The Dawn of Social Prediction

"Economists pride themselves on being the most scientific of social scientists. This leads them to reduce all human motives and behavior to quantifiable variables such as utility, welfare and income. But people are not by nature quantitative, and their motives often have no economic basis. Today’s most divisive issues, from fairness and inequality to national identity and culture, don’t have economic solutions."
~ Greg Ip.  August 26, 2017.  "In Defense of the Dismal Science." Wall Street Journal.

On a recent bucket-list trip to Peru, despite seeing fascinating ruins and cultural attractions, what made an impression on my children was neither the ruins of Macchu Picchu nor the amazing food - it was the poverty.  They could not understand the disparity in development between home and Peru.  They repeatedly asked me to explain: "Why are people in Peru poorer than at home?"  Theirs was an objective and guileless question.  And this wasn't the first time they had posed such a question.  Last summer - in juxtaposition - we visited Vancouver and a similar question was repeatedly brought up: "Why is San Francisco so much dirtier than Vancouver?"

I tried to summarize for my children the idea of property rights and corruption.  For example, Peruvian economist and Nobel laureate Hernando de Soto theorized that some societies languish in poverty because they have not formalized private property rights and/or the judicial system is corrupted.  As a result, property is not used by bank as collateral for loans and credit, and development is impaired.  But de Soto's theory begs the question, "why?"  And my children brought this up..."Why?"  Why do some societies have corrupt legal systems and weakly enforced private property rights?  

In answering that deeper question I noticed that there are dozens of anecdotal explanations that shift the blame among:  a legacy of power-based exploitation (caste system), low pay for civil servants (so taking bribes is more appealing), weak educational traditions (and illiteracy), disease (tropical illnesses and malnutrition), agricultural history, etc.  These all may be true to some extent, and in the following years economists and philanthropists - such as Bill and Melinda Gates - may produce more evidence-based results to work with.  

Yet despite the value of economic approaches - and there is tremendous value there - something important is missing.  As Greg Ip notes in the quote above (from this article), even if we think we have answers based on the work of economists, psychological factors often break economic models.  And consider that much of the public does not trust economists, so politicians are not penalized if they choose not to follow their advice.  Whether in the behavior of crowds, the willed ignorance of politicians, the corruptibility of courts, or the psychological foibles of economists themselves - it is the emotional side of the discussion - the animal spirits - that is both ill-defined and essential for understanding social progress.

In today's newsletter we discuss the psychological elements of social progress through the lens of media analytics.  We examine whether memes present in media can be used to predict the development (or decline) of a society, its economic activity, and its financial markets.  I don't have concrete answers in this newsletter, but I do think we're tantalizingly close to a quantitative understanding of the social phenomenon that animate societies, for better or worse.

Meme First

"The American company that promoted the Internet hardest in its early days, Sun Microsystems, conducted research in 1997 into how people read on the Web and concluded simply, 'They don't.' They scan, sampling words and phrases. Why? In part because any one page, on which the fluttering user happens to have lighted momentarily, competes for attention with millions more."
~ James Gleick. 1999. Faster: The Acceleration of Just About Everything.  Vintage.  p.87.

People gain impressions, fleeting feelings, from what they read in the media.  And over time the discourse in societies often coalesces from fleeting feelings and impressions into broad themes.  In such cases they are called memes - infectious ideas.  A meme is an idea, behavior, or style that spreads from person to person within a culture.  

Memes unconsciously seep into our core beliefs and assumptions about the world.  They animate our interactions. 

 The influence of memes can be measured in impressions and emotion-laden expressions.  Anger - whether directed at the government, corporate management, or the "elites" - is one type of meme.  Economic optimism and trust in strangers (e.g., in relation to immigration) are two other types

In the USA traditional positive memes include "winning," "strength," "innovation," and "freedom."  Negative memes like "racism", "crime", "inequality", and "terrorism" are also present.  Sometimes these memes lodge deep into memory, associate with specific behaviors, and are triggered when cues are encountered.  Sometimes they are contagious, resonating with deep feelings of hope or alienation, transmitted like viruses from individual to individual.  

These feelings can coalesce into social movements such as Brexit or the election of Trump.  Specific publications can capture and channel the memes that move large swathes of society.  As noted in a recent study (see Fig 1 and Fig 2), specific media outlets can influence and even germinate the memes that are most influential over voting behavior.  In the above study, Breitbart was identified as a key meme trend-setter in the period before the U.S. elections.

The Social Operating System

Memes influence us in different ways in different contexts.  They are like modules in software code - activated when needed and idled when irrelevant.  To borrow a concept from Bryan Johnson, they comprise fundamental code of our social operating systems.  

