Bryan Caplan  

I Win My Recession Bet

Government Innumeracy... Monetary policy: The view from...
Two years ago, I made the following bet with a brave young economic pessimist, Jackson Taylor:
$200 on whether or not America will have a recession, defined as two consecutive quarters of negative real GDP growth, in the next two years. Q4 of 2017 being the end of the period. But if that quarter happens to be negative, then we can look to the next one to see if we were or were not in the middle of a recession.
Fourth quarter U.S. GDP numbers are now in, showing +2.6% annualized GDP growth.  Since growth was positive in every preceding quarter since the bet, I have clearly won.  The commendable Jackson prepaid, so we are all settled up.

Why did I make this bet?  As I explained at the time:
Base rates.  U.S. quarterly GDP growth is about 3%, and there's a high short-run positive correlation for quarterly growth.  So we're extremely unlikely to have negative GDP growth for the next two or three quarters.  The chance we actually get two consecutive quarters of negative growth before the clock runs out therefore seems well below 50% to me, making this a good bet.
By my count, this leaves me at 16 wins, 0 losses for my public bets.

Comments and Sharing

COMMENTS (11 to date)
robc writes:
this leaves me at 16 wins, 0 losses for my public bets.

Which means you aren't taking enough risks, you are leaving money on the table.

I am 2-0, so ditto, although my sample size is lower (and mine are weird sports related prop bets).

Thaomas writes:

Where do you find people to make these crazy bets? I recall having dozens of people predicting all kinds of disasters during the Obama years, but none would ever put money behind their predictions. :(

Francisco Garrido writes:

Would you be willing to make a meta-bet? Something like "$100 dollars at even odds that you won't lose any of your next 10 bets currently closest to completion".

David R Henderson writes:

@Francisco Garrido,
That’s not a great bet for Bryan. If his probability of winning each bet is 0.9, then his probability of losing the bet to you would be 1 - 0.9^10 = 1 - 0.35 = 0.65.

robc writes:


If his probability of winning each bet is .9, then there is only a 18.5% chance that he would have won the first 16.

His probably on each is more likely to be at least .958, which gives him a 65% chance of winning the next 10.

Tom Jackson writes:

So where do we go to view Bryan's current unresolved public bets?

Chris H writes:

@Francisco Garrido/David Henderson/robc - That meta bet may be inordinately impacted by his "no country leaving the EU" bet. From what I understand, Brexit seems very unlikely to be reversed in part because there are countries outside of the UK that would veto any attempt to keep them in. If that's within his next 10 bets that might seriously throw the chances in favor of the "Bryan loses one of his bets" side.

James writes:


Have you ever made a bet that some number would deviate from its base rate? If not, that says something a bit bleak about economics. You have a PhD and are probably more knowledgeable than most professors in the field and yet your performance as a forecaster is no better than an amateur relying on a table of long term averages.

magilson writes:


That should tell you something about the economy; not Bryan.

Francisco Garrido writes:

@David Henderson

Well, if the winning probability is 90% then his current track record (16 wins and no loses) has a 19% chance. Not exactly impossible, but unlikely.

If we pump up the winning probability to 95% the chance of his track record more than doubles to 44%, and this would make the chance of winning the meta bet equal to 60%. Not exactly a no brainer at even odds, but pretty decent.

However, If one bet is already compromised, as @ChrisH points out, then this probability could be a lot lower.

RPLong writes:

@ Francisco re: meta-bets

Bryan should take Curry's Bet: If Bryan loses this bet, then Germany borders China...

Comments for this entry have been closed
Return to top