Monday, April 27, 2009

The Nature of this Crisis

This is an overview post to give context (based on my own perspective) on a couple other topics I am writing up. I don't think there is anything novel about the content.

There are many different opinions on the nature of this crisis: it is a crisis of confidence that will heal when people come to their senses, it is just a bad version of the types of recessions we are used to since WWII, we are headed for stagflation or hyper-inflation, it's all the government's fault, it's all the fault of greedy bankers, it's contagion from the "subprime crisis", etc.

I am among those who view it as a crisis of excessive leverage (household, corporate, and financial) and believe Hyman Minsky's theories on financial instability (and a few related theories such as Fisher's debt deflation) provide the most relevant explanation. Since the early 1980s (but especially in the 2000s) unsustainable debt was accumulated by individuals and companies and ignored (and even encouraged and made worse) by regulators and central bankers, and much of the debt only made sense (i.e., would be serviceable or repayable) in the context of continually rising asset prices. This is like a Ponzi scheme. Minsky described such dynamics as intrinsic to human nature, for example our tendency to base expectations of the future on the recent past. He suggested that when risk looks low people take on more and more risk and this can reinforce booms in asset prices (e.g., housing) for a while due to positive feedback loops, so people keep taking more and more risk as their actions seem to be rewarded. But eventually reality takes over and the asset bubbles run out of momentum. This dynamic seems to have occurred repeatedly over the last several hundred years each time the last generation that experienced it leaves the picture, the institutional memories are lost, and those newly in power arrogantly believe they are so much smarter than the last generation that such past crises could never happen again!

A lot of the wealth seemingly created over recent years (and decades) was illusionary and unsustainable, based on expectations of asset price growth continuing indefinitely at a significantly faster pace than underlying economic growth. But money and financial markets alone cannot create wealth, they can only facilitate the exchange of property, knowledge, and services between people. Real wealth comes from economic growth which comes from productivity improvements and expansion in demographics. So there should be no expectation that if only confidence recovers, then stocks and housing will quickly bounce back to their former valuations (or even close).

This crisis shares the same underlying causes and dynamics as both the Great Depression and Japan's Lost Decade, though the current debt levels are much larger. There are two commentators in particular who I think make this case very well -- i.e., they explain their reasoning in great detail rather than simply asserting their conclusions and leaving people to wonder which of the many assertions out there to believe.

Steve Keen, an academic economist living in Australia who has studied financial crises since the 1970s, offers what is in my opinion the best description of the causes and nature of this crisis. He has many great blog posts and papers, but here are some to start with:
Keen's work builds on the theories of Hyman Minsky and Irving Fisher among others. The essence of the theory is that the markets, not the government, are really in control of the credit creation process. With debt now at saturated and unsustainable levels and consumers in shock and undergoing a generational shift in mindset toward frugality, the market will now lead to a lower aggregate private debt level, not higher, despite the government's desperate efforts to prevent deleveraging and expand borrowing further. Keen does not expect inflation to come roaring back despite the best efforts of governments to reflate, due to the sheer overpowering magnitude of the deleveraging process relative to the amount of money being "printed". Eventually we should have a slow return to inflation (absent large supply shocks or other events external to the money and credit dynamics of the economy) but that can take a long time — see Japan.

The US private debt-to-GDP ratio was around 300% at the start of this crisis. At the start of the Great Depression, it was around 180% (it spiked to 250% when GDP crashed). In Japan in 1990, private debt-to-GDP was around 108% of GDP based on Keen's data (I am not certain of what is included in this calculation as I have seen other reports that put it around 190% of GDP at that time). The charts below are from Steve Keen but the US one is available in many forms in other places.

[UPDATE 7/14/2009: It appears that the other sources I'd seen were correct and Keen's data on Japan was incomplete. Here is a more accurate graph showing Japan's private debt-to-GDP at 219% in 1989 and around 274% in 1996. I'll leave the above one in so as not to butcher this post too much, but please disregard it.
]

Even if the government prints more base money, that does not magically invoke the historic money multiplier and cause broad money supply and inflation to expand accordingly, because for expansion in aggregate lending to occur there have to be willing and qualified borrowers as well as willing and well-capitalized lenders. But in the experience of both the Great Depression and Japan, individuals and companies chose to pay down debt faster than taking on new debt, so broad money and credit on aggregate actually contracted, even with money seemingly being pumped into the economy. Contracting credit combined with slowing velocity of money (as economic activity slows down) both contribute to deflation.

UPDATE 5/22/2009: Astute analysts such as David Rosenberg point to the "secular shift in attitudes toward credit", i.e., a permanent reduction in willingness and ability to take on debt.

