A mild December helped reduce construction unemployment for the month, according to analysis by Dr. Bernard Markstein of the Associated Builders and Contractors.

Not seasonally adjusted (NSA) construction unemployment rates for the country and 32 states were lower than in December 2014. Rates were unchanged in three states.

For the year, national employment in construction increased by 273,000. Meanwhile, the year-over-year change in the national NSA construction unemployment rate continued its series of consecutive monthly declines dating back to October 2010. The rate was down 0.8 percent in December 2015 from a year ago.

“NSA unemployment rates generally have a seasonal pattern with the national NSA construction unemployment rate typically increasing from November to December,” according to Markstein. “While the national construction unemployment rate did increase from November to December, as it has every year from 2000 to 2014, the 1.3-percent increase was below the 2-percent average increase for those years.

The top five states with the lowest construction unemployment rates in December in order from lowest rate to highest rate were Colorado, Georgia, Virginia, New Hampshire and Texas.

Four states—Colorado, Georgia, New Hampshire and Virginia—were among the top five in November. “The unseasonably warm weather in the East helped Georgia, New Hampshire and Virginia turn in low construction unemployment rates,” according to Markstein. “Texas also benefited from relatively mild December weather.”

Colorado’s No. 1 ranking, with a 4.3-percent estimated construction unemployment rate, was up from the fourth lowest rate in November based on revised data.

The five states with the highest construction unemployment rates (from lowest to highest) were Illinois, New Mexico, North Dakota, Wyoming and Alaska.

Four of the five states with the highest estimated construction unemployment rates in December—Alaska, Illinois, New Mexico and North Dakota—were among the five highest in November.

Read more of Markstein’s analysis here.