Security events are the cruelest tail risk. Coups, wars and shifts in alliances or political strategies can have a massive impact on asset prices but are devilishly difficult — if not impossible — to predict. To call two of the biggest risk events of 2014 correctly, you had to know the mind of Russia’s Vladimir Putin or Ali al-Naimi, the Saudi oil minister. Good luck with that.
Yet with geopolitical risks exerting greater influence over markets these days, global macro traders have to take a hard-nosed view of a wide range of issues. Back in June, the London- and New York–based traders I survey dialed down their assessment of risk on a number of fronts. They put the probability of a major maritime clash between China and Japan or China and members of the Association of Southeast Asian Nations at 19 percent, down a touch from 20 percent six months earlier; the odds of a major North Korean military provocation at 18 percent, down from 23 percent previously; and the risk of a strike on Iranian nuclear facilities at just 11 percent, down from 18 percent. They were more pessimistic elsewhere, estimating a one-third risk of major violence (defined as more than 1,000 deaths) in Egypt and Ukraine and a 58 percent probability that Iraq would further dissolve into full-scale civil war.
As it turned out, there was no major maritime clash between China and any of its neighbors over contested borders. Tensions were cooled by a perfunctory meeting between Prime Minister Shinzo Abe and President Xi Jinping at the Asia-Pacific Economic Cooperation summit meeting in Beijing in November. There were not a lot of smiles for the cameras and not a very warm handshake, but the two sides managed to forge a diplomatic “agreement to disagree” about the status of the Senkaku Islands, as Japan calls the East China Sea islands, or the Diaoyus, as they’re known in Beijing. Tensions also cooled in the South China Sea as the Chinese withdrew their drilling rigs from contested areas.
North Korea never fails to disappoint, with three military provocations in 2014 — two shelling attacks and a naval intrusion into the Yellow Sea’s contested Northern Limit Line — but none of these caused South Korean casualties, unlike previous bloody incidents. The biggest casualties of Pyongyang’s belligerence were the hapless hacked executives at Sony Pictures Entertainment in sunny Culver City, California.
We all know from the headlines just how much major violence took place in Ukraine, with the United Nations estimating 4,000 people killed. As for Egypt, there is no definitive number for the body count caused by the repression of the Muslim Brotherhood and their liberal friends by the government of President Abdel Fattah al-Sisi, but it probably surpasses the toll of 1,000 that our survey defined as “major violence.” Wikipedia estimates that 183 Egyptians were killed by police in political demonstrations during 2014, 529 members of the Brotherhood were sentenced to death by a military court in March, and an unknown but no doubt sizable number of the estimated 19,000 civilians arrested and jailed during the political demonstrations of 2013 died in prison or were murdered by the security services.
Iraq is engaged in a civil war by any definition, with four overlapping parties to the violence: ISIS in the west, Kurds to the north, Sunni tribes in the west and south and the Shia government with its army and militias in the east. The U.N.’s bleak estimate of 10,000 Iraqis killed in the first nine months of 2014 certainly constitutes “major violence.”
What lies ahead? The traders polled in December assigned a mean probability of just 9 percent to an armed clash over sea borders between China and its neighbors in the first half of 2015, down from 19 percent previously. They slashed the odds of a large land border clash between India and Pakistan to just 14 percent from 52 percent previously. The mean odds remained relatively unchanged for a North Korean military provocation (15 percent), a strike on Iranian nuclear facilities (9 percent) and a major terrorist attack in a country of the Organization for Economic Cooperation and Development (20 percent, down from 24 percent previously) but increased moderately for a major cyberattack (to 39 percent from 31 percent).
“One bet I would make in 2015 is a major cyberattack on a bank,” says one Hong Kong–based trader. “If they could infiltrate Sony, which is meant to be good at this stuff, they could really hurt a bank.” Such an attach could cause panic on the part of the targeted bank’s counterparties, particularly if it were to take place during stressed market conditions and a “flight to quality.”
As for the perceived risk of two “new” security concerns: The average forecast for ISIS being contained within its Syrian and Iraqi borders was 45 percent, and for a serious Ebola outbreak outside of West Africa at 1 out of 10. It’s not clear that either risk would have a major impact on markets. “My concern is about the contagion risk of a geopolitically significant event,” muses Michael Hintze, CEO and senior investment officer of London-based hedge fund firm CQS. “Terrible things have happened and have had little or no effect on financial markets globally. There needs to be a transmission mechanism into the financial system to have a material effect on markets.”
Also from this series:
- Global Macro Traders Are Cautious After 2014 Surprises
- Global Macro: Big Risks and Opportunities in Emerging Markets
- Global Macro: Expectations Decline for U.S. Rates, China Growth
Getting It Right and Wrong
Why did the global macro traders get it so wrong on U.S. interest rates, Japanese growth and the price of oil last year, and why were they on the money on the euro zone, China, emerging markets and (by and large) on most of the security risk wild cards?
The first explanation of this mixed bag of predictions is model accuracy: If we have accurate models, we get accurate predictions.
For this to be true, the models for forecasting economic outcomes in China, emerging markets and the euro zone have to be better than the models used for forecasting U.S. interest rates, the Japanese economy and the price of oil. I find this idea dubious. We’ve had decades to build up dynamic stochastic general equilibrium models for the macro economies of Japan and the U.S., and these are more robust than those built for, say, China. And there are probably a hundred finely tuned regression equations for Brent crude, with a very deep data series, and yet virtually everyone blew it.
