Empathy, bubbles and mistake of Polish monetary policy

“Investors are concerned, not what an investment is really worth, but with what the market will value it at…three months or a year hence”…”We have reached the third degree where we devote our intelligence to anticipating what average opinion expects the average opinion to be and there are some, I believe, who practice the fourth, fifth and higher degree”

J.M.Keynes , The General Theory of Employment, Interest and Money (1936 London page 152)

Keynes paid a great attention for market psychology and higher orders expectation. In his “General theory…” he compares asset markets to a beauty contest where participants have to chose the faces that other competitors find the most beautiful. His beauty contest may be also helpful in understanding uncovered interest parity failure.

Speculative capital moves in search of the highest total return. Let’s assume for simplicity that the total return has two elements: the interest rate differential and the return on exchange rate appreciation. In words to make money you should borrow in low yielding currency i.e. USD and put deposit in high yielding currency i.e. PLN (Polish zloty) and pray for exchange rate to depreciate slower than the interest rate differential.
Classical economy claims that uncovered interest parity (UIP) holds and expected future exchange rate is equal to the interest rate difference. Meaning that higher yielding currency should depreciate faster, flattening out the possible profit. But classical economy assumes that the valuation process is passive whereas the capital inflow is motivated primary by expectations about future exchange rates.

If sufficiently large group of investors expect that the market will value the high yielding currency too high next period, they will buy it pushing exchange rate down vs. lower yielding currency. This self validating process will be in place as long as capital will continue to flow in and as long as there will remain investors non confident in appreciation of high yielding currency.

This expecational error has finite time singularity where the crash probability is highest. Nowhere is Keynes beauty analogy more relevant than in the characterization of the crash hazard rate, because the survival of the bubble rests on the overall confidence of investors in the bullish trend.

In that view bubbles are result of positive feedback among investors and imperfect information processing. LPPL model developed by D.Sornette is just a tool which is designed to catch the coordinated swing in market opinion. I will not here describe it again ( you my find it in Sornette papers or in my presentation about sugar bubble)

Here I want to concentrate on macro signs/”fingerprints” of positive feedbacks which leads to bubbles and crashes.

Below charts show some stylized facts of EUR/PLN exchange rate fluctuation. I cut the whole decade (1999-2010) into 5 periods (3 bubble formation periods ( I 1999-2001, II 2004-2008, III 2009-2010,)and 2 – anti-bubble / crash periods).

It’s clear in the charts below that every single period of rapid appreciation of PLN (and failure of Uncovered interest parity ) was triggered either by large FDI flow or expectations of future FDI inflow (or both). The first chart indicates that FDI inflow was the initial seed for the bubble which triggered self-reflective capital inflow expectations.

Self-reflective character of speculative inflows is well visible on the second chart (blue line). In 2004 portfolio inflow was not only constant but was even accelerating along zloty appreciation ( the stronger zloty was the more was demanded). This may suggest supply and demand curves were not independent but self-reflective.

Again same situation we could observe most recently. Large interest rate differential between Polish and US rates and expected large FDI inflow created initial bias in expectations for zloty appreciation. This was followed by portfolio inflow which to keep zloty appreciate has to accelerate. This is well visible in the fourth chart which shows that at the beginning of the year inflows were become stronger as foreign holding of Polish treasury bonds jumped to record high level.

The charts also suggest that most recent rapid appreciation of PLN was feed by foreign buying of Polish treasury bonds. To keep zloty appreciate the inflow had to be not only continued but also accelerate which amid PIIGS crises was unlikely. This finding supported LPPL estimation results which indicated that the exchange rate went into “critical” period. But as I said in the previous post the full inflating of the bubble is rather unlikely as only half of this year expected FDIs has been executed.

