Family Income Distribution in the Philippines, 1985-2009: Essentially the Same

0.  From the following data and discussion we may be able to form conclusions on whether the distribution of income has become more equitable/equal over the past five (5) decades.   With percentile data, I will present different ways of grouping families according to income obtained from the Family Income and Expenditures (FIES) surveys, the latest of which is for 2009.  I will divide the distribution at the median; look at the top one (1) percent families; examine the coefficients of variation (CVs) at the percentiles; compare Gini coefficients; and, even though lacking in statistical rigor, impose the socio-economic classification of classes A, B, C, D and E on the percentile distribution.
A.  Median incomes

1.  The median income from 1961 to 2009, nearly half a century, splits the upper 50 percent with an 80 percent share of income and the lower half, with 20 percent.

As of 2009, the distribution appears to be the same at the end of Martial Law days.

Table 1.  Median Income and Income Distribution, 1961 – 2009
Family Income 1961 1985 2000 2003 2006 2009
Median income (x P1,000) 1 20 89 95 111 135
% Income Share of upper 50% families 82 80 82 81 81 80
% Income Share of lower 50 % families 18 20 18 19 19 20
Source: National Statistics Office.  Website: http://www.census.gov.ph; Family Income and Expenditures publications; Unpublished percentile data on incomes.

B.  Incomes of Top 1% families

2. In 1985, right before EDSA 1, the families in the top 1 percent (numbering about 100 thousand) of the income distribution earned an aggregate income of PhP 31.4 billion.  This is nearly what the combined 3.15 million families (or 32 percent) in the lower brackets of the distribution earned, which amounted to PhP 31.3 billion.

3. In 2000, right before EDSA 2, the top 1 percent families (numbering about 150 thousand) in the income distribution earned an aggregate income of PhP 251.2 billion.  This is nearly what the combined 5.8 million families (or 38 percent) in the lower brackets of the distribution earned, which amounted to PhP 249.6 billion.

4. In 2003, before the end of the first term of Mrs. Arroyo, the top 1 percent families (numbering about 165 thousand) in the income distribution earned an aggregate income of PhP 235.0 billion (hard to imagine that this declined by 6.4 percent from 2000 but this is the official figure).  This is nearly what the combined 5.3 million families (or 32 percent) in the lower brackets of the distribution earned, which amounted to PhP 227.1 billion.

5. In 2006, before the national elections, the top 1 percent families (numbering about 174 thousand) in the income distribution earned an aggregate income of PhP 256.3 billion.  This is nearly what the combined 5.2 million families (or 30 percent) in the lower brackets of the distribution earned, which amounted to PhP 257.9 billion.

6. In 2009, before the last national elections, some 185 thousand ‘top 1 percent’ families earned the equivalent of what 5.5 million ‘bottom 30-percent’ families collectively earned.

7. The 1:30 ratio in 2009 remained, or stabilized, at the same ratio in 2006.

Table 2.  Top 1% Families and Bottom % Families – Income Comparison
1985 2000 2003 2006 2009
Number of Top 1% Families (x1000) 100 150 165 174 185
with
Aggregate Income (PhP billion) 31.4 251.2 235 256.3 342.7
Equivalent to
Number of Families (in millions) 3.15 5.8 5.3 5.2 5.5
% of Total 32% 38% 32% 30% 30%
with
Aggregate Income (PhP billion) 31.3 249.6 227.1 257.9 343
Source: National Statistics Office.  Unpublished percentile data on incomes.

These results raise even more concern when one looks at the top individual taxpayers of 2009 released by the Bureau of Internal Revenue (BIR) in accordance with Section 71 of the National Internal Revenue Code of 1997.  These individuals may not have been covered by the survey as their transactions would be categorized in statistical parlance as ‘rare events’ and thus would have little chance or probability of being selected as samples.

