Forecasting and time horizon
For long-term studies it may be relevant to consider future scenarios rather than simple extrapolations of past trends.
An example of qualitative forecasting is that of the European electricity market where a continuation of current trends could be expected: 1) the trends of harmonisation and liberalisation of the market to continue. 2) The transmission capacity to continue to be expanding according to the market demand and 3) the current boundaries between electricity markets in Europe to cease to exist, leading to a single European market.
Another method is to apply political scenarios for a market are compared (such as a free market vs. a market that is regulated to include environmental externalities in the decision-making). Such scenario forecasting may, however, be unnecessarily sophisticated for medium or short-term forecasts and more specific, uncomplicated situations.
Sometimes official forecasts are available that may be applied. For instance, in the case of electricity such forecasts are available for several time horizons, e.g. 5, 10 and 20 years into the future. Then the question arises of which time horizon is the most relevant.
When you study a decision to be made at the current point in time, the applied time horizon should reflect the time gap (delay) between the decision and the actual availability of the new capacity. In the power sector, a 5-10 year delay between a decision and the availability of new capacity is realistic.
The question we are really trying to answer is “what technologies will be installed as a consequence of my additional demand in year 0?” Or more concretely, assuming you have a time-series of future investments during the next 15 years: “will my demand in the current year influence the investment decisions in year 0-5, 5-10, and 10-15?” When you formulate the question in this way, it becomes more obvious that what is relevant is only the period immediately after the demand change, including the delay between decision and availability, i.e. 5-10 years.
The additional capacity installed as a result of an additional demand now will of course still be in place in 5 years from now. So if we want to model a change in demand 5 years from now we would have to look at the additional technology on top of the one already installed as a result of the current demand change. That would be the installations in the year 10-15 from now.
In a specific study, you can have activities that are taking place immediately and some that involve operation and disposal maybe 20 years into the future. For the latter you should in principle use the long-term marginal for 20 years in to the future. This does not change the fact that the actual time period to look at is 5-10 years from the point of time of the actual activity, i.e., the difference in forecasts for year 20 and year 25-30 from now, since this reflects the horizon for the investment decisions made at year 20.
The logic described above for choosing the of time horizon of a forecast implies the assumption that our change to the current marginal technology (materialising in year 5-10) does not influence what is identified as marginal technology in the later periods (10-15 years, etc.).
Further readings:
B P Weidema (2004) Geographical, technological and temporal delimitation in LCA. UMIP 2003 method. København: Miljøstyrelsen. (Environmental News 74). http://www2.mst.dk/Udgiv/Publications/2004/87-7614-305-8/pdf/87-7614-306-6.PDF
How to cite this: Consequential-LCA (2020). Forecasting and time horizon. Last updated: 2020-10-01. www.consequential-lca.org.