Guest Post: Apocalypse Not: The Dollar

Submitted by JM

Happy Holidays and God bless.   If the season brings you peace, congrats.  If the year has been painful on the job or financial front, accept my sincere wishes that the coming one brings better times.

It’s the holiday season:  whoop-dee-doo and conspiratorially enjoying KOSHER wine, not getting up early to watch Bloomberg (sorry Scarlet), and in general dispelling gloom.  I can’t offer anyone gloom-dispelling economic data, because I don’t see much but government socialism coming out of Pandora’s Box.  I do offer data that moderates fears of a specific kind of apocalypse:  a currency crisis.

The apocalyptic flavor of the month is dollar crisis. One should take the possibility seriously.  The data does offer reasonable assurance that it won’t happen anytime soon.  Yes, even in spite of massive (but not unprecedented) fiscal and monetary craziness, a socialist president, a populist legislature, and seething people just itching for the whole outhouse to go up in flames.  Why doesn’t it make sense that the dollar should be out on its rear while gold or oil and their devotees dance in the street?

Because those distinctly American circumstances don’t incorporate a wide enough range of vision.  The issue isn’t about the United States in a vacuum.  It is about the United States as a reserve currency country whose debt acts as the world financial system’s risk-free collateral.  Before any imminent implosion of the dollar, there is a whole zoo of more flawed economies like Dubai and Greece and dozens of others that would fall.  As old as history itself is the truth that the last ones standing claim the spoils of victory.  The United States has the 61st highest public debt ratio in the world. 

I built a model to understand currency crisis events better using data from fourteen financial crises occurring in the last 25 years.  The data used admits no examples of total military demolition or war, although such crises naturally accompany conquest. 

The model specifies two mechanisms that drive tectonic currency shifts and a big residual.  This means that the quantifiable probability of a currency crisis hinges on:  1) deposits to GDP and 2) public sector debt to GDP and 3) the Unquantified.  

The Cast

It isn’t just a rouges gallery of prodigal sons under consideration (UK and USA excluded, because their stories aren’t over by a long shot).  It’s an eclectic mix of times, places, Swedes, and Thais.  And yes, everyone’s favorite sad sack—Argentina—shows up four freaking times.  In broad brushstrokes, the case studies present three crisis profiles, although there is no doubt that a lot of intermixture is present in nearly all countries.

  • Bank credit booms are common factors in 6 of the financial crises (Argentina ’95, Finland ’91, Japan ’97, Korea ’97, Philippines ’97, Sweden ’91)
  • Bank insolvency and system-wide bank runs were the common factor in (Argentina ’81, Indonesia ’97,  Korea ’97, Ukraine ’98) 
  • Unsustainable fiscal imbalances and current account problems were the cause of four very nasty currency (and sovereign debt) crises (Argentina ’89, Argentina ’02, Russia ’98, Thailand ‘97).

Bank credit booms end in bank credit busts that typically require serious debt repudiation and nationalization (maybe not in name) of the banking system.  Bank insolvencies not remedied through capital injections or bank balance sheet repair end in a run on the country’s financial system.  Repairing the balance sheets is not easy, and often associated with the residue of prior government direction of credit.  The nastiest of all financial crises are the result of government failure to show fiscal restraint.  More information can be found here and here.

The Goods

The whole idea of a currency crisis is a study in extremities:  pure tail risk explanation.  I explored these extremes using the database of Luc Laeven.  Yes, I know predictive models are not a perfect fit for reality… the logistic model I used is just an objective way to organize available information free of emotional and a chunk of cognitive bias.  I didn’t use all available data, as some of it is ill-conditioned; some is too complicated; and with some it just didn’t seem worthy of violating simplicity.

The story the data tells. 

  • “Creditor’s rights” from 1 to 4 (4 = best) is just subjective garbage and it shows.    (Pr  > χ2= .6882 sucks in every way)
  • Current account doesn’t matter much in the model precisely because there are actually three distinct channels through which currency crisis are born.  Korea, Japan, and others had no balance of payments problems but still had financial crises.  (Pr  >  χ2 = .9174… even this sucks)
  • Depositor insurance really has no value in this model at all.  This is probably the result of mixed effects—simultaneously positive because depositor insurance fosters confidence in the financial system, negative moral hazard problems because the buyer doesn’t beware when she should. (Pr  >  χ2 = .9449… totally sucks…)
  • Now we are getting somewhere:  public sector debt/GDP.  It is a clear signal of a government financial health, i.e., living within its means.                                     (Pr  >  χ2= .1563… not utterly lousy)
  • Deposits to GDP is quite interesting.  As the ratio goes up, the probability of a currency crisis goes down.  Governments (elected or autocratic) have strong incentive to protect the saved capital of its citizens, meaning its currency.  The more deposits stored in the financial system, the stronger the state protection of the currency.  (Pr  > χ2 = .1322… not horrible for a logistic regression)

Hard Core Data Porn

The model fit wasn’t great.  Even the most significant parameters indicate only a weak contribution to implied probability.  Some combination of the variables and more sophisticated calibration could very well result in greater significance, but model risk and over-fitting are things to avoid.  You can see the parameter results in Table 1.

