In the finance world of 2020, politics matter. However, investors still disagree whether political research gives them an edge in terms of performance. Some remain skeptical, while others merely use geopolitics as a retroactive excuse for poor performance.¹

The criticism I've heard most often is that “political analysis is a nice-to-have, but it is not a must-have.” It is marginal to the decision-making process – at best, a handy tool when an exogenous event threatens one's strategic decisions. Another way to put it is, “We will cry geopolitics when something blows up, just so we have a scapegoat to blame for our faulty predictions.”

My answer to this view is: “Don't bother. When something blows up, buy everything.”

Other than the 1973 Yom Kippur War, few geopolitical events since the Second World War have been disastrous for the markets (Figure 2.1). That near-perfect predictability is why this book is not about Black Swans or hoarding gold for an impending apocalypse.² It is about incorporating geopolitical analysis into your investment process, as you would any other macro factor.

Geopolitics Is Not a “Nice-to-have”

In 2011, I arrived in Montreal to start a job at the world's oldest and most-respected independent investment research firm: BCA Research. Since 1949, BCA Research has offered its clients an interesting proposition: getting the macro forces in the economy right is more effective than trying to pick individual stocks. In the 1950s, this big-picture approach was novel. I owe a lot to the firm and my colleagues, as they taught me everything I know about macro investing.

The macro DNA behind BCA's success is credit cycle analysis. Getting the credit cycle right is still the backbone of macro investing. Most investors pay attention to it, and a few – like the $160 billion behemoth Bridgewater – have turned it into a moneymaking science.

When I joined BCA Research, I hardly knew what a credit cycle was. Nor did I understand much about finance or economics. I was hired for a simple reason: the CEO saw the geopolitical paradigm shifts coming and wanted someone on staff to speak on it.³

Several colleagues taught me the basics of macroeconomics and the markets. They probably contemplated self-immolation as I struggled with basic financial concepts such as valuation, discount rates, and logging into a Bloomberg terminal.⁴

I learned a ton in the first six months on the job. But perhaps the most eye-opening experience was navigating the firm's library and archives. Yes, even in 2011, BCA Research had “stacks” of academic journals and books, some going back over a century. I asked the librarian on staff to give me a collection of BCA Research reports, some of which were penned in the 1950s. I slowly moved through the decades. I felt like an archeologist exploring the records of some long-lost civilization. Each decade's core theme took shape while hints of impending paradigm shifts came into focus.

At the end of this apprenticeship period, I sat down with the mild-mannered gentleman who was my mentor. It was time to sell my strategy and earn my place at the shop.⁵ I was full of ideas on how to incorporate geopolitical and political analysis into the firm's process, as I had just consumed hundreds of BCA reports (and thousands of Investopedia pages).⁶ But the business plan my mentor proposed was a letdown:

“Every time something blows up, you should whip up a piece. Focus on events – ‘Black Swans' – that can upend the firm's major asset allocation views.”

I was disappointed. I had a gut feeling that most Black Swan events were irrelevant, a feeling later confirmed by quantitative research. I had also come to BCA with a Jerry Maguire memo in hand. I had proof of concept: the 2010 Euro Area sovereign debt crisis proved that geopolitical analysis could get ahead of the narrative flow that moved markets and generated constant alpha.

Most critically, I had just spent the entire Quebec winter (all six months of it) reading through BCA Research's archives.⁷ From my noneconomist perspective, the firm's own research confirmed that geopolitics were integral to most major economic trends of the past fifty years.

“But that would mean that I am an appendage … extraneous to the firm's process.”

My mentor's reply was wreathed in ice: “Well, yes. But this is how you get paid here. After all, your research has to sell. Otherwise …”

He left the rest unsaid, but I let my imagination fill in the blank.

To this day, I am glad I said no. I mean there I was, in finance. I'd leveled-up from a glorified blogger job and a dead end in academia. If this guy wanted me to write about Black Swan events, why rock the boat? That job is simple enough: take my colleagues' view of the markets as a fait accompli and apply some probability profile from the geopolitical risk smorgasbord.

