Can we trust our economic statistics?
Many are impressed by the glowing growth rates of the Philippine economy of late.
And yet many others are frustrated that such high growth has not effectively translated to large reductions in poverty.
This seeming disconnect is puzzling since economic growth and poverty generally move in the opposite direction around the world. It is no coincidence that the great strides made against poverty in places like China and Sub-Saharan Africa in recent decades occurred during times of high economic growth.
Of course, it may be the quality of such growth that is at fault. This much we learned from Prof. Winnie Monsod in the first lecture of her Development Economics class. Examples of “bad” growth, she mentioned, included voiceless growth (i.e., growth in a non-participatory society), exclusive growth (i.e., growth accompanied by severe inequality), and jobless growth (i.e., growth with high unemployment).
Yet another possible reason for the Philippine growth-poverty anomaly is that these two variables have simply been measured wrongly. That is, we may have gotten too rosy a picture of the economy, at least in the past decade. We present two studies with findings to this effect: not only have growth rates in the past decade been possibly overestimated, but poverty rates were also possibly underestimated.
The first study started with a simple observation: Since the devastation that followed the Asian financial crisis in 1997, most Asian economies experienced slower economic growth on average. Yet Philippine growth became faster on average in the crisis’ aftermath.
The authors investigated this seeming anomaly by decomposing the sources of GDP growth since that time. By looking at the growth of GDP’s major components (namely growth in consumption, investment, government spending, and net exports), they found that it was only the growth of consumption that rose after the crisis. The growth of all other components of GDP fell (and, together, by a larger magnitude than the rise in consumption growth).
Thus, by all indications, one should expect a decline in the country’s growth rate similar to the experience of the rest of Southeast and East Asia. The only reason GDP growth could have risen (as was the case) is if the growth of consumption rose by a disproportionate amount, large enough to offset the growth reductions of the other components of GDP.
Getting consumption data right is crucial since consumption comprises such a great bulk of Philippine GDP (more than 75% ever since the 1950s). And yet the authors showed that there were, indeed, inconsistencies in the way consumption was factored into GDP figures.
First, the authors compared such GDP consumption data with consumption data from a separate national household survey called the Family Income and Expenditure Survey (FIES). Data from these two sources, though generated by different methodologies, normally move together and show similar trends in consumption over time. But after years of mimicking each other well, the two measures started to diverge after 1997, and since then GDP consumption data had failed to reflect the considerable decline in consumption growth registered using the FIES.
Second, the GDP consumption data (unlike the FIES) was also found to violate a well-known empirical regularity in economics called Engel’s law. Engel’s law states that, holding other things constant, growth in food consumption should not exceed growth in total consumption over time. Yet around 2006 this was exactly what had happened based on GDP consumption data. For the authors, this made the GDP consumption data less reliable than the FIES.
All in all, the authors concluded that consumption data used to estimate GDP since 2000 were most likely overestimated, leading to similarly overestimated economic growth figures. In other words, the relatively robust growth of the Philippine economy seen after the crisis could’ve been a statistical illusion: It was not so much a reflection of sound fundamentals but of the country’s weak statistical system.
The National Statistical Coordination Board (NSCB), the agency in charge of preparing and releasing GDP data, responded to the study. They argued, among others, that overestimating consumption will not necessarily overestimate GDP. Also, GDP consumption data have, in fact, been underestimated more times than they had been overestimated from 1998 to 2008.
The second study looked into the consistency of yet another important set of development data, namely poverty incidence.
Poverty incidence simply means the proportion of the population living below a certain threshold called the poverty line. The poverty line, in turn, is the minimum income required for a family or individual to meet basic food and non-food requirements.
Basic food requirements are directly determined by the Food and Nutrition Research Institute (FNRI), which periodically comes up with regional and provincial “food menus” that contain at least 2,000 calories per person per day and satisfy at least 80-100% of the recommended nutritional intakes of vitamins and minerals. Non-food requirements, in turn, are indirectly determined using the FIES, and include clothing, water, housing, and education, among others.
An important consideration in the construction of poverty lines is the effect of inflation. We know that inflation tends to eat up the purchasing power of income received by households and individuals. Thus, in order to see how poverty moves over time, poverty lines must adjust and reflect changes in overall prices as well.
The problem is that there is some evidence (found in the abovementioned study) that after 1997, the official poverty lines have not adequately reflected changes in overall prices. In other words, whereas the data might tell that a person needs at least P17,000/year to live above the poverty line, in reality, inflation has eroded the peso’s purchasing power such that living outside of poverty requires, say, at least P18,000/year.
This divergence between changes in poverty lines and changes in prices implies that not only should there have been more poor people than what the official data showed, but also that the rate of poverty reduction since 1997 was slower than previously thought.
The study suggested that revisions in the methodology of measuring of poverty lines may have compromised the consistency and comparability of our poverty statistics. As such, the government’s antipoverty programs, especially in relation to the Millennium Development Goal of halving extreme poverty by 2015, may have been informed by inaccurate data.
The NSCB, aside from being responsible for official GDP statistics, is also responsible for official poverty statistics. The agency had also responded to claims of inaccuracies in its poverty statistics, saying that the 3 major refinements in their methodology since 1987 were warranted and in line with global standard practices. Also, despite the revisions, the old and new methodologies have nonetheless shown similar trends in the magnitude and incidence of poverty over time.
Importance of good statistics
Recent Philippine economic history has shown a puzzling disconnect between economic growth and poverty reduction. While indeed there may be structural and policy-related challenges involved, such disconnect might also be partly a reflection of the country’s faulty statistical system.
We have shown two studies finding evidence that, in the years following the Asian financial crisis, GDP growth figures may have been overestimated while poverty incidence may have been underestimated. Taken together, these studies suggest that we may have been getting too rosy a picture of the Philippine economy, at least in the past decade.
But a stronger point made by the studies above is the vital importance of getting the data right, especially for purposes of policymaking.
Unfortunately, there is reason to suspect that inadequate funding of our statistical agencies may have compromised the quality of our official statistics. For instance, the data show that the budget allocations for the government’s statistical agencies from 1991-2005 have consistently constituted less than 1% of the total budget for national government agencies. Such allocations were also subject to very wide fluctuations.
This lack of sizable and stable funding very likely hurts such agencies’ efforts at improving their methodologies and data generation systems. One can only hope that the present administration improves on this chronic lack of funding, especially since evidence-based policymaking is largely a garbage-in-garbage-out process. And while not sufficient, good statistics from our statistical agencies are necessary for good evidence-based policymaking.
So next time you encounter economic statistics, it may do well to refrain from immediately accepting anything at face value. Only a thoughtful consideration of the sources and methodologies used will yield the proper interpretation of such data. – Rappler.com
The author is a summa cum laude graduate of the University of the Philippines School of Economics. His views are entirely his own and do not in any way reflect the views of his affiliations.