adjusted earnings excludes all non-recurring, extraordinary items and realized and unrealized gains/losses on securities, derivatives, swaps etc. 

One part I don't get comfort with is that it uses "embedded value" (EV) for the life insurance segment. The change in EV is the adjusted profit in that segment.  Also for adjusted BPS, the life insurance segment is valued using EV, which is basically like the 'gain on sale' concept where you present value the profits you expect to make over the life of a policy.  EV is this present value of profits plus net assets. 

Also, for the P/C segment, "adjusted earnings" exclude reserves for catastrophes and these reserves are added back for adjusted BPS too (so BPS becomes much higher).

Anyway, it can still give you a feeling of the business over time excluding the stock market and other gains and losses.  The message is still the same; TMH is a low return on equity business but pretty consistently profitable.

Here's a table that includes the adjusted earnings and ROE from a May 2012 presentation:

A U.S. GAAP Bonus? 
I noticed that during the years 2003-2006, the net assets reported in Japanese GAAP and shareholders' equity under U.S. GAAP had a difference of about 1 trillion yen.  Here are the figures:

                       Japanese GAAP           U.S. GAAP
  (bn yen)       net assets                      Shareholders' equity        Difference
2003               1,805                            2,824                              1,109
2004               2,311                            3,408                              1,097
2005               2,305                            3,433                              1,128
2006               3,210                            4,440                              1,230

I am no accounting expert, but on the 20-F, there is a reconciliation of total assets from Japanese GAAP to U.S. GAAP, and the large items are:

• adjustment to recognize deferred policy acquisition costs (around 500 bn yen)
• adjustment to present prepaid reinsurance premiums on a gross basis (around 300 bn yen)
• adjustment to present reinsurance recoverable on losses on a gross basis (around 300 bn yen)

This is not listed in all of the 20-Fs, so I just looked at 2005 and 2006 and it seems that these are the large items.  There are other adjustments and some may see something else, but to me these looks like the bulk of the difference. 

Of course, this is stuff from 2005-2006 so I wouldn't count on Japanese GAAP understating value by 1 trillion yen today or anything like that, but it is a possibility.  Even so, there is no reason why the market should reflect this if it's not 'visible' (the stock does seem to track net asset value based on Japanese GAAP pretty well.  Valuation might have been a little higher during the U.S. GAAP years, but that might have been due to the good times back then, and the Japanese stock market was trending upwards too).

Current Valuation
OK, so what's all of this worth?   If you just look it up, TMH looks like it's trading at 0.77x BPS, pretty cheap.  And if the equity portfolio is as big as BPS, then that's really awesome.  You can buy a Japanese stock portfolio at a 23% discount and get all the other income (P/C insurance, life insurance) for free!  (It's not a stub trade, though, since the equity portfolio constitutes part of 'capital' of the insurance companies so is not all separatable).

But of course, with so much stock on the balance sheet and a big decline in the market since March-end, we have to adjust this book value to the current level.  I will just assume that the portfolio moves the same amount as the index.

The Nikkei 225 index closed last night at a around 8,640 versus 10,084 at the end of March.  That's a 14.3% decline, so we will just adjust the equity portfolio down 14.3%. 

At the end of March 2012, TMH had 1,857 billion yen in the stock portfolio.   That is a haircut of 266 billion we need to shave off.  With 767 million shares outstanding, that's 347 yen/share we have to knock off of the net assets per share as of the end of March, which was 2,399/share.  

That leaves us with a current net assets per share of 2,052 yen/share.  The stock is currently trading at 1,836/share so it's trading at 0.89x net assets per share.


Conclusion
I am no expert on the insurance industry, and especially the Japanese insurance industry, but TMH is generally a well-regarded company.  They seem to do well, at least by Japanese standards.  There are risks going forward as the insurance business is very competitive.  TMH is also expanding internationally and that's always a risk; it may not go well.   Time will tell.

Also, they have gotten through the recent disasters OK (large losses from the earthquake was averted due to earthquake reinsurance from the government etc.).

Although I am not as excited about Japan as I want to be, this is definitely an interesting vehicle to play it.  If they do well in the insurance businesses and the Japanese stock market comes back to life, there can be some significant upside here.

The discount is modest so not so large to make this a pound-the-table interesting idea, but it's certainly not a bad one. 

I don't own any here, but I will certainly keep an eye on it.

The Banks' Real Nightmare

Interest rates are plunging yet again in the U.S., and there is talk now of QE3 again.  I'm happy to see things being done to help the economy, but one thing I fear about QE3 is that it will destroy the banking industry.

What Worrys Me Most About Banks
The JPM trading loss doesn't worry me at all.  The multi-trillions in derivatives outstanding doesn't worry me. Dodd-Frank and Volcker Rules don't bother me too much.  A double-dip recession where housing prices go down again?  Nope.  Doesn't worry me too much.  After the crisis, I think loans are written at much better standards and with better loan-value ratios etc.   So I don't worry about bad loans on a double dip recession as much.  Banks also have much more capital and their risk management is much better than pre-crisis.

So what worries me more?    Take a look at Japan.

Japanese Banks
Check this out.  This is some data I pulled from the 20-F filings of Mitsubishi UFJ Financial Group (MTU)  in Japan, which I think is the largest bank there now.  It's pretty much regarded as the best bank, kind of like JPM or what C used to be.

Profitability Ratios of Mitsubishi UFJ 1999-2011

ROAA is return on average assets, ROAE is return on average equity, NII is net interest income to average interest earning assets and average equity / average assets is exactly that (average shareholders' equity to average total assets).

This data only goes through the year-ended March 2011 because they don't file their 20-F until July or so, and they only publish U.S. GAAP results in the 20-F, I think.

You can see why Japanese banks are cheap.  The ROE averaged 1.1% since 1999 and even being generous and starting in 2003 (eliminate the loss years of 1999-2002), the ROE averaged only 4.4%.

