It’s easy to make fun of the idea that economic anxiety had anything to do with people voting for President Trump. Unemployment was almost at a 10-year low back then, and, adjusted for inflation, median income growth was nearly at a 20-year high.

But beneath the surface of these placid economic statistics, Columbia historian Adam Tooze points out, blue-collar workers were being buffeted by intense self-doubt amid what was for some of them a difficult environment. In fact, according to numbers Deutsche Bank has put together, workers with a high school education or less were more worried about losing their jobs at the start of 2016 than they had been during the depths of the financial crisis in 2009. There were, of course, plenty of other reasons that these elusive “Obama-Trump” voters swung from supporting the Democrats in 2012 to Republicans in 2016 — take your pick of racial or gender backlash, Russian interference, or former FBI Director James Comey’s last-minute letter — but it was at least somewhat the economy, stupid.

Why, though, were blue-collar workers so scared in 2016 when 2015 had been one of the best years of the recovery? Three reasons: the strong dollar, the weak oil business, and the risks that China’s at-that-time wobbly economy might hold everyone else back.

Now, the first thing to understand here is that our recovery had picked up enough steam in 2014 that the Federal Reserve felt like it could afford to stop buying bonds right before the European Central Bank started doing so itself. This divergence between the United States, where markets were anticipating rate hikes in the not-too-distant future, and the rest of the world, where interest rates figured to stay at rock-bottom levels for a long time to come, put a lot of upward pressure on the dollar. The idea being that there’d be more demand for dollars than euros — raising their price — when U.S. bonds were paying people positive interest but European ones were not. The result was that, from the middle of 2014 to the beginning of 2016, the dollar went up 23 percent on a trade-weighted basis against a broad basket of other currencies. Which, in turn, put a major damper on our exports by making them more expensive overseas. It’s no surprise, then, that after increasing by about 200,000 jobs a year during the rest of the recovery, manufacturing employment flatlined in early 2016, and actually fell a bit by the end of the year.

Things weren’t any better for oil and gas workers either. The shale revolution that had turned North Dakota’s Bakken, Texas’s Permian Basin, and, as would be especially important in the election, even Pennsylvania’s Marcellus patch into the boomtowns of a new gold rush was a bit too successful for its own good. There was so much more oil supply that when OPEC opted not to cut production in late 2014 in a bid to put pressure on all these debt-financed shale drillers, prices didn’t just fall, but collapsed. It took a little while for this to filter through into hiring decisions — drillers wanted to see if this was just temporary first — but by the end of 2015 it was really starting to hit workers. Between then and Election Day, oil and gas extraction jobs fell from around 195,000 to 155,000, erasing all the gains of the previous six years.

But this wasn’t just a micro story about blue-collar jobs being in trouble. It was also a macro story about a possible slowdown coming out of China. It had had its own stock bubble burst, you see, and was struggling to keep its economy growing at the 6 or 7 percent clip it wanted — and Beijing thought it needed to avoid unrest — without running up too much debt. All these questions, together with a corruption crackdown, had made people move so much money out of the country that the government had to burn through a trillion dollars' worth of reserves just to keep its currency from falling fast. Even then, they had to devalue a little bit in August 2015, and might have had to even more if the Fed had followed through on its plan to raise rates three or four times in 2016. That would have left all the Chinese companies that had a lot of dollar debts in a pretty precarious position, and put the world economy in danger of losing what had been its engine of growth. It was enough to send U.S. stocks tumbling at the start of 2016, and create enough of a recession scare that the Fed backed off its aggressive rate-hike plan just a month after it had raised rates for the first time in over a decade.

The point is that the economy wasn’t great for blue-collar workers in early 2016, and looked like it might not be so good for anyone if China wasn’t able to figure out a way to solve its own problems. It’s easy to forget that, of course, when the Fed did put its plans on hold long enough to keep a recession that could have been made in China from being exported here, but we shouldn’t. The fact of the matter, as Tooze points out, is that the dollar shock and the oil crash were already enough to make blue-collar workers have a pretty negative view of the economy, and the uncertainty around how long and how well the recovery would keep going were real concerns at the time.

Now, this isn’t to say that this was these voters' only motivation. Nor was it the most important story of the election: That was probably the way so many well-to-do Republicans were willing to vote for Trump despite whatever qualms they might have had about him in hopes of getting the tax cuts and conservative judges they wanted. But in a race that was decided by such a razor-thin margin, you could say that a lot of little things were the decisive one.

A few bad months for these swing voters at the worst time for Democrats certainly played a part.