Table 3. BIR Top Individual Taxpayers 2009
Rank Taxpayer Tax Due
1 Elaine B. Gardiola P59.54 million
2 Wilfredo B. Revillame P57.25-million
3 Ronaldo R. Soliman P36.70 million
4 Ramon S. Ang P26.44 million
5 Oscar M. Lopez P25.70 million
66 Henry Sy, Sr P25.18 million
7 Carlos D.C. Ejercito P25.02 million
8 Bonifacio D. Gumboc, Jr P24.74 million
9 Ma. Teresa Caridad P. Gallego P24.45 million
10 Felipe L. Gozon P22.20 million
500 Hitoshi Goto P 3.57 million

Thus this is evidence that the families in the top 1 percent in the income distribution would be under-represented in the survey.  And these should have a higher income share, than is reflected in the FIES, and would further skew the distribution.

C. Coefficients of Variation of the Percentiles

7.  The coefficient of variation (CV) is the standard error expressed in terms of the arithmetic mean (average).  It is a measure of dispersion, a measure of disparity.  The coefficient of variation is useful because the standard deviation of data can be better understood in the context of the arithmetic mean of the data.  The following graphs chart out the CVs of income percentile data obtained from the NSO over many years.

8.  There are no significant changes aside from those at the tails, both at the lowest and highest ends.  The general outlook of the distribution is that of a ‘flat-liner’, bereft of activity showing change.  The family incomes are clustered closely together.  In 2009, eighty-nine (89) 0f the 100 percentile CVs were no greater than 0.1 percent.[2]

Table 4. Distribution of Percentile CVs, 2009
CV (in %) Frequency
0.01 -0.1 89
0.11 – 0.2 6
0.21 – 0.3 2
0.31 – 0.4
0.41 – 0.5 1
0.51 – 0.6
0.61 – 0.7
0.71 – 0.7
0.81 – 0.9
0.91 – 1.0 1
1.01 + 1

Groupings based on a cut-off, for instance, a point/line representing the poverty threshold imposed on these charts would appear to be insufficient.  Income alone would not be a valid indicator of poverty classification because of the observed ‘homogeneity’ of incomes.

D. Gini Cofficient

9.  The Gini coefficient is a measure of the inequality of a distribution, a value of 0 expressing total equality and a value of 1 maximal inequality.  The Gini coefficient is usually defined mathematically based on the Lorenz curve, which plots the proportion of the total income of the population (y axis) that is cumulatively earned by the bottom x% of the population.

10.  However, a low coefficient does not always mean an ideal condition.  It could be that many incomes are similar (either low or high).  In the Philippine example the acknowledged ‘income-poor’ Autonomous Region of Muslim Mindanao has the lowest coefficient followed by the ‘richer’ regions, such as the National Capital Region (NCR) and Central Luzon (Region III).

11. The ARMM had the lowest Gini ratio while Regions 8, 9 and 10 had the highest ratios.

Table 5. Gini ratios, 2009
Region Ratio
A R M M 0.2948
REGION III 0.3727
N C R 0.3953
REGION IV B 0.4004
REGION IV A 0.4063
REGION I 0.4086
REGION V 0.4164
REGION VI 0.4197
C A R 0.4212
REGION XI 0.4275
REGION II 0.4425
REGION XII 0.4425
Caraga 0.4595
REGION VII 0.4601
REGION X 0.4737
REGION IX 0.4738
REGION VIII 0.4841

Table 6. Gini ratios, 2006
Region Ratio
A R M M 0.3113
REGION I 0.3953
N C R 0.3988
REGION III 0.3994
REGION XII 0.4006
REGION IV A 0.4082
REGION IV B 0.4106
REGION II 0.4216
REGION XI 0.4225
REGION VI 0.4326
C A R 0.4418
REGION V 0.4428
Caraga 0.4452
REGION VII 0.4639
REGION X 0.4806
REGION VIII 0.4828
REGION IX 0.5054


11.  Nevertheless, the movement of the coefficient at the national level showed an indication of more equality or less inequality over the years, with the highest being in 1997 and 2000.

