Periodic Reporting for period 4 - HKADeC (Human Capital Accumulation in Developing Countries: Mechanisms, Constraints and Policies)
Reporting period: 2021-05-01 to 2022-10-31
The main objectives of this project were:
(i) A better understanding of the early years human development in developing countries and the role of different inputs in the process, including parental behaviour and centre based care. We wanted to study how the dynamics of the process changes with age to identify windows of opportunities that policies could exploit. We wanted to study the interaction in the development process of different dimensons, such as cognitive and socioemotional skills and health; (ii) What drives parental investment; (iii) To accomplish the points above, we recognised the need of high-quality measurements.
We often piggy backed on existing data collections around the evaluation of policy interventions, funded by other grants. Studying the impact of different ECD interventions and the mechanisms that drive such impacts was key in addressing points (i) and (ii). Collecting new surveys allowed us the possibilities of designing novel measures of both child development and its drivers.
To understand parental behavior, we modeled parental choices in the early years in a variety of contexts and shown that these choices might interact with policy interventions. We show that parents might underestimate the productivity of parental investment and the fact that it might be more productive in conjunction with other inputs. We studied how parental beliefs affect parental investment. We designed tools to elicit parental beliefs about the process of child development. In India, we collected parental beliefs longitudinally, which had never been done before. This will allow us to study how parental beliefs change over time.
At the start of the project we used extensively the Young Lives longitudinal data from India, Peru and Ethiopia. In paper [5], we use the Indian part of the data and develop a novel econometric to relate available measures to the different dimensions of development. This approach allows us to model the evolution of child development in different dimensions as a function of different (initial) levels of development, parental investment and other factors. We use the same methods to analyse the data from Peru and Ethiopia in [1] and a high frequency data set from Colombia in [8]. These papers identify the role played by different inputs and in particular parental investment, how their importance change over time and how different dimensions interact with each other, features key to the design of effective interventions. [6] used similar techniques to estimate the process of child development in Colombia using the data used to evaluate a stimulation intervention in small towns. This allowed to argue the observed impacts arose through an increase in parental investment. In modelling parental investment, we stress that subjective beliefs about the process of child development are a key driver of parental choices. Paper [4] proposes a method to elicit parental beliefs and uses the resulting data showing they are predictive of actual parental investment and that parents tend to underestimate the productivity of parental investment. We are starting to analyse the India beliefs data.
On measurement research we have:
- Developed new measures of child development in different dimensions (such as cognition, language, socioemotional skills), piloted in Tanzania, that could be deployed at scale.
- In Ghana, we used a short test often used (IDELA) and integrated it with a few items designed by Harvard psychologist Elizabeth Spelke. We are finishing the study using these data;
- extended the use of statistical techniques in Tanzania, Ghana, Colombia, and India to aggregate measures of child development and its determinants in coherent synthetic measures to study the process of child development;
- developed new measures to elicit parents’ subjective beliefs about the process of child development (in Colombia and India). The approach to elicit these beliefs, using questions about hypothetical scenarios, is original. In India, these data were collected longitudinally.
On the process of child development, we have:
- extended statistical techniques to estimate the process of child development flexibly, allowing the identification of its main determinants, i.e. the effect of initial conditions (in various dimensions – cognition, socioemotional skills, health), parental investment and other variables on subsequent development;
- studied how different dimensions of development interact dynamically and how the process changes over time; finding it becomes more persistent with age and that the impact of parental investment is larger at younger ages;
- used the estimated processes to interpret the short run impacts of a stimulation [6];
- used the estimated processes to identify windows of opportunities, i.e. ages where interventions are more useful to foster child development;
- In Colombia and Ghana, we looked at the impacts of child care centres improvements on child development. In Ghana (in a paper not complete yet) we found that cheap interventions can have sizeable impacts. In Colombia, we found mixed results ([2][16]); the message being that process rather than structure is more important for child development.
On parental investment we have:
- analysed in Colombia how parental investment is driven by different factors (such as local prices) that are unlikely to affect directly child development. This strategy allows us to identify the causal link between parental investment and child development [6]. Estimates taking into account parental investment endogeneity are larger than those that ignore it.
We have disseminated this research extensively, with publications, working papers and presentations at workshops. In Ghana and India, we presented the results we obtained to government and NGO. In Ghana this led to the scaling up of the intervention we studied. In India, the NGO we worked with is expanding the intervention to three states, funded by USAID.