The results can be visualized in Chart 1, which graphs the deposits-to-GDP-ratio to the estimated probability of a currency crisis.  Countries with low deposit to GDP ratios uniformly show higher default probability.  There is a clear logic in this:  your elected and appointed officials don’t like, respect, or even give a shit about you.  But they do fear for their jobs.  Seems governments of all composition are somewhat careful to stop screwing around before you get real pissed.  Ninety-nine percent of respondents are against raising the debt ceiling?  Fed audits?  Glass-Steagall coming back?  We’ll see what the mid-terms heave up. 

Stop Worrying and Love the Dollar: The Risks

With a deposit to GDP ratio in the United States 80% and rising, the probability of a currency crisis by government tinkering with the financial system has a likelihood fading like Lindsay Lohan’s career.

A more realistic channel is through explicit government failure to correct an unsustainable fiscal position, and as a result goes crash and burn like Russia, Thailand, Argentina (’89), and Argentina (’02).  To me, this is a question of the central bank balance sheet, as they have been sustaining the current fiscal hubris. 

The “balance sheet” risk of a dollar crisis reduces down Federal Reserve solvency.  The chart below shows a theorized Fed balance sheet with insolvency condition at the bottom (hat tip:  Willem Buiter).  That steaming pile of MBS isn’t the Fed’s only risk.

So long as seignorage income exceeds the cost of business plus payments to the Treasury all is well.  The problems are:

  • Seignorage is limited when the central bank must repay TIPS, because they are CPI linked.  If TIPS issuance becomes too large, recapitalization by the Treasury is the only way out of insolvency.  The Fed becomes Citi, and the dollar gets really scared.
  • Defaults on the (non-performing?) MBS the Fed acquired through direct purchases and unsecured lending to the private sector is the other concern.  Problems arise when seignorage can’t generate sufficient income to cover the losses.  Money printing does not always get a free ride. 

These are non-trivial risks.  But given that the Fed has been able to borrow billions of dollars at essentially 0%, and the MBS securities probably are generating at least some yield, I’m not losing any sleep yet.  The greatest benefit of having the reserve currency of the world is that the United States issues exclusively dollar-denominated debt.  Don’t screw it up by issuing a lot of TIPS, idiots.

The risk I most concerned with is that nature of the universe is unforgiving of mistakes.  It is an unquantifiable risk:  a powerful take-away of the model is its insufficiency.  Quantitative exercises may be free of emotional bias, but they are not free of assumptions.  And the big assumption spoken here is linearity.

Here’s hoping Benny, the squid, and Santa’s ‘lil hermaphrodite (Timmy Geithner) don’t base their lives on linearity just because it is tractable and easy.  The diagram below shows why.  The dashed blue line represents assumed linearity:  the implication is that Fed can manipulate the dollar exchange rate however it wants by applying appropriate incremental interventions.  If they screw it up, they’ll just intervene differently until the situation is reversed.  However, if the world reacts to interventions in a non-linear way, then applying incremental interventions can result in sudden and irreversible shifts from reserve currency status to toilet paper (the red arrow).  The world almost surely lies somewhere between these two regimes.  But nobody knows the exact functional form of the world, so hedge accordingly.

AAA Super-Senior Nihilism?  Come On, Guys…

It is a strange human perversity that bad times make some welcome complete annihilation as a transformative event.  Perhaps it’s just a desire to escape mediocrity—a reset switch that puts everyone back to ground state.  I guess such a reset seems so much easier than the pain-staking process of wealth-building and just plain dealing with the Latvian gambit front and center.  It is not. 

Are there risks in a long dollar position?  Sure.  But there are risks in eating smoked herring and choking to death.  The real issue is the characterization of that risk, and the expected cost incurred if the event is realized.

But as far as the data suggests, we are not at the culmination of the Kali Age when the earth is wanting in tranquility, strong in anger, rising in power, and soon falling.  It could have happened already.  It’s not like the twentieth century wasn’t thoroughly awful what with concentration camps and gulags and the Cultural Revolution… I’m stopping now. 
Cosmic carnage as a fine diversion!  Pity that mundane reality sours the fantasia. 


No comments yet! Be the first to add yours.