=Paper=
{{Paper
|id=Vol-1668/paper9
|storemode=property
|title=Multi-Agent Based Ethical Asset Management
|pdfUrl=https://ceur-ws.org/Vol-1668/paper9.pdf
|volume=Vol-1668
|authors=Nicolas Cointe,Grégory Bonnet,Olivier Boissier
|dblpUrl=https://dblp.org/rec/conf/ecai/CointeBB16
}}
==Multi-Agent Based Ethical Asset Management==
https://ceur-ws.org/Vol-1668/paper9.pdf
Multi-Agent Based Ethical Asset Management
Nicolas Cointe1 and Grégory Bonnet2 and Olivier Boissier3
Abstract. The increasing number of ethical investment funds shows The contributions of this article are the following: mapping of an
how the need of ethics in asset management is growing up. In the ethical judgment process in a BDI architecture and instantiating the
same time, in some markets, autonomous agents are managing a components of this model to the asset management domain. The pa-
larger number of financial transactions than human do. If many per is organized as follows. In Section 2, we introduce the asset man-
philosophers and economists discuss the fairness of different ap- agement domain and present what are ethical considerations in such
proaches for responsible investment, there is no strong proposition a domain. In Section 3, we present a BDI agent architecture in which
today about the implementation of autonomous agents able to take the ethical judgment process presented in [6] is embedded. Thus an
into account ethical notions in their financial decisions. This article autonomous agent can decide on actions to execute based both on
proposes an approach to represent morals and ethics in a BDI archi- ethical principles and preferences and on moral values and rules.
tecture and illustrates its use in the context of ethical asset manage- This BDI agent architecture is instantiated to the asset management
ment. An analysis of a first experimentation on a simulated market is domain. Finally, in Section 5, we offer an agent-based simulation to
given. analyse the system’s behavior.
1 INTRODUCTION 2 ETHICS & ASSET MANAGEMENT
The increasing use of IT technologies in today financial markets is In this section we motivate and identify the needs of introducing eth-
no more limited to the use of communication and automatic match- ical dimensions in the autonomous decision making supporting asset
ing mechanisms but is invading also the decision layer where au- management. Firstly, we briefly present what is morals and ethics,
tonomous algorithms make decisions. In this paper, we are interested then we present asset management domain. Then, we present the
in asset management domain where such a transformation in the trad- main concepts to understand ethics in such a domain.
ing of assets generates several practical and ethical issues4 . The ob-
jective and contribution of this article is use a BDI approach to embed
autonomous trading agents’ decisions with ethical considerations, re- 2.1 Morals and ethics
gardless the speed or efficiency of the trading strategy.
Morals consists in a set of moral rules which describes the compli-
Some people consider the use of automatic management decision
ance of a given behavior with mores, values and usages of a group
as the origin of several bad effects such as market manipulations,
or a single person. These rules associate a good or bad value to some
unfair competition towards small investors and flash crashes by cas-
combinations of actions and contexts. They could be specific or uni-
cading effects. Others argue that it reduces volatility, increases trans-
versal, i.e. related or not to a period, a place, a community, etc. This
parency and stability with a lower execution cost [3]. As shown by
kind of rules grounds our ability to distinguish between good and
some reports [5], ethical investment funds are even more growing
evil. Morals can be distinguished from law and legal systems in the
and taking a significant position on the market. However, werehas the
sense that there is not explicit penalties, officials and written rules
performance of such funds can be measured objectively, their ‘èthi-
[10]. Moral rules are often supported and justified by some moral
cal” quality is more difficult to determine as it determines at least in
values (e.g. transparency, responsibility, ecology). Psychologists, so-
part on the values of the observer.
ciologists and anthropologists almost agree that moral values are cen-
Decisions by autonomous agents to whom human users delegate
tral in the evaluation of actions, people and events [15].