Say the house is bullish on the economic growth outlook and recommends investors short duration (i.e., sell long-dated bonds). What do I add? Got it: a random pandemic is a threat to that view as US Treasuries will rally on risk to global growth. Done. Get paid, go home, watch the kids grow!

But BCA's own archives screamed for a more sophisticated approach. Each decade's research had an unspoken geopolitical anchor, even if the macroeconomic analysis danced around it:

• 1945–1961⁸: The immediate post–World War II years saw five recessions in quick succession. The Great Depression and the calamity it birthed was still raw in policymakers' minds. As such, they relied on private-sector relevering to get out of each slowdown. Their over-reliance ushered in what BCA Research later called the “Debt Supercycle” (which, incidentally, didn't end until 2008).⁹
• 1960s: The 1960s saw a period of secular stagnation. Policymakers grew complacent as inflation failed to react to low unemployment, so they became emboldened to pursue fiscally profligate policies. Between the fiscal outlays of the Vietnam War and President Johnson's Great Society entitlement expansion, deficits soared outside of a recession for the first time since World War II. Sound familiar? In 2020, we are living in a 1960s rerun with the COVID-19 pandemic putting the last nail in the coffin of the Washington Consensus and one of its central tenets: prudent fiscal policy.
• 1970s: The 1960s' demand-driven policies led to the stagflation of the 1970s. Yes, the 1973 oil crisis contributed to inflation – the rare example of a significant geopolitical Black Swan – but political and geopolitical decisions made in the 1960s had the most lasting impact on 1970s economic woes. An oil price spike in a deflationary context would have had only a fleeting impact.
• 1980s: The disaster of demand-driven policies in the 1970s caused a political revolution at the end of the decade. As noted in Chapter 1, Reagan and Thatcher gave supply-side policies a chance, while the academics running central banks had a popular political mandate to crush inflation.
• 1990s: The geopolitical paradigm shift of the Soviet Union's collapse complemented the 1980s' supply-side political revolution. These events combined in the 1990s to form the twin tailwinds of globalization and American hegemony. Together, they tamed inflation and altered the political power balance between labor and capital. Great Moderation ensued.

  

• 2000s: The party did not end with the 2001 recession. Policymakers reinflated the Debt Supercycle with gusto, due in part to fear that the 9/11 terrorist attack would stifle consumer confidence. But growing income inequality – which, yes, was present before QE, to the surprise of many investors waking up to inequality only today – led to a massive levering-up by the household sector. Households overspent to keep up with the Joneses and maintain their present buying habits, given the ever-higher cost of middle-class goods (Figure 2.2).¹⁰

From the above decade trends, I concluded that treating politics and geopolitics as an externality to markets made no sense. The only way to incorporate geopolitics was to recognize its pervasive influence. It affects everything it touches – from market forces down to the individuals who analyze them. I realized it was important to investigate my investment views – to assess whether the political assumptions that underpinned them were correct. Seeing this entanglement and determined to untangle it, I declined to pursue the proposed business plan and suggested an alternative:

“We are going to make market calls and forecasts based not on what policymakers want to do, but on what they must do given their material reality.”

My mentor flashed me a look like he was the Cheshire cat gazing at Alice from his tree perch.

“And how do you intend to do all that?”

The First Pillar: Materialist Dialectic

Niccolò Machiavelli's The Prince is the foundational text of modern political theory. Machiavelli posits that governance is an interplay between Fortuna – fate and all things beyond the control of the Prince – and Virtù – the Prince's ability to navigate Fortuna.

Fortuna is a river “which, when enraged, inundates the lowlands, tears down trees and buildings, and washes out the land on one bank to deposit it on the other. Everyone flees before it, everyone yields to its assaults without being able to offer it any resistance.” But not all is lost. If the Prince prepares for the flood and makes “provisions during periods of calm by erecting levees and dikes to channel the rising waters when they come,” the Prince can restrain Fortuna.¹¹

Machiavelli saw Fortuna and Virtù as equals: “… since our free will must not be denied, I estimate that even if fortune is the arbiter of half our actions, she still allows us to control the other half …”¹² The Prince has the free will to manipulate fate.

