What we learn about global systemic risk with neurosciences

A. F. Rocha
RANI – Research on Natural and Artificial Intelligence
Rua Tenente Ary Aps, 172 – 13207 110 Jundiaí – Brazil
armando@braineconomics.com
http://papers.ssrn.com/abstract=2316765

Abstract

Financial market crises are frequent events that trigger large financial losses and
represent important global systemic risk (GSR). GSR forecasting is, therefore, a
necessity to avoid the collapse of the global financial market. Understanding GSR
dynamics is imperative if GSR is to be forecasted and controlled. Traditional finance
theories have developed many tools for risk management that proved to be unreliable in
the case of the 2008 Crisis. Recent results on financial decision-making provided by
Neurosciences have being used to model the stock market dynamics. This approach is
used, here, to model stock price evolution in 20 bourses during the period between
January, 3, 2007 and September, 9, 2011. Present results show that: a) the market
humor, calculated as a function of the conflict associated with the stock benefit and risk
evaluations and with the stock volatility, provides an adequate measure of a systematic
global systemic risk, and b) humor threshold variation reflect unsystematic global
systemic risks.

Key words: Financial Crisis, Global Systemic Risk, Neurosciences, Finances,

Neurofinances, Market Humor

1
1. Introduction

Financial market crises are frequent events that trigger large financial losses that
represent important global systemic risk (GSR). For instance, the recent global financial
crisis sent shockwaves through the world economy that initially triggered widespread
liquidity stress among financial institutions in the US. Later, because of the
interconnected nature of their business and operations at the global level, these waves
resulted in strain that impinged upon the affected institutions and transcended national
boundaries at an extremely rapid rate of speed and provoked important financial losses.
(Gai et al., 2011; Nakamia et al., 2011; Silippo, 2011; Upper, 2011, Wong and Fong,
2011). These instabilities are caused or exacerbated by idiosyncratic events or
conditions in financial intermediaries that are, generally, imposed by interdependencies
in the financial system or market (Kaufman and Kenneth, 2003).
GSR forecasting is, therefore, a necessity to avoid the collapse of the global
financial system or market. But this implies the understanding of its dynamics.
Traditional finance theories assume that markets whose prices reflect all the
available information are deemed to be efficient (Fama, 1967) and that betting on one
sole strategy is riskier than diversification (Markowitz, 1952). Prices do random walks
approximate the mean and are unpredictable. Because of this, rational agents do not
achieve higher returns ad infinitum above average market returns on a risk-adjusted
basis (contrarily, lower returns below average market returns) (Fama, 1967). Any
diversified portfolio must take into account the upside of an asset against a systematic
risk or overall market risk, which is represented by the beta (β) within a determined
industry. They must also consider the expected return of the market and the expected
return of a theoretical risk-free asset, which is usually an interest rate practiced within
an economy, such as Fed funds. Diversification must include the selection of assets
having negative correlations to reduce the systematic risk and to be vulnerable only to
unsystematic market instabilities.
The Value at Risk (VaR) is defined as a threshold value that is given a certain
level of significance (usually 90%, 95% or 99% in certain cases), which is the
probability that the loss on a certain portfolio over a given time period can exceed this
threshold value, assuming normal markets and no trading in the portfolio. VaR has been
the most frequently-used risk measurement tool (e.g., De Vries, 2005; Diamandis,