Even Microsoft CEO Steve Ballmer seems to share this view. As quoted by Keen:
We’re certainly in the midst of a once-in-a-lifetime set of economic conditions. The perspective I would bring is not one of recession. Rather, the economy is resetting to lower level of business and consumer spending based largely on the reduced leverage in economy.

The second commentator I want to highlight, Richard Koo, is the chief economist of Nomura research in Japan, and has experience with multiple crises. He was a US economist at the Federal Reserve in the 1980s (involved in the Latin American debt crisis resolution) and has first hand with Japan's experience since 1990. He has a book describing his experience with "Balance Sheet Recessions", and there are two valuable talks with slides available online:
One of his key claims is that Japan was successful in preventing a depression. (With some reservations, I'll tentatively agree at least with respect to the medium term outcome — their long term outlook and ability to repay their government debt is debatable). Japan's relative stagnation has been mocked by the US in particular since 1990, but few stop to think what the alternatives might have been to their current outcome. They lost 3 times their annual GDP in housing and stock market wealth (equivalent to a loss of $45 trillion of US wealth now, see Exhibit 12), yet their GDP never fell sharply like the US's did during the Great Depression, and GDP actually continued rising slowly. Unemployment didn't get anywhere near Great Depression levels. But note that their stock market and housing prices are still down dramatically from peak after 20 years. And Japan's sometimes exaggerated (though real) positive impacts from exports and negative impacts from demographics don't change the core underlying comparison of debt-driven asset bubbles as the root cause of crisis. Granted, Japan's stock market bubble was bigger as measured by P/E ratio, but the US has had a larger earnings bubble (unsustainable "financial innovation" boosting apparent profits to record levels in the last several years), and the real estate bubbles are huge in both cases.

Unfortunately Steve Keen is not optimistic about our situation today compared to Japan's, given the significantly higher debt levels today. [UPDATE 7/14/2009: Note that his post used incomplete private debt data for Japan (see above) so the comparison is not as bad as it looks in his post (thankfully!)... though the US private debt is still higher than Japan's was, and the problem is still more global today.] And today's situation is worse in other respects than just debt magnitude.

One of Richard Koo's most important slides is his explanation of how different types of banking crises differ (Exhibit 19):
This crisis truly is different than most other crises (as well as normal business cycle recessions) due to the weak demand for new loans, the systemic breadth, and the underlying dynamic of collapsing asset bubbles, excessive debt, and fragile balance sheets. And the macroeconomic aggregates of money supply, potential growth, etc that conventional economists often use don't properly account for insolvency at the balance sheet level within the economy.

Here is a graph (Exhibit 7 in Koo's March slides) showing one measure of demand for loans falling in the US today, to reinforce his point in the "types of crisis" exhibit above.


Richard Koo discusses policy solutions as well as underlying causes. My focus is on highlighting his data and analysis on the experience of Japan since 1990 and the US during the Great Depression — his prescriptions and conclusions about the ultimate success of Japan are more controversial and I am somewhat unsure as to my position on those arguments.

While government policy can make some difference in addressing this type of crisis, it cannot solve it, and there are limits to what even the most perfect policy could achieve. The Federal Reserve responses so far have been primarily focused on supporting liquidity (which is important if done in an appropriate way, not entirely true so far) and have so far prevented a systemic meltdown, but liquidity provisions can't solve a crisis of solvency. They can help buy time, which can help in partially repairing balance sheets via earnings. Other government responses to date (TARP, the stimulus bill, etc) target the solvency issues, but their magnitude is much smaller than the size of the crisis, and political limitations are already in play. Note I'm not advocating that a larger government response is necessarily better or more effective over the long run — there are big costs to weigh and no easy answers. Ultimately GDP must reset to a lower level (though I don't know how much lower that will be than current levels) because previous spending levels were supported by an unsustainable growth in aggregate debt that is in the process of reversing.

A valid question is whether all the necessary adjustment has already occurred. This seems very unlikely. The IMF has recently raised its forecast of global losses within the banking system to $4.1 trillion dollars, and it has so far been too optimistic in each forecast it has made. Banks globally have recorded around $1 trillion in losses so far, i.e., only a quarter of the current estimate. UPDATE: Here is a visual representation of the losses taken so far (from this T2 Partners presentation):


The total US private debt-to-GDP ratio has barely fallen, if at all (see the recent US debt-to-GDP chart from Steve Keen above, still at 350% or $50 trillion). UPDATE (5/15/09): This can also be seen in the latest Fed Flow of Funds (Z1) data. As of the end of Q4 2008, household debt had shrunk only $70 billion from peak (i.e., a contraction of 0.1% of total debt outstanding), while both business and financial sector debt were still rising (though rising more slowly). Also, US bank credit (this includes bank loans but not other kinds of debt such as bonds) has barely begun to contract, as shown in the Federal Reserve TOTBKCR data (see chart below).


