Scheduling a Team Across Latitudes

Managing a team across several countries usually means managing a spreadsheet of clocks. The overlap window gets found, the meeting gets placed inside it, and the geography is considered handled. That covers when people are awake. It does not cover when it is dark where they are, how hot it is on the floor they work on, or how far their legal clock sits from their local sun, and those three vary independently of the time zone and of each other.

Each of them has evidence behind it. The evidence is uneven in quality, and the popular version of it is considerably more confident than the studies are, so it is worth being specific about which claims survive contact with the sources.

Position inside a time zone beats latitude

The most robust geographic finding is not about north and south at all. A time zone is a political object with straight edges, and the sun ignores them, so two towns a few miles either side of a border keep clocks an hour apart while seeing near-identical sunrises. Comparing across those borders isolates the effect of clock position cleanly.

Doing that across US time-zone borders, an additional hour of natural evening light reduced sleep by an average of 19 minutes and made insufficient sleep more likely. Income was around 3% lower on the later-sunset side once counties in commuting zones spanning the border were excluded. Income is an economic outcome rather than a count of work completed, so it supports a cost interpretation without measuring anyone's efficiency directly.

Operationally this means a site's country tells you very little. Spain and Poland share a time zone across roughly 25 degrees of longitude, and the sunrise difference between their edges is well over an hour. Comparing local sunrise against actual start times is the version of this that a roster can act on, which is what counting the dark starts at each site gives you.

Latitude sets the swing, not the level

Latitude does one specific thing: it decides how much day length moves through the year. Near the equator the day stays close to twelve hours; at 60 degrees north it runs from under six hours to over eighteen. That swing is real and it is what people mean by the northern winter.

What it does not do is set the absolute level. A 06:00 shift starts before sunrise on all 365 days in Singapore, which has no winter, and on 227 days in Oslo, which has a severe one. Anchorage at 61 degrees north records fewer dark 06:00 starts than Madrid at 40 degrees north, because Madrid's clock sits so far ahead of its sun. Latitude and clock position are separate variables and a distributed team usually has both.

The health evidence attached to latitude is thinner than its prominence suggests. The clearest study followed 32 indoor day workers in Arctic Sweden across a summer week and a winter week: sleep onset came 39 minutes later in winter, mid-sleep shifted later, and sleepiness was higher, with more morning light associated with earlier timing. Thirty-two people, measuring sleep and mood, not output.

Heat is the effect with real magnitude

If any of these exposures has a large measured effect on work done, it is temperature. In a national panel of Indian factories, annual plant output fell by about 2% per degree Celsius, and worker-level data showed productivity declines of 4% to 9% per degree on hot days, with absenteeism rising alongside.

Two details in that finding matter more than the headline number. The first is that climate control mitigated the productivity losses but did not remove the absenteeism, because cooling a shop floor does not cool the commute, the home someone slept in, or the illness burden of a hot week. Heat is a coverage problem before it is a comfort problem, and an operation that has air-conditioned its floor has solved one of the two channels.

The second is the qualifier that goes missing when these numbers travel. US time-use evidence found the labour-supply reductions concentrated in industries with high exposure to climate, which is why the net effect on total US employment was modest. A per-degree elasticity from weaving sheds does not describe a conditioned office, though it describes a warehouse or a forecourt reasonably well anywhere in the world. The distinction is set out in the evidence on heat and output.

Because temperature is forecastable several days ahead, this is an unusually plannable exposure. A heatwave is a coverage event with a lead time, and can be rostered against the way any other predictable spike is.

The claim to stop repeating

A recurring recommendation for distributed teams is to hand analytical work to whichever region is in winter and collaborative work to whichever is in summer, on the grounds that sustained attention peaks near the summer solstice and working memory near the autumn equinox.

Those peaks are peaks in brain activity, from a study of 28 young adults held several days without seasonal cues. Performance on the same tasks did not change significantly across the year. A separate and much larger analysis did find seasonal variation in cognition, in 3,353 older adults on clinical batteries, where the gap was equivalent to 4.8 years of age. Neither describes working-age staff doing their actual jobs, and we hold the two apart in the evidence on seasonal cognition.

Allocating work types by hemisphere is cheap to announce and awkward to withdraw, and it currently rests on a misreading.

What a defensible policy looks like

• Compare local sunrise to local start times, per site. Not country, not time zone. The same shift is a different job in Madrid and in Warsaw.
• Treat forecast heat as a coverage event. Plan for absence as well as slower work, because cooling the floor addresses only one of them.
• Apply heat elasticities only to exposed work. Warehouses, kitchens, outdoor and delivery roles, not conditioned desks.
• Test start-time changes rather than mandating them. The mechanism has support; no study has tested a winter start policy and measured what happened.
• Standardise measurement, not schedules. Absence, lateness, incidents and throughput per site across a full year will tell you more about your operation than any published effect size.

The useful reframing is that a distributed team is not one operation running in several places. It is several operations that happen to share a company, each with its own sunrise, its own summer, and its own distance between the clock on the wall and the sun outside. Planning as though a shift means the same thing everywhere is the assumption worth dropping first.