The years with large ROE are years where they had big gains in investments, foreign currency or some such other non-recurring things (as far as I can tell from a quick reading). 

Their ROE is low because their ROA is low.  Their return on assets since 1999 has averaged 0.03%, and since 2003 (again, being nice and using a favorable time span) has averaged only 0.14%. 

So why is ROA so low?  Of course, because interest rates are so low over in Japan and has been for many years there.

Net interest margin since 1999 has averaged around 1.2%, and it's been pretty consistent.  Notice that Mitsubishi is more than two times levered than U.S. banks too, having an equity to assets ratio of less than 5%.  U.S. banks are more like 10 times levered now.

So let's look at some of these figures for large U.S. banks.  I'll just look at a couple as they are more or less the same.

                                     ROA        ROE         NIM               Equity/assets
Wells Fargo                  1.25%       11.9%       3.9%                9.9%
J.P. Morgan                  0.86%       11.0%       3.3% (core)      8.2%
M&T Bank                   1.16%        9.7%        3.7%               11.9%

Look at this and compare it to MTU's above.


It Can't Happen Here
I remember having many discussions in the late 90's and early 2000's about the U.S. bubble and how we can follow Japan.   The usual response was that something like Japan can't possibly happen here.  Bernanke also said it can't happen because the U.S. Fed will respond differently.  Everything in the U.S. is way better than it is in Japan (disclosure, regulation etc...)

If you said that short term interest rates will be zero or near zero in the U.S. like Japan for an extended period of time a few years ago, people would have laughed at you.  In fact, I didn't predict anything, but people did laugh at me when I suggested that it could happen.

If you said then that long term treasury rates would go below 2%, nobody would have believed it. 

The Fed will print money like crazy, causing inflation and interest rates would go up, not down.  So a sub-2% interest rate is impossible, they said.

During the financial crisis, the 0% short term rate was deemed a short-term, panic/fear driven interest rate that won't last too long.  But here we are, years after the worst of the crisis with short rates still down here and long term rates much lower and hedge funds still trying to pick the top in the bond market.

This is so reminscent of Japan all throughout the 90s.  Shorting JGB's must be one of the deadliest hedge fund trades ever done (on a cumulative loss basis).  The U.S. treasury short trade is looking a lot like that now.

10 Year Government Bond Rates
So anyway, even though most would have thought it impossible even a year ago, here we are with U.S. 10-year treasury rates at 1.46% or so.

The Japanese (JGB) 10-year rate was 1.26% as of March 2011.  Look at that rate, and then look at the net interest margin of Mitsubishi Bank.  Now the JGB yield is 0.82%.

Now look at the U.S. bank net interest margins.  Hmmmm....

There are factors that is better in the U.S. than in Japan.  Demographics, for example, is favorable in the U.S. while it is a negative in Japan.   I think (despite the financial crisis) that U.S. banks are generally much better managed in the U.S. than in Japan (look at how the subprime blowup occured in the U.S. and Japan avoided most of the U.S. problems, but MTU still lost a lot of money in 2008 and 2009 while JPM didn't even have a single quarterly loss).

Dimon has been asked about this on conference calls, and he feels that although he can't predict interest rates, he thinks it will eventually get back to more 'normal' levels.  I think Buffett said the same thing; he feels that the bond market is the biggest bubble of all time.

This sounds right to me, but I just can't get over the Japanese JGB bubble that has been ongoing for 20 years!  And if it continues here for longer than many think, what happens to the banks?  Don't even open up an income statement of a U.S. bank and try to plug in a 2.0% or 1.5 net interest margin to model potential earnings (or losses).  Your blood pressure will go up.

This is not my primary, base-case scenario.   But to me, this is a bigger risk factor than what most people worry about; risk of European exposure and all that.


Inflation is Inevitable
Everybody keeps saying that inflation is inevitable, but this Japan scenario keeps bugging me.  I know they say Japan is different (debt internally funded etc.), but each time we say it can't happen here, it seems we take a step closer.

Many hedge funds are positioning themselves for inflation but I think Prem Watsa of Fairfax is the only one I know of betting outright on a Japan-like deflationary scenario.  (Of course, all bond investors are betting on deflation even if they are not actually thinking about it and buying bonds by default).

Maybe the tail hedge here is actually long treasury zeros instead of long gold.  As they say in the trading world, if it's obvious, then it's obviously wrong.

Anyway, I don't know how this inflation/deflation resolves itself.  I suspect both sides will be right over the next few years; it's just going to be tough to know when which one happens first etc.  (deflationary collapse followed by hyper-inflation, or maybe we just muddle through...)

Loews Corp: Recent Returns, Adjusted Book Value etc.

People always talk about L in terms of sum-of-the-parts and whatnot, and I just realized that I never really looked at their returns over time and in different time periods.  This may not be necessary as you know the Tischs' have created value over time as they usually illustrate in their annual reports, and if you buy L at a discount to it's intrinsic value, you can hardly go wrong.

But as I gathered up some information to create my own long term data, I noticed something.   First of all, here is what it said in the 2011 annual report:

"Over the past 50 years, the price of Loews common stock has grown at an average annual rate of approximately 14% compared with an approximate 6% growth rate for the S&P 500."

That's a long term outperformance of 8%/year, pretty impressive especially over 50 years.

In the 2010 annual report, they made a similar comparison.  They compared the 50 year average total return from 1960-2010.  The S&P 500 annual total return was +9.6%/year and Loews' annual total return was +17.6%/year.   That's also an outperformance of 8%/year over 50 years, which is pretty astounding; of course this is not surprising as the two 50-year periods share mostly the same data points.

The 2009 annual report had a simliar comment and data, saying that the L common stock appreciated by 15.9%/year versus 6%/year for tthe S&P 500 index.  That's 9.9%/year outperformance over 50 years.

The 2008 annual report said that the stock price appreciated 16.1%/year versus +5.7%/year for the S&P 500 index.  So that's +10.4%/year in outperformance.

OK, so where am I going with this?