[UPDATE 7/14/2009: The Q1 2009 Fed Flow of Funds data shows private borrowing contracting faster than public borrowing is growing, so it appears total debt has begun the process of contracting. More on this later.]

So it seems likely that the crisis so far has been primarily driven by credit crunch, loss in housing and stock wealth and income (e.g., disappearing mortgage equity withdrawal), financial sector losses, and slowing velocity of money — but that deleveraging via debt reduction is only just beginning and will likely run for years or possibly even decades, resulting in a continuous drag on GDP growth and a strong deflationary force that has likely not come anywhere close to its peak impact yet. One of the goals of policy makers will likely be to slow down deleveraging as much as possible, but Steve Keen (who predicted this crisis in detail and whose judgement I trust) thinks there is no chance of them stopping it. This makes sense given how large the loss of wealth so far has been and the resulting need to repair balance sheets and increase savings to replace the previous strategy of "saving" by relying on expanding asset price bubbles (especially housing but many others also).

The Tentatively Bullish Case for Treasuries

Overview

I'm writing this post to communicate some reasons why I think treasuries may not be the giant bubble that many people claim. My thoughts here should not be taken as investment advice — you should use your own judgment and decide where I might be wrong.

There are four different reasons most commonly given as to why treasuries might perform badly in coming years (note my focus is on years, not month to month fluctuations):
  1. Rising inflation or hyperinflation, i.e., the prospect of inflation-adjusted returns on current bonds being negative.
  2. Threat of default as government debt grows substantially and investors question whether it can be repaid.
  3. Ongoing new treasury supply from budget deficits and bailouts swamping the bond markets and forcing up interest rates.
  4. China deciding to stop buying our treasuries.
I'll provide some thoughts on each, though my perspective on #3 I have not seen detailed anywhere else, which suggests it is novel, wrong, or simply something that no source I read has bothered to explain in detail because they thought it obvious. On the prospect of wrong — with so many other ideas in economics proving utterly fallacious, I'm not going to rule out it being correct until I see good evidence, especially since the conclusions are in line with historical experience.

The Nature of this Crisis

Please see the related post that explains my perspective on this crisis. It is key context to the arguments here as it presumes no quick "return to normal" is feasible given the magnitude of the asset bubbles that are unwinding. It assumes a long "balance sheet recession", contracting aggregate private debt, and that this is at the core a crisis of solvency not liquidity. If you disagree with this outlook even after reading the explanatory sources I reference, you will likely disagree with much of this post.

The Historical Record

Both in the US during the Great Depression and in Japan since 1990, government bond yields were a strong investment and yields actually declined over each time span. Here are some graphs of yields from Richard Koo's March 2009 presentation:


It's not clear to me why so many commentators (even those who believe this crisis shares the same underlying dynamics as these historical precedents and will not end quickly) ignore the historical record by asserting treasury issuance will force yields and interest rates up. I have yet to see an explanation of which specific differences this crisis has that would lead to a completely different result for longer term interest rates. Yes, the magnitude of the unsustainable debt load is far larger this time, but does that necessarily imply a tipping point in what happens to interest rates? I have not seen such a tipping point described, other than prospect for default (to be discussed).

Amazingly Japan's government bonds even rallied (i.e., yields declined) during its crisis in the context of a declining household savings rate [graph] (the rate dropped from around 15% in 1990 to around 3% in 2008), though Koo's graph above suggests that when corporate debt repayment is added, private sector savings increased on aggregate. The US household savings rate is expected by many [e.g., see CR] to increase from recently being negative to roughly the historical 5-10% range, which may be a relative benefit for the US compared to Japan in terms of finding buyers of treasuries, depending on the size of changes on the US corporate debt side.

Inflation?

Steve Keen explains his inflation outlook (which I agree with) in detail in his Roving Cavaliers of Credit post. An excerpt:
To make a serious dent in debt levels, and thus enable the increase in base money to affect the aggregate money stock and hence cause inflation, Bernanke would need to not merely double M0, but to increase it by a factor of, say, 25 from pre-intervention levels. That US$20 trillion truckload of greenbacks might enable Americans to repay, say, one quarter of outstanding debt with one half—thus reducing the debt to GDP ratio about 200% (roughly what it was during the DotCom bubble and, coincidentally, 1931)—and get back to some serious inflationary spending with the other (of course, in the context of a seriously depreciating currency). But with anything less than that, his attempts to reflate the American economy will sink in the ocean of debt created by America’s modern-day “Roving Cavaliers of Credit”.
According to this view, inflation will take years to return, and when it does, it most likely will return slowly as individuals and companies start to gradually borrow more again, expanding aggregate private debt. [UPDATE 7/14/2009: In Japan, private borrowing has yet to begin expanding again since it peaked — see the blue line in the graph below. Note, I removed Steve Keen's chart that was included here previously as his Japanese private debt data may be incomplete, and replaced it with the chart below.] A key point is, when the private sector is deleveraging, printing money is actually deflationary as it is used to accelerate repayment of debts, unless you print truly huge amounts and effectively target those with the highest propensity to spend. But those are usually the same people with existing debts and even if you give them all mall gift cards, money is fungible so they can still use the cards to buy food and then pay down debt with the money they saved.