2
2011). However, recently, it has been called into question (Adrian and Brunnermeier,
2008; Daníelsson, 2002), and new approaches are being proposed. Among them,
CoVaR, which was pioneered by Adrian and Brunnermeier (2008), is a measure of
market instabilities that is essentially a reduced-form approach for assessing the degree
of intensification of the economy’s risk.
In contrast to traditional finance theories, Neurofinances considers that investors
are not always rational in their financial decisions (e.g., Kuhen and Knutson, 2005;
Felnner and Maciejvsky; Sanfey et al., 2006; Huettel et al., 2006); that is, they do not
always try to maximize their profits (Sanfey et al., 2006; Rocha and Rocha, 2011).
Emotional influence on decision making has been proposed to explain the irrationality
of the investor’s decision (e.g., Bernenhim and Rangel, 2004; Camerer et al., 2005;
Loewentein et al., 2001; Rocha and Rocha, 2011).
Some studies using functional magnetic resonance imaging (fMRI) shown that
decision making involves the participation of the orbitofrontal cortex (COB), the medial
prefrontal cortex (CPFM), the amygdala (Amig), the nucleus accumbens (NAC) and
other neural structures (eg, Breiter et al, 2001; Camerer and Loewenstein, 2004;
Frederick, Loewenstein and O'Donoghue, 2004; Kuhnen and Knutson, 2005; Knuston
et al, 2003, 2007; McClure et al, 2004). Kuhnen and Knutson (2005) found that nucleus
accumbens activation preceded risky choices as well as risk-seeking mistakes, while
anterior insula activation preceded riskless choices as well as risk-aversion mistakes.
Preuschoff et al. (2006) documents that dopaminergic systems are correlated with the
mathematical expectation conditioned reward and calculated risk as the variance of
reward.
Time elapsed between necessity detection and its possible solution influences
the evaluation of expected reward. The theory of hyperbolic discounting formalizes this
effect (Camerer and Loewenstein, 2004; Frederick, Loewenstein and O'Donoghue,
2004) and McClure et al, 2004 proposed that it depends on interaction between
cognitive and emotional factors mediated by widespread neural circuits.
Two aspects of future thinking influence decision making (Fellow and Farah,
2005; McClure et al, 2004): 1) how steeply rewards are devalued as their delivery is
pushed into the future, a phenomenon known as temporal discounting, and 2) the
perceived dimensions of future time, sometimes labeled ‘future time perspective’.
Although these two aspects of future thinking seem similar, they are not equivalent.
Future time perspective measures a spontaneously chosen time period, which would not

3
necessarily affect the way a person evaluates an event at a specific time in the future
when explicitly cued to do so. Similarly, the rate at which reward decays across a
specified delay may differ across individuals, even if they have a similar future time
perspective (Fellow and Farah, 2005).

Figure 1 – The neuroeconomic model for decision-making

Based on knowledge provided by Neurosciences, Rocha et al. (2009, 2013)

proposed a neuroeconomic based decision-making model (Figure 1) that is dependent
on the evaluation of expected rewards and risks, assessed simultaneously, in two
decision spaces: the personal decision space (PDS) and the interpersonal decision space
(IDS). Motivation to act is triggered by necessities identified in the PDS or the IDS. The
adequacy of an action (e.g., selling or buying a stock s i ) to fulfill a given necessity
(e.g., saving money for retiring) is assumed to be dependent on the expected reward and
risk that are evaluated in both the PDS (savings for oneself) and the IDS (savings for the
family). Conflict generated by expected reward and risk influences the ease (cognitive
effort) and the future perspective of the decision making (short- versus long-term
investment). Finally, the willingness to act (sell or but stock s i ) and willingness not to

act ( or holding the stock s i ) within a future perspective are proposed to be function of

the expected reward and risk associated with s i , as well as of the adequacy and ease of

decision-making regarding selling or buying the stock s i . In this context, willingness to

act increases as expected benefit augments in respect to calculated risk. In the same line

4
of reasoning, willingness not to act increases as calculated risk augments in respect to
expected benefit. Indecision in deciding is determined by amount of conflict. Because
time allocation sets a temporal limit to the decision about implementing or not the
action a i then

possibilit y of acting + possibilit y of not acting + indecision about acting = 1 (1)

and

probabilit y of acting = possibilit y of acting

probability of not acting = possibility of not acting + indecision about acting (2)

whenever conflict tends to zero

probability of not acting = possibility of not acting (3)

because conflict decreases and reduces the indecision to decide.