Pre-2007 Annual Reports
And then I get to the 2007 annual report and I notice that their long term outperformance chart/graph is a 25 year one.  (I've been reading these reports in real time for years, but I didn't really think about this at the time).

In the 2007 annual report, the long term information was that L stock appreciated +14.8%/year vs. +9.8%/year for the S&P 500 index, and BPS grew +13.4%/year in the past 25 years.  That's a stock price outperformance of 5%/year.  Not bad at all.

Here are the comparisons in the annual reports over time:

                          L price          S&P           BPS             Time period
2005 AR:          +14.9%           +9.7%       +12.7%        25 years  (exclude dvd)
2006 AR:          +16.1%         +10.3%       +14.0%        25 years  (exclude dvd)
2007 AR:          +14.8%           +9.8%       +13.4%        25 years  (exclude dvd)
2008 AR:          +16.1%           +5.7%                           50 years  (exclude dvd)
2009 AR:          +15.9%           +6.0%                           50 years  (exclude dvd)
2010 AR:          +17.6%           +9.6%                           50 years  (incl dvd)
2011 AR:          +14%              +6%                              50 years

So the respective outperformance of L stock versus the S&P 500 index was:

                    L stock price
                    outperformance vs. S&P 500
2005:             +5.2%      (25-year comparison)
2006              +5.8%      (25-year comparison)
2007              +5.0%      (25-year comparison)
2008            +10.4%      (50-year comparison)
2009              +9.9%      (50-year comparison)
2010              +8.0%      (50-year comparison)
2011              +8.0%      (50-year comparison)

Part of the reason why they switched to 50-years in 2008 is that it was the 50th anniversary of L's listing on the New York Stock Exchange, and I think it was meant to be a "looking back since listing" sort of thing.

But it also turned out to be a really timely shift. 

Let's see what would have happened if the long term comparison stayed at 25 years instead of switching to 50 years.

                                                                     Relative performance
                   L stock         S&P 500               over 25 years
2008           +11.0%         +7.0%                 +4.0%
2009           +10.6%         +7.9%                 +2.7%
2010            +9.0%          +7.4%                 +1.6%
2011            +8.5%          +6.8%                 +1.7%

 (these returns are excluding dividends since that is how L presented the performance when they showed 25 year comparisons).

So instead of showing an 8-10% long term outperformance, if they kept the "long term" time frame at 25 years, the 5-6% outperformance shown in 2005-2007 would have shrunk to 1.7%/year over 25 years through the end of 2011.

I hold the Tisch family in the highest regard, and they are the straightest, most honest people in business.  So I'm sure they're not trying to fool anyone.

But it is still a little annoying.  I know it's a little tricky to show long term performance when book value is not the best measure of performance, and the stock price performance too can be misleading over different time periods.

BRK and LUK both show their data going back to inception so we can quickly do our own calculations for various time periods and not be 'fed' return figures for periods that look good (I'm not accusing anyone of anything, but that was a pretty timely switch so you can't blame anyone for wondering!).  There will never be any confusion about long term performance with BRK/LUK because the data is all right there on page one.   The average return since inception is also right there.

It doesn't take a whole lot to do that (but it sure is a pain to assemble ourselves with the splits and all that!).    I understand there is a reservation about book value per share (may not reflect accurate value of the firm), and the stock price may be misleading at times too.

But if book value per share (which they used to show a 25-year chart of) and stock price is important or relevant enough to mention in the shareholder letter, then it may be good enough to present in an easily accessible form.

I know, I know, I don't want to turn this into a "Why can't you be more like BRK?" post (too late?).

Recent Performance
So here's the data the way I've looked at other capital allocators recently.  I know book value isn't the best indicator of value for L as they have large, listed subsidiaries (that are consolidated and therefore not marked to market).  But as L themselves have demonstrated, the stock price and book value have tracked closely over time.

L Book Value Per Share (incl dvd) versus S&P 500 Index (incl dvd)

These are the returns for the respective time periods:
                                                                                                +/-                    +/-
                                     L BPS      L Price         S&P 500       BPS basis        Price basis
Five year average         +10.3%     -1.3%         -0.2%           +10.5%              -1.1%
Ten year average          +11.8%    +8.3%        +2.9%           +8.9%               +5.4%
Since 1989                    +11.5%    +7.0%        +8.2%           +3.3%               -1.2%

On a book value basis, L has done pretty well over time, especially over the past five to ten years which is very encouraging, but hasn't done too well on a price basis; if you owned L stock since 1989, you haven't outperformed the S&P 500 index, which is kind of surprising.  Over the past five years you haven't done too well either.

L Stock Price versus L BPS

The long term stock performance is hurt by the fact that L was trading at close to 2x BPS back in 1989.  There may have been some value not reflected in book value like in 2007.  Back in 1989, CNA was trading at 1.5x book and CBS was probaby on the books for less than market value.  The 10-K's only go back to 1994 so I can't see what value there was not included in BPS back in 1989.

Even still, it's a little surprising especially given the 50 year outperformance figure that I have been taking for granted when shown to me in the annual reports.


Loews Valuation
Anyway, Loews is looked at usually as the sum of the parts.   Loews presents it nicely in their annual report and presentations.  

But sometimes, when someone says something is trading at below the sum-of-the-parts, you want to know how the sum-of-the-parts actually did in the past.  We looked at L's performance on a BPS and price basis, but I wanted to see what it looked like on a sum-of-the-parts basis.  

First of all, I will say that a sum-of-the-parts analysis is fine and a one-time snapshot is enough as long as there is motivation to close that gap by management, management is competent in allocating capital and the parts are good businesses (or otherwise solid assets); you know the value is there and it will eventually be realized.

But it's also interesting to see, if possible, what the stock price did versus that analysis.

Here, I am going to do a quick analysis by adjusting the BPS of Loews going back. 

Here is what I am going to do (I will only go back to the year 2001).

