Some of the big news out of Europe this week surrounded Mario Monti’s upcoming resignation from his current post as Prime Minister of Italy. Recall that Monti became prime minister on November 11, 2011 as the leader of a technocratic government. Monti’s government was charged with beginning to dig Italy out from significant economic turmoil, including record bond spreads and fear that Italy could be the next Greece.
Well, as we learned on Monday morning, Monti is resigning as soon as he passes his budget through the Italian parliament. Why did this occur? Due to the withdrawal of support from Silvio Berlusconi’s People of Freedom Party. Financial markets were clearly surprised – and that’s rarely a good thing. The key question from the perspective of my ongoing work is whether or not this was predictable.
This is not just an academic question. As part of the work I’ve been occasionally blogging about on forecasting and international politics, I have been submitting my own forecasts as part of the forecasting tournament. There is an open question in the forecasting tournament about Monti that basically asks whether forecasters think he will exit office prior to 1 January 2013. The question has been open for months.
Over the weekend, prior to the release of the news about Monti, I had the probability of his exit from office before 1 January 2013 as extremely low – around 10%. Clearly I was wrong, but why? Rather than just navel-gazing, this gets to the question of the underlying assumptions that drive the predictions we make about the world. In the case of Monti, I thought two things made the probability of his exit from office very low:
- I thought that the scandals that finally drove Berlusconi from office in fall 2011 would keep him on the sidelines [sidenote: He made an odd offer yesterday to not stand for office if Monti would lead his own party in the upcoming elections].
- Credit markets had clearly responded positively to Monti’s economic reforms, and I thought that pressure from credit markets and other European governments would ensure that Italy’s government remained stable through early 2013, at the very least.
Essentially, I made a system-level assumption about the power of credit markets and other countries, along with an individual-level assumption about Berlusconi (and probably a unit-level assumption about Italian politics). What did I get wrong? Maybe Berlusconi’s hubris and desire to get back into power? Maybe Berlusconi’s hold over his own political party, which I mistakenly had not accurately factored in?
Regardless, the broader question this raises is about uncertainty in international politics and the difficulties of making predictions under uncertainty. It also gets to the notion of “close calls”, or things that happen which could have easily gone in the other direction. In this case, all it would have taken is Berlusconi waiting a few more weeks to pull out the rug from Monti and my prediction would have been fine (and who knows – maybe something else will change next week). How should that influence our overall thinking about accurate forecasting? Should we think about the fundamental accuracy of my mistaken prediction differently from someone who thought Monti would exit office the whole time, but for the wrong reason (e.g. because his reforms would fail and credit markets would punish Italy and drive him from office)? I think this issue is interesting and it is just one of the issues we are looking at examining moving forwards. For those reading – how would you think about operationalizing and measuring the “closeness” of outcomes in international politics?