One period which especially needs attention is period between mid 2007 and end of 2008. For me this period is so important because the bubble which was created then triggered most rapid deprecation of polish local unit ever. What stands out is that at that time Polish monetary policy remained firm on tightening (interest rates up) course (inflation fears) despite the fact that the US monetary policy stance has been changed and also ECB was became dovish at this time suggesting that interest rate cuts were in the pipeline. Very high interest rate disparity and large FDI inflows generated initial bias for PLN appreciation. Because stance of local monetary policy at that time was still hawkish the speculative capital was not flowing through portfolio window (Treasury bonds priser are falling when the interest rates are increasing) but through short term FX instruments (options, SWAPS, Loans). Cumulative inflow to Monetary and Financial Institution (MFI) in that period of time top around EUR40bn!!! Polish exporters to withstand competition amid falling global demand started to buy EUR PUT PLN CALL FX options and financing the premium by selling EUR CALL PLN PUT FX options. As the chart is showing the stronger zloty the more it was demanded a clear self-reflective relationship. To keep zloty appreciate capital had to flow in even faster but process has singularity. Capital inflow slows down in mid of 2008, bubble has busted and a Polish corporations bankrupted as EUR CALL FX options were executed. It would be unfair to blame NBP MPC for that but its shows that inflation targeting framework has to be extended and financial stability has to be also targeted by central banks.

Read More......

worst preforming currency of the month - Polish Zloty (lucky me)

Polish zloty bubble has burst or at least we going through large bifurcation (more likely). A month ago I posted here a short note saying that the bubble is lurking in Polish zloty. At that time I also said that a more detailed analysis will follows however… Since then I was fully concentrated on trading trying building and executing strategy which would utilize findings of my research. Today I have closed EUR/PLN longs and option position. Now It’s time to celebrate but no later than on Monday I will post some short presentation with LPPL estimates for Polish zloty and some stylized facts

Read More......

Latvia - (small) Lat devaluation on cards

Earlier this week heavily exposed to the Baltic region Swedbank reported its third straight quarterly loss. Swanbank also acknowledged that it was planning for a possible currency devaluation but said it’s seeing signs that credit quality in the Baltic region is beginning to stabilize. Indeed the worst may be over but its cleaver to prepare for devaluation in Latvia as the move would help country to recover faster and possibly would limit credit losses.

Just to give a brief characteristic of the country I should say that Latvia peg its currency just after the country won independence from the Soviet Union in 1991. In the recent past on the Latvia’s acceptance into the EU, the economy experienced a huge boom with the growth rates well above 10% per annum. The boom was driven by very strong credit growth financed by FDI inflow mainly into nontradable sectors ( real estate, retail and financial services). As time passes the imbalances built up. The wages went strongly up, the inflation level exceeds the EU level which further eroded the country competitiveness, current account gap top 25% of GDP in 2007.

There were several seeds of economic instability (bad mix policy – too loose fiscal policy combined with lack of monetary sovereignty) but among them most important was positive feedback mechanism from banking sector to private non-financial sector which fueled the housing bubble.

Simply credit growth was stimulating the home prices which were taken as collateral for loans. The more quickly the credit was flowing in the higher were the house prices. This process continues until the credit was unable to grow fast enough to further stimulate the house prices. At that time the private indebtedness top 125% of nominal GDP. (More detailed analysis of that process you may find in the previous post here or in Adrian and Shin paper ). The bad news is that this process works in reverse too (lower prices of homes lead to liquidation of loans which makes the decline more precipitous). In case of Latvia the credit bust was compressed in short time period which had catastrophic consequences on home prices. The figure below is a chart of home prices in capital of Latvia. Now the household’s holds negative equity as the home prices deflated but the nominal debt remained unchanged which suggest that more credit related loses lays ahead. As the households remains heavily indebt the recovery is likely to be very sluggish (L-type recovery is possible)

To get a better insight in situation I just implemented a simple balance sheet approach to Latvia. This analytical approach is different than ESA95 approach and it’s focused to examine balance sheet of the country and potential misalignments and vulnerabilities (more on that methodology here). In particular case of Latvia I focus on credit and currency misalignments so this is not a comprehensive approach. Figure 2 is a excel balance sheet of Government sector, Commercial banks sector and Private –non bank sector (click to enlarge the figure).