In a recent review of economic decision making, Seymour and McClure (2008)
show that people are extremely susceptible to manipulation of their expectations and
evaluations of prices. People judge options and prices in relative, rather than absolute,
terms; and they use them as anchored prices. For instance, the value of the stock on the
trading floor is dependent on the buying and selling price offers; if the difference
between these prices is acceptable, then negotiation occurs, and seller and buyer may
converge to a final trading price (Rocha and Rocha, 2011). Contrarily, if selling and
buying price difference is not acceptable, no trade occurs. Buying and selling prices are,
in turn, anchored on the closing prices of the preceding trades (Rocha and Rocha, 2011).
Anchored prices are noisy but trendy. Prices p d on trading day d are equal to

the prices p d −1 at day d-1 plus an amount ∆ d of money, which is dependent on the
difference between the buying and selling price offers (Rocha and Rocha, 2011). They
−
are noisy because ∆ d is noisy. The mean ∆ d , d + k value of ∆ d for the interval of k days
determines the price trend; it is positive for the bull market, negative for the bear market
and null for a stationary market.

5
 Anchored prices contain most of (but not all) the information investors need
because they are trendy and noisy, as well as because ∆ d is not only dependent on the
difference between buying and selling price but also on the market’s humor (Rocha and
Rocha, 2011). This humor, in turn, is dependent on the conflict associated with the
benefit and risk estimation, as well as on price volatility that determines the strength of

the anchoring process. If humor is greater than a given threshold ς d , then ∆ d is

negative; otherwise, it is positive. External forces may change the actual value of ς d ,
which is influenced by media news, government decisions, national and international
events, contagion, among other forces (Rocha and Rocha, 2011, Rocha, 2013).
The purpose of the present paper is to use the neuroeconomic model proposed by
Rocha and Rocha (2011) to model stock price evolution in 20 bourses during the period
between January, 3, 2007 and September, 9, 2011, to show that the market humor,
calculated as a function of the conflict associated with the stock benefit and risk
evaluations and with the stock volatility, provides an adequate measure of a systematic
global systemic risk, and humor threshold variation reflect unsystematic global systemic
risks.
For these purposes, the present paper is organized as follows. Section 2
discusses data about BMFBovespa price evolution and proposes a model for stock
pricing. Section 3 discuses and formalizes the concepts of humor threshold and market
humor and uses them to simulate stock price evolution. Data from BMFBovespa and
other 19 bourses are used to test the model. In Section 4, Principal Component Analysis
is used to disclose the covariation of data among 20 bourses and, based on these results,
global humor (GH) and a global humor threshold (GTH) evolution were calculated to
generate a simulated world (geral) bourse index (GI). It is proposed from this analysis
that: a) evolution of mean price disagreement, conflict, humor threshold and market’s
humor support the hypothesis that humor may be a good measure of the systematic
global system risk; and b) that humor threshold may provide a measure of the
unsystematic global system risk. These are the conclusions proposed in Section 5.

6
2 Pricing assets

In this section, the process of pricing the assets traded in a bourse is discussed taking
into account trading data at BMFBovespa.

2.1 Data

The evolution of BMFBovespa index (IBov) from January, 2007 to September,

2011 is shown in Figure 2. IBov had a positive trend from January, 2007 to May, 2008
reaching the maximum of 73,520 points. This bull market phase was briefly interrupted
in March, 2007 and experienced significant reductions from June, 2007 to August, 2007
and from October, 2007 to April, 2008. From May, 2008 to November, 2008, IBov
experienced a huge reduction that characterized the 2008 Crisis. After that, it began to
increase again and peaked at 71,900 points in March, 2010. This positive new trend was
briefly interrupted in February, June and October, 2009, and also in January, 2010. IBov
abruptly fell during the so-called Greek Crisis in April, 2010 and recovered much of its
value from May to October, 2010, when the Brazilian Presidential Elections took place.
From the moment the election of Dilma Roussef became a certainty, IBov began to
experience a negative trend that was greatly accelerated in August, 2011.

2.2 Analysis and Modelling

The closing value of a stock on the trading floor is dependent on the buying and
s
selling price offers. On any trading day t , the seller expects to get a price p si for

b
selling the share si , while the buyer hopes to buy the same stock at a price p si ; and

they use these prices as anchors to converge or not to a trading price. Therefore, the
willingness to trade (to sell and/or buy a stock) is dependent on stock price

disagreement ( ∆ i = p si − p si ) increasing as ∆si decreases.

s b s

However, it is known since pioneer works of Webber and Fechner in 19th

century, and Stevens (1957), that magnitude perception is a non-linear process,

7
therefore willingness to trade is better modelled assuming to be dependent on the
perceived price disagreement (ppd) calculated as

ppd = f (∆si ) (4)

where f is a non linear function as in case of Figure 2 mapping ∆ si into the closed

interval [0,1], such that ppd=0 means no disagreement (or full agreement) and ppd=1
means full disagreement (or no agreement).

