Border apprehensions in March 2017 fell a lot compared to March 2016. This is taken as evidence that Trumpian enforcement is scaring off potential migrants.
But as Kevin Drum pointed out, March apprehensions have been falling for a very long time. The red line shows March apprehensions for the corresponding year; the blue line shows February.
Then again, that long fall in March apprehensions stopped in 2011. They were stable until 2015, when they fell again, followed by stability in 2016, and then another big fall this year.
So is it Trump? Well, that depends on what you think happened in 2015. If something happened to start migration falling, then 2016 is the blip and this year is just the resumption of the trend. But February apprehensions (before Trump unleashed the CBP) fell in 2015 but did not rise year.
That makes 2015 seem like a one-off. If 2015 was a one-time drop, then 2017 might also be an independent one-time drop.
So what could have caused a sharp drop in apprehensions in 2015? Take a look at the purple and green lines. Those represent Mexican apprehensions of Central Americans headed north in March and February respectively. Those apprehensions jumped in 2015, when Mexico began to crack down on Central American migrants.
Now, the rise in Mexican apprehensions is less than the fall in American apprehensions, but the Mexican crackdown was brutal and likely dissuaded many people from heading north in the first place.
Which means that we do have a one-off policy to explain 2015: a one-time upward shift in Mexican enforcement efforts.
To sum: apprehensions fell steadily from 2007 to 2011 due to the Great Recession. They then stabilize. In 2015, apprehensions shift downward when Mexico steps up enforcement. Then they stabilize again until 2017, when Trumpian enforcement starts to scare people off.
We won’t know until the numbers for April (and May and June and July; gotta see that the drop holds) are in, but it seems quite plausible that the current angry policies have reduced migration. Simply drawing a trend line from 2000 to date can be misleading.