First observe that government indebtedness is low(ish) both in local and in foreign currencies. Second somehow modes net liability position of commercial banks masked the fact that the majority of the commercial bank assets consists around EUR 8.5bn (Lat12bn) in FX loans to the domestic nonbank sector. In words FX-risk of the banking system simply has been transferred in credit risk which was quite “natural“ as majority of the local banking sector assets are in foreign hands.

Without the coordinated IMF/EU/WB programs the country balance sheet gaps would not be sustainable would end up with currency crises. As I read IMF lead program as financing bridge to adopt euro (in 2012?) and extricate Latvia from currency risk. However this strategy is very difficult to implement as the government has to cut spending to meet the Maastricht criteria among dramatic economic downturn. In words this strategy is unlikely to win sustainable political support. But this is the minor problem. The major problem is fixing LAT to euro at too low level (maintaining overvaluation i.e. Portugal peg its currency too low to euro which had negative implication for the GDP growth). Latvia to repay the debt would need to increase the competiveness to be able to generate the current account surplus. Amid the crises the Latvian C/A deficit shrunk rapidly but this was achieved by collapse of domestic demand and import. Unfortunately amid weak external demand export also drop but leaser than import which ends up in small C/A surplus.

The easiest way to increase the country competitiveness would be allow currency to depreciate if not then the competitiveness may be raised by adjustment on the real side of economy (decreasing wages etc. which is rather painful process). In case of Latvia because of large FX misalignment devaluation is not a cost free solution but implemented orderly would increase the quality of the credit portfolio in the medium term as it would it would promote higher employment. The orderly devaluation would not necessarily jeopardize Latvian plan to join euro soon(ish) as in ERM2 mechanism currency can fluctuate -+15%. from the central parity. An orderly and small Lat devaluation seems to be a plausible solution.To make it orderly both banks and government has to be prepared. If the Swedbank is preparing for possible devaluation maybe Latvian government is preparing for devaluation too but this is only my interpretation and I may be wrong.

Read More......

IMF Needs to Spot New Investment Bubbles: Geithner

CNBC just reports that U.S. Treasury Secretary Timothy Geithner called on the International Monetary Fund to provide rigorous surveillance to spot new investment bubbles (link here). IMF pays a central (and successful) role resolving the current crises and politicians keep to brothering the role of the fund. Despite the Fund success in dealing with the crises it failed to send a warning signal before the crises broke out. Still may be questioned whether rare high impact events may be predicted but I think is worth trying to broad the fund surveillance toolbox with complex system tools which may give some better insight into the stability of the system as a whole. Unfortunately so far the Fund keeps working with models which prove to be wrong indicators ahead of the recent crises (i.e. chapter 3 of new WEO). I think that there is urgent and general need to extend the economists toolbox with the out-of-equilibrium models

Read More......

The leverage cycle , tipping points, black swans and red dragons

Until recently economists were certain that finance has absolutely nothing to offer macroeconomics on the level theory or practice. In the center of economic theory were money supply and interest rates. However In real world when a homeowner takes out a loan and using say a house as collateral, he must negotiate not only just the interest rate, but also how much he can borrow. In other words equally important to interest rate is leverage rate. Let’s take a look on a simple example (I took this example from John Geanakoplos paper The Leverage cycle who together with Hyun Song Shin deserves credit for excellent analysis of leverage and collateral for functioning the financial system)

If the house costs USD100k he borrows USD80k and pay USD20k in cash we say that the collateral rate is USD100k/USD80K=125% the leverage is USD100k/USD20k =5 loan value is USD80k/USD100k = 80%. If then the value of the house raises let say by 20% collateral rate rises to 150%, the loan value drops to 67% and the leverage drops to 4.

In general leverage falls when value of total assets rise. For households, the change in the leverage and change in size of balance sheet size are negatively correlated. However the picture for financial intermediaries is very different. (This is well documented in the series of papers by Hyun Song Shin. Here you can find a link to her web page at Princeton University).

First let’s assume that the bank or other financial intermediary wants to maintain a constant leverage ratio of 10. It keeps securities worth 100 finances is assets with debt worth 90.