There are basically three listed subsidiaries that make up the biggest adjustment factor to BPS; CNA, DO (Diamond Offshore) and BWP (Boardwalk Pipelines).  I will ignore Lorillard as that has no net impact on BPS adjustment over time because the value realized is reflected in BPS (it wasn't a spinoff; the IPO was a sale so L got cash proceeds and the spin-off was actually an exchange whereby L received L stock in exchange for LO stock, like a LO sale / L share repurchase done simultaneously).  LO would have had an impact in between the tracking stock IPO and the complete spin-off/exchange, but only on a mark-to-market basis in between.  Net-net, through time, the value is realized and included in BPS.

In the back of the 10-K's,  there is segment information and the assets are also broken down by segments.  In that section, you can see the shareholders' equity of these listed subsidiaries; what the book value of the holdings is on L's books, as well as minority interest (which is the portion not owned by L). 

So all of these listed subsidiaries are carried on L's books at book.  The adjustment we have to make is to add an amount if the listed subsidiary is trading above book value and deduct from L book value if it is trading below book value. 

This is just another way to sum-of-the-parts this thing.  The advantage is that it leaves all other areas of L in the valuation at book.

March 2012 Adjusted Book Value per Share
Anyway, I'll walk through the adjustments for March-end 2012.

BPS of L at March-end was $48.96/share.

Here is the data:

                            Equity of subsidiary          P/B ratio               adjustments
CNA                     $10.3 billion                     0.64x                    -3.7 billion 
DO                          $2.2 billion                    1.90x                     +2.0 billion
BWP                       $2.0 billion                    1.64x                     +1.3 billion
Total adjustments:                                                                      -0.4 billion

The equity of the subsidiary is the equity portion that L owns excluding minority interest.  So this is what the subs are carried at on L's books.  The P/B ratio is the valuation of the listed subsidiaries' stock.

If a stock is trading above book, we have to adjust the equity upwards to reflect a higher valuation.  If it is trading below book, then we deduct.  

So for CNA, it is on the books at $10.3 billion, but to adjust that to market value, we have to take a 36% discount as the CNA stock is trading at 0.64x book value.

Doing this to all the listed subsidiaries results in the above table.  A total adjustment downward of $400 million is needed to mark these subs to 'market'.

With 397 shares outstanding at L, that's a $1.00/share downward adjustment.  

So with March-end BPS at $48.96, the adjusted BPS would be 47.96/share.   The stock closed today at around $39/share, so the stock is trading at around a 19% discount to this adjusted BPS.  That's cheap.

But then, how has this adjusted book value behaved over the years?   Has it always been trading at a discount to this?  Who needs to own something at a discount if it is declining?

Let's do the above exercise going back to 2001.  I went back and calculated all the p/b ratios for each year-end for each listed sub and did the above adjustment for every year.

L BPS and Adjusted BPS

From the above table, we see that the stock price actually does track the adjusted BPS pretty closely.  The big divergence from book value back in 2007 was mostly due to the rise in DO stock.  And that also explains the poor performance over the five year time frame.

It is encouraging that the adjusted book value has grown at +11%/year over the past ten years, pretty much in line with the growth in book value (so maybe BPS is a good indicator after all).

The chart below shows how closely the stock price tracks the adjusted BPS.

BPS, Adjusted BPS and Stock Price of L

Over the past ten years, the stock has traded at 0.93x this adjusted book value based on yearly data.  So it is fair to say that at $39/share,  L is trading cheaply and can be expected eventually to trade at the adjusted book value.  It looks like the discount started after the financial crisis, so this may just be part of the anti-equity sentiment (many financials and conglomerates like BRK are also trading cheap versus historical patterns).

Anyway, this is just one way of looking at the sum-of-the-parts.  I thought it would be easier to just adjust the book value like the way I did above.  I think it makes sense, but there may be factors that I am missing that makes this not such a good adjustment.  But it seems to track well, and with my very elementary understanding of accounting, it seems at least roughly right so...


Conclusion
So, I do like L and have a lot of faith in the management there.  They are as straight and honest as can be, and this could be a stock I may recommend to people who are going away for a few years on a mission to Mars and can't access the internet or otherwise buy/sell anything.  You would want to own a solid company with good management that can allocate capital well.  L would fit that bill easily.  I would bet on L over most mutual funds, the S&P 500 index etc...

Having said that, this is sort of a pain to analyze.  Sure, the structure is really simple on a one-time snapshot basis.

But we get different performance results depending on if you look at stock price, book value or adjusted book value.  I wish they would put a good summary page on the front like BRK and LUK does.  It would be really helpful.

I was one of those that just took it for granted that L has done well over 25 or 50 years but really didn't have a feel for how they have done over different time periods, and part of that is because it was such a hassle for me to try to figure out.

I haven't changed my mind about L; it's still a great company run by great people, but I feel like I understand it a little bit better now.

Duquesne Family Office Gold Correction

Someone pointed out that the $859 million GLD position was in the form of call options; I missed that (don't go through filings late at night!)  So the increase in the 13F portfolio from $1 billion to $2.2 billion is largely not 'real'; just an increase due to the reporting of the notional amount of the underlying GLD represented by the call options.

That's still heck of a lot of gold, but the exposure is a bit different than owning GLD outright.

Sorry for the error, and thanks for the heads-up!

Duquesne Family Office: Druckenmiller Likes Gold

This is not a blog that tracks the buys and sells of hedge funds or anything like that, but I just stumbled on this so I thought I'd post it.  (Of course, a hedge fund manager liking gold is hardly breaking news, but...)

We all know Stanley Druckenmiller; he needs no introduction (if you do, just google him and there is plenty of short articles about who he is).  He ran the Quantum Fund (Soros' Fund) and then his own hedge fund Duquesne Capital (which he actually started before joining Soros and continued to run during his time there) but closed down Duquesne in August 2010.

I noticed that websites that track hedge funds dropped his investment activities as Duquesne Capital no longer files 13F's.  Well, that makes sense as it no longer exists.

