Monday, March 10, 2008

SWS and NSCB Hunger and Poverty Data Actually Show Overall Downward Trend


The National Statistics Coordination Board (NSCB) reported last week that Poverty worsened between 2003 and 2006, yet taken as whole, their own data for the period between 2000 and 2006 actually shows a long-term downward trend in poverty incidence!

2000 .275 .123
2003 .244 .102
2006 .269 .11

Writing for the Philippine Daily Innuendo, SOCIAL WEATHER STATIONS Chief Mahar Mangahas entitled his reaction to the NSCB announcement "Admitting that poverty can also rise" --but will he acknowledge that his own public opinion polling data shows a long term definite downward trend to poverty?

The column labeled Poverty shows the percentage of Filipinos who described themselves as poor in that year averaged over four quarterly Social Weather Station surveys each conducted on 1200 respondents from NCR, Luzon, Visayas and Mindanao. The column labeled Episodic Hunger shows the percentage of SWS survey respondents who said their household had experienced hunger from a lack of food at least once during the preceeding three month survey period. The column labeled Chronic Hunger shows the percentage of survey respondents reporting that their households always or very often suffer hunger for lack of food to eat.


In order to compare the statistics of SWS and NSCB on Hunger let me first say something about data collection methodology. The NSCB bases all its work on an official determination of the so-called minimum food threshold, which is the income required by a family of five to feed itself at a subsistence level. Families and individuals who do not earn this minimum income are considered to be below the SUBSISTENCE level and are therefore expected to be chronically hungry. It's data comes from the Family Income and Expenditure Study, which takes three years to complete one cycle.

The SWS on the other hand conducts public opinion polls quarterly and asks 1200 respondents whether or not their households have experienced hunger at least once in the quarter because they had no food to eat. The percentage who experience hunger like this all the time are considered to be in SEVERE HUNGER, while the rest are in MODERATE HUNGER. SWS reports the sum of these two figures as TOTAL HUNGER.

But moderate and severe hunger, which I call "episodic" and "chronic" respectively, behave very differently in the data of SWS:

CLICK TO ZOOMOne thing for sure, the SWS data on self-reported SEVERE HUNGER agrees with that of the NSCB's Family Income and Expenditure Study data in the downward trend.

The most questionable of the SWS data is its MODERATE HUNGER data, which in the past I've analyzed and shown contains SEASONALITIES and PERIODICITIES that indicate that experiencing hunger episodically ("at least once in the last three months") is not a good measure of hunger as such, even if it's great for selling HEADLINES.


A central theme raised by Mahar Mangahas is that even though the economy is indubitably expanding and developing as shown by GNP and GDP figures, it is claimed that the benefits are not really being felt by "the poor and hungry." Anecdotally there are of course too many poor and hungry people around here, but the "no trickle down" criticism is a repressed memory of Ronald Reagan brought back to life by reformers with agendas. It flies in the face of SWS's own severe hunger data, as now confirmed, at least in its downward trend, by the NSCB. Moreover, the real cause of economic improvement--the OFW repatriations at over one billion dollars per month--is a direct injection of funds to families at the lowest levels of society. That's DIRECT INJECTION as far as I can tell, not trickle down.


sparks said...

In a mini-survey we did last year, 37% of our respondents said they remitted money for "daily subsistence" of their family back home. Not investments, not for small business ventures. Subsistence.

Your finding fits the overall picture that our migrant labour is keeping this country from imploding.

DJB Rizalist said...

The NSCB sets the subsistence threshold for each of the study periods, so it increases with inflation. They use a coefficient of variation (stdev/mean) to get a dimensionless measure of the "spread" of each implied distribution.

You should compare the "subsistence" level assumed by your survey participants and compare that value with the NSCB's. It's likely to be much higher and what OFW's call subsistence is the higher NSCB threshold for "poverty".

Thanks for the info Sparks!

blackshama said...

The data never lies. It is the interpreters of the data who do. The PhDs fall into this trap too. I think DJB you have fallen into the trap.

The practice of good science relies on

1) empiricism
2) objectivity
3) repeatability
4) predictability
5) falsifiability

Given these principles, let me look at the issue at hand

DJB you seem to be downplaying episodic hunger,nonetheless it is more objective and empirical to use the term " chronic food insecurity". Food insecurity represents the whole phenomenon of which hunger is just a part. Several indices can be designed to describe this phenomenon. SWS and the social scientists here haven't really talked with environmental practitioners on this issue.

