Should data analysis help determine foreign policy decisions?

RASHMEE ROSHAN LALL November 5, 2019

Political scientist Robert Pape analysed data of sanctions leading up to the attack on Pearl Harbour in 1941. Corbis

In 1998, Madeline Albright, then president Bill Clinton’s secretary of state, described the United States as “the indispensable nation”. In 2019, Russia is said to be becoming an indispensable nation, particularly in the Middle East. And last week, a new Atlantic Council report titled Global Risks 2035 defined a “new bipolarity” — the result, it said, of competition between the US and China.

All of the above are valid in terms of a superficial assessment of the shifting contours of global power. But empirical means — that is to say, data analysis — can throw up a slightly different picture of international relations. Some of the metrics of foreign policy data analysis actually suggest that America had less power, even during its era of unquestioned dominance, than is generally thought. What is more, data analysis shows that some of the foreign policy tools used by the US for more than 70 years were neither particularly sharp nor forceful.

Consider some of the work done by CPost, the Chicago Project on Security and Threats at the University of Chicago. CPost’s founder, political scientist Robert Pape, researched economic sanctions and examined the outcomes of more than 115 cases of economic sanctions since the First World War, not just by the US but other countries or regional blocs as well. He found that sanctions worked very well for easy goals such as trying to cut a trade deal. But for tough foreign policy aims such as regime change or requiring a country to pull back a military offensive, Mr Pape found sanctions worked less than 5 per cent of the time — and in about 5 per cent of cases, they resulted in “catastrophic failure”.

He cites the example of Japan in July 1941, South Vietnam in 1963, Iraq after 1990 and Iran’s current defiance of the US. Tuesday marked a year to the day since the US imposed what it described as “the toughest sanctions regime” ever on Tehran to “alter Iran’s behaviour” and persuade it to change its policies, including support for regional militant groups and the development of long-range ballistic missiles.

It could be argued that Iran hasn’t changed its ways very much in the months since the reimposition of sanctions. Data-analyst political scientists might even say that the tough curbs on Tehran’s economic activity have actually moved the needle closer to risk-taking actions that Mr Pape sees as similar to the desperation of Tokyo in 1941, before the Pearl Harbour attack. All of which suggests that data analysis could help inform difficult foreign policy decisions, simply by showing that the usual tactics do not necessarily produce the desired outcomes.

So, why isn’t data analysis more of a factor in the 21st-century practice of diplomacy? The inertia of habit, perhaps. Or, more particularly, the mindset that regards data analytics technologies and techniques as best suited and most useful to commercial sectors, in order to enable organisations to make informed business decisions and maximise profits.

Change is underway, even if it doesn’t make headlines and certainly not the front pages. CPost itself has been around 15 years and other governmental and think-tank initiatives have come to incorporate data analysis in foreign policy. One of the more significant examples is the Finnish government’s support for something it calls “the potential of big data’s contribution to diplomacy”. In 2017, Finland organised two events in Geneva, over the span of several months, with the aim of generating dialogue about “the potential, limitations and challenges related to big data for foreign policy”.

One of the reports summarising the deliberations acknowledged that “big data can be extremely helpful for international affairs [because it] can pinpoint trends, patterns and correlations”. But it also cautioned against the “limited predictive power” of data. “The volatility of international affairs,” the Finnish foreign ministry report noted, “can hardly be summarised into all-explanatory formulas and causal patterns are generally hard to uncover.” It said this was mostly because of confirmation bias and because data often “over-represents those who have access to the internet and digital devices”.

Caveats are all very well, but the Finnish data-in-diplomacy venture throws up interesting insights. For instance, Graham Nelson, head of the Open Source Unit of the UK Foreign and Commonwealth Office, offered examples of how data helped identify trends in world affairs, helping his unit corroborate or challenge long-held assumptions. And Rafael Prince, a secretary at the Brazilian embassy in Helsinki, who holds a doctorate in the area of big data analysis, highlighted his work on large-scale text analysis to better understand multi-stakeholder negotiations in the Internet Governance Forum. Both argued for data analysis in the realm of international policy, even though Sini Paukkunen, the Finnish ministry’s head of policy planning and research, stressed that diplomacy will continue to adapt but the average diplomat “is most likely better with narratives than with numbers”.

Data merely adds another element to the traditional approach to history ­- dates, people in power, wars, treaties — the required knowledge base of those who practice diplomacy

What all of this means is obvious.

Nearly 60 years ago, EH Carr delivered six lectures at Cambridge University on the question: “What is history?” The series was later published in book form. Carr examined historians’ influence on our understanding of the past, the nature of historical facts and the subject matter of history. He too offered caveats about bias but because he lived in a different time, Carr didn’t suggest data analysis as a tool to understand historical trends.

But data merely adds another element to the traditional approach to history ­- dates, people in power, wars, treaties — the required knowledge base of those who practice diplomacy. Data uses numbers to compute correlation, but not necessarily causality, from historical events. It is not foreign policy by numbers but it uses a different statistical language to highlight the previously unregarded lessons of history.

Originally published at