But I did notice by chance that Druckenmiller started filing a 13F again but this time as "Duquesne Family Office LLC".  So this is his own money.

I find this interesting for a couple of reasons.  Of course, Druckenmiller is one of the all-time great traders/investors, so it's always interesting to see what he is up to.  But what is really interesting now is that he is doing a lot of the work for the family office himself.  His colleagues at Duquesne Capital left to start their own hedge fund (which Druckenmiller invested $1 billion in according to one article).

So the 13F stocks show what Druckenmiller himself really likes.  It's not his other portfolio managers' picks.  These are *his* picks.

You will notice that it's a much smaller list than what they used to file as Duquesne Capital.  That makes sense.

Anyway, here is the 13F that was filed in May for the portfolio as of March-end 2012  (Look up Duquesne Family Office at sec.gov).  See table below.

It is a small portfolio in terms of number of stocks and there are many familiar names in there.  It sort of feels right to me as I feel like I can get a sense for why these stocks would be interesting; Wells Fargo is a great bank, YUM Brands is a great way to get exposure to China without having to deal with Chinese companies/fraud etc..., Chipotle is just recreating the fast food business, American Express is a transaction driven high return on capital business (paid on number of transactions, not loans outstanding etc.) etc...

SPDRs!
But then here's the whopper:  5.3 million shares of GLD, the SPDR Gold Trust.  As of March-end 2012, that position was worth $859 million.  The total U.S. equity portfolio on the 13F is $2.2 billion, so that's a whopping 40% of the portfolio invested in gold.    Einhorn has 10% of his assets invested in gold.  40% is pretty big.

[Correction (added later):  The GLD position is in the form of call options and the $859 million is just the notional amount; I didn't notice that the first time I looked at the filing.  ]

As of September 2011, Forbes had Druckenmiller's net worth at $2.5 billion or so.  So this $2.2 billion portfolio constitutes most of his wealth.  Given that, this 40% exposure to gold is pretty big.

This is not like some billionaire that files a 13-F, but the U.S. equity portfolio is only 10 or 20% of his net worth or anything like that.  It looks like this is a substantial portion of Druckenmiller's wealth is represented here.  Of course, we don't know the real exposure as Druckenmiller probably has futures and options positions that won't show up here.  We have no idea what his S&P 500 index futures and gold futures positions are.  Also, I am leaning on Forbes' estimate of his net worth too and I have no idea how close those are, never having been a billionaire with net worth estimated by them.

13-F for Duquesne Family Office LLC filed on May 14, 2012 for Quarter ended March 31, 2012

And here's the other puzzle.  The first filing the family office made was in February of this year and there is no GLD position there.  Here it is:


13-F for Duquesne Family Office LLC filed on Feb 9, 2012 for Quarter ended Dec 31, 2011

Again, this is a nice, small (in terms of number of names), manageable portfolio of stocks.

First thing I notice is that Apple is here but not in the later filing, so Druckenmiller sold his Apple shares in the first quarter during that Apple frenzy.  I haven't tracked back old Duquesne Capital filings, but I think they've owned Apple for a long time. 

Also, he owned JP Morgan as of the end of 2011 but was out by the end of the first quarter.  Did he dump it because he was hearing on the street about the whale and what Dimon would eventually call a tempest in the teapot and then later an egregious mistake?

The other thing, of course, is the GLD position.  It's not there.  So did he just buy gold in the first quarter of 2012? 

The problem with these hedge funds is that you never know what their futures positions are from these filings and there are no futures 13-F's (that I know of). 

So it's possible that Druckenmiller had physical gold in storage at the Fed, had a big futures position he kept rolling or whatever.  And for whatever reason, he converted that into SPDRs.  This is certainly possible. 

That would explain the other thing I notice;  the grand total of the portfolio is $1.075 billion in the December 2011 quarter filing but $2.2 billion in the March-end 2012 filing. 

Since the 13-F is for only U.S. listed equities, there could be a lot of reasons for this.  It could be as simple as the above gold position conversion into an 'equity' position.  It could be due to sales of foreign stocks (that are *not* in the 13-F's) and purchase of U.S. equities (which *are* in the 13-F).

It could also be that he took some money back from people he had run some of his money, but there is no proof of that. But if he had assets managed by others, that would not show up in the 13-F.  So that is certainly a possibility.  It could be any of the above reasons (or another reason I didn't think of).  If I let my imagination run wild, it almost looks like he just yanked a billion or so out of a macro fund he invested in and just put it in GLD's instead.  But that is probably not what happened.   It just sort of fits, but we really have no idea what is going on here (other than that he likes gold). 

[ Correction:  Since the GLD position is just the notional amount of the call option, the above is irrelevant; there was no addition in the dollar value of the portfolio. ]

Anyway, this post is neither here nor there, but I thought it was a little interesting.  I am a fan of Druckenmiller, even though I haven't really followed too closely what he has been doing over the years (except reading about him in the papers). And I normally don't look so closely at 13-F's and try to "diff" the filings from one period to the next. 

Again, this is just something that I stumbled upon, and I think it's sort of a gem in the sense that we get to take a peak at the "pure" Druckenmiller; what he likes with his *own* money doing his *own* thing (and not his staffs), and not under pressure to perform from clients.  It's like peaking into someone's PA (personal account), and that's always fun to do. 

CNA: One-of-the-Parts (of Loews)

Loews (L) has been, for years, a popular sum-of-the-parts play by value investors and it seems that's it's always trading at a discount.  I suppose that's normal for a conglomerate.

I am a big fan of L and have a lot of confidence in the Tisch family but one thing has always been nagging me about L, and that is:  What is up with CNA?  I'm sure I'm not the only one wondering about this.  So I decided to take a closer look.

Now L is a disciplined value investor and stays away from silliness, thinks long term and does all the right things.   They sound a lot like Buffett in many ways.  They concentrate their bets, are very shareholder friendly and all that.