Table 7. Gini Coefficient, Philippines
Year Coefficient
1 1985 0.4466
2 1988 0.4446
3 1991 0.468
4 1994 0.4507
5 1997 0.4872
6 2000 0.4822
7 2003 0.4605
8 2006 0.458
9 2009 0.4484

12.  Of 135 countries and dependencies listed in the World Fact Book of the Central Intelligence Asia (CIA), the following rankings can be obtained.  It is clear that the Gini ratio is not always reflective of state of a country’s development[3].

Table 8. Countries with the lowest Gini Ratios
Country Gini ratio Reference Year
Sweden 23 2005
Norway 25 2008
Austria 26 2007
Czech Republic 26 2005
Luxembourg 26 2005
Malta 26 2007
Serbia 26 2008
Slovakia 26 2005
Albania 26.7 2005
Germany 27 2006


Table 9. Countries with the Highest Gini Ratios
Country Gini ratio Reference Year
Brazil 56.7 2005
Colombia 58.5 2008
Bolivia 59.2 2006
Haiti 59.2 2001
Central African Republic 61.3 1993
Sierra Leone 62.9 1989
Botswana 63 1993
Lesotho 63.2 1995
South Africa 65 2005
Namibia 70.7 2003

13.  Among the Association of Southeast Asian Nations (ASEAN), it was Laos with the lowest Gini, and Singapore with the highest..


Table 10. ASEAN Countries’ Gini Ratios
Country Gini ratio Reference Year
Laos 34.6 2002
Vietnam 37 2004
Indonesia 39.4 2005
Cambodia 43 2007 est.
Thailand 43 2006
Philippines 45.8 2006
Malaysia 46.1 2002
Singapore 48.1 2008
Myanmar N/A N/A

E.  ABCDE Socio-economic classification

14.  Market/opinion researchers classify according through proxies of wealth/assets, aside from measure of income to segment the (consumer) market.  These proxies may include conditions in the community where the residence of the respondent is, the types of materials used for the house, household furnishings, ownership of house and/or lot.

15.  From the 16 April 2007 release of Pulse Asia, its nationally-representative sample has seven (7) percent making up classes A, B, and C; sixty-seven (67) percent, class D; and twenty-five (25) percent, class E.  This breakdown has a sampling error of +/- 3 percent.  [Statistically speaking, classes ABC may be 4 to 10 percent of the population; class D, 64-70 percent; and class E, 22-28 percent.]

16.  In 2010, the breakdown became: 9 percent for class ABC; 62 percent for class D; and 29 percent for class E.  Class ABC can be further subdivided into class AB, 0.3 percent, and class C, 8.6 percent, although Pulse Asia estimates an undercount of class AB.

.

Table 11: Percent Distribution of Families, by Socio-Economic Class
Socio Economic Class Percent Share of Families to Total
2007 2010 My guess-timate*
ABC 7 9 10
of which: AB n.a. 0.3** 1
C n.a. 8.6 9
D 68 62 60
E 25 29 30
Source: Pulse Asia, in consultation with Dr. Ana Tabunda 

Note: * – Rounded off but within +/- 3% standard errors of 2010 figures

** -Undercounted due to refusals of AB respondents

.

17   While statistical rigor will not be as robust, we can apply the above percentages [my guess-timates] to the income distribution and find out how much income these classes earned in during the reference years.