Figure 2 – BMFBovespa Index variation, Selling-Buying price difference and

perceived price disagreement. Price data were normalized for the purpose of a
better graphic display.

Rocha and Rocha (2011) proposed that price variation between trades depends
not only on ppd (see Figure 2) but also on the conflict associated with the expected
stock benefit and calculated risk, as well as on the stock volatility.
. Traditional finance theories propose different ways to calculated risks and
benefits of each stock or of a portfolio. However, as discussed above, human brain
operates with perceived instead of real magnitudes (Vlaev et al, 2011, Walsh, 2003). In
addition, Kuhnen and Knutson (2005) investigated whether anticipatory neural activity
would predict optimal and suboptimal choices in a financial decision-making task. They
found that nucleus accumbens activation preceded risky choices as well as risk-seeking
mistakes, while anterior insula activation preceded riskless choices as well as risk-

8
aversion mistakes. These results clearly show that emotion is a key element on decision
making as well as cognition is. Preuschoff et al. (2006) also describe that dopaminergic
systems is correlated with the mathematical expectation conditioned reward and risk
calculated as the variance of reward. Coates and Herbert (2008) based on a small group
of male traders from London, and during 8 days of analyze, found that when the
participants' testosterone level is higher than their median level they made greater
profits than on days when their testosterone was below their median level. According
the authors the higher profitability is partly related with testosterone's effect on risk, but
the test does not measure how the brain behaves during the decision making process.
These results show that financial decision making is a complex issue in the brain that
recruits even complex molecule networks.
No matter how investor calculated expected benefits and risks, it may be
proposed from the above, that risk and benefits are scaled in a non-linear way to buying
and selling prices. Because of this, Rocha and Rocha (2011) proposed that

perceived benefits = f (∆si ) (5)

perceived risks = f ' ∆si ) (6)

are the neural currency used in financial decision making and are calculated from
different points (selling or buying) of view ( f and f ´ ). However, if trade occurs, then
both seller and buyer will tend to have similar risk and benefit perceptions, because they
converge to equal selling and buying prices. Rocha and Rocha (2011) used this
approach to calculated benefits and risks for each stock traded at BMFBovespa during
the period from January, 2007 to September, 2011.
Conflict in deciding about trading a stock is low either if perceived benefit or
risk is low or high. In contrast, conflict is high if the ratio risk/benefit approaches 1. In
this line of reasoning:

conflict = ω c * benefit * risk , 0 < ω c < 4 (7)

such that conflict decreases as benefit or risk approximates zero and tends toward
ω c / 4 if the ration risk/benefit approach 1. In this context, conflict has values varying

9
from 0 to ω c / 4 ( 0 ≤ conflict < 1 ), and the actual value of ω c determines how the

individual evaluates stress. High values of ω c characterize people that are stress-prone

and low values of ω c characterize people that are stress resistant.

Figure 3 – BMFBovespa Index, Conflict and Volatility

Figures 2 and 3 show that conflict calculated as above, increases during periods
of high ppd and decreases when ppd is low. Conflict and volatility are also correlated as
shown in Figure 3. Volatility is negative when conflict is high and positive when
conflict is low. Conflict and volatility are essential components of market humor (Rocha
and Rocha, 2011). In this context, perceived price variation ( ppv ) is proposed to be
positive if market humor is above a given threshold and to be negative otherwise. In
other words:

total conflict = f (conflict, volatility) (8)

humor = humor threshold - total conflict (9)

and

10
ppv = f ( ppd , humour) (10)

In these conditions, stock price evolution and the stock market index may be
modeled as depicted in Figure 4. Given the closing price at the trading day 0, the selling
and buying offers difference, and the expected benefit and risk, the acceptable price
variation for the next trade is estimated and price evolution is calculated. As simulations
progress, volatility is also calculated and considered to compute market humor.

