Not Your Average Job : Measuring Farm Labor in Tanzania

Arthi, Vellore ; Beegle, Kathleen ; De Weerdt, Joachim ; Palacios-Ló ; pez, Amparo

World Bank, Washington, DC

关键词: recall error;    measurement error;    farm labor;    agricultural productivity;    survey bias;   

DOI  :  10.1596/1813-9450-7773
RP-ID  :  WPS7773

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

A good understanding of the constraintson agricultural growth in Africa relies on the accuratemeasurement of smallholder labor. Yet, serious weaknesses inthese statistics persist. The extent of bias in smallholderlabor data is examined by conducting a randomized surveyexperiment among farming households in rural Tanzania.Agricultural labor estimates obtained through weekly surveysare compared with the results of reporting in a singleend-of-season recall survey. The findings show strongevidence of recall bias: people in traditional recall-stylemodules report working up to four times as many hours perperson-plot relative to those reporting labor on a weeklybasis. If hours are aggregated to the household level,however, this discrepancy disappears, a factor driven by theunderreporting by recall households of people and plotsactive in agricultural work. The evidence suggests thatthese competing forms of recall bias are driven not only byfailures in memory, but also by the mental burdens ofreporting on highly variable agricultural work patterns toprovide a typical estimate. All things equal, studiessuffering from this bias would understate agricultural labor productivity.