the power to sell/buy assets have consequences in real life [7] and
A set of moral rules and moral values establishes a theory of the
as some investment funds are interested to make socially responsible
good which allows humans to assess the goodness or badness of a be-
and ethical trading, we are interested in the definition of mechanisms
havior and theories of the right which define some criteria to recog-
for making financial agents able to follow ethical principles, moral
nize a fair or, at least, acceptable option. Indeed, humans commonly
values and moral rules. In order to achieve this objective, we use a
accept many situations where it is right and fair to satisfy needs or
model of ethical judgment process proposed in [6] mapped into a
desires, even if it is not acceptable from a set of moral rules and val-
BDI agent model. Such agents can decide to trade assets based on
ues. Those theories are also respectively named theory of values and
the moral and ethical preferences or values of their stakeholders.
theories of right conduct [16].
1 Institut Henri Fayol, EMSE, LabHC, UMR CNRS 5516, F-42000, Saint- Relying on some philosophers as Paul Ricoeur [14], we admit that
Etienne, France, email: nicolas.cointe@emse.fr ethics is a normative practical philosophical discipline of how hu-
2 Normandie Univ, UNICAEN, ENSICAEN, CNRS, GREYC, 14000 Caen,
mans should act and be toward the others. Ethics uses ethical princi-
France, email: gregory.bonnet@unicaen.fr
3 Institut Henri Fayol, EMSE, LabHC, UMR CNRS 5516, F-42000, Saint- ples to conciliate morals, desires and capacities of the agent. Philoso-
Etienne, France, email: olivier.boissier@emse.fr phers proposed various ethical principles, such as Kant’s Categorical
4 http://sevenpillarsinstitute.org/
Imperative [11] or Thomas Aquinas’ Doctrine of Double Effect [12],
which are sets of rules that allow to distinguish an ethical option from fact is part of an asset policy and the second one is part of a mar-
a set of possible options. ket policy. Those policies can be viewed as a set of moral rules. As
Indeed, the core of ethics is the judgment. It is the final step to moral rules cannot be satisfied in all contexts, ethical asset managers
make a decision and it evaluates each choice, with respect to the use ethical principles to make their decisions. By instance “Always
agent’s desires, morals, abilities and ethical principles. Relying on execute the most profitable action which violate as few as possible
some consensual references [1] and previous work [6], judgment is rules” is an example of ethical principle for an ethical asset manager.
the faculty of distinguishing the most satisfying option in a situation, Finally, an asset manager needs to be able to judge that the asset
regarding a set of ethical principles, for ourselves or someone else. exchanged and the modalities of the transaction are both compliant
Finally, if an agent is facing two possible choices with both good with his morals and ethics. To this end, some institutions as authori-
and/or bad effect, the ethical judgment allows him to make a deci- ties, non-governmental organizations or journalists observe markets,
sion in conformity with a set of ethical principles and preferences. funds, asset managers, and companies. From those observations, they
provide evaluations that may be used by funds, companies and as-
set managers to make ethical decisions. For instance, the ethiscore5
2.2 Asset management
is a tool proposed by some journalists to rank a set of hedge funds
The asset management is the art of selecting financial assets (e.g. regarding a given set of values as ecological, political or social con-
equities, bonds, currencies, merchandises and so on) to be bought siderations. According with this tool, a company quoted on a market
and be sold in order to manage a capital, respecting regulatory and may satisfy some criteria as producing sustainable products, having
contractual constraints, and applying an investment policy defined a socially responsible management method and so on, depending on
by the owner of the managed portfolio (a set of assets) in order to the values of the investors, to be considered in an ethical investment
optimize his profit, considering a chosen level of risk. portfolio.
