It sounds like your old meter was a "clockwork" style meter and the "smart" meter is of course electronic. An electronic meter will register any and all wattage used while the old style meter might not register any usage if the wattage is small enough...just not enough oomph to overcome the stiction of all the gears.
Here in Canada, "Measurements Canada" is THEY are the ones that tell the power companies what meters to send in for testing and they set the accuracy limits. Residential meters here have a tolerance range of plus or minus 3% (and since it is unrealistic to test ALL meters, samples were taken of those which had been in service for at least 5 years) and commercial ones are plus or minus 1.5% (and ALL of those were tested every five years). MC determine what meters get sampled and tested. If more than a certain % of those meters are outside the tolerance ranges, ALL the meters in that manufacturing batch get changed. (They keep records of that for that specific purpose) It was part of my job to exchange meters for sample testing purposes and recall purposes, besides installing new meters also. Some of the old clockwork style meters were up to 60% slow (no typo). You can easily understand how the new meters would give the impression to the customer that he was betting billed for much more than he was using when in fact he was just starting to get billed what he actually used.
Out of my 35 years as a Journeyman Lineman, I worked 16 of those as a District Operator. A fair part of the District work centered around meters and investigating customer complaints about large bills from meters that the customers thought were registering too much..."fast' meters. In ALL my time working for the provincial electrical utility, I heard of just two meters that were indeed "fast". One was on a farm transformer pole that had been hit by lightning...we assumed it was damaged by the strike...and I actually found one that was fast in an apartment and for that one, I have no idea why it was fast, it just was. I also had a couple of meters that the hands would slip on the shafts that they were mounted on. That made for some weird readings and they were a real bear to figure out what the cause was. The reads would go way high and way low with no rhyme or reason. Over the years, literally thousands of meters passed through my hands, hundreds of meter accuracy verifications also, but only just a couple of actual verified inaccurate meters that caused the customer to be overbilled. The vast majority of the clockwork meters would cause the customer to be underbilled by varying percentages for years.