Table 12. Percent Distribution of Families and Incomes, by Socio-Economic Class, 1985
CLASS Families Cumulative Income Average Income
Number Share Amount Share
(x 1000) % (x PhP 1 million) % (x PhP 1000)
ABC 985 10 111,420 36 113
D 5,908 60 165,857 54 28
E 2,954 30 28,498 9 10
Total 9847 100 305,775 100 31
Table 13. Percent Distribution of Families and Incomes, by Socio-Economic Class, 2000
CLASS Families Cumulative Income Average Income
Number Share Amount Share
(x 1000) % (x PhP 1 million) % (x PhP 1000)
ABC 1,507 10 838,445 38 556
D 9,043 60 1,174,919 54 130
E 4,522 30 173,886 8 38
Total 15072 100 2,187,250 100 145
Table 14. Percent Distribution of Families and Incomes, by Socio-Economic Class, 2003
CLASS Families Cumulative Income Average Income
Number Share Amount Share
(x 1000) % (x PhP 1 million) % (x PhP 1000)
ABC 1,648 10 884,478 36 537
D 9,888 60 1,346,581 55 136
E 4,944 30 206,191 8 42
Total 16480 100 2,437,250 100 148
Table 15. Percent Distribution of Families and Incomes, by Socio-Economic Class, 2006
CLASS Families Cumulative Income Average Income
Number Share Amount Share
(x 1000) % (x PhP 1 million) % (x PhP 1000)
ABC 1,740 10 1,082,478 36 622
D 10,442 60 1,669,309 56 160
E 5,221 30 254,316 8 49
Total 17,403 100 3,006,104 100 173
Table 16. Percent Distribution of Families and Incomes, by Socio-Economic Class, 2009
CLASS Families Cumulative Income Average Income
Number Share Amount Share
(x 1000) % (x PhP 1 million) % (x PhP 1000)
ABC 1,845 10 1,343,697 35 728
D 11,071 60 2,117,478 56 191
E 5,536 30 343,150 9 62
Total 18,452 100 3,804,325 100 206

18.  When class ABC is further subdivided into class AB and class C, it becomes apparent that class AB could be the top 1 percent, with an income share equal to that of class E.

Table 16-A. Percent Distribution of Families and Incomes, by Modified Socio-Economic Class, 2009
CLASS Families Cumulative Income Average Income
Number Share Amount Share
(x 1000) % (x PhP 1 million) % (x PhP 1000)
AB 185 1 342,736 9 1,857
C 1,661 9 1,000,960 26 603
D 11,071 60 2,117,478 56 191
E 5,536 30 343,150 9 62
Total 18,452 100 3,804,325 100 206

19. In summary, the shares of income of class ABC ranged from 35-38, class D, from 54-56, and class E, from 8-9 percent during the past, nearly a quarter-century, period from 1985-2009.

20.  The good news is that the income distribution has not worsened.  The bad news is that it has remained essentially the same..

G.  Summary

21.  From the following data and discussion we can surmise that development efforts for the past five (5) decades have failed to effect an equitable/equal distribution of income.

  • The median split has been at 82:18 to 80:20 in favor of the families at upper fifty (50) percent over the past fifty (50) years.
  • The top one (1) percent families earned income equivalent to income earned by 32 percent of the families at the bottom of the income ladder in 1985.  This peaked to 38 percent in 2000, was replicated in 2003, and moved down to 30 percent in 2006 and 2009.  In twenty-five (25) years the top 1 percent gave up two (2) percent to the families at the bottom rungs.
  • The CVs show very little variation at the percentiles except those at the extreme ends, indicating little spread of income across the entire distribution.
  • The Gini coefficient, with its measure of inequality subject to misinterpretation, had moved up during the ‘Baht’ financial crisis, and down from then on.  The Gini ratio of the Philippines is neither among the highest nor the lowest in the world, including ASEAN.
  • The shares of income of class ABC ranged from 35-38, class D, from 54-56, and class E, from 8-9 percent the past twenty-five (25) years from 1985-2009

22.  There is also utter lack of information on the distribution of family income which the government, particularly the National Statistics Office (NSO) and the statistical system, need to address.  Perhaps one of the reasons why the distribution has generally remained unchanged is because even if many think that this is so, there has been insufficient empirical evidence to establish its extent and chronicity.

