Showing 1–9 of 9 results for author: Daoud, A

Deep Learning With DAGs

Authors: Sourabh Balgi, Adel Daoud, Jose M. Peña, Geoffrey T. Wodtke, Jesse Zhou

Abstract: Social science theories often postulate causal relationships among a set of variables or events. Although directed acyclic graphs (DAGs) are increasingly used to represent these theories, their full potential has not yet been realized in practice. As non-parametric causal models, DAGs require no assumptions about the functional form of the hypothesized relationships. Nevertheless, to simplify the… ▽ More Social science theories often postulate causal relationships among a set of variables or events. Although directed acyclic graphs (DAGs) are increasingly used to represent these theories, their full potential has not yet been realized in practice. As non-parametric causal models, DAGs require no assumptions about the functional form of the hypothesized relationships. Nevertheless, to simplify the task of empirical evaluation, researchers tend to invoke such assumptions anyway, even though they are typically arbitrary and do not reflect any theoretical content or prior knowledge. Moreover, functional form assumptions can engender bias, whenever they fail to accurately capture the complexity of the causal system under investigation. In this article, we introduce causal-graphical normalizing flows (cGNFs), a novel approach to causal inference that leverages deep neural networks to empirically evaluate theories represented as DAGs. Unlike conventional approaches, cGNFs model the full joint distribution of the data according to a DAG supplied by the analyst, without relying on stringent assumptions about functional form. In this way, the method allows for flexible, semi-parametric estimation of any causal estimand that can be identified from the DAG, including total effects, conditional effects, direct and indirect effects, and path-specific effects. We illustrate the method with a reanalysis of Blau and Duncan's (1967) model of status attainment and Zhou's (2019) model of conditional versus controlled mobility. To facilitate adoption, we provide open-source software together with a series of online tutorials for implementing cGNFs. The article concludes with a discussion of current limitations and directions for future development. △ Less

Submitted 12 January, 2024; originally announced January 2024.

$ρ$-GNF: A Copula-based Sensitivity Analysis to Unobserved Confounding Using Normalizing Flows

Authors: Sourabh Balgi, Jose M. Peña, Adel Daoud

Abstract: We propose a novel sensitivity analysis to unobserved confounding in observational studies using copulas and normalizing flows. Using the idea of interventional equivalence of structural causal models, we develop $ρ$-GNF ($ρ$-graphical normalizing flow), where $ρ{\in}[-1,+1]$ is a bounded sensitivity parameter. This parameter represents the back-door non-causal association due to unobserved confou… ▽ More We propose a novel sensitivity analysis to unobserved confounding in observational studies using copulas and normalizing flows. Using the idea of interventional equivalence of structural causal models, we develop $ρ$-GNF ($ρ$-graphical normalizing flow), where $ρ{\in}[-1,+1]$ is a bounded sensitivity parameter. This parameter represents the back-door non-causal association due to unobserved confounding, and which is encoded with a Gaussian copula. In other words, the $ρ$-GNF enables scholars to estimate the average causal effect (ACE) as a function of $ρ$, while accounting for various assumed strengths of the unobserved confounding. The output of the $ρ$-GNF is what we denote as the $ρ_{curve}$ that provides the bounds for the ACE given an interval of assumed $ρ$ values. In particular, the $ρ_{curve}$ enables scholars to identify the confounding strength required to nullify the ACE, similar to other sensitivity analysis methods (e.g., the E-value). Leveraging on experiments from simulated and real-world data, we show the benefits of $ρ$-GNF. One benefit is that the $ρ$-GNF uses a Gaussian copula to encode the distribution of the unobserved causes, which is commonly used in many applied settings. This distributional assumption produces narrower ACE bounds compared to other popular sensitivity analysis methods. △ Less

Submitted 22 August, 2024; v1 submitted 15 September, 2022; originally announced September 2022.

Comments: 12 main pages (+8 reference pages), 4 Figures, Accepted at Probabilistic Graphical Models (PGM) 2024. Oral Presentation

To What Extent Do Disadvantaged Neighborhoods Mediate Social Assistance Dependency? Evidence from Sweden

Authors: Cheng Lin, Adel Daoud, Maria Branden

Abstract: Occasional social assistance prevents individuals from a range of social ills, particularly unemployment and poverty. It remains unclear, however, how and to what extent continued reliance on social assistance leads to individuals becoming trapped in social assistance dependency. In this paper, we build on the theory of cumulative disadvantage and examine whether the accumulated use of social assi… ▽ More Occasional social assistance prevents individuals from a range of social ills, particularly unemployment and poverty. It remains unclear, however, how and to what extent continued reliance on social assistance leads to individuals becoming trapped in social assistance dependency. In this paper, we build on the theory of cumulative disadvantage and examine whether the accumulated use of social assistance over the life course is associated with an increased risk of future social assistance recipiency. We also analyze the extent to which living in disadvantaged neighborhoods constitutes an important mechanism in the explanation of this association. Our analyses use Swedish population registers for the full population of individuals born in 1981, and these individuals are followed for approximately 17 years. While most studies are limited by a lack of granular, life-history data, our granular individual-level data allow us to apply causal-mediation analysis, and thereby quantify the extent to which the likelihood of ending up in social assistance dependency is affected by residing in disadvantaged neighborhoods. Our findings show the accumulation of social assistance over the studied period is associated with a more than four-fold increase on a risk ratio scale for future social assistance recipiency, compared to never having received social assistance during the period examined. Then, we examine how social assistance dependency is mediated by prolonged exposure to disadvantaged neighborhoods. Our results suggest that the indirect effect of disadvantaged neighborhoods is weak to moderate. Therefore, social assistance dependency may be a multilevel process. Future research is to explore how the mediating effects of disadvantaged neighborhoods vary in different contexts. △ Less

