This is Part 3 of an ongoing look at CO2 offsets. Here are parts 1 and 2. If you’re just joining the discussion, I’d suggest reading those first.

In this post we’re going to look at the effect of “time-shifting” of carbon offsets using Dell’s Plant a Tree for Me program as an example.

Time shifting means that we are going to emit CO2 over one period of time, and offset those emissions with savings or sequestration over a different period of time. In Dell’s case the program assumes that a PC is operated for 3 years, emitting 1/3 of the CO2 each year, and that the offset is planting of trees which will sequester CO2 over a 70 year lifetime. For simplicity, I’m assuming that the trees sequester 1/70th per year, which is not true for the early years when the trees are small. However, it makes the calculations easier and any error is in favor of the program.

So lets do some analysis of how CO2 is offset over time. I’ve attached a spreadsheet (Open Document and Microsoft proprietary (Excel) formats) which is used in all of the results I’ll show below. Feel free to try it out yourself. In the spreadsheet I’ve highlighted three years: 2025, 2050 and 2075 as milestones that we’ll look at. 2025 is interesting because it is getting to the end of the 10-20 year window that the Stern Review has identified as the period where emissions must peak and head downward. I’ve also used Dell sales numbers from here, though the specifics aren’t that important to this analysis.

In the simple case, lets assume Dell runs this program for one year. Once the 3 year lifetime of the PC is up the cumulative amount which is offset gets pretty linear: by 2025 27% of the CO2 has been sequestered, 2050 is 62% and 2075 is 99% (just short of the 70 years). So basically, during this critical period, only 27% of the CO2 has really been offset.

But what happens next year? Let’s assume that Dell continues the program indefinitely. Now the sequestration is having a harder time catching up, and the results at the three dates are 15%, 33% and 51%. So we’ve sequestered even less in the critical period, and even by 2070 only half is sequestered. This makes sense, since every year we’re emitting more and counting on more future sequestration.

This model assumed that Dell doesn’t grow however, which is not the case. Dell was recently growing at 4% unit growth. At 4% CAGR in units, the numbers drop to: 13% (2025), 24% (2050) and 31% (2075). And if we add in a 10% factor to account for the ever increasing energy usage of PCs, the numbers drop to 10% (2025), 13% (2050) and 13% (2075). So in the critical period we’ve offset 10% of emissions, and by 2075 we’re still only up to 13%. (Note: obviously power usage can’t grow at 10% forever. Take this as an indication of the effects of compound annual growth of any kind, not necessarily power).

This is just one example, but having played around with these numbers, you can start to draw some general conclusions:

- Time shifted offsets may have limited impact on next 10-20 years, a period many feel is critical in terms of global warming.
- Time shifted offsets are not a good fit for long-running programs as new carbon emissions continue to dampen the effect of offsets that are finally taking effect.
- Time shifted offsets can become totally ineffective for long-running programs that involve annual emissions growth. This suggests that time-shifted offsets are not long-term solutions in rapidly growing economies such as India or China.

This is one example, and my treatment is far from comprehensive. I’d really like to see a sustainable business research organization publish a more thorough study. The other thing that would be interesting is some kind of standard CO2 net present value calculation. Clearly reduction ton of CO2 in 2067 is not as valuable as reduction of a ton of CO2 in 2007, but by how much? Can I take a time series of reductions and create their NPV?