But I just never really got CNA.  I'm sure they bought CNA well in 1974, but the performance there has been less than stellar. Yes, you have to think long term and not worry about short-term problems there. (I was a little surprised that they took such a hit back in 2008, but those were mostly marks against them and BPS has bounced back quickly so maybe it wasn't as bad as it looked at the time; but still, you don't want to see that sort of book volatility!).

So anyway, thinking long term is exactly what we have to do so that's what I did.  Since I can't see long term into the future, I decided to take a long term look at CNA going back. 

And this is what I found:

CNA Book Value Per Share versus S&P 500 (w/dvd)

This is the book value per share (BPS) of CNA versus the S&P 500 index since 1987.

Here is how they compare:

                                      CNA BPS                 S&P 500 (incl dvd)
5 year average                +4.4%                      -0.2%
10 year average              +1.8%                     +2.9%
Since 1987                     +4.5%                      +9.5%

CNA book value per share over the past five years has beaten the S&P 500 index and that is certainly encouraging, especially given what happened in 2008 (they gained back the big loss in 2008 and then some).  (CNA BPS change includes dividends) 

But if you look at CNA over the past ten years and since 1987, it's been pretty dreadful.

A quick calculation shows that if CNA did just as well as the S&P 500 index since 1987, the book value of L today would be $82.40/share instead of $47.49/share (through year-end 2011).  The sum-of-the-parts of the listed holdings plus cash and investments at the holding company would be $80/share instead of $42.59/share (from 2011 annual report).

But of course, we can't just look back and say that; every portfolio is going to have a dog.  We can't say, well, if we owned this instead of that, we would've been better off.  So that's certainly not fair.

But still, from such a prominent value investor, we'd expect better than this for such a large part of the portfolio.

Here is the comparison between CNA (book value growth w/dividends) and the S&P 500 index return (w/dividends) over rolling five year periods.  This is the measure used by Buffett in a recent annual report showing how BRK should retain earnings if they can keep outperforming the index.

CNA versus S&P 500 Index by Five-year Periods

It's kind of stunning how thoroughly CNA has been outdone by the S&P 500 in both up-markets and down-markets. Reading the annual reports, you'd think that it was all about the poor insurance market, but the data shows constant underperformance since 1992 which should have included some good years in the insurance market, economy etc... But you see no trace of that here.


CNA is a Key Value Driver
L has owned most of CNA since 1974 and it's been a large part of the portfolio since then.  Below are some figures that show how big of a factor CNA has been in the value of L over time.

CNA as Percentage of Loews by Market Cap

           %CNA              CNA                 Loews                    CNA %
           owned by L       market cap       market cap             of L mkt cap
1994    84%                 $4.0 bn             $4.8 bn                   70%
2000    87%                 $7.1 bn             $10.2 bn                 61%
2011    90%                 $7.2 bn             $14.9 bn                 43%

CNA as Percentage of Loews by Book Value

             L equity          CNA equity     L share of CNA      CNA equity % L equity
1994        $5.4 bn             $4.5 bn           $3.8 bn                  70%
2000      $11.1 bn             $9.6 bn           $8.4 bn                  76%
2011      $18.8 bn           $11.6 bn         $10.4 bn                  62%               

You can see that CNA is big part of L, but it is getting smaller as other parts grow.  CNA has spun off Lorillard, sold part of DO etc... 

But CNA's holdings still account for 43% of L's market cap and 62% of value based on book value.


CNA Has Done Well in Past Five Years
So it seems like CNA has done well over the past five years, despite their large loss in 2008.  So why is CNA stock so cheap? Shouldn't it get at least book value per share for their outperformance?

To see what's going on, I looked at some other insurance companies.  For comparables, I just took the list  in their proxy statement and didn't include companies that didn't seem like it was in the same business (Ace Ltd is reinsurance, Allstate is mostly retail (auto and home-owners), Lincoln National is life etc...).

Book Value Growth (incl dvd) Since 2001

(CB = Chubb, TRV = Travellers, AFG = American Financial Group, HIG = Hartford Financial)

This is how other insurance companies have done over the past ten years compared to CNA.  CNA is the dark, fat line at the bottom of the chart.  You notice that CNA has done demonstrably worse than the other insurance companies and even underperformed the S&P 500 index (as we already saw earlier).

Book Value Growth (incl dvd) of Various Insurance Companies

So yes, CNA has outperformed the S&P 500 index over the past five years, but when compared to other insurance companies, it doesn't look so good.  It looks good compared to HIG, but that is not such a great comp. CNA underperformed even HIG over ten years.

Reading through the CNA annual reports over the years, I am struck at how often the management talks about the tough competitive environment and horrible pricing and things like that.   They also talk about awful economic environment and disasters that cost them money but these are things that other insurance companies had to deal with too.

Also, management often talks about how they have a great balance sheet,  great relationships with agents, are controlling costs and getting more efficient, focusing on more profitable business lines and things like that, but it's sort of been the same story since at least 1994 (read all of the letter to shareholders and you'll see what I mean (there are some LTS missing since they are not in some of the 10k's).

CNA is Cheap
CNA is certainly cheap at 0.65x book or some such.  But if you look at the chart below, CNA is sort of always trading below book value, and the subpar performance of the business is probably the reason for that.

CNA BPS versus Stock Price

During one of the L earnings conference calls, I think last year, someone asked Tisch what has the most potential upside in the coming year or two; what business has the most potential for value gains at Loews?  (kind of a similar question was asked at the Leucadia annual meeting and Steinberg didn't want to answer the question due to Reg FD).

Tisch responded by saying that CNA stock is really cheap so if that gets revalued, that would be an upside boost to L.  He also said that CNA will get hurt in a rising interest rate environment as they have to own bonds against their insurance reserves, but Tisch said that a stronger economy and improving business conditions in insurance would be a positive offset to that.

So let's see.  What is it going to take to get CNA back up to book value?

Since 1987, CNA has averaged 0.94x book.