23.  I also urge the government to come up with an official definition of the often-used ABCDE socio-economic classification and the ‘generic’ low-middle-high income classes.   in cooperation with the academe and private sector.  These are terms that many policy and decision-makers and the general public have come to accept and use rather than deciles, quintiles and percentiles and the government can respond by standardizing these and help improve the statistical literacy of society, in this case on income distribution.


[2] On the basis of this observation, a review of the sampling scheme may be valid since it appears that the sampling size can be reduced with the very low CVs; this may be the case of surveying ‘more of the same’.  The soundness of sub-national results may also be evaluated by examining relevant CVs at the percentiles/ /quintiles/deciles.  The FIES questionnaire should also be reviewed in this light; it has 950 items, with 234 on income, 677 on expenditures and others, 39.

[3] It has been pointed out that Gini coefficients can be computed using income or expenditure /consumption data and this should be considered when comparing country coefficients.

Advertisements

23 Comments

Filed under income distribution, Philippine development, Philippines, statistics

23 responses to “Family Income Distribution in the Philippines, 1985-2009: Essentially the Same

  1. Joseph H. Francia

    You have filled a need. Using socio economic classes ABCDE is easier to understand and more appealing to a general than deciles. I hope the NSO responds to your suggestions for presenting their data in this format.

  2. Joseph H. Francia

    Correction. I should say “general audience.”

  3. katherine

    any data during Martial Law? that would have been interesting since people in the younger generation like me don’t have any idea what was it like during martial law and honestly speaking im quiet confused. i can hear people saying “it was better during martial because you can have piso-busog”. i just want to know if there’s any basis to that claim “statiscally”. it would be nice to compare the martial law data to our present condition. good job by the way, im not into stat but i find your article really valuable for my term paper… gracias! 🙂

    • makuhari

      Unfortunately many key statistics came out only after martial law. Some say one governs better during authoritarian rule by keeping the citizenry either misinformed or uninformed. Am glad to be of some help to you.

  4. Charity

    Do you know the population for class AB from ages 25-35?

    • makuhari

      Not really. But…

      from Table 16-A, there are about 185 thousand AB families. Assuming 5 as family size, 185 thousand multiplied by 5 = 925 thousand. From the 2007 population distribution by age, those 25-34 years old were about 15% of total population. Hence 15% of 925 thousand would be about 140 thousand.

      I hope that these estimates help.

  5. Good day sir, can I quote your data on the socio-economic classification?

  6. Tony

    Hi there, any data specific to rizal province? If none how can I possibly come up with a “guess-timate” to break the data you have presented according to region, province down to cities or municipalities. I hope you will allow me to do that, you really presented it well. God Bless.

    • makuhari

      Sorry. I have not checked my blog for quite a while. There is scant information on income distribution lower than the national level. Perhaps generating tables at the regional level may yield useful results, but this would take a lot of work and effort.

  7. rollypolly

    Hi sir, how did you get the cumulative income amount in table 16-a? thanks!

  8. pau

    Hi sir. Can I use your data for our case study? Thank you!

  9. Hello! Do you have information on the income polarization of the AB segment (the extremes)? Basically a breakdown of the average illustrated on Table 16-A.

    • makuhari

      Sorry, I don’t have those figures. I think the respondents at that bracket were so few even with a total sample of more than 50
      thousand, and so figures could not be presented, without violating the strict rules on confidentiality.

  10. its hard to operationalise the socio-economic status, like how many properties could AB have, i mean the assets and other properties, i’m doing it for my thesis.

  11. mamu

    Sir gud pm. Can u answer this, it is quite hard for me. Please..
    filipino profile in terms of population status
    filipino profile in terms of distribution accdg to region
    and filipino profile in terms of rate of population growth.

    Thank u.

    • makuhari

      You will need to define your terms.
      1. Filipino? citizen, language?
      2. population status?

      These can mean many things.

  12. holylele

    Thank you for sharing your knowledge kind sir.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s