--------ASSETS ---LIABILITIES
Securities 100---- 10 Capital
------------------------90 Debt

Now let’s assume that value of securities increases by 1% to 101. Now the balance sheet looks like this:

-------ASSETS ---LIABILITIES
Securities 101 ---11 Capital
-----------------------90 Debt

Leverage drops to 9.18 but the banks wants to keep it constant at 10 and it has to purchase securities worth 9 and finance this purchase with additional debt. Thus, an increase in the price of the security leads to an increased holding worth 9 So the demand curve is upward sloping

---------ASSETS ---LIABILITIES
Securities 110 ---11 Capital
------------------------99 Debt

This mechanism works in reverse too. If now the securities prices fall so the value of assets kept on balance sheet inches down to 109 the leverage now is to high 109 to 10= 10.9. The bank has to adjust down its leverage by selling securities worth 9.

---------ASSETS--- LIABILITIES
Securities 109 ---10 Capital
-------------------------99 Debt

Thus a fall in the price of securities leads to sales of securities which mean that the supply curve is downward sloping. Leverage targeting entails upward sloping demands and downward sloping supplies.
But in real world financial intermediaries are far from being passive in adjusting their balance sheets. Leverage sector (banks, brokerage houses, hedge funds etc.) uses Value at Risk models to adjust their balances sheets. Adrian and Shin showed that procyclical leverage can be traced to the counter-cyclical nature of VaR.

In other words greater demand for the assets tends to put upward pressure on its prices, and then there is a feedback effect in which stronger balance sheets feed greater demand for the asset which in turn raises the asset price and lead to stronger bigger balance sheets. This process if unchecked can lead asset prices up to the shy and then down to the hell (as this process works in reverse too)

(chart reproduced from Adrian and Shin )

That’s why in absence of intervention leverage becomes too high in boom times and too low in bad times (bust). As result , in boom times asset prices are too high, and in crisis time they are too low. This is the leverage cycle.

Eastern Europe boom and bust

Investment bank portfolios are marked to market but the commercial banks balance sheet is not directly affected by changes in asset prices. The reason behind is that usually the biggest item on the balance sheet of commercial banks are loans to consumers (credit) which are not mark to marked. Figure 1 is a scatter chart of the change in leverage and change in total assets of 5 biggest Ukrainian banks (2000-2008). I choose Ukrainian banks as Ukraine and Baltic countries (Lithuania, Latvia, Estonia) suffers in the aftermath of financial crises the worst GDP contraction in whole CEE region. Although I had access only to limited number of observations but most of the observations are concentrated along the vertical line which suggest that Ukrainian commercial banks were targeting constant leverage ratio.

Even in absence of asset price changes (loans, credit) there is potential room for feedback between the balance sheet size and asset prices through shifts in perceived risk.
As I mentioned earlier when the homeowner takes out a loan and using say a house as collateral, he must negotiate not only just the interest rate, but also how much he can borrow. If then the home price increases the collateral ratio raises which may give false impression that the credit risk is falling.

Figure 2 is a chart of average flat prices in capitals of Ukraine (Kiev), and Estonia (Tallinn). I choose home prices as a “natural estimator” of collateral value as credit boom in CEE was mainly driven by surge in number of mortgages. The chart shows that the prices not only raise but also its pace of growth accelerated as time passes reaching maximum in 2008. This was an effect of slowly (at the beginning) accelerating credit growth, which pushes up the collateral ratio and then shifts down perceived risks. Loan officers were more willing to lend as the collateral ratio keep exploding. The more credit was given the higher was the value of collateral the higher was the credit needed for purchasing home which further erodes the lending standards.
Positive feedback process between the credit and collateral value starts slowly at the beginning the credit is soundly based. But as the amount of debt accumulates, total lending increases in importance and begins to have leading appreciable effect on collateral values.

Figure 3 is a chart of commercial bank assets to GDP ratio in Estonia , Poland, and Ukraine. In Estonia and in Ukraine the ratio grows very rapidly which underlines growth of credit in leading role in the process of asset prices/value collateral surge.