Now I'll admit that 30 years is a long time and a lot can happen in that timeframe, so long term treasuries are the riskiest duration with respect to possible negative inflation-adjusted return. However, it's not a clear cut case given the historical precedents of Japan and the Great Depression. Plus, markets have proven over and over that they do not always properly foresee and price in long term outcomes! So hunger for "safe" yield could cause them to rally for years before eventually declining many years from now.

Another relevant quote that summaries the inflation outlook well is from David Rosenberg:
Too much slack in the economy to worry about inflation: The fact that the Fed can state this view, knowing full well that it has dramatically expanded its balance sheet and the money supply, is a testament to the view that the central bank has been leaning against the winds of deflation rather than creating inflation. In our view, the latter will be practically impossible to do in an environment where the underlying unemployment rate is approaching 16% and capacity utilization rates are at all-time lows of 66%. There is simply too much slack in the economy, in our view, for us to be worried over the prospect of inflation or a sustained bear market in bonds.

[UPDATE 5/12/09: Here is a good post from Pension Pulse via NC that also argues that deflation will persist and be supportive of treasuries.]

[UPDATE 5/18/09: Two very useful posts from ZeroHedge that provide a marketplace liquidity/velocity/leverage perspective, including David Roche's liquidity pyramid: The Exuberance Glut and Chasing the Shadow of Money. One relevant quote:
"...the securitization of debt, and creation of derivatives amounted to a huge virtual printing press, primarily fueled by a massive increase in risk appetite which allowed for a huge expansion in the value of claims on financial assets and goods and services. It is worth pointing out, that the Fed has little to no control over this "printing press" at this point, which at last count was responsible for over 90% of the liquidity in the system."
]

Potential for Default

Looking at the historical record, Japan started its crisis around 1990 with a government debt-to-GDP ratio of around 55%. That has since tripled to around 180% (I'm not sure why Richard Koo's graph above shows 140% as I always read elsewhere 170-180%), yet yields are much lower now than in 1990, and the credit default swap market has been pricing the risk of Japanese sovereign default as very low, generally around the lowest of G7 nations. In the Great Depression, US government debt-to-GDP rose to around 130%, again concurrent with interest rates declining. Britain ended World War II with a ratio of 250%! (See the graph below). Yet it managed to pay down this debt. (Note that I read somewhere, though haven't confirmed, that some UK bond coupon rates were forcibly repriced lower, which some have called a form of default, though as far as I know the full principle was returned, plus partial interest).

The US started this crisis with a 60% government debt-to-GDP ratio (or 40% debt-to-GDP actually held by the public). If it rose to Japan's level of 180%, that could reflect an additional $17 trillion of treasury issuance ($14 trillion GDP * 120% additional debt). The US private debt-to-GDP ratio is around 300%, so there is certainly no way that the government can or should swap all private debt for public debt. The biggest threat to treasuries in my opinion would be for the government to truly take on too much, of the order of $20-$30 trillion or more. But that seems very unlikely given how difficult it was to pass the stimulus bill and the TARP. Congress and the public have a very limited appetite for bailouts, so far. PPIP (Public-Private Investment Program) is an attempt to bypass Congress (in part by using the FDIC to lend, since the public won't want to see the FDIC fail therefore it will get any extra funding it needs) but it remains to be seen whether PPIP works and how big it can get.

To put a little caution to my own historical examples — I think a 250% public debt-to-GDP would be much harder to grow out of in the 21st century given how much demographic growth has slowed since mid 20th century, but the point is that 250% is still much bigger than 40-60%. One more related caution — it seems likely that Japan's high government debt is a drag on their growth at this point, whether or not adding to it in response to their crisis was the right thing to do, as simply servicing it (let alone paying it off) must eat up a lot of tax revenue. Yet the markets certainly aren't pricing in Japanese government bond defaults.

This bailout tracker shows the following expenditures versus total commitments:
  • Federal Reserve: $1.6 trillion out of $6.2 trillion
  • Treasury and miscellaneous: $1.1 trillion out of $4.3 trillion
  • TOTAL: $2.7 trillion out of $10.5 trillion
Not all of the commitments so far will be used as some reflect guarantees and not all guaranteed assets should lose 100% of their value. Note how much smaller these actual expenditure numbers are so far than Japan's eventual expenditure as a percentage of GDP. Of course, much more will inevitably be committed in coming months and years.