Figure 4 – Stock price evolution algorithm

3. Market Humor

Data from Figures 2 and 3 shows that price disagreement and conflict increased
during the periods when IBov experienced negative trends. Volatility also experienced
important changes during these periods. The increase of perception of price
disagreement (ppd) during these periods augments the perception of risk and decreases
the expectation of benefit. These two facts increase conflict about prices and change
price volatility. As a consequence, market humor moves from a positive feeling,
characterizing a bull market, to a negative perception, characterizing a bear market.
These humor changes are observed in Figure 5.

11
 Figure 5 – IBov (real and simulated) and Humor

BMFBovespa price evolution (simulated IBov) was simulated according to the

c
model in Figure 4, assuming that closing price pd at trading day d depends on the

acceptable price variation apvd , which is turn is dependent on humord and on

perceived price disagreement ppdd −1 in day d -1:

humord = humor threshold - f (conflict + volatility) (11)

apvd = ∆sdi * ppdd −1 * humord

(12)

pdc = pdc −1 + apvd (13)

where f is a power law function.

Figure 5 shows the evolution of the simulated IBov, assuming a constant humor
threshold of 0.52. Whenever the total conflict (see equation 11) is less than this
threshold amount, humor is positive and prices increase. On the contrary, if the total
conflict is less than the humor threshold, price trend becomes negative. These results
clearly show that although IBov simulation, under the hypothesis of a constant humor
threshold, provides a good approximation of the real IBov evolution, some important
discrepancies are observed, mainly in the period after the 2008 Crisis, when the
simulated values were consistently below the real IBov.

12
 However, the simulated IBov may be better adjusted to fit the real IBov, if the
humor threshold is briefly modified during specific periods as shown in Figure 6. These
modifications were either a reduction of the threshold at the beginning of some periods
of the bear market or an increase of this threshold during most of the price recovery
periods that followed the bear market. These periods of increased threshold
predominated after the 2008 Crisis, and they are important determinants of the IBov
evolution after 2009. These ad hoc humor threshold adjustments are assumed, here, to
be necessary in order to simulate unsystematic humor variation due to external causes.

Figure 6 – IBov, Humor threshold and Adjusted Humor

4. Global financial market

The index evolution, between January, 3, 2007 (trading day 0) and September, 9,
of 19 other bourses in world was simulated in a similar way as described above. The
results are shown in Figure 7 and indicate that the model proposed here is adequate to
formalize trading behavior irrespective of these specific markets.

13
 Figure 7 shows the real and simulated data for some of the 20 bourses studied in
this paper. The results clearly show that the proposed model for financial decision
making and stock pricing seem to adequately describe the dynamics of the trading
at these bourses.

Principal Component Analysis (PCA) was used here in order to study the
interconnected nature of these markets (Gai et al., 2011; Nakamia et al., 2011; Silippo,
2011; Upper, 2011, Wong and Fong, 2011). This analysis showed (Figure 8) that the
indices of the 20 studied bourses covaried according to two well-defined patterns,
having eigenvalues of 12.1 and 3.4 and explaining 92% of the data variance. The first
group of bourses is composed by the markets of New York, Paris, France, Germany,
Ireland, Spain, Italy, Japan and Shanghai. The second group is composed by the markets
of Brazil, Canada, Mexico, Argentina, Singapore, Korea, India, China and NASDAQ.
Although two patterns of index covariation were identified, strong interactions among
all the studied markets is also supported by Principal Components Analysis because
when just one pattern of covariation is forced, all markets except BMFBovespa (load of
0.38), have loads higher than 0.6. These results support the hypothesis that bourses
around the world are interconnect in a global market and subjected to common
influences of worldwide events and decisions.
These strong interactions allowed us to calculate, as normalized means, both the
Global Index (GI) to represent the index variation of these 20 bourses as well as the
corresponding global price disagreement (GPD), as illustrated in Figure 9. The
evolution of these two indices is very similar to the evolution of IBov and BMFBovespa
Selling-Buying Price differences shown in Figure 2.
Once again, a clear correlation between the evolution of price disagreement and
market index is observed. Price disagreement peaked immediately before each

14
important negative GI variation and attained low values whenever GI trend is positive.
PCA showed that price disagreement at most (14) of the 20 bourses had loads higher
than 0.6 and explains 80% of data covariance. This means that conflict about price
offers in the different bourses was correlated and induced similar humor oscillations in
the worldwide interconnected stock market.