The Prince is not a treatise on forecasting politics and geopolitics. It is a manual for the acquisition and preservation of power. In its original context, Virtù is important, but I never understood how to incorporate it into a forecast. Furthermore, while following global events in my early career, I realized that policymakers of even the highest quality succumb to the material reality of Fortuna. On the other extreme, even the most incompetent policymakers are ultimately pushed to do the right thing by the constraints of the market, economics, and politics. As economist Herb Stein said, “if something cannot go on forever it will stop.”¹³

Machiavelli's analogy of Fortuna and the river has stayed with me since I first read the book 25 years ago. If forecasters could predict the flow of that river, would they not be more than halfway to predicting policymaker behavior?

Imagine a giant holding a large mug of ale over a hill. The giant tips the mug over and allows the beer to gush down the decline. If I know the hill, I can predict the ale's course: it will follow the path of least resistance, because it is influenced by the material reality of the hill's terrain.

As Machiavelli claims, the flow of Fortuna may be equally as important as the Prince's Virtù in determining the final outcome. But the forecast must start with its course – Fortuna, rather than the policymaker's reaction to it; the reaction is a derivative of the flood. And the only way to know the flood is to know the terrain.

And Karl Marx knows the terrain.

Marx – as an analyst, not a prophet of doom – is an essential teacher in analyzing the terrain upon which Fortuna flows, and his analysis is the first pillar of the constraint method.¹⁴

In Das Kapital, Marx explains how the post–Industrial Revolution world works. He takes a complicated and somewhat nebulous concept – capitalism – and breaks it into its most basic components. Using this materialist approach, he rarely delves into the qualitative world of ideas. In the first nine chapters of Das Kapital, Marx decomposes capitalism into its constituent, material, elements. He focuses on the material realities that underpin capitalism: price, money, labor, means of production. After exposing the system's inconsistencies, Marx concludes with a forecast: a crisis is coming.

While the conclusions of Das Kapital are interesting, and much ink (and blood) has been spilled debating them, it is the engine that powers these conclusions that is most relevant to the constraint framework. This engine is “dialectical materialism.”

Marx's dialectic stands in opposition to the Hegelian dialectic, which many know as the oft-quoted adage “thesis, antithesis, synthesis.” Both dialectics are an attempt to make sense of human history and how society defines “truth.” For Hegel, the starting point of the search for truth is human thought: ideas. Like Machiavelli, Hegel sees the human actor as possessing agency. According to Hegel's worldview, the ideological preferences of both the powerful few and the masses influence history.

Or, as John Keynes famously said,

The ideas of economists and political philosophers, both when they are right and when they are wrong, are more powerful than is commonly understood. Indeed, the world is ruled by little else. Practical men, who believe themselves to be quite exempt from any intellectual influences, are usually slaves of some defunct economist.¹⁵

The innovation of Marx's Das Kapital is that it rejects the notion that ideas dictate human history. Marx and his collaborator, Friedrich Engels, posited that the material world, not ideas, must be the starting point of analysis.

What makes the materialist dialectic a dialectic? In Marxist thought, the material world – the society's modes of production – is the concrete foundation upon which all thought, norms, values, and institutions ultimately rest.¹⁶

According to Marx, the European feudal system was not a product of human thought but of the means of production available in the Middle Ages. All nonmaterial aspects of feudal society – the culture – reinforced the hegemony of thought created in service of material modes of production. Ideas serve materials. If Hegel's theory says the idea egg came before the chicken, Marx's counters that it was the material chicken that got the idea egg rolling.

For the sake of not putting anyone to sleep, I kept this discussion brief, and it is woefully inadequate to meet the standards of graduate courses in political theory.¹⁷ The bottom line is that the material dialectic is the main pillar of the constraint framework. The framework's starting point of analysis is the material world, not the world of ideas. Material conditions create human reality. Thought systems (philosophy, religion, political parties, etc.) develop around this material condition and are therefore a derivative of it. People cannot “think” or “prefer” their way out of material constraints.

