Skip to content

“Fiji First Party, Statistical Lies, FBS and World Bank” (ed. in FT 15/5/2022)

15/05/2022

The Fiji First Party (FFP) has on its Facebook page this graph which claims, alongside a beaming PM Bainimarama:  “World Bank confirms Fiji Poverty Rate down to 24.1%”.

            To the average person, the numbers on the graph and the continuous downward sloping red line appear to show that Fiji’s Poverty Rate (as measured by the FBS) declined from 35% in 2002-03 to 31% in 2008-09, to 28% in 2013-14, and declined even further to 24% for 2019-20.  This 24% is supposedly a correction made by World Bank from the earlier estimate of 29% (another strange story).

            But it is far more important to note that the Graph is totally misleading since the FBS/WB 2019-20 estimates are using a completely different methodology from their own earlier three estimates.

            So where is the “statistical lie”, FFP supporters will demand to know? Let me deal with that first as it is a serious allegation.

The statistical lie

The graph implies that the Rate of Poverty of 24% for 2019-20 can be compared to the higher Rates of Poverty for the previous three surveys (35%, 31% and 28%).  But these four numbers cannot and must not be placed on the same line. Don’t take my word for it.

            The Fiji Bureau of Statistics and World Bank themselves emphasized and warned in their correcting Joint Press Release a few weeks ago, in a section subtitled “Comparability of poverty and consumption in 2019-20 to previous years”.  “Fiji’s 2019-20 poverty estimates cannot be compared directly to 2013-14 or earlier estimates due to changes in methodology”.

            There were two reasons for this statement.  The first reason was that “poverty estimates in 2019-20 are based on consumption, while those from previous years (2002-03, 2008-09 and 2013-14) have been based on income” (I will come back to this later).The second more important reason was that “the national poverty line was also recalculated [reduced] in 2019-20 using consumption and the CBN methodology”.   The FBS/WB Press Release was absolutely clear that: “Due to these changes, the 2019-20 estimates of consumption and poverty are not directly comparable to those from previous years.”

 Given that the 24% was plucked out of this latest FBS Release, it is shocking that the FFP (and the all powerful Minister for Economy) deliberately ignored the above very important warning highlighted by FBS and WB, and presents a graph to the public with all  four Poverty Rates on the one line, even though they are not comparable. 

The FFP is deliberately misleading (through a statistical lie) the public as far as that last number (24%) is concerned. Moreover, I show below using FBS’ own statistics, that the real comparable rate is probably twice that.

Understand the Poverty Rate analysis

It is not hard for the ordinary public to understand how the FBS calculates the Poverty Rate or the percentage of Fiji’s population that is below some “Basic Needs Poverty Line standard”.

Essentially, the FBS first ranks all the households in Fiji by their Income per capita (or Adult Equivalent which counts children <14  years old as half an adult) as was done for 2002-03, 2008-09, 2013-14.

Or the households can be ranked by Consumption Expenditure per capita as was done by the FBS/World Bank for the 2019-20 HIES.

The FBS then estimates a “Basic Needs Poverty Line pc” which essentially is how much (a decent minimum) a family would need per capita, to satisfy its “basic needs” of Food and Non-Food items (housing, water, education, health etc). Of course this should be in reference to Fiji needs and income levels- it cannot be determined by the absolute poverty in some African or Asian countries.

            For the 2002-03 HIES, the Basic Needs Poverty Line was estimated very sensibly using a locally suitable nutritious Food Basket (designed by the Fiji Food and Nutrition Centre) and the Non-Food value arbitrarily chosen as the average Non-Food Expenditure for the bottom third decile or 25% percentile household, considered then to be just at the margins of poverty in Fiji then).

            For the 2008-09 HIES and 2013-14 HIES, this Basic Needs Poverty line was adjusted upwards by the change in the Consumer Prices Index. For the 2013-14 HIES, this was adjusted further by the GS Mr Waqavovono, again by the CPI change.

There was therefore inbuilt consistency and comparability between the Poverty Rates calculated for 2002-03 and 2008-09 (by me and WB), and for 2013-14 (by FBS and Mr Waqavonovono).