Figure 8 – Principal Component Analysis of 20 Bourses’ Indices

Figure 9– Global Index and Global Price Variation. Data are normalized for a
better graphic display.

As predicted by the model proposed here, Global Total Conflict (GTC) nicely
correlated with GI, too (Figure 10). Principal Component Analysis demonstrated that
GTC strongly load (loading > 0.5) on one factor for most of the markets studied.

15
 Figure 10 – Global Index and Global Total Conflict

As prices rose before the 2008 Crisis, the GTC increased and attained a value
approximating 0.8 at the end of July, 2007, after experiencing a sharp increase in
March, 2007 that was accompanied by a temporary reduction of the GI (Figure 10).
During August, 2007, the GI sharply decreased and this reduced GTC to values of
approximately 0.5 that were maintained until the end of May, 2008. At that point in
time, GTC abruptly increased and the GI started to fall in anticipation to the huge bear
market that was established in October, 2008, when the GTC again increased abruptly,
to reach a peak on October, 7, which is associated with the huge GI fall in the beginning
of October. In the sequence, the GTC decreased and attained a minimum value
approximating 0.39 at the end of March, 2009. At that precise time, the GI started the
recovery phase that was interrupted by the Greek Crisis in May, 2010, which was also
associated with an initial GTC increase. After this new crisis, the GTC increased again
until August, 2011, when once more the GI started to experience important losses.
Global Humor Threshold (GHT) data shown in Figure 11 demonstrate that
changes in the humor threshold had a worldwide covariation different that observed for
GTC (see Figure 10). GHT covaried at only 7 of the bourses studied, considering those
loadings greater than 0.5. GHT mean values clearly differed from zero (or the
normalized value of 0.4) at the beginning of the 2008 Crisis and at the beginning of its
recovery, as well as during this August, 2011 Crisis. The fact that GTC and GTH load
differently in PCA suggests that they may be under different influences. While GTC is
mostly dependent on price disagreement and volatility, GTC depended on ad hoc humor
threshold adjustments that were independent for each studied bourse.

16
 Figure 11– Global Index and Global Humor Threshold.

PCA showed that Global humor (GH) had loadings greater than 0.5 for 13 of the
20 studied bourses and data shown in Figure 12 demonstrate that GH is well correlated
with GI evolution during the entire studied period. It was greater than zero (or above the
normalized value of 0.4) when GI was increasing and negative in the bear market
periods.

Figure 12 – Global Index and Global Humor.

It follows from its definition that GH provides the same information given by the
GTC plus the information furnished by the GHT. GTC is dependent on conflict and
volatility, which in turn are dependent on price perception disagreement and stock
volatility. GHT is dependent on ad hoc humor threshold adjustments to better fit
simulated to real index variation. In this context, GTC evolution is very predictable,

17
while GHT is not. Therefore, GH provides an adequate measure of global systemic risk,
incorporating both a systematic component, described by the GTC, and an unsystematic
component, described by the GHT.