Let's take a look at some other insurance companies:

Price-to-Book Value Ratio Comparisons

So we see that CNA has pretty much always traded much cheaper than the other insurance companies (excepting HIG; they are both in the doghouse now).

Obviously, the better performing insurance companies trade at a high p/b multiple.

Expected Returns
So we looked at historical returns of the various insurance companies (book value growth including dividends is the same as comprehensive income for this purpose.  Net income is not so great as realized gains go through the income statement but not unrealized gains, even when unrealized gains increase the value of a company.  Unrealized gains are booked in AOCI (accumulated other comprehensive income) and does change book value of a company.)

So we know that CNA is cheap, but it's cheap also because it hasn't done too well versus peers.

What is the proper value of CNA, then?

For this I was going to do a scatter plot, but realized that five companies isn't going to make a great chart, so I did something simpler.

Expected Returns based on Historical Book Value Growth (plus dvd) and P/B Ratios

In the table above, the average return is just the average growth in book values including dividends (my "comprehensive income"), the p/b ratio is exactly that, and the average expected return is simply the average return divided by the average p/b ratio.  If a company earns an ROE of 20% and you pay 2x book value for it, your expected return is 10%.

So the table above just shows the expected return based on average returns and valuations.  We see that the higher the comprehensive income return on equity, the higher the valuation.  It seems that even though CNA has been cheap averaging 0.7x book in the past five years, that is still not cheap when considering that it only earned an average of 4.4%/year.  So the expected return assuming a 4.4%/year return and paying 0.7x book is still 6%, much lower than the 12% you can earn on CB paying 1.2x book if CB continues to earn the average 14-15% comprehensive return on equity.

If CNA continues to earn 4.4%/year and we need to get expected return to 10% or more like the other insurance companies (except HIG), then CNA would have to trade at 0.44x book value.  This means that CNA may, even at 0.65x book, be 50% overvalued.

For reference on current valuations, here are the P/B ratios based on book value as of March-end 2012 and current stock prices:


                            Current     Current                 Expected return based on:                
             BPS        Price         P/B                        5 yr avg return     10 yr avg return
CNA    $44.48     $28.94      0.65x                        6.8%                   2.8%
CB       $57.37     $72.70      1.27x                       11.4%                  11.8%
TRV    $63.81     $62.92      0.99x                       14.0%                  14.3%
AFG    $45.65     $39.15      0.86x                       13.9%                  15.3%
HIG     $43.25     $17.40      0.40x                        -7.8%                  11.0%


This is the same table with current price and P/B ratios.  Assuming CNA can do as well as it has in the past five years, paying 0.65x book for it will yield an expected return of 6.8%.   So we see that unless there is substantial improvement in the operations of CNA (which is questionable due to the long history of promising comments from management with little improvement in 20+ years), it's hard to see how CNA can get valued much higher.

Paying a premium for CB and even TRV and AFT would yield higher expected returns.

There is a problem in using historical book value growth (or comprehensive returns on equity) as interest rates are very low now and many insurance companies still report investment income in their fixed income portfolios of 4-5% range.  This is because even though interest rates have declined substantially, it will take time for the old bonds with high yields to roll off and be replaced by lower yielding bonds.

But this applies to all insurers, not just to CNA.  So going forward, returns may be lower (unless conditions normalize). Even then, CNA will go lower from a lower base and could even make them unprofitable if rates stay here and the insurance operations keep going as it has been.


Underwriting History
OK, so let's just take a quick look at CNA's underwriting history.  There wasn't an easy way to get historical combined ratios and "float" figures (I didn't want to add up all the different things), so I just used a quick proxy for this.

Operating Profits Excluding Investment Income

So for a proxy of underwriting gains, I just used income excluding investment income (and some others).  I took net earned premiums and other revenues and then deducted "claims, benefits and expenses".  So that may not be the exact combined ratio or underwriting profit, but these figures were easy to pull out of the reports.

So we see that excluding investment income and realized gains on investments, CNA is pretty deeply unprofitable.  That's OK, many insurance companies run at underwriting losses over time.

The columns on the right is "Total Investments" and the column next to that is the operating losses excluding investment income as a percentage of Total investments.  Total investments include float and CNA's own shareholders' equity so is not quite cost of float.

But it does tell you how much the investment portfolio has to earn for CNA to be profitable.  The two basic income streams here is insurance operating profits and investment income.

So you see here that if interest rates continue to go down and investment income goes down, CNA can look a lot worse. 

Limited Partnerships
The other thing that may offer some upside to CNA is their investments in limited partnerships.  They do have $2.2 billion or so invested in them.  They are invested in 79 partnerships that breaks down into 81% hedge funds, 14% private debt and equity and others including real estate.

The hedge funds seek "gains from differentials between securities, distressed investments, sector rotation or various arbitrage disciplines".  46% of the hedge funds are in equity related strategies, 32% in multi-strategy, 19% in distressed and 3% in fixed income.

People have asked on conference calls who CNA has given money to and they won't disclose that.

Anyway, here is some data from their 10K's:

They have $2.2 billion in limited partnerships as of the end of 2011.  I don't think you can calculate the rate of return on these funds as we don't know the cash inflow/redemptions from these funds.  All we know for sure is the investment income that comes from the partnerships, and that is listed in the table above.  That comes to an average of 7.5% since 2001, but that probably understates the actual return.  In 2011, there was $560 million in undistributed earnings included in the $2.2 billion figure above.  I suppose you can calculate the change in undistributed earnings, but that isn't disclosed going that far back.

In any case, I think as a whole, these strategies would probably earn 10%/year over time.  So with $2.2 billion invested, that's $220 million or so in earnings.   Against $12 billion or so in equity, this would add almost 2.0% to pretax return on equity.  If these funds can earn 15% over time, that can add 3% or so to pretax return on equity.

But keep in mind that the above returns are already included in all of the historic return calculations of CNA.  To make it additive to what is already in the history, they would have to do that much better.

Having looked at a bunch of listed hedge funds and private equity companies and their performance, I wonder if they can earn high returns going forward.

