Month: February 2016

Will more than 1 million refugees and migrants arrive in Europe by sea in 2016?

Here are a few general points to provide context: The BBC reports that “More than a million migrants and refugees crossed into Europe in 2015.” Specially 1,294,000, 78% (just over 1 million) of whom arrive by sea, and 37% of whom seek asylum in Germany. The most common entry points are in the Mediterranean: Spain, Italy, and – especially – Greece [1]. Currently, Syrians make up about half of the migrants and refugees crossing into Europe, though about a quarter come from Afghanistan.

Importantly, only 283,532 migrants and refugees entered Europe in 2014[2]. So 2015 represents a roughly 450% increase in the number of migrants and refugees entering Europe. Over the long run, I expect the number of sea-borne migrants and refugees entering Europe (hereafter “M&Rs” – note that I’ll take the qualification “by sea” for granted and not repeat it in order to avoid some tedium) to return to previous historical levels. But the current situation in many way does not resemble the historical situation. Trying to estimate the number of men 18-30 who would die in Europe is usually best done by looking at earlier historical levels. But from 1939-1945, the previous historical levels would have been pretty useless. I think we’re in a situation more-or-less like that with regard to M&Rs in Europe.

There often seem to be two sides to the equation. One is a demand side; here, M&Rs seeking entry to Europe. The other is a supply side; the EU and individual European nation-states (including non-EU nation-states like Switzerland) who can grant entry to Europe.

The demand side of the equation looks to remain very high. To engage in a bit of reckless generalization, the situation in places like Syria and Afghanistan (which made up about 70% of M&Rs in 2015) does not appear to be getting any better. Similarly, places like Iraq and Iran continue to generate sizable numbers of M&Rs. None of these countries appear to have been emptied of potential M&Rs either. So on the demand side, last year is probably the best guide for this year. Since the number of M&Rs in 2015 was a bit over 1 million, I’d bet that there is a high probability (70% or so) that the number will be between 850,000 and 1,250,000 for 2016, with smallish tails (15% or so) on both sides. That translates to about a 52% chance that over 1 million M&Rs.

Obliviously, I’ll want to continue to update the demand numbers as more information becomes available. And the UNHCR reports that there have been 130,110 (call it 130,000) M&Rs so far (2/29) in 2016. That’s about 13% of 1,000,000 M&Rs, about 16% of the way through the year. So it looks as if we’re a bit off the pace of 2015. But in fact the number of R&Ms went up considerably in later months of 2015, peaking in October. If I’m reading the data right, there were only 5,550M&Rs in January 2015 and 7,271 M&Rs in February 2015[3]. So, if I’m reading the situation rightly (and I really wonder if I am since the result is so surprising), we’re an order of magnitude ahead of the pace set in 2015. Here’s the graphic (which is from the UNHCR):

Refugees_Migrants_Emergency_Response_-_Mediterranean_-_Regional_Overview

Another way to put it is this: At the end of February, 2015 the total number of M&Rs was 1% of 1,000,000. At the end of February, 2016 the total number of M&Rs was 13% of 1,000,000. All of this suggests that the probability that demand side will be over a million is higher than 52%. I’d call it something closer to 75%, with that probability rising if the number of M&Rs remains so high in March.

But what about the supply side? The EU and individual European nation-states could limit the number of places for M&Rs. Austria[4], Hungary[5], and Slovenia[6] have done so, as have some other European nation-states. But these are largely barriers to entry (or migration internal to Europe) by land. These barriers will not (directly) prevent entry by sea. There has been some talk of the EU “ring-fencing” Greece[7], which would probably make an already difficult situation in that troubled country much worse. But Greece is still a part of the EU, so a landing there is still a landing in Europe, which is all that matters for the purposes of this question. Perhaps some think that ring-fencing Greece would force the Greek government to discourage M&Rs further. However, I doubt Greece has the resources necessary to do much more than they already are. (BTW, the human cost to the M&Rs of being subjected to – what to say? – a low cost effort to prevent landing in the EU would likely be horrendous[8].) In general, the EU has just not (yet) shown that it can change much on the supply side of the equation, though that might change.

So a 75% likelihood that there will be 1,000,000 or more M&Rs in 2016 looks right to me for now.