The assets are commonly exchanged on a marketplace, i.e. a sys- Knowing those concepts, our proposition consists in representing
tem designed to match bid and ask orders at the best price and the them explicitly (asset policies, market policies and evaluations) and
best frequency. Different types of matching methods are available, as integrate them in autonomous agents’ decision process in terms of
auctions or order books, and those methods accept different types of values, morals and ethics.
orders, as cancellable or dynamic orders. Marketplaces are actually
more than simple interfaces for buyers and sellers because they also
provide a variety of functionalities: 3 BDI AGENT ARCHITECTURE FOR
ETHICAL ASSET MANAGEMENT
1. to finance companies and institutions by the emission of bonds,
warrants or equities; In this section, we first provide a global view of our architecture and
2. to increase liquidity of the exchanges, i.e. minimizing the impact then focus on the two main components for making agents able to
of a bid or ask order on the price; produce ethical behaviours: goodness and rightness processes.
3. to indicate the value of the assets in real time;
4. to increase the control and monitoring on the economy, by con- 3.1 Global view
tributing to the transparency with the publication of detailed bal-
ance sheets and number of analyses. The agent architecture in which we introduce the necessary repre-
sentations and mechanisms to have agents able to produce ethical
Each asset manager composes the orders to put on the marketplace behaviours is based on a BDI approach [13]. In this approach, the
with a set of options as a possibility of cancellation, the duration of its behaviour of an agent is the result of a deliberation designed to issue
validity, a limit price, an investment strategy and so on. To decide the intentions, to bring about or to react to some world states with respect
best order to perform, the asset manager needs to be well informed to the agent’s evaluation of the situation (represented by a set B of
on the state of the market, through raw data and various indicators. beliefs) and the agent’s goals (represented by a set D of desires).
To be able to produce an ethical behaviour, the basic BDI delib-
2.3 Ethical dimensions of asset management eration cycle must be enriched with a process to evaluate the good-
ness of a behaviour (represented by a goodness process named GP )
Ethical asset management, also called responsible investment or so-
and with another process to evaluate the rightness of a behaviour
cial investment, considers new information in the management de-
(represented by a rightness process named RP ) resulting from the
cision process, as sectors, labels or any indicators on the impact of
execution of actions. To this end, agents are equipped with an ac-
these assets and their underlying on the society. Thus, the morals of
tion knowledge base A and four other knowledge bases that define
an agent (combination of moral values and rules) may be defined
value supports V S, moral rules M R, ethical principles P and ethi-
by an asset policy (e.g. trading nuclear-free assets or never trading
cal preferences e . Moreover, agents are equipped with an ontology
in the defense sector). Moreover, the manner to trade is important
O = Ov ∪ Om of moral values Ov (e.g. carefulness, ecology or
too. In the last decade, the introduction of autonomous agents on the
transparency) and moral valuations Om (e.g. moral, quite good or
marketplaces comes with new harmful practices (e.g. layering, quote
immoral). The global architecture is given in Figure 1 and is issued
stuffing, spoofing). Therefore, the morals of an agent may also rely
of the judgment process proposed in [6].
on transparency, honesty or avoidance of any manipulation of the
In our agent architecture, each action of A is described as a pair
market. Such policies are not about assets, but about the morality of
of conditions and consequences bearing respectively on beliefs and
agents’s behaviors on the market.
desires. Perception P er and communication Com functions update
For instance, an ethical asset manager in Islamic finance may
beliefs and desires from, respectively, perception of the environment
both agree on the fact to “exclude stocks of companies that pro-
duce/distribute prohibited goods/services regarding the Shari’ah” [2] 5 http://www.ethicalconsumer.org/buyersguides/money/
and the fact to “prefer to deal with other Islamic agents”. The first ethicalinvestmentfunds.aspx
and communication with other agents. From its beliefs B and de- In addition to moral values, an agent must be able to represent
sires D, an agent executes an Evaluation Process EP to assess both and to manage moral rules. A moral rule describes the association
desirable actions, Ad ⊆ A (i.e. actions that allow to satisfy the con- of a moral valuation m ∈ Om to actions or moral values in a given
sequences of the action), and executable actions, Ac ⊆ A (i.e. ac- situation. A moral rule is a tuple hw, o, mi ∈ M R where w is a
tions whose conditions are satisfied on the current beliefs about the situation of the current world described by w ⊂ B ∪ D interpreted
world). The evaluation process EP produces desirable actions Ad as a conjunction of beliefs and desires, o = ha, vi where a ∈ A and
and executable ones Ap from B and D. At the end of the process, we v ∈ Ov , and m ∈ Om is a moral valuation that qualifies o when w
find a classical deliberation function that generates the intentions to holds. For instance, some rules may be represented as follows:
execute given the right actions of Ar .