A second possible explanation is that cyclical changes such as normalization of U.S. ten-year Treasury yields are easier to predict than outcomes involving structural changes, such as the third arrow of Abenomics or Chinese macro rebalancing. Cyclical outcomes by definition have happened before, and there are a data set and some kind of regression model to work with, whereas structural changes can be one-off or sui generis.
Yet this explanation isn’t much help in explaining why our group got some things right and some things wrong. We’ve seen several oil cycles before, but traders missed the Brent collapse; we’ve seen several emerging-markets financial crises before, and they accurately called this one in 2014. Conversely, fundamental structural changes are driving the trajectory of GDP growth in both China and Japan, yet our mid-2014 survey called China right and Japan wrong.
A third explanation for the relative accuracy of the June 2014 survey is that favorite catch-all, globalization. The increasing mobility of capital, labor and technology across national borders has created tighter causal links between these event outcomes, including between economic and security events. Some deep drivers, such as Chinese GDP growth, increasingly determine an outcome like the price of Brent crude or the probability of a balance of payments crisis in Brazil. So those who get the important deep drivers correct have a better chance of getting the other outcomes correct too. Conversely, if you get a deep driver wrong, the rest of your scenarios are likely to be flawed too.
I went back to the data from the June survey and tested for correlation between accuracy in forecasting Chinese growth and accuracy in forecasting EM currency crises. Alas, there isn’t any meaningful statistical relationship. Yet I think there is still a nugget of truth in this explanation.
The fourth explanation looks not to the model, cycles or globalization but to the decision-making process itself. Outcomes that are decided by relatively small groups of people — such as the Federal Open Market Committee in Washington, Vladimir Putin’s inner circle in the Kremlin, Kim Jong-un in some palace in Pyongyang or Abe’s coalition in the Japanese Diet — are harder to predict than those involving hundreds, thousands or millions of people voting with their feet or their wallets, such as the Ukrainian protesters, ISIS, global oil consumers or Chinese real estate investors.
This strikes me as inherently plausible. Elite decision making can be cracked if your intelligence is good enough. That’s a hard problem when those elite decision makers are former spies themselves, like Putin and most of his henchmen from the former Komitet Gosudarstvennoy Bezopasnosti (KGB), whose operational security is very high. Getting inside Kim’s head is equally difficult and, at least in my opinion, even scarier, having spent some years closely observing the murderous Kim clan when I was at the CIA. By contrast, we now have new methods of tracking and aggregating the “wisdom of crowds” by listening to the vast digital conversations that are unfolding across social media, from Facebook to Twitter and Wikipedia, and extrapolating the probability of outcomes from these digital signals. I am spending an increasing amount of my time and mental energy in refining these methods of predictive data analytics. Yet these factors don’t explain the June 2014 survey’s relative accuracy: elite and mass outcomes are on both sides of this divide.
A fifth possible explanation for relative accuracy looks not just at the decision-making process but also at the decision makers themselves. Traders think about political events in terms of optimal decision making by rational actors; they expect efficient outcomes. Yet what if these decisions aren’t rational but are driven instead by some volatile brew of fear, ideology, religion or nationalism? If you can’t estimate the objective value of the different outcomes, you can’t calculate the optimum path down the policy decision tree.
I’m afraid this explanation doesn’t get us very far in understanding the relative accuracy of our traders’ June 2014 survey. True, identity politics and nationalism played a key role in the violence in Ukraine and Egypt, which was relatively underestimated, but they also played a key role in the chaos in Iraq and in continued Indian-Pakistan clashes, which our survey accurately foresaw. If you dissect the political calculus behind all of the fiscal, monetary and security actions that shaped the outcomes the traders got right and the ones they got wrong, you find that all of those actions were driven by the rational objective of leaders determined to hang onto power — be it the Standing Committee of the Politburo of the People’s Republic of China, the council of ministers of the European Union or Supreme Leader Kim in Pyongyang.
The sixth possible explanation is that our survey participants and I have been deluding ourselves, and that these outcomes are actually just a “random walk,” as my onetime Princeton professor, Burton Malkiel, eloquently put in back in 1973. In other words, global macro predictions are as scientific as the zodiac.
“Market forecasting and prediction is a waste of time,” says Florida-based portfolio investor Tim Melvin, dismissively. “If someone tells you that they know what the market is going to do and can make you money predicting stock and bond prices, take a deep breath, kick them somewhere sensitive and remove yourself from the scene. Every once in a while someone makes a lucky guess, and immediately and with great fanfare is anointed Wall Street’s newest genius.” Such soothsayers, he adds, “spend the next several years losing a ton of money for people unfortunate enough to fall for the pitch.”
When Stanford economist Ezra Solomon returned to Palo Alto from service on Richard Nixon’s Council of Economic Advisers, he reluctantly concluded, “The only function of economic forecasting is to make astrology look respectable.” So we will see how the global macro stars align in June, when the rosso corsa Ferraris are finally pulling up at 432 Park’s elegant porte cochere.
James Shinn is lecturer at Princeton University’s School of Engineering and Applied Science (jshinn@princeton.edu) and chairman of Teneo Intelligence. After careers on Wall Street and Silicon Valley, he served as national intelligence officer for East Asia at the Central Intelligence Agency and as assistant secretary of defense for Asia at the Pentagon. He serves on the boards of CQS, a London-based hedge fund, and Predata, a New York–based predictive analytics firm, and serves on the advisory board of Kensho, a Cambridge, Massachusetts–based financial analytics firm.