[1] http://www.bbc.com/news/world-europe-34131911

[2] https://en.wikipedia.org/wiki/European_migrant_crisis

[3] http://data.unhcr.org/mediterranean/regional.php

[4] https://en.wikipedia.org/wiki/Austrian_border_barrier

[5] https://en.wikipedia.org/wiki/Hungarian_border_barrier

[6] https://en.wikipedia.org/wiki/Slovenian_border_barrier

[7] http://www.ft.com/intl/cms/s/0/c40504cc-c12b-11e5-9fdb-87b8d15baec2.html#axzz41an3ZbDM

[8] http://www.theguardian.com/world/2015/apr/22/could-australia-stop-the-boats-policy-solve-europe-migrant-crisis

Advertisements

Who will win the Republican Party nomination for the US presidential election? (update)

More evidence that Trump is being accepted by mainstream-ish Republican authority figures:

Sen. Jeff Sessions of Alabama, Senate’s leading immigration opponent, endorsed Donald Trump today onstage during a rally in Alabama. Sessions is the first senator to endorse Trump. Along with the endorsements of Govs. Chris Christie of New Jersey and Paul LePage of Maine on Friday, and of former Arizona Gov. Jan Brewer on Saturday, it’s another sign that many in the Republican Party are beginning to accept that Trump will likely be their nominee.[1]

Trump’s unwillingness to repudiate hate groups[2] and his proclivity for retweeting the folderol of fascists from days gone by[3] might well hurt him in the primaries (and certainly in the general election), but I don’t think we’ve yet seen evidence that this is happening, so let’s wait and see.

  • 17% Ted Cruz
  • 32% Marco Rubio
  • 49% Donald Trump
  • 2% John Kasich

[1] http://www.vox.com/2016/2/28/11130298/donald-trump-sessions-klan

[2] http://www.theatlantic.com/politics/archive/2016/02/why-wont-donald-trump-repudiate-the-ku-klux-klan/471345/

[3] http://www.bbc.com/news/world-us-canada-35682844

Will Montenegro become a NATO member in 2016? (update)

“48 percent of Montenegro citizens would like to see Montenegro be member of the western military alliance, according to a new opinion survey”[1]. No idea (yet) what sort of methodology the poll used, so I’m sticking with the status quo ante(46%). That said, I’d say there is some more reason to worry about Montenegro joining NATO in 2016 if less than half the public is in favor of it. Montenegro’s current ruling party might want to spend some time ginning up support before it makes the plunge. Maybe.

[1]http://www.dtt-net.com/en/index.php?page=view-article&article=9452

If Donald Trump wins the Republican nomination, which party will win the U.S. presidential election?

I currently rate the likelihood of Trump winning the Republican nomination at 47%, so the urgency of this question is only a coin flip away.

An unweighted average of polls taken since the beginning of 2016 has HRC over Trump by 7.7%, with a standard deviation of 6.4%[1]. Hence, HRC wins 88% of the time, and Trump wins 12% of the time. And an unweighted average of polls taken since the beginning of 2016 has Bernie over Trump by 8.1% with a standard deviation of 5.9%[2]. Thus, Bernie wins 91% of the time, and Trump wins 9% of the time. Currently, I think HRC has 91% chance of winning the Democratic primary, with Bernie following at 9%. (It’s just a coincidence that the 91%/9% split shows up twice.) So that means we can carve up the probability space into four unevenly sized quadrants:

  1. HRC wins the Democratic nomination (0.91) and beats Trump (0.88) = 0.80
  2. HRC wins the Democratic nomination (0.91) but loses to Trump (0.12) = 0.11
  3. Bernie wins the Democratic nomination (0.09) and beats Trump (0.91) = 0.08
  4. Bernie wins the Democratic nomination (0.09) but loses to Trump (0.09) = 0.01

The result is that Republicans have a 9% chance of winning the election, if Trump is the nominee.

End of story, right? Probably not. My average of polls was, as I said, unweighted, and not all polls are equal. I didn’t use a weighted average because (a) the off-the-shelf weighted averages from, for example, RealClearPolitics and Pollster disagree quite sharply in some cases and (b) I don’t have a reliable weighting algorithm of my own (nor am I likely to have one any time soon). Furthermore, there are very few polls to average, so sample sizes are low. And, polls at this stage of the game have limited predictive value. Finally, the presidential contest is only indirectly (though strongly) related to individual preferences for one candidate or another. It’s possible to lose the popular vote but when the electoral college, as has happened within living memory.

So how to take all of the problems with the information we have into account without just throwing up our hands? I propose that following: The principle of insufficient reason[3] suggests that we should assign equal likelihood to states when the probability of their occurrence is unknown. Extending the principle a bit, we can say that we should assign equal likelihood to states when the probability of their occurrence is perfectly unknown, but we should assign a weighted average of equal likelihoods and likelihoods based on what evidence that we have when the probability of their occurrence is imperfectly unknown. For instance, an outcome might be unknown imperfectly to a degree of 75%. We could estimate this in this way: (0.75)(Outcome A*0.5 + Outcome B*0.5) + (0.25)(Outcome A*known probability of Outcome A + Outcome B*known probability of Outcome B). And so on. Call this discounting a forecast by the principle of insufficient reason, where we can discount by any real number between 0 and 1, inclusive.

