Forecasting

Miscellaneous Alternatives

May 09, 2021

In my previous blog “Demystifying Residential Miscellaneous Usage” I described how the miscellaneous end use is defined as a set of specific technologies and a larger “Other” category. Virtually, all the growth in miscellaneous usage is due to Other. It is the primary driver behind increasing residential electricity usage in the long-term.

Miscellaneous Other is derived by subtracting all identified end uses from total residential electric usage.  In other words, it is a residual. According to the Energy Information Administration (EIA), the forecast of Other is based on real household income projections. There are no saturation or efficiency assumptions involved in developing this forecast.

In the Statistically Adjusted End-Use (SAE) framework we explicitly account for household income in the model specification. If we incorporate household income into the model structure, do we even need to include Miscellaneous Other in the model variables? 

To test the impact, we first estimated residential SAE model that includes Miscellaneous Other. The resulting model parameters and forecast are shown below.

Next, we excluded Electric Other from the Miscellaneous leaving only the specific or identified end-uses and repeated the exercise. The result is depicted below.

There is a slight improvement in model fit when miscellaneous Other is dropped from the model. The standard error falls from 33.34 to 32.93 and in-sample MAPE falls from 2.77% to 2.70%. But the most significant difference is that the negative shift variable in the model (Aft15 – years after 2015) turns positive and statistically insignificant. This suggests that the negative shift variable is only compensating for the increase in Miscellaneous Other usage. The long-term average use forecast is slightly lower without Miscellaneous Other.

Starting with the 2021 SAE files, we will separate specific Miscellaneous end-uses from the Other Miscellaneous end-use. We encourage our Energy Forecasting Group (EFG) members to evaluate their forecast models with and without the Other Miscellaneous end use. If you would like to learn more about the EFG or long-term forecasting, we encourage to visit Itron’s forecasting website or contact us at forecasting@itron.com.

By Oleg Moskatov


Forecast Analyst


At Itron, Mr. Moskatov is responsible for providing forecasting support to clients in the electric, natural gas, and water utility industry. This includes conducting research, developing client-focused statistical analyses used for forecasting long-term sales and peaks, and writing supporting documentation. This also includes providing on-site support and software/forecasting training. Mr. Moskatov also works with the Energy Information Administration (EIA) to provide annual updates to the Statistically Adjusted End-Use (SAE) electric and natural gas end-use indices for the residential sector. As a graduate student at Suffolk University, Mr. Moskatov focused on the development of financial markets in Central/Eastern Europe and conducted research for the IPO aftermarket study.