5. Conclusions

Global systemic risk (GSR) is the possibility of partial or total financial system
or market collapse due to financial system instability (potentially catastrophic), caused
or exacerbated by idiosyncratic events or conditions in financial intermediaries that are,
in general, imposed by constructed interdependencies in the financial system or market
(Kaufman and Kenneth, 2003). For adequate GSR forecasting and control, the
understanding of its dynamics is crucial. Here, we presented model for financial
decision-making based on knowledge provided by Neurosciences that closely simulated
the evolution of the indices of 20 different bourses from January, 3, 2007 to September,
9, 2011. During this period, the world financial markets experienced five short-lived
periods of bear market in addition to a catastrophic crisis in 2008, and it is now
experiencing another period of instability and important losses.
As predicted by the model, bear markets were characterized by augmented total
conflict (GTC) due to the stock price increase during the preceding bull market, when
GTC began to rise. Although GTC may be a good measure for Global Systemic Risk,
the model was unable to closely reproduce the indices’ evolution. After the 2008 Crisis,
simulated GI was always smaller than the real GI (e.g., see Figure 5) if GH (see
equation 9) was calculated assuming a constant humor threshold of 0.5. It was necessary
to ad hoc adjust humor threshold (see, e.g., Figure 6) to obtain the best index fit of GI
by the simulated data (see Figures 11 and 12).
Although the threshold adjustment was ad hoc, at least in the case of
BMFBovespa, this adjustment correlated with the ratio between negative and positive
media news about the Brazilian Stock Market (Rocha and Rocha, 2011). In addition,
Rocha and Rocha (2011) showed that the humor threshold also has a seasonal
dependency in the case of commodities. Despite these dependencies, GTH evolution is
less predictable than that of GHC.
From the above discussion, it is proposed here that GH is an adequate measure
of the Global Systemic Risk in the World Stock Market, and it is composed of the

18
systematic component represented by GHC and the unsystematic component
represented by GTH.
In this context, a financial crisis is proposed to be preceded by the increase of
GTC (see Figure 10), which if above 0.7 predisposes the market to the influences of
negative economic events or news that decreases the market’s humor threshold GHT
(see Figure 11). The size of this GHT reduction seems to correlate with the crisis
intensity or the experienced losses (see Figures 11 and 12). In contrast, the GTC
decrease promoted by the GI reduction triggered by the crisis (see Figure 10),
predisposes (if below 0.5) the market to influences of positive economic events or news
that increases GHT (see Figure 11). The size of this GHT increase seems to correlate
with the degree of market’s crisis recovery (see Figure 11 and 12). Because of all these,
we propose, here, that GH provides an adequate description of the evolution of the
global systemic risk of the world financial system (see Figure 12).

6. References

Adrian, T., Brunnermeier, M.K., 2008. CoVaR. Federal Reserve Bank of New York,
Staff Report 348.

Allena, F. and E. Carletti (2011) New theories to underpin financial reform Journal of
Financial Stability xxx (2011) xxx– xxx

Alessi, L and C.,Detken (2011) Quasi real time early warning indicators for costly asset
price boom/bust cycles: A role for global liquidity European Journal of Political
Economy 27 (2011) 520–533

Breiter, H. C., Aharon, I., Kahneman, D, Dale, A. and Shizgal, P. (2001) Functional
imaging of neural responses to expectancy and experience of monetary gains and losses.
Neuron, 30:619-639

19
Camerer, C. and Loewenstein (2004) Behavioral Economics: Past, Present, Future. In
Advances in Behavioral Economics, (Camerer, Loewenstein and Rabin, eds). Princeton
Univesity Press, Princeton, pp. 3-52.

Coates, J, Herbert, J (2008) .Endogenous Steroids and Financial Risk Taking on a

London Trading Floor. Proceedings of National Academy Science 105(16), 6167-72.

Daníelsson, J. (2002) The emperor has no clothes: Limits to risk odelling Original
Research Article Journal of Banking &Finance, 26: 1273-1296

De Vries, C. G (2005) The simple economics of bank fragility

Journal of Banking & Finance 29:803-825

Diamandis, F. P., A. A. Drakos, G. P. Kouretas and L. Zarangas (2011) Value-at-risk

for long and short trading positions: Evidence from developed and emerging equity
markets. International Review of Financial Analysis, Volume 20, Issue 3, June 2011,
Pages 165-

Dufrénot G., V. Mignon and A. Péguin-Feissolle (2011) The effects of the subprime
crisis on the Latin American financial markets:An empirical assessment Economic
Modelling 28 (2011) 2342–2357

Fama, E. (1970). Efficient Capital Markets: a review of theory and empirical work.
Journal of Finance, v. 25, n. 2. New York, 1970. p. 383, 389-413.