Tipping points, Black Swans and Red Dragons

According to standard economic theory markets are fully efficient and only revelation of new important information can cause a crash. But standard economic theory neither can explain creation of the bubbles nor bust of it. The leverage much better explains the economic process of asset price boom and bust. Question is whether it is possible to predict coming asset price bust? According to Nassim Taleb author of famous book “ the black swan” high impact rare events are unpredictable as their share all characteristic of smaller events.

However Didier Sornette in series of papers showed that high impact rare events may be predictable as they belong to a statistical population, which is different from a bulk of the distribution of smaller events, requires some additional amplification mechanism. He argues that extreme events are even “wilder” than predicted by extrapolation of the power law distribution. He calls these outliers a kings or dragons. Emergence of these wild events requires a positive feedback mechanism which leads to increase cooperative of the whole system , leading to development of endogenous instabilities.

(chart reproduced from D.Sornette presentation)

This is complex system approach which is quite different forms “analytical” economics approach which favors decomposing system in components and analyzing it separately to understand the functioning of the whole. “Analytical” approach is destroying information about interactions/linkages between the parts of the system which may lead to some surprising “emergent” properties of the whole.

The emergent property of leverage sector (i.e Banks) comes competition which leads to synchronization of risk exposure. Let’s imagine a borrower who is coming to the bank for mortgage and says that another bank down the block offers him a credit in value of 110% of real estate he wants to buy instead of 50% he is getting here. If the bank would not follow competitors excessive lending policy its balance sheet would not grow fast enough to stop competitors from takeover, management could be replaced for unimpressive profits etc… Pretty soon both banks eases the lending standards which keep feeding positive feedback process from balance sheet size to asset prices.

The process continues until a tipping point is reached where total credit cannot increase fast enough to continue stimulating asset prices but the whole system is sharing similar risk (real estate). At that time system as a whole is very sensitive to even a small shock but the source of the crash lies in its emergent fundamental instability.

This tipping point is analogues of critical points studied in statistical physics and system before reaches that point may sending precursory signals which can be modeled by LPPL(Log –periodic power Laws.)

Read More......

The Chinese equity bubble - UPDATE

Dieter Sornette and his team just released an updated version of its earlier paper on Chinese equity bubble. This version of the paper includes more details i.e. description of bubble detection tools etc. Link here

Read More......

World stock market: approaching trend reversal?

Another interesting implementation of LPPL model has been just published today. Stanislaw Drozdz and Pawel Oswiencimka in short paper predict that core stocks indexes will face significant correction. A quote:

Based on our ”finance-prediction-oriented” methodology which involves such elements as log-periodic self-similarity, the universal preferred scaling factor 2, and allows a phenomenon of the ”super-bubble” we analyze the 2009 world stock market (here represented by the S&P500, Hang Seng and WIG) development. We identify elements that indicate the third decade of September 2009 as a time limit for the present bull market phase which is thus to be followed by a significant correction

Read More......

Sugar bubble ready to burst

Over the past few weeks, sugar has been on such upward spiral, hitting a 28-year high. Growing demand in Brazil for sugar to be turned into ethanol, coupled with a sharp fall in Indian and Brazilian production, have both sent sugar prices sky high. Many think that the supply shortfall will extend through 2010. Then from fundamental perspective valuation current sugar valuation may be correct as the supply fall short behind the demand. But to speak of supply and demand as if they were determined by forces that are independent of the market participant’s expectations is quite misleading.

It’s true the situation is not so clear cut in case of commodities where supply is largely dependent on production and demand on consumption. But the price that determines the amounts produced and consumed is not necessarily the present price. On the contrary, the market participants are more likely to be guided by future prices, either as expressed in future markets or as anticipated by themselves. In either case it is inappropriate to speak of independently given supply and demand curves because both curves incorporate expectations about future prices.

Those expectations may become self-referential which will lead to the positive feedback process (the higher the price or the price return in the recent past, the higher will be the price growth in the future). Positive feedbacks, when unchecked, can produce runaways which beyond a certain point become unsustainable ending in crash. In short LPPL models developed by Sornette aims, at detecting the transient phases where positive feedbacks operating on some markets or asset classes create local unsustainable price run-ups. The result of the analysis is summarized bellow in figure 1.