As for Marx's forecast of a proletariat revolution, he was not entirely wrong. His description of capitalism remains cogent, particularly the tension between what employees earn, what employers accumulate in profit, and what that does to the aggregate demand level in an economy. Some – including myself – would say that the world is at an unsustainable extreme between the share of the economy going to corporate profits and the share going to labor (Figure 2.3).

By focusing on the material world, Marx forecasted the tumult that would dominate the twentieth century. Yes, he got the final outcome wrong, and he was hopelessly prescriptive, but he correctly surveyed that the status quo of the nineteenth century – in which laborers had scant protections or political power – would not last.

Second Pillar: Diagnosticity of Constraints

For investors used to quantification, a focus on the material world should be welcome. Constraints are observable and therefore empirical.

To find these constraints, investors should observe as much as possible about the real world, right? Wrong. Not all observable data is created equal, and more information often does not produce superior forecasting results. The quality of the data matters more than its quantity, especially in circumstances where a complete data set is impossible to obtain.

Richards J. Heuer, Jr., author of the CIA methodology manual Psychology of Intelligence Analysis, spent his life improving the judgment of intelligence analysts: “Judgement is what analysts use to fill gaps in their knowledge. It entails going beyond the available information and is the principal means of coping with uncertainty.” Heuer goes on, “While the optimal goal of intelligence collection is complete knowledge, this goal is seldom reached in practice.”¹⁸

Heuer dealt with political analysis and forecasting throughout his nearly 50-year tenure at the CIA. He thrived on uncertainty, paucity of high-quality information, and concept-driven (as opposed to data-driven) analysis. His career in intelligence forced him to rely on the fuzzy, the qualitative, and the soft, as opposed to the clear, the quantitative, and the hard, data. Instead of shying away from these challenges – or trying to fit the qualitative square peg into a quantitative round hole – Heuer developed a systematic approach to intelligence analysis.

While Heuer does not address material constraints directly, he rightly asserts that the intelligence analyst rarely has complete information.

Given this limitation, Heuer introduces two concepts for all political and geopolitical analysts to keep in mind. First, having more information does not necessarily help one's forecast. More information sometimes only contributes to a higher level of conviction, not necessarily forecast accuracy.

Second, the quality of data is what matters. And the key determinant of quality is diagnosticity, the second pillar of the constraint framework. Diagnosticity is “the extent to which any item of evidence helps the analyst determine the relative likelihood of alternative hypotheses.”¹⁹ To illustrate, Heuer uses the example of a patient with a fever. Because a body temperature in excess of 38 °C is consistent with so many reasons to be ill, it has limited diagnostic value.

Paucity of information and diagnosticity of intelligence are critical concepts for investors and those without access to government-funded intelligence agencies. Without access to near-unlimited budgets, satellite imagery, mass surveillance, and a web of assets across the world, agency outsiders always operate with limited information. As such, it is all the more important to understand what information is actually diagnostic.

Diagnosticity helps analysts eliminate unlikely, or competing, hypotheses (a process known as “competing hypotheses analysis”). A nondiagnostic variable is one that does not help eliminate any of the hypotheses.

Preferences are not diagnostic variables because they are optional; the policymaker chooses whether to act on them. Such variables are nondiagnostic because it is impossible to eliminate hypotheses based on a variable that may not affect the outcome at all. In contrast, constraints are the gatekeepers that determine whether preferences affect the outcome. Constraints have high diagnosticity because preference-based outcomes are subject to them.

Diagnosticity in Practice

Take President Donald Trump's preference to have the Affordable Care Act, “Obamacare,” repealed. In 2017, analysts could easily conclude from Trump's statements that he had a strong preference on this topic. However, it was impossible to gauge his sincerity, commitment, and pain threshold for pursuing such a policy. Furthermore, Obamacare is an entitlement, and American policymakers' track record of repealing entitlements is poor.