hBel(sector(α, medicine)), hbuy(α), _ i, morali
Data flow
Awareness Process hBel(going down, α), h _ , caref ulnessi, quite goodi
VS MR Evaluation Process
Goodness Process A moral rule can be more or less specific depending on the situa-
Rightness Process tion w or the object o. For instance “Transparency is good” is more
Knowledge base
B ME Am State general (having less combinations in w or o, thus applying in a larger
Function number of situations) than “To sell an asset in a quantity superior than
Com the available bid between the current value and the moving average
A CE Ac EE Ae J Ar Del I minus five percent is immoral”. Classically, moral theories are clas-
Per sified in three approaches using both moral values and moral rules as
P e defined above, we can represent such theories.
D DE Ad
• A virtuous approach uses general rules based on moral values, e.g.
“Ecology is moral”,
Figure 1. Ethical BDI agent architecture • A deontological approach classically considers rules concerning
actions in order to describe as precisely as possible the moral be-
havior, e.g. “Buying an asset of an eurolabel certified company is
moral”
3.2 Goodness process • A consequentialist approach uses both general and specific rules
concerning states and consequences, e.g. “Investing in an asset of
The goodness process GP identifies moral actions Am ⊆ A6 given an company that will practice animal testing is not moral”.
the agent’s beliefs B and desires D, the agent’s actions A, the agent’s
value supports V S given moral values and M R moral rules knowl-
edge base. It is defined as: 3.3 Rightness Process
From the sets of possible (Ap ), desirable (Ad ) and moral actions
GP = hV S, M R, Am , M Ei
(Am ), we can introduce the rightness process RP aiming at assess-
ing the rightful actions. As an ethical agent can use several ethical
where M E is the moral evaluation function:
principles to conciliate these sets of actions, we consider a prefer-
M E : 2D × 2B × 2A × 2V S × 2M R → 2Am ence relationship between those principles. Thus, a rightness process
RP produces rightful actions given a representation of the agent’s
In order to realize this goodness process, an agent uses knowledge ethics. It is defined as:
that associates moral values to combinations of actions and situa-
tions, meaning that the execution of the actions in these situations RP = hP, e , Ar , EE, Ji
promotes the corresponding moral values.
where P is a knowledge base of ethical principles, e ⊆ P × P an
We represent this knowledge through value supports. A value sup-
ethical preference relationship, Ar ⊆ A the set of rightful actions
port is a tuple hs, vi ∈ V S where v ∈ Ov is a moral value,
and two functions EE (evaluation of ethics) and J (judgment) such
and s = ha, wi is the support of this moral value where a ⊆ A,
that :
w ⊂ B ∪ D. Here, the precise description of a moral value through a
EE : 2Ad × 2Ap × 2Am × 2P → 2E
value support relies on the language used to represent beliefs, desires
and actions. For instance, from this definition, carefulness supported where E = A × P × {⊥, >}.
by “do not buy any asset α if the volatility V is over a limit Vlimit ”
may be represented by: J : 2E × 2 e → 2Ar
hhbuy(α), {Bel(V ≥ Vlimit )}i, ¬caref ulnessi An ethical principle is a function which represents a philosophical
theory and evaluates if it is right or wrong to execute a given action
where α represents any asset, Bel(V ≥ Vlimit ) is a belief represent- in a given situation regarding this theory. For instance “It is right to
ing the context for which executing the action buy(α) does not sup- do the most desirable action which is, at least, amoral” may be a very
port the value caref ulness. A moral value may also be a subvalue simple principle. Formally, an ethical principle p ∈ P is defined as:
of another more general one, i.e. all its value supports also support
the more general one. p : 2A × 2B × 2D × 2M R × 2V → {>, ⊥}
6 A
m * Ad ∪ Ac because an action might be moral by itself even if it is not The ethics evaluation function EE returns the evaluation of all
desired or feasible. desirable, feasible and moral actions (resp. Ad , Ap and Am ) given
the set P of known ethical principles. Given a set of actions issued executed parts are removed from the CLOB. At the end of the inser-
from EE, the judgment J selects the rightful action Ar to perform, tion, the new best bid is p − 1 and the new best ask is p. All these
considering a set of ethical preferences (defined as a partial or total changes are perceived by the agents.
order on the ethical principles). For instance, a principle P 1 ∈ P Agents get a set of beliefs describing the market and their port-
may be “if an action is possible, desirable and motivated by a moral folio, making them able to represent and reason on the current sit-
rule, it is right to do it” and a principle P 2 ∈ P “if an action is uation. Agents also perceive each minute a set of statistics on the
possible, desirable and at least not immoral, it is right to do it”. If activity of each asset: the volume v (the quantity of exchanged as-
P 1 e P 2, the agent will select a right action according with P 1 sets), two moving average prices mm and dblmm, respectively the
and, if it is not feasible, a right action regarding P 2. The right action average price on the last twenty minutes and on the last fourty min-
Ar is transmitted to the classic deliberation function to choose the utes, the standard deviations σ of prices, the closing prices on this
intention I to execute. period, and the up and down Bollinger bands (respectively mm + 2σ
and mm − 2σ).
The agents’ perception function provides the following beliefs
4 AGENCY FOR ETHICAL ASSET from the environment:
MANAGEMENT
indicators(Date,Marketplace,Asset,Close,Volume,
This section describes the experiment used to to illustrate and eval- Intensity,Mm,Dblmm,BollingerUp,BollingerDown)
uate the use of the architecture presented in the previous section.
We have implemented a multi-agent system that simulates a finan- onMarket(Date,Agent,Portfolio,Marketplace,
cial market where some autonomous ethical trading agents exchange Side,Asset,Volume,Price)
assets. This system has been implemented using the JaCaMo plat- executed(Date,Agent,Portfolio,Marketplace,
form where agents are programmed using the Jason language and Side,Asset,Volume,Price)
the market place is based on artifacts from Cartago.
The ethical agents are initialized with a set of beliefs about ac-
tivities of the companies (e.g. EDF7 produces nuclear energy) and
4.1 Financial market modeling some labels about their conformity with international standards (e.g.
We consider a marketplace where autonomous trading agents have Legrand8 is labeled FSC).
the possibility to manage portfolio of assets and to sell or buy assets
(both currencies, i.e. money, and equity securities, i.e. part of a capi- 4.2 Ethical settings
tal stock of a company) on the market. The set of actions that an agent
can execute on the market are “buy”, “sell” or “cancel” orders. They The ethical agents know a set of organized values: for instance “envi-
respectively correspond to the exchange of an equity for a currency, ronmental reporting” is considered as a subvalue of “environment”.
the opposite way and cancellation of a proposition of exchange if They are declared as :
this order is not yet executed. These actions will be the ones consid-
ered in the ethical deliberation cycle of an agent. Agents can specify value("environment").
subvalue("promote_renewable_energy","environment").
a limit price or can accept the current market price. Each equity is subvalue("environmental_reporting","environment").
quoted in a state-of-the-art Central Limit Order Book (CLOB) [3]. subvalue("fight_climate_change","environment").
