SAN FRANCISCO (Reuters) - An increase in the number of hot days as climate change warms the globe would likely damage the U.S. economy over the long-term, according to research published on Tuesday by the Federal Reserve Bank of San Francisco.

"Our findings suggest that, under a scenario with no large-scale efforts to reduce carbon emissions, future increases in extreme heat would reduce the capital stock by 5.4% and annual consumption by 1.8% by the year 2200," wrote Stephie Fried, a senior economist at the San Francisco Fed, and co-authors Gregory Casey and Matthew Gibson, both professors at Williams College.

The researchers used scientists' best estimates for the number of days per year where working outdoors would cause heat stress, estimated to rise from 22 days in 2020 to 80 in 2100.

They then projected the likely drain on labor productivity in construction, where - unlike most of the services and manufacturing sectors - air-conditioning cannot counter the impact of hot days.

They focused on construction because it makes up a larger share of overall economic output and U.S. investment than other sectors like agriculture or mining where workers are also vulnerable to heat.

"Decreases in construction productivity slow capital accumulation and therefore have long-lasting effects on macroeconomic outcomes," they wrote.

Using a less-likely alternative scenario under which the number of extreme-heat days rises to 125 in 2100, the authors found much larger consequences from a decline in construction productivity, with capital accumulation projected to fall by 18% and consumption by 7% in 2200.

(Reporting by Ann Saphir; Editing by Andrea Ricci)