Also, many of the Federal Reserve programs (loans, etc) lead to an increase in the banks' reserve accounts with the Fed, on which they are now paid interest. Paying interest on reserves (see some details on this in the context of the Fed's balance sheet at econbrowser) is an effective sterilization alternative to issuing additional treasuries via the Treasury Supplemental Financing Program, and means not all Fed programs (perhaps only a few?) lead to an increase in treasury debt issuance.

Overall though, this seems to me to be the factor that could end up being most bearish for treasuries, if political limitations are overcome and the government goes overboard with bailouts. However, I think many treasury bears may not realize just how much smaller our current public debt to GDP is than many historical precedents, and what a historically small fraction of GDP (relatively speaking) the bailouts so far amount to. (Just to be clear, I am not taking a position for or against bailouts here.)

[UPDATE: One other factor may be relevant to whether treasury markets could feasibly price in the expectation of government default: if investors abandon treasuries for fear of default, they need to put those dollars in some other US dollar denominated investment (and even if the investors exchange the dollars for foreign currency, those who bought the dollars face the same choice). If investors choose cash, what is to stop much of that being backed by short term t-bills, effectively still funding the government (even if forcing the government to shift more debt issuance to the short end of the curve to keep yields steady)? If all the money tries to move to other asset classes, what is going to look like a better investment for $5+ trillion dollars? In the context of an economy bad enough that the treasury looks likely to default, is corporate debt going to look healthier? Perhaps not impossible, but has a government's debt ever yielded more than private debt (denominated in the same currency) before in history? (I don't know the answer.) As to the gold option, from what I understand the gold market is not big enough to support this large a shift, and it is a purely speculative investment that offers no yield, but I know there is no shortage of people out there who will claim gold is the answer. I am skeptical.]

[UPDATE 5/21/2009: Felix Salmon discusses "Can America's debt get too big to pay?" and concludes that the answer is no. He also notes that "there really isn’t much of a correlation between the level of a country’s debt and the probability that it’s going to default — especially not when that debt is wholly denominated in domestic currency."]

Too Much Treasury Supply?

[UPDATE: Since this section is especially long, here's an up front summary of the ways that when added together suggest that treasury demand is likely to keep pace with supply:
  • Shrinking aggregate private debt allows public debt to fill the "gap" without crowding out.
  • Increased domestic savings increases demand for treasuries.
  • Bailout money does not disappear into a void — it persists on balance sheets (despite write-downs) and since it most likely replaces fixed income assets it will have a high propensity to generate demand for new fixed income assets.
  • Money markets (cash) buy a lot of Treasury bills.
  • Quantitative easing generates treasury demand and reduces treasury supply.
  • Increased flight to safety as the Great Deleveraging progresses.
  • Arbitrageurs maintaining "logical" spreads between various fixed income types (corporate, government, etc), though in the real world in which the efficient market hypothesis is deeply flawed, this factor probably doesn't contribute much.
]

The idea that the growing treasury bond issuance will compete with and starve other asset markets and/or force up interest rates seems to assume two things that I question. First, it seems to assume that all this money raised by government debt issuance disappears into the void and no longer participates in asset markets. Second, it seems to depend upon the notion of us returning to a "normal" environment in the near future in which qualified private demand for loans continues expanding like it has during the last several decades, leading to aggregate expansion in bond issuance that will compete with treasury debt issuance. But if instead private debt on aggregate is shrinking as Steve Keen and Richard Koo and others expect, the increase in government debt is likely to simply replace the disappearing private debt, and not have a large effect on asset prices or interest rates. In other words, with the same aggregate supply of debt assets (private debt plus government debt) and the same broad money supply (it may shrink but quantitative easing may offset that), then the price of assets relative to debt wouldn't necessarily change over the medium term unless other factors cause a repricing. Yes, prices are set on the margin, and I'll discuss below why marginal demand may keep up with marginal supply, but I think the aggregate supply of assets relative to money supply also matters, especially when measured over years rather than months (a lot can fluctuate short term). Among the separate factors that can of course change are expectations on what prices should be and hence relative supply and demand for the assets at each given price point. But that factor is somewhat independent.

I'll go through some scenarios as a thought exercise to illustrate how new treasury supply may have a high propensity to increase treasury demand. These are generalized examples not exactly matched to the specific programs in place now, but I believe the principles carry over from a balance sheet perspective.

Calculated Risk recently provided a couple good graphical primers on how balance sheets work here and here. Mine will not be so pretty so his may be worth a look first. Also here is a guide to how write-downs work from CreditWritedowns.