            But it was all changed by the WB for the 2019-20 HIES, I suspect with the FBS going along.

Why the WB Changes for 2019-20?

Any sensible local person who wants to know how poverty is changing in a country would advise its Government Statisticians to use the same methodology for all the HIES so that the results can be correctly compared over time.

            But the WB, and I suspect purely because of its global practice of comparing Rates of Poverty of one country to another, decided to totally change their methodology to their international practice, with no apparent concern for maintaining continuity of analysis within Fiji.

First, instead of using income as the criterion of poverty, they decided to use “consumption expenditure”.   I suspect that did this because in many of the poorest developing countries, expenditure is often more accurately monitored than income (which households can easily hide from Bureau of Statistics staff).

            But the World Bank manuals themselves (for example the 2009 WB book by Haughton and Khandker) strongly advise that as in developed countries, income should be used where possible and consumption expenditure as a last resort.  I have personally found (and I am sure the WB will agree) that the FBS household surveys are extremely accurate and consistent for both income and expenditure.

            You may ask yourself: would you judge a person’s state of poverty or wealth by his income or by what he chooses to consume or spend?  You know as well as I do that there are many rich people in Fiji with high incomes, whose consumption expenditure is incredibly low (even below the poverty line, especially if they are vegetarians) and they choose to save more because they are frugal. Under the older FBS methodology, they would not be classified as “poor”. Under the WB methodology used for the 2019-20 HIES, they could well be below the poverty line.

WB also reduced Basic Needs Poverty Line

The WB also decided to use a lower BNPL poverty line for 2019-20, which even Fiji economists would struggle to understand.  Perhaps good Fiji economists like Dr Neelesh Gounder might like to get an excellent academic article out of the sheer idiocy of WB using computations of Basic Needs Poverty Lines which may be suitable for basket case countries in Africa or Asia, but not for Middle Income countries like Fiji.

            But first note that the poverty line used by FBS/Waqavonovono for the 2013-14 HIES was $52.31 pc pw.  If you adjusted that by the change in the CPI (roughly 13% between 2013-14 and 2019-20), then Basic Needs Poverty Line for 2019-20 would have risen to $59.11 pc pw.

            But for 2019-20, the WB (and the FBS) used the value of $49.91 which is even lower than that for the 2013-14 HIES.

            You do not need to be an Einstein to guess that if you reduce the poverty line from $59.11 pc pw to $49.91 pc pw, you will ipso facto tend to reduce the Poverty Rate.

Using the old BNPL

The public might want to know what would have been the Poverty Rate had the FBS used the old FBS 2013-14 BNPL adjusted by the CPI to the 2019-20 value of $59.11 (using FBSs’ own data).  Luckily the FBS HIES Report for the 2019-20 HIES allows us to do that, even if very roughly.

            Row 1 (Table 1) gives the WB/FBS “corrected” Poverty Rate of 24.1% using the BNPL of $41.91 pc.

Table 1 BNPLPoverty Rate
1      BNPL (FBS)$41.9124.1%
2      BNPL + 20% (FBS)$50.2939.1%
3     CPI adj. BNPL (Mine)$59.1150.7%
4     BNPL + 50% (FBS)$62.8758.1%

But the FBS 2019-20 HIES Report also states (on page 30) that if the BNPL were to be increased by 20% (to $50.29) this would throw another 15% of the population into poverty (i.e. the Poverty Rate would be (24.1% + 15%) = 39.1% (Row 2 of the Table).

            It also states on the same page that if you increased the BNPL by 50% to $62.87 pc pw, that would throw another 34% into poverty, i.e. (24.1% + 34%) = 58.1% (Row 4).

            So in Row 3, I use the 2013-14 BNPL adjusted by the CPI change of (13% roughly) to $59.11 to estimate very roughly the resultant poverty rate for 2019-20 using FBS consumption data:  50.7% or just over a half of the population, and double the official figure.