You say episodic hunger is not a good measure of hunger per se. That is if you apply a linear model to predict total hunger. Recall that the equation that SWS made ( episodic hunger+severe hunger = total hunger) is a simplification. The two terms in the equation have different temporal characteristics and may necessitate different analytical techniques.

As an environment practitioner I am more interested in episodic hunger since this has a lot of possible autocorrelative factors with environment and even corruption ,provided we come up with measurable indices.

Anyone who has taken a course in time series analysis can get my point. Since you have a physical sciences background then I think this isn't new to you.

There is also a danger in interpreting trendlines. If two point estimates are compared (2003 and 2006), the conclusion that poverty has increased is indeed plausible. But a trendline just averages the estimates and may be misleading. Also the temporal component of the model is quite short, trendlines may allow for erroneous conclusions.

But since these statistics deal with people, the moral dimension for interpretation comes in. In moral terms there really isn't much difference between episodic food insecurity and total food insecurity.

We really can't blame the Philippine Daily Innuendo. Their journalists and editors are so science illiterate and one thing you and I can do is to teach them how to do science and to think scientifically.

DJB Rizalist said...

I was always mroe interested in the episodic hunger data because it jumps around so much. In various treatments of the hunger and poverty surveys I 've analysed that data to death I think, though I can't pretend to understand it entirely.

Of course, it is entirely possible that at least some of increase in episodic hunger must be due to the decrease in chronic hunger. But the numbers can't account for increase.

What do you make of the "inflection point" that seems to have occurred in the episodic hunger plot in year 4?

cvj said...

It could be that the chronic hunger figure is going down because people are already dying of hunger. You cannot stay chronically hungry for several years without it affecting you somehow.

DJB Rizalist said...

I'm sure the mortality and morbidity among the chronically hungry is higher than others. Of course the other possibility is that more of the chronically hungry are joining the moderately hungry, which is why that number seems to be growing. both processes probably occur, eh?

cvj said...

I agree, those scenarios are also possible.

DJB Rizalist said...

I'm just amazed at Mahar Mangahas. This data I am presenting comes directly from SWS, without any modifications == just a calculation of the power law trendline associated with each set.

Why is is necessary to distort the interpretation?

Whatever happened to the Truth?

Anonymous said...

What truth? More like whose truth!

Sabi ni Itay, Oh mag-ingat kang magmaneho at baka masama ka sa statistics!

Meaning stats means crap to those whose suffer waht its numbers support or not support.

Anonymous said...

Another problem is that the poverty threshold may be adjusted to make things look better. For example, sometime ago one government study set a PhP6,500/mo poverty threshold for a family of six in Metro Manila. I think that amount is too small, given the definition that one is above the poverty threshold if he lives in a shelter that follows local building codes, has more than enough money to pay not only for private health care (because public health care is lacking) but even expensive treatment for heart condition, kidney failure, etc., 2,000 calories of food a day for everyone and meals that have nutrition that follow a local RDA, sufficient community resources such as an adequate ratio of health care workers, police officers, teachers, and even librarians to people, access to public schooling with an average class size of at most 30 (not 60 or 120), more than enough money for school supplies, including updated textbooks (not books riddled with errors, reprints, or outdated materials), schools with electricity, blackboards, potable water, and even principals, and many, many more requirements.

We're not talking about a wealthy or even a middle class society, just access to basic health care, education, food requirements, security, and local employment that doesn't involve having to leave the country.

Given all that, I'd say the official poverty rate is highly understated, and the same goes for any good news given by the government, including employment (I even read one time that employment was defined as having the intention to work for at least one hour a week).

DJB Rizalist said...

Even assuming that the government cheats its data, how do you explain the undeniable fact that the SWS data itself confirms the downward trend?

In fact, the downward trend has always been there in the SWS data, yet they never gave you that impression did they?

So I guess they are both cheating eh?

Anonymous said...

To DJB, if you adjust standards then you can make trends go up or down. For example, if the poverty threshold is PhP6,500 for a family of six, and one thinks that that amount will be able to buy all of the requirements that I mentioned, then a downward trend will appear.

It's not an issue of cheating on both sides but understanding what one can buy with a certain amount of money.

