Yes.

Michigan left

This page (pdf here) is very informative, though you really have to dig to get what you want. In a sub-section of 10.2.2 Median U-Turn Crossover, I found this (A "Michigan left" is referred to as a "median U-turn crossover"):

A study on a Michigan corridor used simulation to compare median U-turn crossovers with two-way left-turn lanes (TWLTL). The study showed that during peak hours, the corridor with median U-turn crossovers had a lower travel time by 17 percent and a 25 percent higher average speed than the same corridor with a TWLTL. However, vehicles made more stops on the arterial with median U-turn crossovers. In nonpeak hours, the median U-turn crossovers had the same efficiency as the TWLTL, even though a higher delay for left-turning vehicles had been expected due to the higher travel distance a vehicle must cover to turn left using a median crossover.

So that's a yes for stopping congestion during peak hours. More information on that specific simulation can be found under found under footnote 149, which isn't too easy to find. Other simulations reportedly found similar results:

Simulation studies using a range of intersection configurations (number of through lanes on the major and minor street) and volumes from intersections in Virginia and North Carolina suggest a reduction in overall travel time for all movements through the intersection when compared to a conventional intersection: -21 to -2 percent during off-peak conditions, and -21 to +6 percent during peak conditions. The studies also show a general increase in the overall percent of stops when compared to a conventional intersection: -20 to +76 percent during off-peak conditions, and -2 to +30 percent during peak conditions.

The rest of 10.2.2 has some more safety information:

• The collision rate is lowered slightly
• There are less "conflict points" (i.e locations where collisions are likely to happen)

Jersey jughandle

The Jersey jughandle (referred to as simply a "Jughandle") does reduce conflict points, though not as much as a Michigan left. It, too, appears to increase efficiency:

Simulation studies using a range of intersection configurations (number of through lanes on the major and minor street) and volumes from intersections in Virginia and North Carolina suggest a reduction in overall travel time through the intersection when compared to a conventional intersection: -6 to +51 percent during off-peak conditions, and +4 to +45 percent during peak conditions. The studies also show a large increase in the overall percent of stops when compared to a conventional intersection: +15 to +193 percent during off-peak conditions, and +19 to +108 percent during peak conditions.

Is there a clear winner? Both clearly reduce travel time and congestion, so the answer to your question is a definite yes. The Michigan left has many less conflict points than the Jersey Jughandle, which I consider quite the advantage. It also has a lesser increase in stops. I'd give the edge here to the Michigan left, though both are probably improvements over your standard four-way intersection.

Yes.

Michigan left

This page (pdf here) is very informative, though you really have to dig to get what you want. In a sub-section of 10.2.2 Median U-Turn Crossover, I found this (A "Michigan left" is referred to as a "median U-turn crossover"):

A study on a Michigan corridor used simulation to compare median U-turn crossovers with two-way left-turn lanes (TWLTL). The study showed that during peak hours, the corridor with median U-turn crossovers had a lower travel time by 17 percent and a 25 percent higher average speed than the same corridor with a TWLTL. However, vehicles made more stops on the arterial with median U-turn crossovers. In nonpeak hours, the median U-turn crossovers had the same efficiency as the TWLTL, even though a higher delay for left-turning vehicles had been expected due to the higher travel distance a vehicle must cover to turn left using a median crossover.

So that's a yes for stopping congestion during peak hours. More information on that specific simulation can be found under found under footnote 149, which isn't too easy to find. Other simulations reportedly found similar results:

Simulation studies using a range of intersection configurations (number of through lanes on the major and minor street) and volumes from intersections in Virginia and North Carolina suggest a reduction in overall travel time for all movements through the intersection when compared to a conventional intersection: -21 to -2 percent during off-peak conditions, and -21 to +6 percent during peak conditions. The studies also show a general increase in the overall percent of stops when compared to a conventional intersection: -20 to +76 percent during off-peak conditions, and -2 to +30 percent during peak conditions.

The rest of 10.2.2 has some more safety information:

• The collision rate is lowered slightly
• There are less "conflict points" (i.e locations where collisions are likely to happen)

Jersey jughandle

The Jersey jughandle (referred to as simply a "Jughandle") does reduce conflict points, though not as much as a Michigan left. It, too, appears to increase efficiency:

Simulation studies using a range of intersection configurations (number of through lanes on the major and minor street) and volumes from intersections in Virginia and North Carolina suggest a reduction in overall travel time through the intersection when compared to a conventional intersection: -6 to +51 percent during off-peak conditions, and +4 to +45 percent during peak conditions. The studies also show a large increase in the overall percent of stops when compared to a conventional intersection: +15 to +193 percent during off-peak conditions, and +19 to +108 percent during peak conditions.

Is there a clear winner? Both clearly reduce travel time and congestion, so the answer to your question is a definite yes. The Michigan left has many less conflict points (16) than the Jersey Jughandle (26), which I consider quite the advantage (the standard four-way intersection has 32). It also has a lesser increase in stops. I'd give the edge here to the Michigan left, though both are probably improvements over your standard four-way intersection.