"In the same way that computers have operating systems at their core — dictating the way a computer works and serving as a foundation upon which all applications are built — everything in life has an operating system (OS). It is at the OS level that we most frequently experience a quantum leap in progress."
~ Bryan Johnson - Founder, Braintree and the OS Fund

Using text analysis, we can quantify what societies value by detecting memes.  We can track which memes are most shared, praised, or disparaged.  

After quantifying memes, we and other researchers can create predictive models on top of this memetic data.  Such preedictive models can be used to predict elections, true, but more interesting to me (and my children) is whether such models can be used to predict long term national economic development, insecurity, or even disease (e.g., heart disease, diabetes, obesity to name a few).  

And in the long-term, perhaps we can create diagnostic tests using memes.  Like testing an individual's white blood cell count, the presence of certain memes - mistrust, anger, fear, uncertainty, contempt - may signal when a society's immune system is vulnerable.  

If we can predict probable outcomes, perhaps we can intervene.  If the average Greek's baseline mistrust of authority had been considered as Greece borrowed more from German banks, perhaps German lending standards would have been tighter.  And within societies perhaps OS modules can be reprogrammed to create better outcomes, as through public campaigns (e.g., anti-smoking or tax compliance).  In Greece, if government officials paid their taxes in cash in front of TV cameras, as a spectacle to demonstrate their own honesty and self-sacrifice for the society, perhaps more Greeks would have followed their example and tax collections could offset the debt burden.  Perhaps societies can be immunized against memetic "viruses" with media-based interventions.

It is within our power - with less guesswork than ever - to influence social, electoral, and business outcomes by harnessing - or even changing - the most influential memes.

True, this level of social engineering could be used in an Orwellian fashion.  It is already happening.

Social Analytics

"Much of the human experience (knowledge, disease) spreads by proximity, and for any one person the number of elbows in proximity has exploded. In past times, even in the most crowded city, we lived close enough to only a few people to, say, read their journals or track the temperature of their hot tubs. Now, in hordes, they put that information on-line. The multiplication of information pathways leads to positive feedback effects in the nature of frenzies."
~ James Gleick. 1999. Faster: The Acceleration of Just About Everything.  Vintage.  p.86.

At MarketPsych Data our software gathers articles from thousands of global news and social media sources in real-time.  It performs text analysis and measures references to hundreds of topics and "meanings" (including memes) in relation to specific locations.  Is there a crime wave in Cleveland or Mysore?  If the media are reporting it, and if people are talking about it in social media, then our software is identifying it.  In our Thomson Reuters MarketPsych Indices (TRMI), we aggregate location-specific references to the country-level and create a rolling time series of such references (since 1998).  The TRMI data also covers references to currencies, commodities, fixed income products, and companies.  

Representations of this data is visible in the below global heat maps.  A map of social media-reported Social Unrest in 2016 is shown first, below:


The next chart represents the percentage of country-specific references to Innovation in media.  The UAE, Columbia, Estonia, and Canada all stand-out with high levels of innovation references in the media.


Innovation and social unrest are two - very different - components of social functioning.  Using these and 60 other memes, we have developed predictive models for economic activity.

Our colleague Aleksander Fafula PhD created an algorithm to predict economic growth using the TRMI country data.  While operating, on average his algorithm successfully predicted the flash PMI (70% directional accuracy) and final PMI values (80% directional accuracy) for the G-12 nations days in advance, using real-time global media data.  That work is described in a textbook chapter (available upon request).

Housekeeping and Closing

"The first psychometricians, eager as they were to find some real, innate, general quality of mental ability that they could measure with tests, rarely paid attention to speed of thought."
~ James Gleick. 1999. Faster: The Acceleration of Just About Everything.  Vintage.  p.112

Despite the lengthy travel, my children agreed that it didn't feel like we had traveled very far to be in Peru - "like we drove 2 hours."  Perhaps this was a consequence of globalization.  Memes crossing borders in the duration of a cell phone call.

Our online personality tests for retail users and the professionalized Insights version for financial advisors identified Openness to New Experiences as the chief personality success trait of long-term investors.  To thrive in a world of rapidly changing memes, we need such a mindset of openness, flexibility, and adaptability.  An open mindset observes and understands new memes.  An open mind more easily remains non-attached to memes that are harmful for growth and development.  

As investors, we need self-awareness of which memes are influencing our behavior.  In our pursuit to develop better predictions of - and preparations for - the future, understanding global memes in an essential first step.

Our Thomson Reuters MarketPsych Indices are deployed globally to monitor real-time market psychology and macroeconomic trends.  If you're an academic interested in data for research, please reach out for access.  If you represent an institution, please contact us.  The commercial Thomson Reuters MarketPsych Indices dataset covers 45 currencies (including Bitcoin), 61 countries' fixed income products and stock indexes, 12,000+ companies and stocks, 36 commodities, and 187 countries.  

Happy Investing!
Richard Peterson and the MarketPsych Team