            This is more the kind of number (51%) that civil society workers for poverty alleviation like Sashi Kiran (FRIEND), Nalini Singh (FWRM) and Shamima Ali (FWCC) would say reflects the reality on the ground, not the low 24% that FBS and WB have dished out and used erroneously (and deliberately) by FFP on its Facebook page.

            More to the point, the FFP Graph for the lowest part would be shooting upwards somewhere towards 50%,  not downwards as deceptively shown.  I am sure that the FBS statisticians can give you far more accurate numbers than my very rough estimates here, using either consumption or income as the criterion.

            But if you wanted to correct the FFP Graph right at  the top of the article, the FBS would need to work out the Poverty Rate using income as the criterion, not consumption expenditure.   It would have been a very simple exercise which the FBS staff have done previously.

            Why did the FBS (and the WB) not give income based Poverty Rates for 2019-20, as well as their consumption based expenditure estimates?  I suspect that the Political Masters of FBS would not have liked to see the bad result of the Poverty Rate rising severely between 2013-14 and 2019-20.

FBS, WB and  Poverty Analysis

It would be useful for the public might want to know how the different Poverty Rates have been derived by the FBS since 2002-03.

            The analyses of the 2002-03 and 2008-09 HIES were done by me together with the FBS staff (Toga Raikoti and Epeli Waqavonovo). The 2013-14 HIES was analysed by the FBS (and GS Epeli Waqavonovono)

            Most importantly, all these three analyses used Income per capita (Adult Equivalent) of households as the criterion of poverty and the same Basic Needs Poverty Lines, adjusted by the CPI.  The FBS (and the WB) could very easily have done that for the 2019-20 HIES in addition to what they did.

            After all, even the WB manuals clearly state that one of the objectives of quantitative poverty measurement is “to monitor the state of poverty over time, so as to assess the degree of success of past policies and effectiveness of institutions charged with the responsibility for reducing poverty”.

Even the WB itself has previously analysed the 2002-03 and 2008-09 HIES using the income criteria and similar Poverty Line values and arrived at very similar results as FBS and Narsey.  The WB, with their army of massively overpaid experts, could have easily done that for 2019-20, but they did not. But who cares about them.

            More to the point, why did the FBS not do so?

The fear among FBS Staff

The public might want to remember the sacking of the GS Kemueli Naiqama last year, because the FBS released Poverty Rates by ethnicity (with much higher Poverty Rates for indigenous Fijians) which the Bainimarama Government did not want to have publicised.

            The public might want to note that in its more recent Press Release which corrects the earlier Poverty Rate of 29% to 24%, the FBS/WB also give the corrected Poverty Rates by Division and Rural/Urban, but no corrected Poverty Rates by ethnicity.

The FBS staff know what is good for them and their families when it comes to putting food on the table (and quite a bit more food than the miserable amount that WB estimates as the ideal Food Poverty Line for Fiji.

It is utterly sad that the FBS has become totally dependent on an international organization like WB which clearly shows little concern for continuity and local sustainability of poverty analysis in Fiji.  They have never seen fit to involve local independent economists in their dubious analysis.

            The public might want to watch a video on Youtube of a book launching at USP (chaired by FFP MP Dr Mahendra Reddy) by FBS Government Statistician (the late Mr Timoci Bainimarama) who expressed  his great satisfaction at FBS staff co-operating with USP economists in producing policy oriented reports by analysing household survey data, all gladly financed by AusAID/DFAT. There will soon be another video of the late Mr Tim Bainimarama in 2009 launching at USP a monograph on the analysis of poverty.

            I sincerely hope that the FBS can once more begin to collaborate with capable USP quantitative economists (like Dr Neelesh Gounder) so that there is local capacity and continuity in producing solid policy reports for the Fiji public and the Government (whoever they are), without any unhealthy dependence on WB experts with their own international agenda.

            Hopefully that collaboration will be without any political interference by Government Ministers (and their political parties) who want only want favourable statistical results from FBS, and will even happily use statistical lies to suit their own political agenda, as the Fiji First Party Graph so clearly illustrates.

Comments are closed.

%d bloggers like this: