“…we debated whether M2 was mean reverting. The data said yes.”

that is a Steven Kopits commented.

figure 1: M2 in billions, seasonally adjusted, logarithmic scale (blue). Source: Federal Reserve via FRED.

• Use ADF to reject 10% root of unit nulls (intercept): no,
• Use ADF to reject unit root nulls at 10% (intercept, trend): no
• Use DF-Elliott-Rothenberg-Stock to reject 10% root of unit nulls (intercept): No
• Reject unit root nulls at 10% using DF-Elliot-Rothenberg-Stock (intercept, trend): no
• Reject 1% mean stationary nulls using KPSS: Yes
• Reject 1% trend stationary nulls using KPSS: Yes

If there is such a debate, I miss the conclusion (and the general consensus in economics) that Mr. Kopitts is referring to.

Perhaps he was referring to the velocity (M2):

figure 2: The speed of the M2. Sources: Federal Reserve, BEA via FRED, and authors’ calculations.

• Use ADF to reject 10% root of unit nulls (intercept): no,
• Use ADF to reject unit root nulls at 10% (intercept, trend): no
• Use DF-Elliott-Rothenberg-Stock to reject 10% root of unit nulls (intercept): No
• Reject unit root nulls at 10% using DF-Elliot-Rothenberg-Stock (intercept, trend): no
• Reject 1% mean stationary nulls using KPSS: Yes
• Reject 1% trend stationary nulls using KPSS: Yes