Naive projection of poverty

The World Bank (WB) has a new report, “Philippine Economic Update: Making Growth Work for the Poor,” that warrants comment on account of its simplistic, trickle-down, and wishful—in a word, naive—way of projecting poverty.

A section, “Poverty projections through 2016,” says: “A straightforward method to project poverty incidence is to estimate the growth elasticity of poverty. The growth elasticity of poverty refers to the percent change in poverty incidence for every percent change in GDP per capita. The growth elasticity of poverty between 2006 and 2012 is estimated at -0.24, while it is -2.02 between 2012 and 2013. The latter is not unrealistic as the country was able to achieve an elasticity of -2.17 between 1985 and 2000.“ (Report No. 93530-PH, World Bank, January 2015, page 11.)

In brief: Though GDP (gross domestic product) was growing all along, poverty was flat in 2006-2012 (actually, it was 2000-2006), unlike in 1985-2000, when poverty was responsive to trickle-down, according to the triennial official figures. But recently, between 2012-S1 (S1 means first semester) and 2013-S1, poverty fell. Couldn’t the situation of 1985-2000 be renewed in 2014-2016?

I think the WB report makes far too much of 2012-S1 and 2013-S1, the last official poverty reference points. On the other hand, it is oblivious to the lengthy SWS data on poverty, which was semestral in 1986-91, and quarterly in 1992-2014.

The WB staff members are well-trained, and aware of the content and quality of the SWS poverty surveys. Their failure to consider such plentiful data leads me to believe that the WB is prejudiced against subjective indicators—in effect, against listening to the poor—as a matter of institutional policy.

The decline in official poverty between 2012-S1 and 2013-S1, reported on April 29, 2014, by the Philippine Statistical Authority, was no surprise. I wrote immediately (Opinion, 5/3/2014): “[T]he SWS surveys anticipated the official finding that poverty declined from the first semester of 2012 to the first semester of 2013. In addition, they show that the decline: (a) had occurred as early as the second semester of 2012; (b) was maintained into the first semester of 2013; and (c) was partially reversed in the second semester of 2013.”

When Pope Francis says that economic growth has not helped the poor, he is empirically correct. Thus it is naive to project poverty by means of crude elasticities to economic growth.

The most thorough Filipino econometrician on poverty is Dr. Dennis Mapa, dean of statistics and professor of economics at the University of the Philippines. He has made an econometric model with general inflation as the prime factor affecting poverty—mainly after one quarter, but also after two and even three quarters. The model’s secondary determinants of poverty are food-price inflation and underemployment—but not GDP, for lack of correlation to poverty. Dr. Mapa estimated its parameters from the SWS time-series on Self-rated Poverty (SRP); there are no other quarterly data.

Hence I wrote: “[T]he most important empirical determinant of both poverty and hunger is inflation in the cost of living. Even short-term spikes in inflation are extremely painful for the poor.” (“Inflation, enemy of the poor,” Opinion, 8/9/2014)

How can “Yolanda” be ignored? It is naive to project poverty to the year 2016 directly from 2013-S1, without recognizing the effects of Supertyphoon “Yolanda,” which struck in November 2013.

Up to now, there is still no official measure of post-Yolanda poverty. Until then, common sense dictates that poverty researchers should examine the five SWS surveys done from December 2013 to December 2014.

A year ago, I already reported: “Supertyphoon Yolanda was so devastating that it raised Self-rated Poverty by 3 points, Self-rated Food Poverty also by 3 points, and Hunger by almost 1 point, according to the Social Weather Survey of December 11-16, 2013. …

“In December 2013, the Self-rated Poverty of Yolanda victims was 72 percent. It was fully 20 points higher than the 52 percent among nonvictims, which we take as the national situation if Yolanda had not occurred. The 3 point excess of the national Self-rated Poverty percentage of 55 from the nonvictims’ 52 is the estimated effect of Yolanda. From its 50-percent rate in September 2013, Self-rated Poverty would have risen by only 2 points in the quarter, instead of by 5 points.” (“Poverty, hunger and Yolanda,” Opinion, 1/25/2014)

I do not claim that SRP or bottom-up poverty, is more “accurate” than poverty-line-based or top-down poverty. They are measures taken from different perspectives, and so have necessarily different magnitudes.

What I do assert is that SRP is superior to line-based poverty on the criteria of frequency and timeliness. Using line-based poverty is as practical as using a GPS that needs one hour to give a car’s location, on a road trip.

SWS released its latest SRP rate weeks ago: “Fourth Quarter 2014 Social Weather Survey: Families rating themselves as Mahirap or Poor at 52%; Families who were Food Poor at 41%” (12/29/2013). In 2013, the SRP quarterly percentages of families in poverty were 52, 49, 50 and 55, with an average of 50.5 for the first semester and 52 for the year. In 2014, the quarterly SRP rates were 53, 55, 55, and 52, with averages of 54 for the first semester and also for the year.

Thus, according to the SRP instrument, poverty rose from 2013-S1 to 2014-S1 by over 3 points. Let us wait and see what the official figures show about poverty incidence at those two points. In any case, by the time official poverty of 2014-S1 is released, presumably in April, there will already be an SRP figure for the first quarter of 2015.

* * *

Contact [email protected].