South Sudan Poverty - Githubissues

PardeeCenterIFs commented 10 months ago

On November 14th Taylor wrote:

I was poking around in preparation for a call I had with USAID Ethiopia this morning and noticed that poverty in South Sudan is looking pretty peculiar. It falls and then comes back again very quickly before 2025, spends two decades falling gradually, and then around 2045 gets eliminated in the span of a few years:

This happens across poverty thresholds and I don’t see any changes that would explain this rapid elimination in Gini or GDP.

PardeeCenterIFs commented 10 months ago

José replied:

It’s definitely coming from Consumption, specifically Manufactures, but I still haven’t figured what triggers the sharp rise, everything else looks more stable.

PardeeCenterIFs commented 10 months ago

Barry did the following exploration on Nov 15th:

PardeeCenterIFs commented 10 months ago

José replied:

It seems the change happened in our move to 2017 US$, there was an update of data in version 8.02, but that didn’t trigger the change, it only happened in version 8.03 with the change to 2017 US$, and in reality what changed was GDP, not energy exports, GDP dropped from 6.7 to 1.8 with the new constant dollars, and that’s why are % of GDP is so big now.

PardeeCenterIFs commented 10 months ago

Barry addressed the following to Yutang on Nov 16th:

I think we probably have a GDP value and series for South Sudan wrong in recent model versions. According to José, in the transition to $2017 dollars the value in IFs in base year dropped from 6.7 to 1.8 billion dollars. That has caused the export of oil to balloon to more than 100% of GDP, setting up crazy behavior on build-up of foreign assets (not debt, assets) leading to a collapse of poverty levels that Taylor found in the 2040s when household consumption explodes and other problems in subsequent years. Note the IMF data source and economic profile below, which shows current price GDP much more like the value we used to have. The difference between current dollar prices year to year and between $2015 and $2017 or even $2019 cannot be more than a factor of 3, so it looks to me like something got badly messed up. Can you please take a look at this issue?

PardeeCenterIFs commented 10 months ago

Barry added:

Yutang, I pasting on top of my previous message estimates on South Sudan export and import 2021 current dollar values from the CIA factbook (2019-2021). They are pretty balanced and well below our early $6.7 billion value of GDP but way above the $1.8 billion. I suspect we have some data/preprocessor issues around the country’s energy/oil trade also (and much of export value goes into private pockets not official assets), but this reinforces the high improbability of our current GDP figures.

PardeeCenterIFs commented 10 months ago

Yutang replied:

I took a look at the data and I think this is what happened. The exchange rate (South Sudan’s LCU per US$) jumped from 2.99 in 2011 to 113.65 in 2017. This number is essential to our constant price GDP conversion. The dramatic change happened rather back in 2016. The exchange rate was 3.6 in 2015 and shot up to 46.73 in 2016.

This is also reflected in IMF’s World Economic Outlook. South Sudan’s GDP, in current $US, dropped from 15.33 billion to 2.9 billion in 2015-2016. Under the current USD price, its GDP went down even further to 1.8 billion in 2017, which is also the base value for South Sudan’s GDP in 2017 as measured in constant 2017 USD. The GDP growth rates in IFs match with what is shown in IMF’s April release (similar to its latest Oct release).

Also, I thought trade value is expressed as % of GDP in IFs. The values you are showing from CIA’s factbook is expressed in current USD, which is very different than constant price for a country with such big change in its exchange rate and GDP deflator. I attached its GDP deflator below in the table.

2011 | 2012 | 2013 | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- | -- 165.396 | 212.406 | 221.403 | 221.396 | 271.334 | 778.124 | 1,243.85 | 2,740.49 | 3,929.15 | 5,838.23 | 11,461.72 | 24,700.07

Thus, unfortunately, I don’t think there is anything I can do on the data end as long as we are sticking with the way GDP is converted to a constant USD price. There will be a data delivery near the end of this month for IMF’s recent October release but it won’t fix this issue.

PardeeCenterIFs commented 10 months ago

Barry replied on Nov 17th:

Yutang, thank you for the quick and useful response on the forecasts for South Sudan!

All, I attach a document also relevant to that exploration, including some historical and future analysis from the Financial Times group that shows the collapse of GDP in current dollars after 2015 (graphic also below). Our own forecasts through 2022 do not show even as much bounce back as the graphic, but apparently the IMF has not been optimistic. Still with all of the oil and gas potential, some kind of big bounce up in later years (supporting a rapid drop in poverty) would seem at least possible. So our crazy pattern of forecasts still seems all but impossible, but it is hard to know what does seem possible much less likely. We could mess with the model to limit the huge growth of consumption in the 2040s, but ….

Perhaps worth touching on in the Monday model pod. At this point my inclination would be to simply say that there is little or no way to make forecasts for the country that have real credibility.

PardeeCenterIFs commented 10 months ago

The attached document:

https://www.fdiintelligence.com/content/feature/the-promise-of-oil-and-gas-in-south-sudan-81521

mtirfan66 commented 6 months ago

Extreme poverty South Sudan: IFs 8.17: poverty declines but starts picking up

This is coming from consumption picking early and then declining,

The question is why, consumption picks so fast, if it increased gradually, it might not have peaked to an unsustainable value, before starting decline. So, I looked at savings and disposable income. Disposable income is initialized at a much lower value than what we had in the VB6 versions. Anyway, data might have changed. So, I looked at savings. Looks like we are initializing South Sudan and Syria at a very high savings rate. These are poor countries, so the savings rate needs to come down fast. Then there is this oscillation later in the horizon.

Consumption moves at the opposite phase of this savings oscillation, as expected.

So, where is the problem? Why is initial savings so high? Since it is high as a share of income, the problem has to be consumption initialization. Total household consumption is very low. Some of you might have explored this earlier. So, how do we know it is problematic. I have compared, all the CS with VADD for year 1, where all the consumption by sectors are added to obtain total consumption. So, if you compare VADD for services, you'll see it is much larger than CS for services. The services consumption is almost non-existent.

In the pre-processor, I see CCS, foe consumption by sector, which comes from private consumption, pcons by sector, which is computed from VADD, intersectoral flows, export import, government consumption, total private consumption data etc. It is likely that either service sector value added or the i-o matrix data or consumption data - one of this is messed up.

mtirfan66 commented 6 months ago

SS might be exporting all the services, and initializing domestic consumption at the minimum. Could be a x... grouped region issue on GTAP data. I see some historical service export data though, very large amount, could be from Sudan combined..

mtirfan66 commented 6 months ago

Ignore that "large amount" - that's total export. Value added by sector is 32% of GDP, export from the service sector is 3.8% of GDP. Most everything of the domestic use goes as input to other sectors, leaving only a tiny amount of services (half a percent of GDP, .01 billion dollar) for final consumption. Not sure if that's right.

PardeeCenterDU / IFs-Issues-Tracking

South Sudan Poverty #255