A CLOB simply stores and sorts by price the set of “buy” and “sell”
orders (respectively placed on bid and ask sides of the same order They also have a set of value supports as “trading assets of nu-
book) provided by the agents. When an agent put an order on the bid clear energy producer is not conform with the subvalue promotion of
or ask side, the CLOB obey the following rules (see Figure 2): renewable energy”, “trading asset of an FSC-labeled company is con-
form with the subvalue environmental reporting” and “trading assets
• if there is no order to match with, the order is added, of nuclear energy producer is conform with the subvalue fight against
• if there is an order to match with, both the incoming and the climate changes”. Some examples of value supports are:
present orders are filled, and the rest of the biggest, if any, is placed
in the CLOB (and may eventually match with another order). ~valueSupport(buy(Asset,_,_,_),
"promote_renewable_energy"):-
activity(Asset,"nuclear_energy_production").
Price Price Price
Bid side Ask side Bid side Ask side Bid side Ask side valueSupport(sell(Asset,_,_,_),
p+3 Ask5 Ask5 p+3 Ask5 Ask5 p+3 Ask5 Ask5 "environmental_reporting") :-
p+2 Ask10 p+2 Ask10 p+2 Ask10 label(Asset,"FSC").
p+1 Ask6 p+1 Ask6 p+1 Ask6
Bid8 p Ask13 EXE p EXE Ask5 p Ask5
Bid5 Bid5 Bid5 Bid5 Bid5 Bid5
Agents are also equiped with moral rules stating the morality of
p−1 p−1 p−1
Bid8 p−2 Bid8 p−2 Bid8 p−2
environmental considerations. For instance, “It is moral to act in con-
t t+1 t+2
formity with the value environment” is simply represented as:
moral_eval(X,V1,moral):-
Figure 2. Execution of a limit order added on the market valueSupport(X,V1) & subvalue(V1,"environment").
The example on the Figure 2 illustrates the addition of an ask or- moral_eval(X,"environment",moral):-
valueSupport(X,"environment").
der of thirteen assets at the price p. Before the addition, the best bid
is p and the best ask is p + 1. The new order encounter an order on 7 The French national energy producer.
8 A French electric infrastructure producer.
the other side during its insertion, so the biggest is splitted and the
In this example, an ethical agent is now able to infer for instance the agents. They concern the equities “LEGRAND”. The main data
that, regarding its belief, trading Legrand is moral regarding this the- (represented by a candlestick chart) show us the evolution of the price
ory of good, and that trading EDF is both moral and immoral. Fi- on the market and the two moving averages mentioned in section 4.1
nally, ethical agents are equipped with simple principles, such as “It (the line charts in the middle of the candlestick chart) are slowly
is rightful to do a possible, not immoral and desirable action”. The moving up and down. They are used by the desirability evaluation
implementation of this principle and some preferences is: to detect opportunities according to the rules detailed in Section 4.3.
The candlestick chart does not break often the Bollinger bands, but
ethPrinciple("desireNR",Action):-
possible_eval(Action, possible) &
these breaks may happen sometimes. We observe some peaks in the
desire_eval(Action,desired) & volume barchart when the moving averages are crossing each other.
not desire_eval(Action,undesired) & This is due to the number of exchanges performed by the zero-ethics
not moral_eval(Action,_,immoral). and ethical agents because their desirability function is triggered by
prefEthics("perfectAct","desireNR"). this event.
prefEthics("desireNR","dutyNR").
� ����
4.3 Families of agents for asset management ���
� ����
Each agent receives a portfolio (a set of equities and currencies)
������������
at the beginning of the simulation and may exchange it on the ���
market. Three types of agents are considered in this system: zero- � ����
intelligence, zero-ethics and ethical agents. ���
• Zero-intelligence agents are making random orders (in terms of � ����
price and volume) on the market to generate activity and simu- ��
������ �����������������
late the "noise" of real markets. Each zero-intelligence agent is ��
assigned to an asset. Their only desire and ethical principle are the
�
application of this random behaviour. In this experiment, they are ��� ��� ��
� ���� ���� ���� ���
� ���
�
�����������������
used to generate a realistic noisy activity on the market in order to
create opportunities for the other agents.