Here's another example: we can argue that poverty trends worldwide are also moving downward because more people can now make at least a dollar a day. The issue isn't whether or not that's true but whether or not one can receive standard housing, public education with an average class size of 25 (not 60), follow the RDA, and so on with only a dollar a day. If one will need much more than that, and assume that the poverty threshold is not a dollar a day but six or seven dollars a day, then the trends change dramatically. If one assumes a middle class existence, i.e., a vehicle, one's own home, a secure job that can pay for that vehicle, one's home, education for the children, health care, a pension and insurance to cover advanced health care, and more, then poverty rates are much higher.

DJB Rizalist said...

i don't think trends, which are based on changes between measurements at different points in time, are affected by absolute levels or measurement units.

Anonymous said...

Of course they are. Read my last paragraph in my previous post.

DJB Rizalist said...

Maybe you'll understand the point if you ride a roller coaster down.

mbw said...


This post is very informative because it contains inputs/analysis that you've made from a mathemathical viewpoint and those made by the other commenters, which, for a sublime ignoramus like me, are helpful although I'm still grappling with the ultimate question: what measures are being drawn up to decrease the poverty level in the Philippines?

By poverty level, I refer to what Anonymous say (who is MB btw, Manuel Buencamino?) as income fixed at PhP6,500/mo ... for a family of six in Metro Manila.

The other question that is paramount in my mind is the following: Will the current "trend" involving staple food products that we are now experiencing in the West have an impact on the poverty level developping countries like the Philippines (you may not like trending but there is an upward trend for prices of staple products in the West)?

For instance, a BBC report (can't find the link anymore but I have kept a printed hard copy) shows the following statistics on spiralling food prices by Dec 2007(from January):

Dairy: 290%
Oils and fats: 250%
Meat: 150%
Cereals: 220%
Sugar: prices started off in January at 140% (from 100% in year 2000) and fluctuated to stabilise at 130%

Prices rises in a single year: March 2007-March 2008

Corn: up 31%
Soya: up 87%
Wheat: up 130%

The trigger: Rising oil prices and fears over climate change have been seen a massive rise in the use of maize to make bio-fuels, pushing food prices

DJB Rizalist said...

These are terrible times in the world and it will likely get worse before it gets better. But one can see what one wants to see in the news of doom and gloom in the economy.

But this has nothing to do with data analysis. I don't necessarily accept the SWS and NSCB dta are correct, and at any level of comparison of the absolute levels of poverty, their measurements don't agree, precisely because of the thing MB thinks is fundamental: the poverty level chosen.

But as an old data collector and analyst, I know one thing, TRENDS are very hard to cheat, and when two completely different measurement methods agree that the trend is downward, that is very hard to arrange through cheating or prejudice or what not.

The other bit of experience I have is hearing precisely the kind of argumentation of MB from those who didn't like the results of some study or data we collected and think they can argue with the data.

WE scientists and mathematicians, if we are honest, argue with the data even more, check and double check the figures and our analysis and make sure our reasoning and procedures are rigorous.

As a consequence we understand these objections like MB's. And know why they are WRONG objections borne of inexperience and only try to force the conclusion they want.

As I said here, the data never really lies, analyzed properly, only the headline writers do because as you know statistics make for powerful propaganda.

Others may lie with their statistics, but anyone who comes here should know they better know what they are talking about when it comes to this stuff, coz I won't let math or logic or science be abused around here unchallenged, undemolished!

What steps are being taken to reduce poverty, you ask? I suppose you mean by the govt, I suppose. I really don't care because I don't believe government is responsible at all.

It's a billion dollars a month directly injected into the economy by OFWs that's doing it.

BTW, that is the other bit of BS going around: that the economic growth isn't "trickling down." (Remnember "voodoo economics"?)

It's an old anti-reagan criticism. But the economic growth under Reagan was qualitativelty very different than our OFW-fuelled growth, which as I've pointed out is directly injected into the poorest families of the OFWs. monthly, weekly, daily!

It's scary, because it is this kind of intellectual dishonesty--a refusal to acknowledge, ahem, inconvenient truths that seem to contradict political positions or ideology, while they are out of power, that creates the kind of monsters we get in public office once they do have power.

For example, the only reason even Mahar and MB won't acknowledge or highlight the definite downward trend even in his data on poverty is that they think it would make GMA look good. That's crazy. It would give OFWs their due in statistical praise and glory, not deny them of it so MB can bash GMA over the head with his "analysis."

philippine poverty said...

I hope the government would wake up,and do something to overcome this problem.

What is the use of good economic development (based on govt. statatistics, which is obviously "?")if there are millions of filipinos suffering from poverty?