On the basis of all of this, I estimate we should discount the polling data above by the principle of insufficient reason at a rate of 70%. On this basis the probability that a Republican will win, given that Trump is the nominee is currently (0.7)*(0.5) + (0.3)*(0.09)=0.38. Conversely, the probability that a Democrat will win, given that Trump is the nominee is currently (0.7)*(0.5) + (0.3)*(0.91)=0.62. Furthermore, I postulate that the degree of ignorance will diminish by – let’s just say – 10% points per month as we get closer to the election. So at the end of the month, I can (and will!) update the likelihood with new polling data (and anything else I can find), discounting this information by the principle of insufficient reason at a rate of 60%.

(Note: this forecast revisits and updates some material I’ve discussed in forecasting a related question or two.)

[1]Data about individual polls was taken from http://www.realclearpolitics.com/epolls/2016/president/us/general_election_trump_vs_clinton-5491.html. However, I did not use RCP’s weighting algorithm.

[2]Data about individual polls was taken from http://www.realclearpolitics.com/epolls/2016/president/2016_presidential_race.html. However, I did not use RCP’s weighting algorithm.

[3] http://mathworld.wolfram.com/PrincipleofInsufficientReason.html

Will the Republican candidate for president win the party’s nomination on the first ballot, at the party’s convention in July? (update)

Rubio is talking up (somewhat indirectly) the possibility of a contested convention[1], but it sounds more like panic in the face of Trump’s recent successes than something that is probable on its merits alone. I think we’re likely to have a clearer picture of the playing field on Wednesday, though I think it’s highly likely that Trump will walk away with the lion’s share of the delegates. For now I’m bullish on a first ballot candidate: 85%.

[1]http://www.slate.com/blogs/the_slatest/2016/02/26/marco_rubio_s_contested_convention_talk_shows_how_desperate_he_is.html

The 2016 Presidential Contest and the January Economist/YouGov Poll

The Economist/YouGov just published[1] a poll that was conducted in mid-January. Here’s a data array that’s of some interest for this question – namely, “If an election for president was going to be held now, would you vote for….” Here are the results:

  • The Democratic Party candidate: 39%
  • The Republican Party candidate: 46%
  • Other: 2%
  • Not sure: 12%
  • I would not vote: 2%

If we toss out folks for aren’t going to vote and folks who are planning to vote for a 3rd party candidate, then we have Republicans getting 47.9% (=46/96) of the vote and Democrats getting 40.6% (39/96) of the vote, with the remaining 12.5% (=12/96) in play. In order to win, Democrats would have to capture the remaining 5/6 of the undecided votes. All other things being equal, that’s highly unlikely. Very few demographic groups regularly go that strongly for either party by this large a margin, and given that the undecided voters are, well, undecided after all, they are likely to break fairly evenly for either party. Getting 60% of the remaining 12% would be quite a victory; getting almost 85% would be a miracle. This is bad news for the Democrats.

How good is the polling data? Glad you asked. This poll has reasonably good credentials, though some think that it tends to favor Republicans and typically Republican causes[2]. Moreover, the data it presents is a month old, and this olde world moves mighty fast these days. But I should clarify that though the survey as a whole aims to get at the attitudes of adult Americans, the question about who one would vote for was limited to people who self-identified as registered voters. Nevertheless, it’s worth noting that 98% of these registered voters claim that will vote this fall. Fat Chance! In fact, 85% is a more realistic number. In 2012, the rate was a tad higher: 86.3% (=126,144,000/146,311,000)[3]. More generally, voter turnout in U.S. presidential elections has oscillated between 45% and 65% since 1908, and it hasn’t been over 60% since 1968[4].

If those registered voters planning to vote for the Democratic Party candidate turn out at just over 90%, while those planning to vote for the Republican Party candidate turn out at a rate of just under 80%, then the number of voters for each party’s candidates would be very close to equal, so it would come down to who is able to get the largest share of those who are both currently undecided and likely to vote in the fall.

How likely is that to happen? One’s initial reaction might be: Not Bloody Likely. After all, “Republicans are more likely than Democrats to turn out”[5]. But the question isn’t, “Are you more/less likely to turnout, given that you’re a Republican.” The question is, “Are you more/less likely to turn out, given that you’ve said you plan on voting for the Republican candidate.” Only 27% of the respondents self-identified as a Republican, fewer than those who self-identified as a Democrat[6]. That suggests that many of those who intend to vote for Republican candidate (or, more precisely, would do so if the vote were today) are among the 29% who self-identify as independents or the 10% who choose “Not Sure” as an answer.