Trump's preference to repeal Obamacare was subject to two material constraints: the risk of losing popular support, and congressional math. In 2017, President Trump did have a majority in the Senate, allowing him to repeal Obamacare.²⁰ However, it was not a large majority. Due to the political risk to moderate Republican senators – and likely Senator John McCain's animosity toward Trump – the repeal bill failed 49–51 in the Senate.

By the summer of 2017, Republicans had tried to kill Obamacare 70 times since its 2010 implementation. And yet, when all the stars aligned – when they held both chambers of Congress and the presidency – they fell short.

Given the preference of President Trump and the Republican party to eliminate Obamacare, this outcome surprised most investors. They miscalculated because they overestimated the importance of preference and failed to consider its low diagnosticity. As Heuer posits, it is extremely difficult to eliminate a hypothesis based solely on policymaker preferences.

Material constraints, on the other hand, have high levels of diagnosticity. They are diagnostic because they are not optional. President Trump could change his mind about Obamacare, but he could not wish away moderate Republicans Susan Collins, Lisa Murkowski, and John McCain in 2017. Furthermore, Obamacare became popular in mid-2017, right when Republicans finally held the political capital to extinguish it. By the time the GOP circled back to thinking about healthcare in 2018, the midterms were upon them, and support for Obamacare surged (Figure 2.4).

Diagnosticity is the second pillar of the constraint framework. Beyond diagnosticity, Heuer also briefly mentions the third pillar. Later in his book, he points out that “when observing another's behavior, people are too inclined to infer that the behavior was caused by broad personal qualities or dispositions of the other person and to expect that these same inherent qualities will determine the actor's behavior under other circumstances. Not enough weight is assigned to external circumstances that may have influenced the other person's choice of behavior.”²¹

Third Pillar: The Person versus the Situation

The third pillar supporting the constraint framework is the idea of the fundamental attribution error, borrowed from the field of social psychology. I draw the most from The Person and the Situation by social psychologists Lee Ross and Richard E. Nisbett.²²

While Ross and Nisbett drew many insights from social psychology experiments, the one that best illustrates the material constraint framework is their discussion of the Princeton Seminary experiment. In the early 1970s, behavioral scientists John Darley and Daniel Batson constructed a fascinating experiment.²³ The setting of the study was the Princeton Theological Seminary. The subjects? Students studying to become ordained priests.

Seminary students were asked to deliver a sermon in a classroom across campus, where senior members of the faculty would evaluate their performance. They were given some time to prepare the speech. At the conclusion of their preparation, a third of the students were told that they were very late, that their superiors were already waiting on them, and should leave immediately; a third that they were going to be late if they did not leave soon; while a third were told that there was no rush and that they could proceed to the meeting at a leisurely pace.

Darley and Batson designed the experiment to test how time pressure influences behavior. On the way to the sermon, each student encountered a victim in need of help (an associate of the experimenters). Out of the time-stressed cohort of students, only 10% stopped to aid the victim. Out of the cohort that was in a hurry, but not really stressed, 45% offered to help. Finally, 63% of the students in no rush stopped to see if they could help.

Remember: the test subjects are young men looking to dedicate themselves to the service of God and thus presumably predisposed to being good Samaritans. The character and preferences of these students were as conducive to offering aid to a stranger as one was going to get on a college campus.

What was the topic of the sermon they were asked to perform? The story of the Good Samaritan (Luke 10:29–37 in the New Testament). They had specifically been primed to “do the right thing.”

Ross and Nisbett offer other examples and research to prove their overall point that time and time again, the situation is a better indicator of the outcome than the person. The context in which an individual finds himself has more influence on his behavior than his character, background, religion, upbringing, etc.

Given their career paths, the majority of test subjects probably had a preference to help a fellow human in need. And being a Good Samaritan was top of mind, as they had just prepared a sermon on it! However, when subjected to the material constraint of time, their preferences succumbed to the constraint.

Using this and other examples, Ross and Nisbett introduce the concept of the “fundamental attribution error,” a mistake analysts make when they attribute real-world outcomes to characteristics, personality, and moods of individual actors. The individual's psychological profile is given primacy over the external context; the person takes precedence over the situation.