I analyzed sugar#11 future time series between September 2007 and September 3 2009. The y axis is logarithmically scaled so that the exponential function would appear as a straight line. LPPL fit exhibit upward curvature which is clear evidence that the prices were growing “super-exponentially”. The projected crash dates are September 5-15 .It must be noted that a good fit of the model to the data series is not a 100% certainty for a crash, but it clearly points at a bubble formation.

Read More......

The Chinese Equity Bubble: Ready to Burst (NOW)

D. Sornette et all just published interesting working paper on possibility of Chinese equity bubble to burst. A quote from summary:

Amid the current financial crisis, there has been one equity index beating all others: the Shanghai Composite. Our analysis of this main Chinese equity index shows clear signatures of a bubble build up and we go on to predict its most likely crash date: July 17-27, 2009 (20%/80% quantile confidence interval).

Read More......

Asset price misalignments and the role of money and credit

ECB just released working paper on the role of money and credit indicators for detecting asset prices misalignments. This is most likely the second ECB publication on early warning asset bubbles indicators this year. Conclusion are quite intuitive that monetary and credit developments may be very useful in predicting asset bubbles burst. I have no problem with that conclusion because positive feedback loop from credit to house prices and again to credit was behind surge of house prices (also in Eastern Europe). I think that that to reduce price to price feedback (to avoid superbubbles creation) central banks should be more focused on credit growth and also closely monitor interactions between collateral valuation (i.e. house prices) and banks willingness to lend.

Read More......

Irrational Exuberance. LPPL signatures in EUR/PLN

People like to go to excesses. I think that what started out with a tulip, maybe four-hundred years ago, and continued through the South sea bubble and all of those sorts of things ( Tronic bubble in 60’s, new economy bubble in 90’s). I’m not saying here that that all human choices are orthogonal to rational ones but in human nature is to heard. Herding can result from a variety of mechanisms, such as anticipation by rational investors of noise traders strategies, agency costs and monetary incentives given to competing fund managers, sometimes leading to the extreme Ponzi schemes, rational imitation in the presence of uncertainty, and social imitation. Many financial economists recognize that positive feedbacks and in particular herding is a key factor that can push prices upward (downward) faster-than-exponentially which if unchecked can lead to bubbles.

Roughly 10 years ago Johansen and Sornette proposed model that attempts to incorporate those ingredients into a traditional rational expectations model of bubbles proposed by Blanchard.

The Jochansen-Sornette model assumes the financial market is composed of two types of investors: perfectly rational investors who have rational expectations and irrational traders who are prone to exhibit herding behavior. The noise traders drive the crash hazard rate according to their collective herding behavior, leading its critical behavior. Due to the no-arbitrage condition, this is translated into a price dynamics exhibiting super-exponential acceleration, with possible additional so-called “log-periodic” oscillations associated with a hierarchical organization and dynamics of noise traders.

A+B*(t-tc)^C*(1+D*COS(w*LN(t-tc)+O)

This so-called log-periodic power law (LPPL) dynamics given by has been previously proposed in different forms in various papers. The power law A−B(t−tc)_ expresses the super-exponential acceleration of prices due to positive feedback mechanisms. The term proportional to cos(w ln(t-tc)+o) describes a correction to this super-exponential behavior, which has the symmetry of discrete scale invariance. This formulation results from analogies with critical phase transitions (or bifurcations) occurring in complex adaptive systems with many interacting agents. The key insight is that spontaneous patterns of organization between investors emerge from repetitive interactions at the micro-level, possibly catalyzed by top-down feedbacks provided for instance by the media and macro-economic readings, which are translated into observable bubble regimes and crashes.

In previous posts I fit the LPPL formula into several financial time series. This time I fit LPPL into EUR/PLN exchange rate. As it is seen on the chart the formula fits remarkably well into the data. Even more surprisingly the LPPL model explains not only the periods when the positive feedbacks let to the zloty overvaluation but also it fits well to the period when zloty was depreciating (2001-2003). This confirms the universality of the process described by LPPL formula which describes well not only bubbles but also antibubbles.

Read More......