Macroeconomics with Human Sentient and Social Actors

1
Macroeconomics with Human Sentient and Social
Actors

• David Tuckett , Paul Ormerod, Robert Smith and
  Rickard Nyman
• Centre for the Study of Decision-Making
  Uncertainty
• University College, London

2
Human Sentient Agents

• An economy is a complex system populated by
  thinking human beings who have co-evolved in
  human groups.
• Goal of macroeconomics is to understand the
  conditions under which economies grow or contract
  or their future rates of inflation, unemployment
  and capital utilization and formation and to give
  advice about the impact on them of particular
  fiscal and monetary policies or shocks.
• To do it we postulate models with mechanical
  agents in quite simple systems with little
  interaction or uncertainty.
• Models exclude the feared complications and loss
  of rigour that are introduced by
• Radical uncertainty
• Human Psychology and Neurobiology
• Social Life
• In what ways to understand what issues in the
  economy can we construct models with human
  sentient and social actors that work better?

3
Information under Ontological Uncertainty

• Modelling of agent behavior has depended on the
  assumption that information is non-ambiguous and
  present or absent
• no special processes of perception and
  interpretation
• Psychology ? irrationality and bias.
• Psychologically or socially driven action
  introduces individualistic and idiosyncratic
  impulses disconnected from any fundamental
  considerations that might drive outcomes or as
  herding
• But in social and psychological science agents
  can mostly only discover fundamentals through
  constructing the meaning of the information and
  drawing on a wide range of adaptive capacities
  and social signals to do so.
• Ontological uncertainty about information meaning
  occurs when decisions have to be made in
  situations with a long term time horizon, when
  decisions are non-reversible and in non-routine
  situations.
• In such situations action to obtain gain
  necessarily incurs potential loss constituting
  an inherent cognitive and emotional conflict
  between action and inertia. The question how
  does an agent manage to act at all?

4
Emotion, Simulation and Narrative

• What are emotions for? An evolutionary survival
  resource to treat them as error/bias is a
  mistaken simplification.
• Faced with uncertainty actors seek to gain
  control over the situation facing them through
  time and thy do this using their human capacity
  for neurobiological simulation and establishing a
  sense of truth.
• Narrative is a core human capacity for
  constructing and making sense of the world and
  communicating about it.
• It is built in to the development of brain
  structures and language to process complex
  situations (e.g. life in a family and the world)
• Paradigmatic and Narrative sense of truth (Jerome
  Bruner)
• By definition narratives leave things out.

5
Conviction Narrative Theory (CNT)

• CNT is a theory of how agents use human capacity
  to face the future by gaining confidence to act
  at all (animal spirits spontaneous urge to
  action).
• Overcome inertia and anxiety by engaging
  neurobiologically based affective, attachment and
  bonding system.
• CNs developed by simulating the state of the
  organism (body) in future, managing conflict by
  creating enough certainty to act.
• Conviction can be achieved by eviscerating doubt
  phantastic object narratives (PON), divided
  states (DS), groupfeel (GF).
• CNs are socially embedded.
• CNs are psychologically invariant must generate
  more E than A. ? emotion in an economy or group
  not just individual or random noise but an
  indicator.
• ? CNT can be studied through ethnographic
  observation, interview data and document
  analysis.

6
Directed Algorithmic Text Analysis (DATA)

• Algorithm based unstructured document analysis.
• Detect shifts in relations between Anxiety and
  Excitement in CN through time.
• Procedure
• Text with date stamp and (ideally) To and
  From labels (networks)
• Identify excitement (attractor) words and
  anxiety-doubt (words) (Doubt repellers).
• Count them.
• Track shifts in relationship through time.
• Apply statistical tests.

7
Source Reuters News Archive
8
UK

• The animal spirits series in both the US and the
  UK appear to have given clear prior warning of
  the downturn, well in advance of any other
  indicators. In June 2007, the value fell sharply
  in both economies, to -0.113 in the US and -0.406
  in the UK. Compared to the mean values January
  2004 through May 2007, these are, respectively,
  2.55 and 3.49 standard deviations below -
  pre-dates that of existing survey measures by
  several months.

9
Fannie Mae Sentiment Shifts 2003-2013
Text Source Reuters News Archive
10
Using Economic Consensus Forecasts to Predict
Preliminary Monthly MCI
11
Using Relative Sentiment Shift Measure to Predict
Preliminary Monthly MCI. Source Brokers Reports.
12
(No Transcript)
13
New Data Sources

• Bloomberg, HSBC, Reuters broker reports and
  economic commentary.
• Bank of England Market Intelligence or other
  organizational documents.
• Email.
• Interview transcripts eg BoE agents, Regulatory
  inspectors, risk surveys.
• Economic and Financial blogs, social media.

14
More Advanced Analysis

• Better specification of shifts in target emotions
  beyond word lists. (Sematic and Syntactic
  analysis etc.?)
• Introduction of network dynamics.
• Better selection of narratives expected to
  correlate with future plans.
• Correlations with measures of capital formation,
  innovation etc.
• New more complex clustering of words.