Political analysis provides fertile ground for the fundamental attribution error. The news media is partly to blame because it personalizes events and tells human rather than situational stories. To be fair, stories don't sell well without individual actors in them. Journalists are giving their readers what they want, which is a reaffirmation that people matter and that human agency can triumph over the determinism of the situation. Geopolitical analysts have no such excuse.

Fundamental Attribution Error in Practice

Examples of this attribution error abound in both the media and what passes for geopolitical analysis:

“Kim Jong-un Is a Wild Card”: In early 2013, the US media obsessed over the potential danger of a confrontation between North Korea and the US. The young leader of North Korea, Kim Jong-un, was an unknown entity at the time, and analysts feared Kim would do something dramatic to cement his leadership. Rhetoric from Pyongyang certainly fueled the uncertainty and backed up the personality-driven analysis.

However, focus on the situation would have revealed considerable constraints on Kim Jong-un's preferences, whatever they were. In 2013, North Korea had limited ballistic missile technology. Its conventional military capability – aside from artillery – was (and remains) paltry. An equally restrictive constraint came from its nominal ally, Beijing. China opposed – and continues to oppose – any confrontation with South Korea. Such a confrontation would give the US an excuse to establish a greater military presence in Northeast Asia. In addition, the geography and demographics of South Korea make it extremely difficult for North and South Korea to engage in a limited military conflict, as any engagement would expand rapidly. This landscape prevents an escalation in tensions to the point of war. Because Seoul is effectively unprotected from North Korean conventional artillery, South Korea would have to preemptively attack the North at the first sign of conflict, likely ensuring the end of the Kim dynasty. The same logic – and constraints – held during the showdown between President Trump and Kim in 2017. While the media breathlessly accused President Trump of playing with fire, the White House bluff ultimately worked precisely because Kim Jong-un was never holding a strong hand, and his preferences – and even potential irrationality – ultimately could not overcome his constraints

“Israeli Hawks Want to Bomb Iran”: Since at least 2011, a unilateral Israeli strike against Iran has been one of the greatest tail risks to the markets. In early 2012, the media ramped up Israeli rhetoric and oil markets exhibited a considerable risk premium. Investors accepted the rhetoric at face value. They assumed Israeli Prime Minister Benjamin Netanyahu headed a “hawkish” government that considered Iran's nuclear program an existential threat. Observers of the situation made a big deal of the personalities of both Prime Minister Netanyahu and Iranian President Mahmoud Ahmadinejad. The former would never allow a regional rival to threaten Israel's existence; the latter was trying to hasten the return of the Mahdi by causing an apocalypse.²⁴

Investors placed too great an emphasis on the rhetoric coming from both sides, ignoring constraints. Israel did not (and does not) possess a strategic air strike force (i.e., bombers), the lack of which would complicate a military strike against Iran. And the Arab Spring unleashed forces that would ultimately coalesce into the Syrian Civil War and the rise of the Islamic State. Both were much more proximate threats to Israel than an illogical Iranian nuclear strike. The final constraint formed when Iran became an implied nuclear power once its underground – and impenetrable to Israel – Fordow facility went fully operational in December 2011. At that point, an Israeli attack would have all but guaranteed that Tehran would eventually produce a nuclear device.

“Vladimir Putin Wants to Recreate the Soviet Union”: The most obvious example of the fundamental attribution error in the last decade is the prevalent analysis of President Vladimir Putin's strategic thinking. The flawed argument is that since coming to power in 1999, the Russian president, enamored of the Soviet Union, has wanted to recreate the communist empire. Evidence for this argument: the 2008 invasion of Georgia, the 2014 annexation of Crimea, and the subsequent interference in domestic affairs of former Soviet states. The Baltic states must, by this analysis, “be next.”²⁵

Even if Putin's nostalgia motivated Moscow's policies, Russia faces numerous constraints. Its symbiotic economic relationship with Europe is a major constraint. While most pundits see Moscow dominating that relationship, it is actually Berlin that has Russia by the … pipelines. With at least 80% of Russian natural gas exports headed for the EU in 2019 – and about half of that going to Germany alone – it would be economic suicide to turn off the tap to Europe.²⁶

Another subtler constraint is the state of Russia's military. While much improved since the 1990s, Russia's military does not have the capability necessary for the massive power projection required to maintain the borders of a vast empire. It barely had the capacity to intervene in Donbas, where Ukrainian troops held Moscow's mercenaries and unofficial volunteers in check. Ukraine has one of the least equipped and motivated militaries in Europe.

