The Rationale for Socio-Economic Classification

The invocation of socio-economic class smacks of elitism.  In democratic societies, people like to think of themselves as being equals without social or economic barriers.  Thus, it is in bad taste, if not alright divisive, to speak of socio-economic class.

Nevertheless, it seems that most democratic countries seem to have some kind of socio-economic definition that is used in marketing research.  In this article, we will be build up the case as to why socio-economic levels are rational consequences of the need to be more efficient and effective in marketing applications.

As a starting point, we posit that not all products/services will appeal to everybody in a society.  It is true that some products and services have near universal appeal, but not everything is.  Rather, some products and services have appeal to certain sub-populations.  Here are some obvious examples --- hair loss products are useful for people who are losing their hair; Viagra is useful to people who experience erectile difficulty; Rolex watches can be bought by people who can afford them; and so on.  Each product may have its own unique niche, as characterized by demographic characteristics (such as age, gender, income, etc), values, attitudes, lifestyles and so on.  The rational marketer would realize that it is inefficient to attempt to market a niche product to the general population.

Within the experience of market researchers all over the world, the notion of socio-economic class is useful for a large number of products and services all over the world, time and again.  We will present ample examples later.  Thus, market researchers found it convenient to define a socio-economic level system that can be used by everyone.  The common definition permits the comparison of market research studies conducted by different people.

Looking across the world, we find that many socio-economic level systems are based upon the consideration of economic characteristics of products and services as well as social standing (such as profession and education).  In many cases, the system is a precise numerical point-scoring system where a short inventory of products and services are taken (e.g. automobiles, computers, vacuum cleaner) and then points are awarded (e.g. 10 points for one automobile, 15 points for 2 or more automobiles; 2 points for vacuum cleaner, etc).  Eventually, people are placed into groups based upon their point totals.

For the purpose of discussion, we will refer to the Economic Classification Criterion used by the marketing associations ABA, ANEP and ABIPEME in Brazil (PDF file).  The point system is as follows:


Product/Service Does Not Own Has One Has Two Has Three Has Four or More
Color TV 0 2 3 4 5
Radio 0 1 2 3 4
Bathroom 0 2 3 4 4
Automobile 0 2 4 5 5
Servant (paid monthly) 0 2 4 4 4
Vacuum Cleaner 0 1 1 1 1
Washing Machine 0 1 1 1 1
VCR 0 2 2 2 2
Refrigerator 0 2 2 2 2
Freezer 0 1 1 1 1


Educational Background Points
Illiterate/Incomplete Primary 0
Completed Primary/Incomplete Junior High 1
Completed Junior High/Incomplete High 2
Completed High/Incomplete University 3
Completed University 5

After the points are added up, the people are assigned to different socio-economic classes:

Socio-Economic Class Points range
A1 30-34
A2 25-29
B1 21-24
B2 17-20
C 11-16
D 6-10
E 0-5

The Brazilian system is so simple that it seems to be naïve.  Is it possible that just a few simple products and services plus the education of head of household is sufficient to capture the complexity of socio-economic class as applied to all types of marketing situations?  Indeed, we will show that this system is in fact potent.  Furthermore, this is a robust approach in the sense that the details of the system can be changed (e.g. you can take away videocassette and add computers instead).  All across the world, there are systems that look somewhat different but their ultimate performance characteristics and capabilities are quite similar.

As a first step, we show that this particular system captures the essential characteristics of its components.  We will use the TGI Brasil study to illustrate.  This is a survey of 10,624 persons between the ages of 12 to 64 years old conducted in Brazil during 2003.

Following the convention of marketing research, we report the indices for these items by socio-economic class.  Here is an example:  overall, 44% of Brazilian households have cars; within socio-economic class AB (=A1+A2+B1+B2, which accounts for 32.4% of the survey universe), the incidence is 87% for an index of 100 x 87 / 44 = 198; within socio-economic class C (which accounts for 37.8% of the survey universe), the incidence is 37% for an index of 37/44 = 84; within socio-economic class DE(=D+E, which accounts for 29.8% of the survey universe), the incidence is 6% for an index of 100 x 6 / 44 = 13.  If you look at 'automobile' in the following chart, you will see the indices 198, 84 and 13.

(Source: 2003 TGI Brasil)

This chart shows that we can pretty much get to the potential buyers of bathroom interior renovation, automobiles and related services, maid services, vacuum cleaners, washing machines and VCRs (and cassette tapes) through this socio-economic class system.  Color television sets, radio sets and refrigerators are mass consumer durables and hardly need to be treated as niche markets.

In this next chart, we show the indices of the educational achievement of the head of household by socio-economic class.

(Source: 2003 TGI Brasil)

But this is an uninteresting exercise.  We have combined a number of variables to form a socio-economic classification system.  Tautologically, those variables must surely behave that way.  This is a self-fulfilling prophecy.  This exercise is more of a sanity check, because if things didn't turn out this way, then something must have been implemented incorrectly.

