Define countries under study

[jm-rivera]

Which countries should the index consider?

We are framing this as the places where a combination of factors make the food security situation particularly challenging. Based on indicator selection, these countries will not be High Income Countries.

Outside HIC countries, should we narrow things in another way? Key questions to consider (for countries outside HIC):

• For how many countries do we have data?
• Is there any pattern to missing data (income, geography, etc)?
• Is the age of data broadly comparable across groupings?

This issue tracks answers to the above and discussion/decision if needed.

[jm-rivera]

31 deals with a way to store different groups.

[lpicci96]

Dimension 1

1. Headline inflation

Missing - Solomon islands

2. Insufficient Food Consumption

missing overall - 52% continents - 'America': 67%, 'Asia': 48%., 'Africa': 13%, 'Europe': 95%, 'Oceania': 57% income levels - 'High income': 100%, 'Low income': 11%, 'Lower middle income': 6%, 'Upper middle income': 67%

missing low income - Eritrea, North Korea, Syria missing low middle - Belize, Comoros, Papua New Guinea

[jm-rivera]

2. Insufficient Food Consumption

missing overall - 52% continents - 'America': 67%, 'Asia': 48%., 'Africa': 13%, 'Europe': 95%, 'Oceania': 57% income levels - 'High income': 100%, 'Low income': 11%, 'Lower middle income': 6%, 'Upper middle income': 67%

missing low income - Eritrea, North Korea, Syria missing low middle - Belize, Comoros, Papua New Guinea

@amy-dodd Insufficient food is quite key for this. WFP is not collecting data in many UMICS. We could assume that those just mean zeros.... but that will introduce a bias as zero is quite an extreme value (when compared with the range in the data for countries with data). Such a decision wouldn't really be justified by anything but a feeling that no data means no hunger.

Let's see for the other dimensions... Maybe we don't exclude all UMICs, but instead do establish a data completeness threshold (like a country must have data for at least 5/6 indicators to be included).

[lpicci96]

Dimension 2

(values are % missing)

Economist index

overall missing - 35% continents - 'Asia': 32%, 'Africa': 41% 'Europe': 19%, 'America': 28%, 'Oceania': 83% income level - 'Low income': 33%, 'Lower middle income': 39%, 'Upper middle income': 50%, 'High income': 14%

Wasting

overall missing - 10% continent - 'Asia': 4%, 'Africa': 0%, 'Europe': 44%, 'America': 3%, 'Oceania': 8% income level - 'Low income': 0%, 'Lower middle income': 0%, 'Upper middle income': 2%, 'High income': 40%

[lpicci96]

Dimension 3

(values are % missing)

Fiscal Reserves minus gold

overall - 4% missing - 'Benin', 'Burkina Faso', "Cote d'Ivoire", 'Guinea-Bissau', 'Mali', 'Niger', 'Senegal', 'Togo'

Service Spending Ratio

overall - 37% continent - 'America': 38 %, 'Asia': 37%, 'Africa': 9%, 'Europe': 73%, 'Oceania': 33% income level - 'High income': 100%, 'Low income': 4%, 'Lower middle income': 6%, 'Upper middle income': 15%

low income - South Sudan lower middle income - Micronesia, Fed. Sts.', 'Pakistan', 'Palestine

[jm-rivera]

Extremely helpful, thank you @lpicci96!

I think we can definitely rule out HICS.

It's not looking promising for UMICsCould we get an overview of the UMICs that have data for at least 5/6 indicators?

[lpicci96]

UMICs % of total variables that are null

iso_code	country	missing (% of all variables)
DOM	Dominican Republic	0
JOR	Jordan	0
GTM	Guatemala	0
ECU	Ecuador	0
TUR	Turkey	0
BWA	Botswana	0
PER	Peru	0
BRA	Brazil	16.66667
KAZ	Kazakhstan	16.66667
ARG	Argentina	16.66667
PRY	Paraguay	16.66667
ROU	Romania	16.66667
MDA	Moldova	16.66667
ARM	Armenia	16.66667
SRB	Serbia	16.66667
THA	Thailand	16.66667
PAN	Panama	16.66667
MEX	Mexico	16.66667
AZE	Azerbaijan	16.66667
LBN	Lebanon	16.66667
BGR	Bulgaria	16.66667
CRI	Costa Rica	16.66667
COL	Colombia	16.66667
CHN	China	16.66667
ZAF	South Africa	16.66667
BLR	Belarus	16.66667
FJI	Fiji	16.66667
ALB	Albania	33.33333
TKM	Turkmenistan	33.33333
NAM	Namibia	33.33333
RUS	Russia	33.33333
MYS	Malaysia	33.33333
LBY	Libya	33.33333
MNE	Montenegro	33.33333
BIH	Bosnia and Herzegovina	33.33333
CUB	Cuba	33.33333
MUS	Mauritius	33.33333
GEO	Georgia	33.33333
GUY	Guyana	33.33333
GAB	Gabon	33.33333
JAM	Jamaica	33.33333
MDV	Maldives	33.33333
MKD	Macedonia	33.33333
IRQ	Iraq	33.33333
GNQ	Equatorial Guinea	50
SUR	Suriname	50
LCA	St. Lucia	50
TON	Tonga	50
VCT	St. Vincent and the Grenadines	66.66667
DMA	Dominica	66.66667
GRD	Grenada	66.66667
MHL	Marshall Islands	83.33333
TUV	Tuvalu	83.33333
XKX	Kosovo	83.33333

[jm-rivera]

Thanks @lpicci96, I say we keep Fiji and above. What do you and @amy-dodd think?

[amy-dodd]

Thanks @jm-rivera - agree that seems the most sensible approach. I don't think we're losing any 'important' countries by eliminating the others?

[amy-dodd]

Eritrea, North Korea, Syria Also on indicator 2 insufficient food consumption in particular, Syria jumps out as a potentially problematic example of missing countries. Have we come to a conclusion on how we're addressing data gaps like that?

[jm-rivera]

All countries, minus:

• High income countries
• Upper middle income countries
• ["KIR", "FSM", "PRK", "WSM"] for missing more than 2 indicators.

A total of 78 countries