So if – and it’s a big if – this poll tells us something useful about the 2016 election it might be that it could come down to whether self-identified independents and other unaffiliated registered voters show up to vote. And how, I ask you, do we estimate the probability of that?

[1]https://d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/ctucuikdsj/econToplines.pdf – Question 30

[2]https://en.wikipedia.org/wiki/YouGov

[3] http://www.statisticbrain.com/voting-statistics/

[4]https://en.wikipedia.org/wiki/Voter_turnout_in_the_United_States_presidential_elections

[5]http://www.pewresearch.org/2016/01/07/measuring-the-likelihood-to-vote/

[6]https://d25d2506sfb94s.cloudfront.net/cumulus_uploads/document/ctucuikdsj/econToplines.pdf – Question 76

After the Super Tuesday primaries, which Republican candidate will have won the most pledged delegates?

Here’s where we stand as of today (February 25):

  • Trump: 82
  • Cruz: 17
  • Rubio: 16
  • Kasich: 6
  • Carson: 4[1]

In other words, Trump has a head start of 65 delegates. In order to make sure that he remains ahead after Super Tuesday, he only needs to maintain this head start by one delegate. Suppose Trump wins 280 delegates on Super Tuesday. Then Trump ends the day with 362 (82+280) delegates. And suppose that – per impossible – Cruz wins all of the remaining 344 (624-280) delegates. That gives Cruz 361 delegates. In short, Trump needs only get 44.8% of the delegates available on Super Tuesday to guarantee he’ll have the most delegates after Super Tuesday.

How hard will it for be for Trump to win 44.8% of the delegates available on Super Tuesday? Up to this point he’s been winning 61.6% of the delegates in play, and there are now fewer candidates with which to compete, so my initial thought is: Not that hard. As long as Trump performs 73% (44.8%/61.6%) as well as he has in the first 4 primary states, and he is guaranteed to come out ahead on delegates.

And is there any reason to think that Trump will underperform by such a large margin? Not much, as far as I can see. Trump has a 77% chance of winning Alabama[3], a 79% chance of winning Georgia[4], a 43% chance of winning Massachusetts (beating Rubio at 18%)[5], a 43% chance of winning Oklahoma (beating Rubio at 41% – really too close to call)[6] – despite being unable to spell the name of the state correctly[7] – and a 66% chance of winning Virginia[8]. Only in Texas is Trump an underdog. Here Cruz has an 83% of winning with Trump and Rubio each getting about 9%[9]. This last fact is not trivial since Texas has 155 total delegates to pledge.

Suppose that we give Cruz 35% and Rubio 30% of the delegates (giving the larger number to Cruz because of his advantage in Texas) and throw another 5% at Carson and Kasich. The outcome would look like this:

  • Cruz: 17 + 219 (=0.35*624) = 236 delegates
  • Rubio: 16 + 188 (=0.30*624) = 204 delegates
  • Kasich and Carson eat up 32 delegates, but who cares?

That leaves 185 (=624-439) delegates for Trump, and combined with his 82 delegate total coming in, he easily retains his lead with 267 (=185+82). So Trump could come out ahead on Super Tuesday in this scenario by winning just less than 30% of the delegates and coming in behind both Cruz and Rubio in the total number of delegates won. So I think that the probability that Trump will lead in delegates after Super Tuesday is very high indeed.

I admit that the delegate math is a nightmare[10], but I’m not even going to try to make sense of it because life is just too damn short. If it makes a big difference to my forecast, then I wouldn’t be surprised if someone points it out.

Hope I managed to get the math right this time….

  • 7% Ted Cruz
  • 5% Marco Rubio
  • 88% Donald Trump

[1] http://frontloading.blogspot.com/p/2016-republican.html

[2] https://www.washingtonpost.com/news/the-fix/wp/2016/02/25/donald-trump-needs-to-win-only-39-percent-of-super-tuesday-delegates-to-stay-on-pace-to-be-the-gop-nominee/

[3] http://projects.fivethirtyeight.com/election-2016/primary-forecast/alabama-republican/

[4] http://projects.fivethirtyeight.com/election-2016/primary-forecast/georgia-republican/

[5] http://projects.fivethirtyeight.com/election-2016/primary-forecast/massachusetts-republican/

[6] http://projects.fivethirtyeight.com/election-2016/primary-forecast/oklahoma-republican/

[7] http://www.politico.com/story/2016/02/trump-oklahoma-spelling-219775

[8] http://projects.fivethirtyeight.com/election-2016/primary-forecast/virginia-republican/

[9] http://projects.fivethirtyeight.com/election-2016/primary-forecast/texas-republican/

[10] https://en.wikipedia.org/wiki/Republican_Party_presidential_primaries,_2016