--- FISCAL STIMULUS SPENDING: ---

An increase in the US savings rate is contributing to a shortfall in aggregate demand and a corresponding shortfall in tax receipts. While I won't get into the controversial topic of what should be done from a policy perspective, it is feasible for the treasury to increase spending to fill that gap and issue treasuries to fund the spending. The treasuries are likely to be bought by Americans via the increased savings rate. Part of the reason is that there will not be a net increase in private assets looking for buyers in the context of the private sector paying down debt faster than new loans are made. Richard Koo directly addresses (in the video more so than the slides) how increased savings will fund sovereign deficit spending in his October 2008 talk.

Note that the treasury debt is funded not just by bonds and notes with the duration counted in years, but also short term treasury bills that are bought by many money market funds — so as far as I know even cash savings added to money market funds help fund treasury debt issuance. The public debt breakdown on wikipedia shows as of December that $1.9 trillion out of $6 trillion in issuance is treasury bills.

--- TOXIC ASSET PURCHASES: ---

Sample balance sheet of a company before purchase of toxic assets by government:

ASSETS
Cash: $10
Toxic assets: $50
Good assets: $50
TOTAL: $110

LIABILITIES:
Various liabilities: $100
Shareholder equity: $10
TOTAL: $110

Sample balance sheet after purchase of toxic assets by the government (for which government issued $50 in treasuries to raise the cash):

ASSETS
Cash: $60
Toxic assets: $0
Good assets: $50
TOTAL: $110

LIABILITIES:
Various liabilities: $100
Shareholder equity: $10
TOTAL: $110

Now, the question is what happens to that extra $50 in cash on the asset side. My expectation is that it could lead to an almost corresponding increase in demand for treasuries of some duration. This would give the $50 in cash back to the broader market (from which it originally came before the treasury issuance) and put the new treasuries on the balance sheet of the bailed out company. The net effect is essentially the same as if the government directly traded treasuries for toxic assets. I'm not suggesting any of this is fair or commenting on what the "right" solution is — on the face of things there is little good to be said for the taxpayer having to send a stream of future income to companies that made big profits acting irresponsibly during the asset bubble years.

Here are some possibilities for what happens to that money from the government.

If the company keeps it in cash, whatever funding vehicle holds that cash will need to purchase $50 more of whatever underlying instruments the vehicle holds. I don't know what all the typical options are here but at least some could be expected to go to purchasing short duration treasury bills.

If the company wants to earn more than the yield from cash, it will likely look for some other fixed income investment to replace the toxic assets it held. There will continue to be little trust as to the valuation of other private securities on offer in the market, and treasuries are the largest and most liquid market in the world, so that alone provides some support for $50 of treasuries to be purchased. But more importantly, there won't be new private loan assets on aggregate competing for buyers because private debt will still be contracting as people pay back loans faster than taking out new ones. So the $50 increase in treasury supply from when the government raised the cash may temporarily force up treasury yields relative to other assets, which will immediately make them more attractive relative to other assets, and cause an increased propensity for them to be purchased as part of an arbitrage trade by the bailed out company or other market participants. (I acknowledge that arbitrage is unlikely to fully rebalance things as markets aren't completely omniscient/rational, but I would think arbitrage should do better at maintaining spreads within one asset class than prices between different asset classes.) Now it is true that the credit worthiness of the government goes down marginally each time it issues new debt, but I suspect the difference in total government debt relatively speaking at each increment will be so small that the market won't care much in terms of setting relative spread between bond types.

It is possible that the company could choose to purchase other assets (stocks, gold, etc), but it seems that it would have a high propensity to replace one fixed income asset with another fixed income asset. Even if some buyers chose gold, for example, ongoing economic stagnation and a need for income could drive them back to dividend paying fixed income assets. Having assets that provide cash flow is likely especially important for companies such as banks and insurers that will need ongoing income to help keep their balance sheets solvent amid other asset write-downs.

Now a different possibility is that the company would deleverage by paying down liabilities. The balance sheet could then look like:

ASSETS
Cash: $10
Toxic assets: $0
Good assets: $50
TOTAL: $60

LIABILITIES:
Various liabilities: $50
Shareholder equity: $10
TOTAL: $60

Unless they are bank loans (credit created through fractional reserve lending), these paid down liabilities (most likely bonds) will lead to a corresponding increase of the assets of another company or individual that had provided the money.

Sample funding company/individual before bailed out company deleveraged:

ASSETS
Cash: $20
Loan assets: $50
TOTAL: $70

LIABILITIES:
Various liabilities: $10
Shareholder equity: $60
TOTAL: $70

Sample funding company/individual after bailed out company deleveraged and paid off the $50 bond:

ASSETS
Cash: $70
Loan assets: $0
TOTAL: $70

LIABILITIES:
Various liabilities: $10
Shareholder equity: $60
TOTAL: $70

This company/individual then faces the same choices as the one bailed out: invest the cash, buy higher yielding assets, or pay down debt. But even if multiple companies deleverage down a connected chain, eventually the process will stop at companies or individuals with no need to pay down debt, leaving them holding cash or buying other assets.

The point is that the government's $50 treasury issuance led to a corresponding $50 increase in demand for assets of some sort, and in the context of aggregate deleveraging, there seems to be a high propensity for these to be treasuries of some duration. From what I understand, Japan's banks hold a huge amount of Japan's government debt, so I suspect this dynamic probably occurred there. [UPDATE: I have also seen a reference to banks acquiring a large number of government bonds during the Great Depression.]

So there are two places in this process that lead to a "loss" in the amount that marginal treasury demand is increased as a result of the increase in marginal supply:
  • Decrease in bank credit (fractional reserve lending) — however in the US this is the minority of debt (Alea reports 63% of GDP in bank credit and 168% of GDP in direct debt securities. This doesn't quite add up to the 300% of private debt-to-GDP so something is missing).
  • Propensity to put increases in asset side of balance sheet in non-treasury assets. I suspect this will turn out to be low when deleveraging fully takes hold.
The question is what might plug this potential gap in the amount of treasury demand generated as a result of each increase in treasury supply. One answer might be additional flight to safety as the crisis worsens along with an ongoing repricing of risk. Another answer may be found by taking a closer look at the implications of contracting private debt:
  1. Let's assume a situation to start in which the amount of both public debt (treasuries) and private debt are not changing over time. Absent any other external factors, the marginal supply and marginal demand for all debt assets on aggregate will stay roughly balanced with each other, and contribute to setting a price that will be somewhat stable.
  2. Now let's assume the Treasury is issuing more public debt over time. This step will increase marginal supply and push down prices. Then let's say the mechanics explained above come into play (the high propensity for the cash to find its way to treasury purchases) and half (but not 100%, to be realistic) of the bailout cash generates increased marginal demand for treasuries. That will push up prices again but not all the way to the original price level, so they are still lower absent other external factors.
  3. Now let's assume the aggregate supply of private assets is decreasing over time. For example, as homeowners pay down their mortgages, loan assets will convert to cash on the balance sheet of the holders of MBS (mortgage backed securities). With homeowners on aggregate paying down debt faster than adding new debt, the marketplace won't offer a lot of new MBS or other loans opportunities as a replacement. So let's say half of this cash stream is then used to buy treasuries for the same types of reasons discussed previously. This will again push up treasury prices, potentially taking them back to around the original price level.
  4. Perhaps the additive effect of both these factors by which marginal demand for treasuries might increase, combined with flight to safety, is enough to explain why yields declined (i.e., prices rose) during past balance sheet recessions.
The above scenario isn't meant to be mathematically rigorous, it is just a rough sketch of how contracting private debt may play a role in filling the previously mentioned gap.

--- CAPITAL INJECTIONS: ---

Here is a scenario for how a company being recapitalized by the government should similarly lead to an increased demand for treasuries.

Balance sheet before capital injection:

ASSETS
Cash: $10
Toxic assets: $50
Good assets: $50
TOTAL: $110

LIABILITIES:
Various liabilities: $100
Shareholder equity: $10
TOTAL: $110

Balance sheet after capital injection of $50 (for which government issued $50 in treasuries to raise the cash):

ASSETS
Cash: $60
Toxic assets: $50
Good assets: $50
TOTAL: $160

LIABILITIES:
Various liabilities: $100
Shareholder equity: $60
TOTAL: $160

Balance sheet after company writes down $20 of the $50 in toxic assets:

ASSETS
Cash: $60
Toxic assets: $30
Good assets: $50
TOTAL: $140

LIABILITIES:
Various liabilities: $100
Shareholder equity: $40
TOTAL: $140

Note that the writedown shrunk the toxic assets on the left side of the balance sheet and the equity on the right side, but didn't touch the newly injected capital on the left side. So that capital is free to participate in purchases of new assets or in deleveraging by paying off debt, just as in the previous example that covered toxic asset purchases by the government.

Will China Stop Buying?

If China buys less treasuries because the trade deficit shrinks (and thus so does their increase in reserves), then it doesn't really matter to treasuries, as those same dollars are then in the form of American savings rather than Chinese reserves. This is only bearish for treasuries if you believe treasuries are a completely irrational investment that only a foreign central bank should consider. One of the key points of this post is that they may not be an irrational investment. Brad Setser (considered by many an expert on international capital flows) seems to follow a lot of the conventional wisdom but has in recent months been covering this topic a little in his observations. An excerpt from a recent post:
There is a very widespread sense that the US “needs” China more now because it is issuing more Treasuries to finance its fiscal deficit. That isn’t quite true. As a result of the crisis, the US consumer has started to save and American businesses have reduced their investment, so the US “needs” to borrow a lot less from the rest of the world. The US needs to borrow from the world when private Americans do not want to save and the US running a large trade deficit, not when private Americans want to save and the trade deficit is down.