I wouldn't want to lean on this to make the CNA idea work.   If the funds can do well, that would be a nice bonus if the rest of CNA can improve.  But these limited partnerships in themselves isn't a reason to get excited about CNA.

What is Loews Thinking?
So what the heck is Loews thinking with this thing?  It has performed horribly over the years. If you ask them, they will say that they own 90% so noone is more disappointed in CNA than them and noone would like to see them improve more than them.  They will say that they think about it all the time and work very hard in trying to turn it around.

As for selling CNA even at this nominally cheap price (but possibly very expensive depending on point of view), I can see Tisch saying that they can sell it, but cash is earning nothing right now and with CNA, they can at least be owning something that earns 4% at a 40% discount and in this environment that doesn't sound like a good sale.

Fair enough. 

(well, how about then repurchasing L stock with the proceeds?)


Thought Experiment
But I wonder if they can't sell CNA, take a loss and use it to offset a gain on a sale of DO and then buy back a ton of stock?

As of the end of March, 2012, the BPS of L was $48.96.  The sum of the listed entities plus net cash and investments of L was $45/share (this is based on the sum-of-the-parts in the 2011 annual report with the figures updated to March-end.  This excludes L's unlisted subsidiaries etc.)

So I will assume L sells all of it's CNA and DO.  The loss on the CNA sale will offset the gain on sale of DO.

In a conference call, I remember someone asking about the cost basis of L's subsidiaries and I think Tisch said that it is complicated and they haven't worked out the exact cost basis for each sub.

But here, I will just use a simple measure which may be totally off, but that's OK.  This is just a thought experiment and we're allowed some guesswork.

For DO and CNA, I will just use book value as the 'cost'.  Since these entities are consolidated, the book value shows the accounting value of the businesses.  If a business is sold above book, a gain is realized and if it is sold below book, then a loss is realized.  I don't really know much about tax and accounting in these situations but let's just assume that is a reasonable ballpark assumption.

Then if we now sell DO, L would get proceeds of $4.3 billion and a gain of $2 billion (because DO is trading at 1.9x book value). 

If we sell CNA now at current prices, L would get proceeds of $6.9 billion and a loss of $3.9 billion (because CNA is trading at 0.65x book value).

So the transaction above would cause a net realized loss of $1.9 billion pretax and cash proceeds of $11.2 billion.

Let's say further that this $11.2 billion was used to repurchase L stock.  I know, you're not going to buy $11 billion of L stock in one shot. 

Anyway, book value of L at March-end 2012 was $19.4 billion.  With a pretax loss of $1.9 billion, let's call that $1.2 billion net after tax, that brings L shareholders' equity to $18.2 billion.

So assuming $11.2 billion is used to buy back stock, the shareholders' equity would drop to $7.0 billion (loss and cash out for share repurchase), and number of shares will go down to 117 million shares ($11.2 billion can buy 280 million L shares at $40/share).

Shareholders' equity of  $7.0 billion and shares outstanding of 117 million shares would bring BPS to around $60/share.

That compares to the current book value per share of $49/share and sum-of-the-parts (excluding unlisted entities) of $45/share.

A big advantage would be that with a much smaller capital base of $7 billion or so, the universe of potential investments can be much larger.  After this hypothetical transaction, L would still have $3 billion of net cash and investments on the balance sheet.

OK, so this may not happen any time soon, and $60/share is not *that* exciting for such a drastic transaction.   Who's going to buy CNA?  DO?  Who's going to sell L?  But that is the potential value embedded in L (or one way to look at it).

I have juggled a bunch of numbers here on this thought experiment; I may have screwed something up but I think it gives a rough picture of what could be...

Synthetic Restructuring
We know that we can jump up and down all we want and say L should sell CNA, even at a loss and they won't do it (well, actually we don't know that they won't.  They just won't say what they are planning!). This is not to say that they are wrong. There are all sorts of other factors that must be considered, and I'm sure the Tisch's are thinking of all the best options for them and I can't imagine them making anything other than a rational decision (even though sometimes we on the outside can't see everything so can't understand things).

But here's an idea. If you don't like CNA, but think CB, TRV or even BRK is better, you can synthetically swap out the business: Just go long L and then sell short the amount of CNA that L owns, and then go long your preferred insurance company.   L has done very well over the years despite the performance anchor of CNA.  Imagine what would happen with a strong insurance company!

Of course, this is not without risk. There can be a sharp rally in CNA for whatever reason (deal, massive restructuring, sudden hardening of the insurance market etc...) and the L share price may not reflect dollar-for-dollar the increase in the CNA value.

Of course CNA can go up and your choice of insurance company can go down; that's what it seems like always happens on these kinds of ideas.

But the fact is, you can restructure CNA by yourself.

Conclusion
I do like L, and I really respect the Tisch family.  I have no doubt they will do very well going forward.  As for the discount to asset value, this has been the case for many years so I wouldn't get too excited about that gap closing any time soon.

If you own L, you have to believe that L will continue to grow intrinsic value per share over time.  I don't doubt that they will.

On the other hand, part of the reason why I looked so closely at CNA here is because it is so cheap.  I was thinking that maybe CNA on it's own would be a good investment or maybe even better than L.  Any turn in the insurance market, improving market and economic conditions may really boost CNA results and get it trading closer to book value.  

That's what I was thinking initially, but a look at the long term history of CNA has made me wonder about that.  It's true that they are doing a little bit better in the past five years or so (despite the jawdropping decline in book in 2008) but improvements at CNA have been promised before.

So I'm not as excited about the potential of CNA as I thought I would be after taking this deeper look.

I hope I am wrong, though, and hope CNA does turn around.  If interest rates stay low and investment returns start coming down as higher yielding bonds roll off, things can get quite ugly at CNA.  It's hard to imagine that they would be able to make up that lost income with underwriting gains given their underwriting history (not so good).

Of course, this concern about CNA can also be a factor in the discount to asset value for L as a whole.

I hope they figure something out.