Let me list the numbers in terms of electricity generated:
Electricity generated in Giga KWH:
Total: 364 to 341, -23
Gas: 74 to 91, +17
Wind: 9 to 14, + 5
Petroleum: 2 to 1, -1
HydroElec.: 26 to 24, -2
Coal: 171 to 129, -42
Once again, the weather factor was responsible for a -23 drop. Wind energy contributed to another 5, which was weather related too, as the wind was simply stronger, not that there were more wind installations. Combined, the weather factor contributed 28G KWH, accounting for 2/3 of coal usage drop of 42G KWH. Natural gas only contributed 17G, or 40%.
But even those numbers still do not reflect the full picture, Deteriorating of coal energy efficiency by -3.66% means the same coal generated 5G less electricity. While improvement of NG energy efficiency by 2.17% means NG generated 2G extra electricity. Since the efficiency improvement is probably an illusion created by EIA statistical error, the 5G and 2G should be removed.
Once you remove it, Coal usage dropped by 47G KWH, natural gas contributed 15G KWH extra electricity to displace coal demand, which is 31.9%. The rest is attritable to weather and many a bit of economic activity.
Based on actual increased of NG usage, the increase of NG is 112.2 BCF. Based on energy efficiency by EIA, 1 BCF is equal to 0.0696M tons of coal. So 112.2 BCF of NG displaces 7.8M tons of coal, out of the 19.5M tons less coal burned as reported by EIA.
On the surface, his very detailed use of official figures seems to present a strong argument. And I too have faced criticism from Mark Anthony for not using enough statistics. To this I say that insights trump statistics and that the numbers can easily lie or be misinterpreted.
Indeed, the massive use of historical statistics was one of the things at the core of the massive credit bubble and crisis. This happened because the structured credit products were built upon historical default rates for the cohorts with similar attributes as the ones that the originators were then lending to. The logic went that if those attributes led to a given level of default, then as long as you built the product with enough safety margin, you could lend freely to bad credits carrying those attributes. Nobody ever stopped to think that lending freely to such credits would, itself, change the dynamics in the marketplace, reduce the default rates, and feed back into the products saying that there was less risk than thought possible, leading to even more credit to those risky cohorts, to the point where people were originating the famous NINJA loans, which anyone with a brain could tell you would default massively. In short, the numbers led the analyst astray (with a bit of creativity filling those forms at the origination level, granted).
And here, again, we have a fallacious interpretation of the figures presented. At several times, Mark Anthony ascribes the drop in coal (KOL) usage to weather and not to natural gas (UNG) substitution, putting the figure at just 20%-40% substitution and the rest coming from weather. This is wrong.
Why is it wrong, with so many figures seeming to confirm it is right? Well, the problem is that the calculations Mark Anthony did were made as if a drop in power demand had to hit coal disproportionately, when just the contrary should happen. Given the way the power pools are structured (as explained in my article "Little Reason For Optimism On Coal"), the cheapest power sources are usually scheduled first and thus see more utilization and less impact from fluctuations in power demand. And coal is usually baseload and one of the cheapest fuels. So, if there was less demand because of weather, you'd see natural gas dropping the most in terms of usage, since it would be the higher-cost, marginal fuel, more often. But that didn't happen: Natural gas usage even went up! And why did it go up in usage? Because the power plants being idled and not scheduled were the coal plants! What does this mean? It means that the total substitution of coal for natural gas was not just the fraction represented by the increased usage of natural gas, but also the fraction that natural gas would have absorbed because of less power usage overall- but didn't, because it substituted coal.
In short, the substitution that really took place was much larger than the one Mark Anthony calculated. How much larger? That's impossible to know without an incredibly detailed study of the scheduling at each power pool in the country. We just know it was much larger.
As Einstein once said, "Not everything that can be counted counts, and not everything that counts can be counted."
Why is this relevant? It's relevant because one can find false safety in numbers, leading one to risk investing into situations where there's still a lot of potential grief ahead. Can these markets (coal, natural gas) turn on a dime? They can, if for instance we have an incredibly hot summer. But given that natural gas might reach its storage potential, it might also happen that several companies in both spaces will end up in bankruptcy, so it's incredibly dangerous to be fishing in the space now, before those same companies start discounting the most horrible outcome possible - which none really are doing today.
Just pull up a 10-year chart on Peabody Energy (BTU) or Arch Coal (ACI) or even CONSOL Energy (CNX). Do those look like the charts of equities that face potential distress? They don't. Yet the situation right now can clearly evolve into distress. If you're buying under such context, you really want to be paying prices commensurate with it. If you buy a basket of securities discounting a distressful situation, even if one or two go belly up, you should do alright - though not so if you buy them at valuations that don't discount the distressful situation.