This article, and the ECOFYS report to which it refers, are unhelpful and misconstrued.
You (and ECOFYS) are making the classic and straightforward mistake of seeing the CO2 problem as a "flow" problem, rather than as a "stock" problem. While many other commentators do the same thing, being in good company does not necessarily put one on the side of reason.
So what's the "flow" approach, and why is it wrong? This approach looks only at the rate of flow of CO2 into the atmosphere, typically between the present and an arbitrary cut off date (usually 2020 or 2050, but it's not important), and takes the view that anything which reduces the flow of CO2 emissions during this period, must be a solution. Hence, energy efficiency looks like a cheap and useful thing to pursue! At first glance, it's nice and intuitive - and the elegant simplicity of the findings "feels" right to the reader. However, what happens after the cut-off date? Is the problem solved, game over? Here, the flow approach offers nothing. The graph ends in 2020 or whenever, and the assumption is that the trend will continue. Sadly, that assumption is likely to be wrong. After all, the CO2 problem is about how great a stock of CO2 is allowed to accumulate in the atmosphere before CO2 emissions (ie, the flow of CO2 into the atmosphere) comes to a halt. Or is made to come to a halt.
Which brings us to the "stock" approach. This approach sees the CO2 problem from a longer-term perspective. In fact, the CO2 problem is only "solved" when mankind stops emitting CO2 into the atmosphere (either by stopping digging up [fossil] hydrocarbons, or by installing CCS at a large scale) - typically when the cost of unabated burning off hydrocarbons becomes higher than that of renewable or nuclear energy (or CCS). Helpful/effective CO2 policy then acts to bring forward this end point - basically by making the cost of burning hydrocarbons unabated relatively more expensive than that of using renewables, nuclear or CCS.
Interestingly, any explanation of the "stock" approach doesn't rely on a 2020 cut-off to hide the inconvenient details: the CO2 solution, such that it is, is defined as that elusive point when mankind stops the unabated burning of hydrocarbons. I'm reluctant to make predictions as a rule, but here I make an exception: that point is highly likely to lie well beyond 2020, 2050, or even 2100!
So how does energy efficiency look under the stock approach? Sadly, it doesn't look too promising. For one thing, there are two types of energy efficiency. One of them, carbon-augmenting, makes things worse - even if it might appear to make things better by 2020. Carbon-augmenting efficiency improvements increase the efficiency with which mankind uses fossil fuels, while leaving renewables, nuclear, and CCS unaffected. Hence a more efficient internal combustion engine or a higher efficiency domestic boiler can be classed as carbon-augmenting. In the long-term, such efficiency improvements postpone the point at which unabated burning of hydrocarbons becomes more expensive than the renewable, nuclear, or CCS alternative. The total economically viable stock of hydrocarbons that we will eventually burn, increases. Carbon-augmenting efficiency improvements are not a basis of an effective CO2 policy.
The other type of energy efficiency improvement, carbon-neutral, doesn't really make much difference. Here, things like better home insulation or more efficient lighting, make it cheaper to use both unabated hydrocarbons and their renewable, nuclear or CCS alternatives. And it does this in equal measure, such that the point at which mankind stops the unabated burning of hydrocarbons, remains unchanged. Far from being a cheap and obvious solution, carbon-neutral energy efficiency improvements don't really change that much from a CO2 point of view. There is one exception to this, which of course this article and the ECOFYS report are too simplistic to address, let alone be aware of: carbon-neutral efficiency improvements, to the extent that Gigatonne X of cumulative CO2 emissions might take place at a later point in time than otherwise, may allow slightly more time for the price of renewables to decline further (ie, before Gigatonne X of cumulative emissions is reached). Hence, for this reason alone (and it is far from certain to hold), carbon-neutral efficiency improvements might be worth a try - but the benefits, far from being the obvious and easy ones set out in this article, are uncertain and subtle.
To conclude, we don't have that many choices. A CO2 tax, in one form or another, is the only viable solution that economists have come up with so far. By contrast, energy efficiency policy, by itself (as described in this article) is, at best, about as effective as pushing on a rope. ECOFYS and the ACEEE ought to reflect carefully on what they are doing, and whether it is really helping to improve CO2 policy formation. There is a fundamental difference between the practical tools available to reduce CO2 emissions (of which there are many, as outlined by ECOFYS) and the policies available to enable them to work effectively (of which we currently only really have one: CO2 taxes).
Interested readers will find a fuller explanation of these arguments in the European Energy Journal. The article, written by myself, is called "The Limits of Energy Efficiency", and can be accessed through http://www.europeanenergyreview.eu/site/pagina.php?id_mailing=287&toegang=918317b57931b6b7a7d29490fe5ec9f9&id=3773