Steve Wiggins (ODI) | Tracking agricultural transformation – if measuring productivity is hard, should we focus on rural wages?

Share this articleShare on Facebook0Share on Google+0Tweet about this on TwitterShare on LinkedIn0Email this to someone

30 July 2015.

As low income countries (LIC) grow and transform their economies, agriculture plays a key role. It has to raise production to feed increasing numbers living in towns, as well as to provide raw materials to domestic manufacturing — cotton for textiles, hides for shoemakers, palm oil for biscuit and cake makers, etc. For LICs lacking oil, gas, minerals, and substantial manufacturing, agriculture will probably be the largest source of exports to finance imported capital goods. With populations still largely rural, it helps domestic manufacturers if agricultural incomes rise since this expands the domestic market. Last, and certainly not least, agriculture has to free up labour for manufacturing and services — and, depending on the effectiveness of the financial system in rural areas — it may also transfer capital to other sectors.

These functions of agriculture in development, first set out by Johnston & Mellor in 1961, represent a stiff challenge for farming: more bricks from less straw. Yet it is a challenge that was met handsomely during the green revolutions of Asia that began in the late 1960s. Success is marked by:

  • Increased productivity of agriculture, above all of labour and land. This is the only way to increase production, raise farm incomes, and allow labour (and capital) to be released;
  • Labour moving out of agriculture into other work. Most obviously this can be seen in migration to towns and cities. But labour does not necessarily leave the village or household: instead, members of farm households increasingly spend their time on non-farm jobs. These jobs may be carried out at home, elsewhere in the village, or even by commuting to a nearby towns; and,
  • Increasingly active factor markets in rural areas — for seed, fertiliser, chemicals, veterinary drugs, animal feed, irrigation equipment, machinery; technical services; transport; savings, insurance, payments and credit; labour; and land.

Tracking these changes is not easy. Since rising productivity is so important, we would like to be able to make reasonably reliable estimates of agricultural productivity, ideally total factor productivity, but at least partial productivity of land and labour. Trying to measure both outputs and inputs in low-income countries is, however, difficult, as Carletto et al. 2015 explain. Problems abound:

  • Some agricultural output is consumed at home and never formally weighed and logged. Some crops, such as banana and cassava, are often harvested over extended periods, making it difficult to estimate by recall how much in total has been produced. When farmers do count their output, it is often in bags, bunches and bundles: measures that may be only roughly consistent in practice, the definition of which may change from district to district.
  • Small farmers tend to overstate their holdings, large farmers to understate theirs. Cultivable areas on sloping land may be exaggerated — it’s the horizontal plane that matters, not the sloping hypotenuse. Recording area to planted to particular crops can be complicated by intercropping.
  • Labour use estimates can only be rough estimates, when so much farm work involves a few hours at a time on a given field or in attending livestock, through production cycles that last for months.
  • When it comes to inputs, few farmers log their use of seed, fertiliser and chemicals. Capital costs of tools and their maintenance are difficult to capture, as are those of draught livestock services.
  • Most surveys rely on farmer recall which may vary in accuracy, and of course depends on farmers being prepared to reveal what they know. When farmers fear taxes, land redistribution, or exclusion from some social protection or development programme for not being sufficiently poor; or just do not wish their neighbours to become jealous; there are incentives to under-report on all counts. The alternative to recall is to measure: surveying land sizes, taking samples from fields for crop cutting at harvest time, etc. The cost of this, however, is usually prohibitive.
  • Moreover, agriculture varies from year to year as weather, pests and diseases, human health, conditions in markets and so on, mean that production and inputs may fluctuate significantly over time. Hence a very careful survey may just capture an unusual year and not provide an accurate guide to more typical years.

The problems are near intractable. We are, after all, dealing with semi-subsistence, semi-commercial, small-scale and diversified family farming that depends considerably on variable natural and human environments — complex, diverse and risk-prone as Robert Chambers (1989) once described them. All those adjectives and qualifications make measurement difficult.

The Integrated Surveys on Agriculture (ISA) now being added to Living Standards Measurement Surveys (LSMS) surveys are a commendable step forward, but they still rely overwhelmingly on farmers recalling a mass of detail for their plots and livestock, and so will remain subject to most of the shortcomings outlined. Indeed, the main (sole?) technical innovation being offered under ISA seems to be GPS mapping of field boundaries for more accurate assessment of land sizes.

All in all, estimating agricultural productivity is difficult. Even getting simple measures such as yield per hectare or gross value of production per worker is fraught with problems, let alone the valiant but surely vain attempts to estimate total factor productivity that require even bigger guesses to be made about capital inputs.

Does this mean that perhaps we should look for other indicators of agricultural transformation? How about rural labour moving out of farming? This indicator would be powerful, but labour surveys are few and far between, while it is hard to assess time spent on different activities by people who work on a portfolio of activities. Many surveys just record principle occupation, sometimes also recording secondary activities, but rarely reporting the share of labour spent on different activities.

If not labour use, then how about activity in rural factor markets? Almost all of this, however, is difficult to measure and may be sensitive. Take land markets. Few things indicate rural structural change more than land changing hands. But most such changes are informal and unregistered: moreover those involved may be reluctant to reveal such changes especially in countries where formal laws try to govern land transactions, rents, or set ceilings for ownership.

But one aspect of rural factor markets can be observed and could be usefully informative: rural wages, especially those for unskilled work. It is a reasonable bet that when farm productivity rises, when the rural non-farm economy thrives, and when the urban economy flourishes — all conditions for economics transformation — then demand for labour will rise and so will wages. Where, then, unskilled rural wages are rising rapidly, it is very likely that agriculture, rural economies and the overall economy are being transformed. Contemporary Asia has several examples.

Of course, they are imperfect measure of productivity change and transformation. Rural wages could rise when agricultural productivity is stagnant, as people are pulled out of farming by the attraction of better-paid jobs being created where manufacturing is thriving. Wages may, moreover, be in part determined by imperfect markets — monopoly power of either employers or workers, friction in labour markets — as well as by non-economic factors — such as social expectations and public policies. These factors may be more important when looking at wage levels than when looking at changes to wages.

Rural wages may be the canary in the cage. Just as the sick canary cannot tell us what the gas is, nor where it comes from, the silent bird signals something important is afoot. We should track wages, perhaps by establishing sentinel sites to observe them in key locations. When they show significant moves, then check for what the causes may be.

Share this articleShare on Facebook0Share on Google+0Tweet about this on TwitterShare on LinkedIn0Email this to someone