Zeroes, man. Or as most computer scientists say, “It’s a null value problem.” Trying to tally the JODI Oil data with BP data, I encountered the “binary meaning of a zero” question – where you ask “is this really a zero value, or is it a missing value (or null) ?” Nulls are the bane of engineers and social scientists alike. Helpfully, JODI has provided a guide to trying to determine whether a zero is a null or a zero value, in the form of the download JODI-Oil monthly data availability by country (in Excel file) on the Data Downloads web page, but I found one or two problems when I looked into the actual data – where a data point is supposedly available, it is sometimes given as a zero. I also found that some of the totals given in the JODI Oil data were given as zeroes instead of proper values. I am going to report what I found to JODI, but in the meantime, I have improved my C programmes to read the JODI Oil data and produce reports similar to those found in the BP Statistical Review of World Energy.
This time, I have reproduced page 8 from the 2016 BP Stat Rev based on JODI data and then calculated the difference for each value between the two versions. I have used conditional formatting in Excel to create a spectrum of cell background colours to highlight where the two data sets diverge. It seems clear that data problems for North America were fixed in one or other of these two data sources in 2009, but data for the Middle East are still quite divergent. Most of the countries in the Middle East are in the OPEC group, and according to various documents, their figures for crude oil production do not include NGLs – Natural Gas Liquids (or Natural Gas Plant Liquids), so I thought this was possibly where some of the divergence came from. So as an experiment, I looked up the OPEC Annual Statistical Bulletin (ASB) for 2016 to get 2015 crude oil production data for OPEC countries from Table 3.6 “World crude oil production by country (1,000 b/d)” and found some agreement with the JODI Oil data. This was not a total surprise, as OPEC reports into the JODI group. JODI Oil however, is reporting significantly higher than OPEC for several key OPEC countries – so this will take some further investigation.
What this experiment confirms is that BP is probably reporting high for the Middle East because of the NGLs question. But the BP report is showing much higher values for crude oil production for the non-OPEC category as well, and only shows close agrement with JODI Oil for the European Union region and the OECD.
The NGLs question is hard to answer, because OPEC report NGLs for the OPEC group as a whole and not for each individual country. So now I have to go back to the EIA dataset for an independent data source to try to untangle this. I don’t know how independent any of these data sources are, though. They seem to rely on each others’ data reports in some cases, and their reporting cycles are staggered accordingly. OPEC has a table of monthly production from “secondary sources”, which presumably means companies or countries further upstream in oil production, which would also be reporting to JODI.
It looks like I have more to understand about oil production (and consumption) reporting before I can get on to world gas.