The third pillar of the constraint framework is the recognition and avoidance of fundamental attribution error. To forecast politics and geopolitics, analysts have to avoid the fundamental attribution error and focus on the situation, not the person.

The Three Pillars of the Constraint Framework

The constraint framework has helped me make sense of the world, though it is not scientific or quantitative. It is a blend of the materialist dialectic, intelligence methodology, and social psychology. And yes, some voodoo to boot!

My framework's uncertain origins and inexact nature are caveats that I will repeat often, lest someone accuse me of claiming to have invented sliced bread.

And yet, the framework did not come out of thin air. There is a philosophical (materialist dialectic), practical (hypothesis testing), and empirical (social psychology) method in the madness.

To review, the three pillars of the constraint framework are the following:

• The Materialist Dialectic: The starting point for geopolitical analysis must always be rooted in the material world. While that reality does not necessarily need the Marxist emphasis on “means of production,” it does need to be empirical. Investors should always seek the immutable reality – the constraints – in a situation. While the economic constraint is not always the most important one, some other tangible barrier will be.
• Diagnosticity: Given paucity of information and poor data quality, investors (and others seeking to forecast politics) must focus on the data that is concrete and iterative, rather than ephemeral data locked away in someone's brain. To properly falsify a hypothesis, investors should focus on data that has a high level of diagnosticity. Policymakers' preferences are not diagnostic. Constraints are.
• Social Psychology: Social psychology makes a strong case that the context and the situation, not the person, drive human behavior. Therefore, investors and others seeking geopolitical insights should focus on constraints over preferences.

These three pillars support the Maxim That Shall Forever Be Bolded:

Preferences are optional and subject to constraints, whereas constraints are neither optional nor subject to preferences.

And Now for Some Math

No book about forecasting can be without math. Math is often overused in both finance and academia, and I don't want to fall into the trap of giving my framework a false veneer of science. But I feel obliged to have some formal logic in the book if only to preserve my street cred among fellow forecasters. As such, I have decided to load one section of the book with enough mathematics to fill the required quota.²⁷ In this section, I “formalize” the above maxim.²⁸

In the preference model, the probability that a decision-maker, , will accomplish the preferred outcome, , is expressed as . The equation to solve for is

where is the probability that prefers action , and is the probability that taking action leads to 's preferred outcome, . I call this the “preference model” because the decision-maker can control her actions, y, but the action is not the direct, sole variable in determining the outcome. The decision-maker's preferences also determine the outcome by influencing her actions.

The constraint model introduces variables that constrain the decision-maker from taking action . These constraining variables are . For instance, constraint – rain – will likely influence an athlete's preferred action, – to go out for a run. Citing a single preference, associated with an action, ignores the hidden variables, like weather, that influence the athlete's decision.

A world in which the preference model holds true would look like this: a toddler prefers to take the action of buying a pony. The outcome – ? She buys a pony, because material constraints – driving age, parental permission, access to forms of revolving credit, literacy, etc. – don't inhibit her preference from affecting the outcome.

To account for the role of material constraints in the decision-making process – and the final outcome – the second model is

In the above model, I add up all the possible constraints (), while is the probability that is true and affects the outcome, . The other factors remain the same but now depend on these additional constraints.

What is the point here? The additional factors – the constraints – mean that these two equations are not the same, and thus predict different outcomes.

If I think of the preference model as just adding up the preferences alongside each constraint – and – then the probabilities, in the two models would only produce the same outcome when the constraints are irrelevant (, or the total constraints have zero influence on decision-making).

I'll illustrate this with an example.

