Since May, I have been able to spend time in two of the world’s biggest metropolises, Istanbul and Mexico City. They are similar in several ways beside their vastness. Each is the leading city of a substantial and proudly independent nation with a distinctive culture. In both, neighborhoods of poor, traditionally religious migrants from the countryside abut cosmopolitan secular districts.

One significant difference is the crime rate. Three people per 100,00 are murdered annually in Istanbul, versus 13.2 in Mexico City. This is a case where a quantitative difference becomes qualitative. In Turkey, crime is so low that I have seen—admittedly, in a smaller city than Istanbul—a whole shop’s worth of consumer goods left outdoors and unguarded overnight. In Mexico City, conversation quickly turns to avoiding violence. Is it safe to hail a cab? (Generally, the answer right now is: No.) Can you drive safely along a certain road?

According to Philip Mansel (in Constantinople: City of the World’s Desire), Istanbul was notably safe in the 1500s, when it was already one of the biggest—or perhaps the single most populous—city in the world. “Merchants could leave stalls unattended; pastry-sellers trusted members of the public to pay for the wares they left on a small round tray.” Mansel quotes Lord Charlemont’s report from the 1700s: “Housebreaking and street robbery, crimes so unfortunately common in our great towns as to render dwelling in them unpleasant and unsafe, never happen in the Turkish metropolis, and a man may walk its streets at all hours of the night with his pocket full of money, without the smallest fear of danger or molestation.”

Charlemont had an explanation: “the salutary rigour of frequent acts of execution.” It is hard to believe that executions were more frequent in Constantinople than in London or Paris, where a child who stole a roll would be hanged. And even if the Ottomans really used capital punishment more than the English or the French, I doubt the executions of those days can explain the safe streets now.

The “broken windows” hypothesis won’t explain the difference, because Mexico City and Istanbul are both rather chaotic and rife with violations such as illegal construction. Poverty is not a likely explanation, because Mexico City’s per capita GDP is twice as high as Istanbul’s, and its inequality index is not much worse. The proportion of residents under age 20 is virtually identical in both cities. Any explanation of the difference may have to be regional, because Mexico City shares similar crime rates with most other large cities in its region. (With Istanbul, it’s a little harder to say what the region is, and I cannot find crime statistics for possibly similar cities, such as Tehran.)

I’m sure there’s a large literature on this topic, but no persuasive explanation is well-known enough that it has reached me.

I suppose one possibility is that there is no root cause. Crime is part of a cycle: each act of violence begets more violence, puts strains on the state, and makes law-abiding behavior less rational for other people. Once you get into a high-crime pattern, it is hard to get out. But if you can avoid or stop the circle, you may be better off even five hundred years later.