Fellner, G. and B. Maciejvsky (2007) Risk attitude and market behavior: Evidence from
experimental asset markets. J. Economic Psychology 28:351-364

Frederick, S, G. Loewenstein and T. O’Donoghue (2004Time discounting and Time

Preference; A critical review. In Advances in Behavioral Economics, (Camerer,
Loewenstein and Rabin, eds). Princeton Univesity Press, Princeton, pp. 162-222

Gai P., A. Haldane and Kapadia, S (2011) Complexity, concentration and contagion.
Journal of Monetary Economics in press

20
Huettel, S. A., C. J. Stowe, E. M. Gordon, B. T. Warner and M. L. Platt (2006) Neural
signatures of economic preferences for risk and ambiguity. Neuorn 49:766- 775

Kaufman, George G., and Kenneth E. Scott. 2003. "What is Systemic Risk, and Do
Bank Regulators Retard or Contribute to It?" Independent Review 7 (Winter), pp. 371-
91.

Kim Y, S., S. T. Rachev . M.L. Bianchi ., I. Mitov., F. J. Fabozzi (2011) Time series
analysis for financial market meltdowns. Journal of Banking & Finance 35 (2011)
1879–1891

Knutson, B., Fong, G. W, Bennet, S. M, Adams, C. M., Hommer, D. (2003) A region of

mesial prefrontal cortex tracks monetarily rewarding outcomes: characterization with
rapid event-related fMRI. NeuroImage, 18:263-272

Knutson, B. Rick, S., Wimmer, G. E., Prelec, D, and Lowenstein, G. (2007) Neural
predictors of purchases. Neuron, 53:147-156

Kuhnen, C. M. And Knuston, B. (2005) The neural basis of financial risk taking.
Neuron, 47:763-770

Markowitz, H. (1952). Portfolio Selection. Journal of Finance, v. 7, n. 1, 1952. p. 77-78.

McClure, S., Laibson, D. I., Lowenstein, G. and Dochen, J. D. (2004) Sperate Neural
Systems Value Immediate and Delayed Monetary Rewards. Science, 306:503-507

Namakia,A, A.H. Shirazi b, R. Raei a, G.R. Jafari (2011) Network analysis of a

financial market based on genuine correlation andthreshold method Physica A 390
(2011) 3835–3841
Preuschoff, K., S. Quartz and P. Bossaerts (2008). Human insula activation reflects risk
prediction errors as well as risk. J. Neurosci. 28, 2745–2752.

21
Rocha AF, Burattini MN, Rocha FT, Massad EM. (2009). A neuroeconomic modeling
of Attention-Deficit/Hyperactivity Disorder (ADHD). Journal of Biological Systems
17(4): 597-622.

Rocha, A. and F. T. Rocha (2011) Neuroeconomia e Processo Decisório, LTC, São

Paulo

Rocha, A. F. R. I. R. Lima Filho, H. A. X. Costa and I. R. Lima (2013) The 2008 Crisis
from the Neurofinance Perspective: Investor Humor and Market Sentiment. Economics
of Networks eJournal Vol 5, Issue 74, October 11, 2013
http://papers.ssrn.com/abstract=2332200

Sanfey, A. G., Rilling, J. K., Aronson, J.A., Nystrom, L.e., Cohen, J. D. (2003) The
neural basis of economic decison-making in the ultimatum game Science 300:1755-
1758

Seymour, B. and S. M McClure Anchors (2008), scales and the relative coding of value
in the brain Current Opinion in Neurobiology, 18: 173-178
Sharpe, William F. (1964). Capital asset prices: A theory of market equilibrium under
conditions of risk, Journal of Finance, 19 (3), 425-442

Silipo, D. B. (2011) It happened again: A Minskian analysis of the subprime loan crisis
Journal of Economics and Business 63 (2011) 441– 455

Stevens, S.S. (1957). "On the psychophysical law". Psychological Review 64 (3): 153–
181

Upper, C. (2011) Simulation methods to assess the danger of contagion in interbank

markets Journal of Financial Stability 7 (2011) 111–125

Wong, A. Y-T, (2011) Analysing interconnectivity among economies Emerging

Markets Review 12 (2011) 432–442

Vlaev,I., N. Chater, N. Stewart.and G. D.A. Brown(201) Does the brain calculate value?
Trends in Cognitive Sciences, 15:546-554

22
Walsh, V. (2003) A theory of magnitude: common cortical metrics of time, space and
quantity. Trends in Cognitive Sciences, 7:483-488

23