Figure 3. Evolution of the asset LEGRAND. The candlestick chart repre-
• Zero-ethics agents only have a simple desirability evaluation func-
sents the evolution of the price, with the moving averages in the middle and
tion to speculate: if the price of the market is going up (the shortest
the up and down Bollinger bands on each side.
moving mean is over the other one), they buy the asset, otherwise,
they sell it. If the price goes out of the bollinger bands, these rules
are inverted. This strategy is also used by the ethical agents to Figure 4 represents the evolution of the portfolio of an ethical
evaluate the desirable actions. agent during the experiment. It provides information on the behavior
• Ethical agents implements the ethical decision process to take of this agent and it was chosen because it is quite representative of
their decisions. An ethical agent implementing the ethical deci- the portfolios of the other ethical agents. The y-axis shows the value
sion process without any moral value or moral rule and an single of the portfolio and the colors depend on the assets placed in the
ethical principle that simply considers desirability are also ethical portfolio.
agents, more precisely hedonic agents. It is different from a zero-
ethics agent because this agent still has all the ethical decision ��������� ��� ����
����� �������
process and explicitly believes that its action are not moral or im- ������
moral. In this experience, ethical agents have the three following � ���
principles (by order of preferences): “It is rightful to do a possi-
� ���
ble, moral, not immoral and desirable action”, “It is rightful to do
a possible, not immoral and desirable action” and “It is rightful to � ���
������������
do a possible, moral, not immoral and not undesirable action”.
� ���
���
5 EXPERIMENTAL RESULTS �
� ���
This section details and analyzes the results of a simulation executed
with ten zero-intelligence agents per asset, eight zero-ethics agents ��
�
� � �� ��
� ��� ���� ���� � ��� � ��� ���
�
and two ethical agents to illustrate the impact of the ethics described �����������������
previously on the behavior of an ethical agent. This quantity of agents
was the optimal one to generate enough opportunities in the simula- Figure 4. Evolution of the portfolio of an ethical agent
tions with the limited performances of a laptop. You can download
this experience on the internet9 .
At initialization, each agent receives a portfolio containing a ran- Firstly, we can notice that the number of EDF equities in the port-
dom set of assets for a total value of 500e more or less. folio never changes during the simulation. We can easily explain that
Figures 3 and 4 show the results of the experiment. Figure 3 shows by the agent’s ethical settings given in Section 4.2: it is never right-
all volume and price information made accessible by the market to full to trade this equity because the agent thinks that EDF is a nuclear
energy producer and no ethical principle provided to the agent con-
9 https://cointe.users.greyc.fr/download/experience-EDIA2016.zip
siders an immoral action as rightful.
Secondly, we can also observe many periods where the portfolio [6] N. Cointe, G. Bonnet, and O. Boissier, ‘Ethical judgment of agents’
contains the “Legrand” asset. In fact, trading this asset is the only behaviors in multi-agent systems’, in 15th International Conference on
Autonomous agents and multi-agent systems, (2016).
action judged as moral due to its label. So to buy and to sell this asset [7] Directorate-General for Economic and Financial Affairs, ‘Impact of the
is the only way to satisfy the most preferred principle and obviously, current economic and financial crisis on potential output’, Occasional
they are here the most executed actions. Papers 49, European Commission, (June 2009).
Finally, we can notice different stages where the agent put in its [8] J.-G. Ganascia, ‘Ethical system formalization using non-monotonic
portfolio various equities. These equities are bought or sold due to the logics’, in 29th Annual Conference of the Cognitive Science Society,
pp. 1013–1018, (2007).
desirability of these trades and the impossibility to execute a moral [9] J.-G. Ganascia, ‘Modelling ethical rules of lying with Answer Set Pro-
action. gramming’, Ethics and information technology, 9(1), 39–47, (2007).
