One Measure By Which Things Have Recently Gotten Worse

The United Nation’s refugee agency today released its annual report on people displaced by war around the world, and the news is bad:

The number of people forcibly displaced at the end of 2014 had risen to a staggering 59.5 million compared to 51.2 million a year earlier and 37.5 million a decade ago.

The increase represents the biggest leap ever seen in a single year. Moreover, the report said the situation was likely to worsen still further.

The report focuses on raw estimates of displaced persons, but I think it makes more sense to look at this group as a share of world population. The number of people on the planet has increased by more than half a billion in the past decade, so we might expect to see some growth in the number of forcibly displaced persons even if the amount of conflict worldwide had held steady. The chart below plots annual totals from the UNHCR report as a share of mid-year world population, as estimated by the U.S. Census Bureau (here).


The number of observations in this time series is too small to use Bayesian change point detection to estimate the likelihood that the upturn after 2012 marks a change in the underlying data-generating process. I’m not sure we need that kind of firepower, though. After holding more or less steady for at least six years, the share of world population forcibly displaced by war has increased by more than 50 percent in just two years, from about one of every 200 people to 1 of every 133 people. Equally important, reports from field workers indicate that this problem only continues to grow in 2015. I don’t think I would call this upturn a “paradigm change,” as UN High Commissioner for Refugees António Guterres did, but there is little doubt that the problem of displacement by war has worsened significantly since 2012.

In historical terms, just how bad is it? Unfortunately, it’s impossible to say for sure. The time series in the UNHCR report only starts in 2004, and a note warns that methodological changes in 2007 render the data before that year incomparable to the more recent estimates. The UNHCR describes the 2014 figure as “the highest level ever recorded,” and that’s technically true but not very informative when recording started only recently. A longer time series assembled by the Center for Systemic Peace (here) supports the claim that the latest raw estimate is the largest ever, but as a share of world population, it’s probably still a bit lower than the levels seen in the post–Cold War tumult of the early 1990s (see here).

Other relevant data affirm the view that, while clearly worsening, the intensity of armed conflict around the world is not at historically high levels, not even for the past few decades. Here is a plot of annual counts of battle-related deaths (low, high, and best estimates) according to the latest edition of UCDP’s data set on that topic (here), which covers the period 1989–2013. Note that these figures have not been adjusted for changes in world population.

Annual estimates of battle-related deaths worldwide, 1989-2013 (data source: UCDP)

Annual estimates of battle-related deaths worldwide, 1989-2013 (data source: UCDP)

We see similar pattern in the Center for Systemic Peace’s Major Episodes of Political Violence data set (second row here), which covers the whole post-WWII period. For the chart below, I have separately summed the data set’s scalar measure of conflict intensity for two types of conflict, civil and interstate (see the codebook for details). Like the UCDP data, these figures show a local increase in the past few years that nevertheless remains well below the prior peak, which came when the Soviet Union fell apart.

Annual intensity of political violence worldwide, 1946-2014 (data source: CSP)

Annual intensity of political violence worldwide, 1946-2014 (data source: CSP)

And, for longer-term perspective, it always helps to take another look at this one, from an earlier UCDP report:

PRIO battle death trends

I’ll wrap this up by pinning a note in something I see when comparing the shorter-term UCDP estimates to the UNHCR estimates on forcibly displaced persons: adjusting for population, it looks like armed conflicts may be killing fewer but displacing more than they used to. That impression is bolstered by a glance at UCDP data on trends in deaths from “intentional attacks on civilians by governments and formally organized armed groups,” which UCDP calls “one-sided violence” (here).  As the plot below shows, the recent upsurge in warfare has not yet produced a large increase in the incidence of these killings, either. The line is bending upward, but it remains close to historical lows.

Estimated annual deaths from one-sided violence, 1989-2013 (Source: UCDP)

Estimated annual deaths from one-sided violence, 1989-2013 (Source: UCDP)

So, in the tumult of the past few years, it looks like the rate of population displacement has surged while the rate of battle deaths has risen more slowly and the rate of one-sided violence targeting civilians hasn’t risen much at all. If that’s true, then why? Improvements in medical care in conflict zones are probably part of the story, but I wonder if changes in norms and values, and in the international institutions and practices instantiating them, aren’t also shaping these trends. Governments that in the past might have wantonly killed populations they regarded as threats now seem more inclined to press those populations by other means—not always, but more often. Meanwhile, international organizations are readier than ever to assist those groups under pressure by feeding and sheltering them, drawing attention to their miseries, and sometimes even protecting them. The trend may be fragile, and the causality is impossible to untangle with confidence, but it deserves contemplation.

Leave a comment


  1. I was just taking a look at the paper you linked: A Bayesian approach to change points detection in time series. I find it difficult to believe in that paper that at Fig.1 the third row of time series have all the same variances. Even to an untrained eye the variances seem all different; or am I being fooled by appearances? What’s your opinion?

    • I see what you mean, but isn’t the issue whether or not the variance is proportionate within those segments, not what the variance is across the segments?

      • I’m not sure I understand what you mean by “the variance is proportionate”. So maybe I’m missing something.

        What I see is that the paper claims a methodology to detect change points. In the particular example I singled out, they claim to be able to detect the change when mean and standard deviation are the same, but the correlation length is different. However, if in that particular example they were able to detect the change points, thinking of having done so just because correlation length was changing, and in reality it worked because variance was changing, than their conclusion is invalidated.

      • I’m not sure I know what I meant either, but I guess I was seeing the segments as fractal; if you blow up the ones that look tight, they would look similar to the ones that look loose. So, the distributions may be similarly structured but on different scales.

        More important, the method doesn’t originate with this paper, which might not have been the best example to use.

  2. Texas Cowman

     /  July 14, 2015

    At last, someone is actually looking at refugees. Too many experts today gleefully
    talk about war as a video game where one side or the other is winning. In truth,
    in places like Syria nobody is winning. Infrastructure is being degraded to the
    stone age. People are faced with the choice of staying in place and being subject
    to violence and impoverishment from the various warring factions, or fleeing to
    the tenuous safety of a neighboring country.

    But even this analysis is lacking. It deals with clean numbers. Nothing about
    becoming a refugee is clean or safe or even sane. Four million people are now
    refugees from the Syrian conflict. Those people plus the refugees from Libya,
    Somalia, Yemen, and other conflicts have the potential of the Sea Peoples
    from the ancient Egyptian records to cause chaos to flow from their own conflicts
    into more distant countries. Even now the flow into Italy is both life-threatening and
    potentially disastrous for Italy and to a lesser extent the rest of Europe.

    There is also the vast educational loss where the country’s future engineers
    and craftsmen replace the years necessary to reach those levels of education
    with 5 years in a refugee camp scratching their gonads.

  1. 17 ways the unprecedented migrant crisis is reshaping our world - The Washington Post
  2. 17 ways the unprecedented migrant crisis is reshaping our world | AFSDP
  3. Globala trender 2016 – krig, kris och utveckling | Johan Larnefeldt
  4. The Reprivatization of Security (World Peace edition) | Notes On Liberty
  5. The Re-Privatization of Security (World Peace edition) | Isenberg Institute of Strategic Satire

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

  • Author

  • Follow me on Twitter

  • Follow Dart-Throwing Chimp on
  • Enter your email address to follow this blog and receive notifications of new posts by email.

    Join 13,608 other subscribers
  • Archives

%d bloggers like this: