What makes statistics economical reports so varied and different?

Search for Iran's GDP in Google, and what you will see is an image representing GDP of Iran, compared to other neighbor countries. It gives us the value of 393.4 billion USD (2015).

Now in that page simply open some results. From Wikipedia, from World Economics, and other sources. You see values from 300 to 1.3 trillion Dollars.

Values are not coherent with each other. As an engineer who is not an economist but understands figures and diagrams, I think these values are so varied that they can not be used in researches as reliable information.

I'm working on a proposal, and I need information that can be trusted. What makes such a bold and universally known economical value so varied across data sources?

• Notice that the title of your question is way way broader than that of your actual question in the text. I answered the latter, as the former is way too broad (and thus might be off-topic). I suggest you make the title in line with the content of the question. Sep 29 '17 at 12:43
• @luchonacho, I brought forward GDP as a sample. Other samples might be Gini coefficient, or unemployment rate, or any other number that is used as the base of arguments for a wide range of reasons, from political debates to entrepreneurship. Numbers are so varied that you don't know which one to trust. Sep 29 '17 at 15:06
• Yes, but the reasons why a variable has different values across sources is strongly idiosyncratic to the nature of the variable. A general answer that covers all of them is bound to be very useless. Primary sources IMO are by far the best option, although if you want comparability, this is many times not enough. Sep 29 '17 at 18:29
• Dear @luchonacho, yep it was helpful. Thank you. Oct 12 '17 at 21:00
• Haha, great. Please consider marking it as accepted if it solved your question. If you still have doubts, let me know. Oct 13 '17 at 8:09

Regarding the general question in the title, the reasons why "the same" variable vary across sources depends very much on the particular variable you are talking about. Personally, one of my first shocks when I started working after my undergraduate was the high number of ways the same conceptual variable studied theoretically (e.g. inflation) could be measured in practice. These alternative definitions might refer to slightly different phenomena, or might reflect different set of assumptions by those computing the statistics. Depending on your ontological position when it comes to economic concepts (e.g. is there such a thing as "inflation" out there?), you can argue that some assumptions are better than others. But this does not deny the fact that there are potentially many ways to measure the same variable, and different institutions/individual might prefer one over the other.

Focusing in the case of GDP for Iran in particular, there is the "original" GDP (as coming from official or primary sources) in local currency (notice there are several definitions of GDP too, and in the case of Iran, one with and without oil production). For international comparison purposes, this metric needs to be transformed into e.g. current US\$or PPP US\$, process which involves some assumptions (e.g. which exchange rate to use?). Again, different institutions might use different metrics and prefer different assumptions. As said below, for data coming from countries using the Islamic calendar, there are further complications when translating to the Gregorian calendar.

Let us dig deeper in the case of Iran's GDP. First, you need to start with the official data (or primary source), available in the website of the Statistical Center of Iran. If you select the file for GDP at current prices (this file), you get the quarterly data for GDP in current Iranian currency. Notice however, that this data corresponds to the Islamic year, which is different from the Gregorian calendar used in most of the world (and used in the websites you mention). The former starts around the 21st of March of the latter. This is not exactly the end of the first quarter of the Gregorian calendar. Since quarterly data is the minimum frequency of the Iranian GDP, there is no exact way to convert quarterly Islamic-based GDP into annual Gregorian-based GDP. This tells you immediately that the translation exercise is by definition an approximation.

One reasonable approximation is to use the last quarter of an Iranian year and the first three quarters of the next Iranian year to produce a metric of the Gregorian-based year GDP. For 2015, this is:

Quarter     GDP
-----------------
1393_4    439,687
1394_1    432,264
1394_2    450,158
1394_3    433,799


Which add up to 1,755,908 billion Rials (this includes Oil production).

To translate this into \$USD, which is the metric of the World Bank data, you would need to assume something about the exchange rate. Do you use the average over the quarter? Over the year?

For example, the same official statistics website above offer a measure of GDP in USD for the Gregorian calendar (see here):

Unfortunately, they do not mention which exchange rate conversion was used.

Meanwhile, the World Bank offers a slightly different one:

The notes for the latter states:

Dollar figures for GDP are converted from domestic currencies using single year official exchange rates.

Did they use a different conversion rate? It's very likely.

Regarding the IMF data, which is reported in the Wikipedia article (particularly in the graph here), this is measured in PPP terms (and the data is highly consistent with World Bank data of GDP in PPP; see here). PPP adjusts GDP figures not only by the exchange rate but also by a measure of the relative cost of living of Iran with respect to the US. This inflates the GDP of Iran (see PPP rates here).

There are plenty of questions in this site about PPP. For example, see here and here.

• Super interesting answer (and question too!)! Sep 29 '17 at 14:12