Quantitative Easing

The US Federal Reserve has already embarked on quantitative easing, and it is expected by many to continue for a long time and increase in magnitude. Japan tried this several times as well. The expected effects are controversial. One view holds that the ongoing increase in marginal treasury demand from the Fed can help match the marginal supply increase due to new treasuries being issued, and keep yields lower than they would otherwise be. In addition, this operation results in a lower supply of treasuries held by the market, potentially having a lasting effect on prices (i.e., keeping prices higher as those treasuries stay out of trading supply).

A potential piece of evidence that QE may work to help keep yields down (at least relative to no QE, they can still shift for other reasons) is the recent relative out-performance of agencies compared to treasuries, if I understand this ZeroHedge post correctly. This seems to align with the Fed having started upon the planned buying of $1.45 trillion of agencies but only $0.3 trillion of treasuries. [UPDATE: Additional evidence in these graphs from Calculated Risk.]

Some raise concerns that eventually those treasuries need to be sold back to the market and the market will price in this expectation, but if we might be in a lower leverage higher regulation environment for decades to come, it's not certain that the Fed will ever need to do this -- the increase in base money supply might not be inflationary even in the medium term if systemic leverage (i.e., broader money supply and credit) does not returns to previous levels.

Also while the Fed holds the treasuries, all interest payments on the debt are effectively returned to the government, so this lowers the servicing burden of the aggregate government debt and lowers the probability of default being necessary.

Sentiment

Many people believe that sentiment for an asset class is an important indicator. The theory is that if everyone is bullish than there is no one left to buy, and conversely if everyone is bearish than there is no one left to sell. Barrons Spring 2009 Big Money Poll (found via some valuable commentary from David Rosenberg) shows money managers are 3% bullish, 13% neutral, and 84% bearish on US treasuries! If you believe in sentiment indicators, that indicates extraordinary room for a large "flight to safety" as this crisis worsens. But does the fact that this group of "experts" believes treasuries are a bubble suggest they most likely are? Look at what the same group expected in the Spring 2008 Big Money Poll with respect to stocks! (Only 12% were bearish for their outlook through December 2008, and the bears expected the S&P 500 to end the year at 1230...)

Reasons This Tentative Bullishness Could Prove Wrong

  • Though right now it seems politically unlikely, the government could end up trying to swap more private debt for public debt than can ever hope to be serviced by taxpayers. One of Steve Keen's conclusions in comparing the US to Japan was: "On this empirical record [comparisons of the scale of the respective credit bubbles and the fact that Japan's public debt grew faster since 1990 than its private debt shrank], the portents to the USA to be able to get out of this crisis by debt-financed government spending and direct financial sector bailouts–which effectively swap private debt for government debt–are not good." If the government really were to try too broad a rescue, partial or full default might not be out of the question. [UPDATE 7/14/2009: It appears Keen's private debt data for Japan may have been incomplete, so this conclusion may be worth revisiting.]
  • The potential for significant political instability as a result of the crisis could lead to very unpredictable consequences (intentional default on government debt even if it is serviceable, printing such truly massive amounts of money targeted to those who would spend it that it would cause high inflation, etc).
  • A huge supply shock (energy, other natural resources, labor, etc) could potentially ignite meaningful inflation, though for it to be more than a one time jump in price levels, wage inflation would have to take hold as well, which is unlikely in today's globalized labor market (but that could change).
  • [UPDATE 5/25/2009: The portion of past government tax revenue that came from capital gains will not be funded by Treasury bond buying via the increased savings rate — that tax income is likely gone along with the asset bubbles that drove it. This may be one reason current budget deficits seem to be higher than increased savings, and it is a negative pressure on treasuries. It remains to be seen how big a factor it ultimately is compared with the bullish elements discussed previously. But according to the Congressional Budget Office report from 2002, capital gains are usually between 2 percent and 3 percent of total receipts, which suggests this isn't a big factor.]
  • I'm not sure just how large current underfunded obligations like social security really are, and what impact this will ultimately have.
  • Perhaps my logic in this post is just plain wrong.
[UPDATE 6/9/2009: Since I wrote this post, treasury yields have kept rising. This post did not even attempt to make predictions about the short term — it was about the medium term. There are a number of explanations out there for this short term rise that aren't necessarily bearish for treasuries in the medium term. Here is one partial summary. My quick summary of the factors I've seen: (1) inflation expectations, (2) economic recovery expectations, (3) volatility premium, and (4) convexity hedging. Note that the recent rise in sovereign debt yields has been a global phenomenon, not just US.]