A better way to define rural

Posted on July 6, 2014  /  3 Comments

Nalaka Gunawardene asks good questions. So I paid attention when he tweeted:

Based on the ongoing work using mobile transaction generated data (big data), we can give a good answer about areas that are predominantly residential, ans those that are mixed or predominantly commercial. But this will not mesh with the rural-urban divide. For example, it will show the densely populated Colombo North or Colombo’s inner city to be predominantly residential. It definitely is not rural.

Why does one need to know what is rural? Traditionally, it was because rural areas lacked in services; and needed extra attention. But Sri Lankan rural areas have never fit the pattern. We had post offices distributed across the country, at levels that matched those of Italy. We had rural dispensaries and hospitals.

I can see the value of identifying predominantly residential areas. Could someone answer the question as to why we need to identify rural?

Here is Nalaka’s response:


  1. For purposes of Marketing & Social Research Surveys, the use of Sectors (Urban, Rural & Estate) was discontinued many years ago since this variable was no longer considered to be an efficient discriminator of local market segments. In it’s place we now use ‘SEC’ groups. The Socio Economic Classification (SEC) of Households is based on two variables – The Occupation of the Main Earner and the Highest Exam Passed by the Main Earner. The SEC as defined above yields five mutually exclusive and exhaustive sub-groups when applied to a population of households. Each sub-group represents a level of ‘life-style’ of the households in that group. Since ‘Consumption’ is driven by ‘Life-style’, the Socio Economic Classification of a sample of Households will provide a better segmentation of a prospective market with regard to both, Attitude & Behaviour.
    Mr Nalaka Mendis should check out whether there is any significant ‘info-deprivation’ between the higher and lower Life-style groups.
    For a detailed definition of SEC groups, please visit our web-site, http://www.srl.lk

  2. Sri Lanka is very small and densely populated. The difference between urban and rural may thus be very small. The right distinction needs to be country specific and post offices are not really a good indicator on its own. Other options include:
    1) Average size of land a households lives on = higher in rural areas.
    2) Agricultural production distinguished by size. Presence of large scale would indicate rural areas.
    3) Service density: secondary schools, universities, clinics, police stations, super markets, shopping centres, etc.
    4) Income for a particular activity
    5) Share of labour force employed in primary sector other than mining
    The ideal distinction may combine several characteristics.

  3. Rohan Samarajiva

    In a technocratic, not political, approach we’d ask what the problem is and then evolve a remedy. For example, back in 2003, before mobile became cheap and spread to rural areas, we were thinking we needed to subsidize those areas. We did not use the government definition, which included the Kaduwela Pradeshiya Sabha (right up to Water’s Edge). Instead we calculated the population densities of each of the Divisional Secretariat Divisions. Those falling below a threshold were designated as eligible for subsidy. The cost of serving less densely populated areas is higher and revenues from them are lower. Therefore, our definition made sense.

    In a smaller subsidy scheme that we ran from 1999, we simply stayed with the government definition. The cost of developing a different defensible definition was not justified by the size of the subsidy program.

    Mr Uduman, above, has a different purpose, selling FMCGs. The definition he needs is different.

    In political approaches, none of this matters. What matters is the message that is communicated. There is a strong sense of rural romanticism among even those who live in urban areas, so there are lots of political points to be scored by pandering to rural areas. I do not know whether this is still true, but between 1948 and 1977, the value of a vote of a rural resident was greater than that of an urban resident. That is it took more votes to elect an MP in Colombo than in Kahatagasdigiliya. In such weighted systems, it makes political sense to favor rural areas. Certain political parties do well in rural areas and not in urban. Therefore, they too emphasize rural subsidies and programs. A good technocrat will understand these factors and shape the rational argument accordingly.