Which is not to say it isn't an improvement, just that it oversimplifies things only to look at one side of the balance sheet.
This is highly dependent on where you live. At the extreme, North Dakota [1] gets 75% of its electricity from coal burning. But in California, the mix is 35% Natural Gas, 25% Renewable, 10% hydro, 10% nuclear [2], so nearly half of the electricity is carbon-neutral. If huge places like California go electric, they will take other places along with them.
But even in the unlikely scenario of a primarily coal-based grid in the future, the efficiency increase that comes from the centralization of combustion at the coal plant (as opposed to in the vehicle) still results in lower well-to-wheels CO2 emissions [3].
[1] http://www.ndenergyforum.com/expert-facts/
[2] http://www.energy.ca.gov/almanac/electricity_data/total_syst...
[3] http://mydocs.epri.com/docs/CorporateDocuments/SectorPages/P...
The only USC source I could find[1] says that 20% of US emissions are tailpipe emissions, but the question was about global emissions. Globally transport makes up 14% of emissions according to the IPCC.[2]
[1] http://www.ucsusa.org/clean-vehicles/car-emissions-and-globa...
[2] https://www.epa.gov/ghgemissions/global-greenhouse-gas-emiss...
That's a significant percentage. If the oil demand drops, theres probably another 5-10% that could reduce.
So we're looking at probably a 30-35% drop in emissions.
As an owner of an electric (recently bought), I'm fascinated by how amazingly agile my relatively-inexpensive Ford Focus is. If everyone drove an electric for a day, many would likely decide to go electric purely for performance reasons.