I am a big fan of basketball – the National Basketball Association (NBA), to be exact. I've watched it since I was a kid. Back in the day, players used to take swings at one another all the time. Shaquille O'Neal once almost decapitated Charles Barkley. Chris Childs landed a combo on the great Black Mamba (Kobe Bryant). And those were just the 1990s. If you go even further back in history – the 1980s in particular – the brawls were vicious.²⁹

There are lot of situations in the NBA today where it appears that a fight is about to break out. And yet, there are few fights. Instead, there are a lot of “hold-me-back” incidents that only appear aggressive. In fact, they involve two players jawing aggressively but patiently waiting for their teammates and referees to separate them so that they can yell at each other from a safe distance.

How do the two models account for this?

• Preference model: In the preference model, y is 's preference to be in a fight, and is the final outcome of an actual fight breaking out. From the frequent posturing, it seems that players really want to fight. So . However, the final outcome is rarely a fight. So . This disparity between preference and outcome suggests that it is difficult to go from wanting to fight to actually fighting, so must be close to zero () for the model to balance. But there is no place in this model to explain why should be 0, when preferring to fight should lead to a fight outcome (think 1980s basketball). The preference model has no means to describe the teammates who stop the fight, despite 's strong preference to fight.
• Constraints model: The preference for a basketball player, , to fight is actually related to whether or not his teammates are there to stop him. If (there are no teammates nearby), then which means that the bold professional has no desire to fight. If there are teammates present (), then . However, through years of dedicated observation, I know that when teammates are nearby, there is no fight because they “hold back,” so that When teammates are not near – – there can be a fight if that is 's preference. Unlike the preference model, the constraints model enables me to accurately describe what is happening on the court:
  

In the NBA, there are no fights because the preference to fight is anticorrelated with the probability of a fight outcome. Or, to put it bluntly, “Hold me back …”³⁰

Wait, does that sound familiar? It should! There are “hold-me-back” moments all the time in the world of geopolitics. Leaders taunt each other and signal aggressive policy when they may have no intention to follow through on their threats. Sometimes the intention of the rhetoric is the exact opposite – to get one's opponent back to the negotiating table.

I have shown that merely knowing the preference of an actor is insufficient to determine his ultimate action. Preferences are optional and subject to constraints, whereas constraints are neither optional nor subject to preferences. As such, investors need to study the constraints, not the preferences.³¹

Where Do We Go from Here?

As illustrated in the above equation, the constraint framework leaves room for the potential relevance of decision-makers' preferences, including their ideologies, upbringing, culture, and religion. The framework allows for respect of those who delve deep into policymakers' minds. A good biography or a history book is always the first place to start when contemplating a forecast. In order to practice the constraint framework, one has to know things.³²

But the constraints put on this knowledge create much-needed objective distance between the analyst's and actors' personalities. The problem with personality-based (preference-based) analysis is that it is difficult to be empirical about preferences. It is also difficult to operationalize them. I can read The Art of the Deal once to download the supposed crucial insights on President Trump's preferences, behavior, and temperament. But with this task complete, it is difficult to accurately apply that information to forecast whether he will bomb Iran, pass corporate tax cuts, repeal Obamacare, impose tariffs on China, invade Canada, or let COVID-19 burn through the population like a fever. It is difficult because he may have a preference to do all of those, but I do not know the limiting factor that determines what he can and cannot do: material constraints.

As such, a preference-based analysis is the starting point of a forecast. But it is something that analysts need do only once or twice per public figure. It is not an iterative process.

In the next section of this book, I delve into the actual constraints. I define five material constraints that are particularly crucial in forecasting events: political, economic, financial, geopolitical, and constitutional/legal, as well as the constraint wild cards: terrorists and pandemics. I arranged them in order from the most salient to the least. Each chapter discusses how each constraint operates, how to measure it, how it helped make an important forecast in the past, and how it may help investors do so in the future.

But before I embark on the analysis of material constraints, in Chapter 3 I survey the “other” framework of forecasting geopolitical events – the one that focuses on intelligence, or insights.