[10] B. Gert, ‘The definition of morality’, in The Stanford Encyclopedia of
Philosophy, ed., Edward N. Zalta, fall edn., (2015).
6 CONCLUSION [11] R. Johnson, ‘Kant’s moral philosophy’, in The Stanford Encyclopedia
of Philosophy, ed., Edward N. Zalta, summer edn., (2014).
This paper presents an ethical BDI architecture for agents in a multi- [12] A. McIntyre, ‘Doctrine of double effect’, in The Stanford Encyclopedia
of Philosophy, ed., Edward N. Zalta, winter edn., (2014).
agent system. This architecture is not designed to only implement [13] A.S. Rao and M.P. Georgeff, ‘BDI agents: From theory to practice’, in
a given ethics in the decision process, but also to integrate differ- Proceedings of the First International Conference on Multiagent Sys-
ent moral rules and values, or ethical principles as parameters of a tems, June 12-14, 1995, San Francisco, California, USA, eds., V.R.
generic architecture. Lesser and L. Gasser, pp. 312–319. The MIT Press, (1995).
[14] P. Ricoeur, Oneself as another, University of Chicago Press, 1995.
The paper also presents an experiment that illustrates, in a partic-
[15] S.H. Schwartz, ‘Basic human values: Theory, measurement, and appli-
ular use case, how to represent and to use moral values, moral rules cations’, Revue française de sociologie, 47(4), 249–288, (2006).
and ethical principles in a BDI agent in order to describe a rightful [16] M. Timmons, Moral theory: an introduction, Rowman & Littlefield
behavior. The experiment highlights how a few and simple values, Publishers, 2012.
moral rules and ethical principle can influence the behavior of an
agent in order to incite it to prefer a set of rightful actions when they
are available.
Of course, we cannot yet answer some interesting issues such as
how to evaluate the cost of this ethical behavior in terms of finan-
cial performance with a real state-of-the-art trading strategy, or what
is the impact of a given population of ethical agents on a market
behavior. To answer those questions, we need to enrich the knowl-
edge bases of the ethical agents with some logical models of several
famous available principles in the literature (such those modeled in
[4, 8, 9]) and complete the definition of moral values and rules.
Even if the morals and ethics of the agents are only used in this ex-
periment to guide their own decisions, we intend in a future work to
use them to evaluate the behavior of the other agents. Indeed, several
usecases can need this kind of abilities, for instance when an author-
ity wants to monitor the actors on a market, or when an hedge funds
expresses the policy to only cooperate with other trading agents that
satisfy an ethical behavior.
ACKNOWLEDGMENTS
The authors acknowledge the support of the French Agence Na-
tionale de la Recherche (ANR) under reference ANR-13-CORD-
0006. They also express their gratitude to Professor Jomi Fred Hüb-
ner for his theoretical and practical support during the design of this
experiment.
REFERENCES
[1] Ethical judgment. Free Online Psychology Dictionary, August 2015.
[2] O.B. Ahmed, ‘Islamic equity funds: The mode of resource mobilization
and placement’, Islamic Development Bank, (2001).
[3] I. Aldridge, High-frequency trading: a practical guide to algorithmic
strategies and trading systems, volume 459, John Wiley and Sons,
2009.
[4] F. Berreby, G. Bourgne, and J.-G. Ganascia, ‘Modelling moral rea-
soning and ethical responsibility with logic programming’, in Logic
for Programming, Artificial Intelligence, and Reasoning, pp. 532–548.
Springer, (2015).
[5] S. Bono, G. Bresin, F. Pezzolato, S. Ramelli, and F. Benseddik, ‘Green,
social and ethical funds in europe’, Technical report, Vigeo, (2013).