The more interesting question is whether this socio-economic class system can be applied effectively to other variables unseen so far.  In the next chart, we have chosen a list of very common products and services.  In spite of the fact that none of these items were included in the original definition of socio-economic class, they all show some degree with association with socio-economic class.  

(Source: 2003 TGI Brasil)

This is a general phenomenon that is found in different systems all over the world --- a small set of carefully chosen set of variables can predict the consumption behavior in a large number of other products and services.  This is because the heavy consumer is someone with the economic purchasing power AND who is willing to spend; conversely, the light consumer is someone who does not have the economic purchasing power OR has little or no desire to spend.  In 1997, we published a paper (Defining Socio-Economic Level in Latin America) describing the socio-economic levels used in the Los Medios y Mercados de Latinoamérica study.  This was a point scoring system based upon a different set of variables and weights.  This system was then compared with the Brazilian system for the country of Brazil, and the overlap between the two systems is high (Socio-Economic Level: National vs. Regional Definitions).

Once we identify that socio-economic class is a meaningful way of identifying consumers, it is easy to take the next step of reaching these consumers.  For example, the next chart shows media usage by socio-economic class.  Each of these four media is an efficient medium for reaching the AB class in Brazil.

(Source: 2003 TGI Brasil)

Lest we oversell the case here, not all products and services are driven by socio-economic considerations.  In the next chart, we show some examples of foods and beverages that are virtually flat across the socio-economic classes.  In these cases, the marketing researcher must look for other characteristics that define the users (e.g. age/sex).  And if indeed the products and services are intended for the masses, mass media such as television and radio can be used to reach large numbers of consumers.

(Source: 2003 TGI Brasil)

If socio-economic class ultimately refers to economic purchasing power, then it is natural to ask why income was not used directly.  It is true that in the United States, market researchers use household income as de facto definition of socio-economic class.  This is feasible because most people have a good sense of what their annual household income is, as everyone is supposed to file an income tax return each year that will state the official household income for the past year.  Elsewhere in the world, it is by no means easy to get a handle on income even if the survey respondents are dutiful cooperators.  We will not get into the many reasons which deserve a book of their own.  Here we will look at the relationship between socio-economic class and reported income in the TGI Brasil study.  

First of all, we look at personal income.  Here, we ignore the complexity of what constitutes personal income (hint: apart from wages, there are tips, bonuses,commissions, stocks, stock options, subsidies, pensions, interests, gifts, alimony, etc).  In the following chart, we show the indices of personal income categories by socio-economic level.  As expected, the upper person income categories are filled with AB-types.  But something unusual happens at the bottom of the personal income scale.  Who are all these people with little or no personal income but have upper-class consumption patterns?  A simple example is enough --- the wife of the CEO of a major corporation does not work and therefore derives zero personal income, although her full-time job is to shop, shop and then shop some more ...  For this reason, we should not be using personal income to characterise consumption.

(Source: 2003 TGI Brasil)

But what about household income?  That unemployed wife of the CEO would have a very high household income.  We can look at the data in the TGI Brasil study.  Again, we will ignore the complexity of what constitutes household income (hint: apart from the sum of the personal incomes of the household members, there are taxes, dividends, interests, depreciation, capital gains, rental income, barter income, etc).  In the following chart, we show the indices of household income categories by socio-economic level.  As expected, the upper household income categories are filled with AB types and now the lower household income categories are filled with DE types.  

(Source: 2003 TGI Brasil)

However, these two variables are not identical.  Let us we look at the two extreme ends.  On one side, there are people with high household income who do not score high on socio-economic class.  This could be someone who makes a lot of money, but has no interest in consuming the typical goods and services.  Or it could be someone who may earn a lot of money but have further obligations (such as huge debts to pay off) that prevent them from spending the income.  Whatever the reason, we know that these people do not purchase products and services.  On the other side, there are people with little or no income but who appear in the high socio-economic class.  This could be someone who is spending beyond their means (e.g. through credit cards).  Or it could be a non-working person who has a lot of money in the bank and can live in comfort.  In any case, if the purpose is to capture consumerist behavior, it is more logical to base the definition on actual consumption behavior rather than on some hypothetical driver such as income.

The astute readers would have detected a point of contradiction here.  In a study like the TGI (Target Group Index), information is collected on both product and media consumption, in addition to demographics, values, attitudes and lifestyles.  Within this study, it is in fact possible to directly link product consumption with media usage without having to go through some intermediate entity such as socio-economic class.  That is correct.  Unfortunately, a TGI study is an elaborate and time-consuming effort and cannot be easily replicated in other research studies.  Therefore, socio-economic class systems are still the rule across the broad range of research studies.

(posted by Roland Soong on 2/1/2004)

(Return to Zona Latina's Home Page)