Submitted 17 August, 2022; v1 submitted 9 June, 2022; originally announced June 2022.

Counterfactual Analysis of the Impact of the IMF Program on Child Poverty in the Global-South Region using Causal-Graphical Normalizing Flows

Authors: Sourabh Balgi, Jose M. Peña, Adel Daoud

Abstract: This work demonstrates the application of a particular branch of causal inference and deep learning models: \emph{causal-Graphical Normalizing Flows (c-GNFs)}. In a recent contribution, scholars showed that normalizing flows carry certain properties, making them particularly suitable for causal and counterfactual analysis. However, c-GNFs have only been tested in a simulated data setting and no co… ▽ More This work demonstrates the application of a particular branch of causal inference and deep learning models: \emph{causal-Graphical Normalizing Flows (c-GNFs)}. In a recent contribution, scholars showed that normalizing flows carry certain properties, making them particularly suitable for causal and counterfactual analysis. However, c-GNFs have only been tested in a simulated data setting and no contribution to date have evaluated the application of c-GNFs on large-scale real-world data. Focusing on the \emph{AI for social good}, our study provides a counterfactual analysis of the impact of the International Monetary Fund (IMF) program on child poverty using c-GNFs. The analysis relies on a large-scale real-world observational data: 1,941,734 children under the age of 18, cared for by 567,344 families residing in the 67 countries from the Global-South. While the primary objective of the IMF is to support governments in achieving economic stability, our results find that an IMF program reduces child poverty as a positive side-effect by about 1.2$\pm$0.24 degree (`0' equals no poverty and `7' is maximum poverty). Thus, our article shows how c-GNFs further the use of deep learning and causal inference in AI for social good. It shows how learning algorithms can be used for addressing the untapped potential for a significant social impact through counterfactual inference at population level (ACE), sub-population level (CACE), and individual level (ICE). In contrast to most works that model ACE or CACE but not ICE, c-GNFs enable personalization using \emph{`The First Law of Causal Inference'}. △ Less

Submitted 17 February, 2022; originally announced February 2022.

Comments: 8(+6) pages, 3(+3) figures, arXiv admin note: text overlap with arXiv:2202.03281

Measuring poverty in India with machine learning and remote sensing

Authors: Adel Daoud, Felipe Jordan, Makkunda Sharma, Fredrik Johansson, Devdatt Dubhashi, Sourabh Paul, Subhashis Banerjee

Abstract: In this paper, we use deep learning to estimate living conditions in India. We use both census and surveys to train the models. Our procedure achieves comparable results to those found in the literature, but for a wide range of outcomes. In this paper, we use deep learning to estimate living conditions in India. We use both census and surveys to train the models. Our procedure achieves comparable results to those found in the literature, but for a wide range of outcomes. △ Less

Submitted 27 October, 2022; v1 submitted 27 December, 2021; originally announced February 2022.

The International Monetary Funds intervention in education systems and its impact on childrens chances of completing school

Authors: Adel Daoud

Abstract: Enabling children to acquire an education is one of the most effective means to reduce inequality, poverty, and ill-health globally. While in normal times a government controls its educational policies, during times of macroeconomic instability, that control may shift to supporting international organizations, such as the International Monetary Fund (IMF). While much research has focused on which… ▽ More Enabling children to acquire an education is one of the most effective means to reduce inequality, poverty, and ill-health globally. While in normal times a government controls its educational policies, during times of macroeconomic instability, that control may shift to supporting international organizations, such as the International Monetary Fund (IMF). While much research has focused on which sectors has been affected by IMF policies, scholars have devoted little attention to the policy content of IMF interventions affecting the education sector and childrens education outcomes: denoted IMF education policies. This article evaluates the extent which IMF education policies exist in all programs and how these policies and IMF programs affect childrens likelihood of completing schools. While IMF education policies have a small adverse effect yet statistically insignificant on childrens probability of completing school, these policies moderate effect heterogeneity for IMF programs. The effect of IMF programs (joint set of policies) adversely effect childrens chances of completing school by six percentage points. By analyzing how IMF-education policies but also how IMF programs affect the education sector in low and middle-income countries, scholars will gain a deeper understanding of how such policies will likely affect downstream outcomes. △ Less

Submitted 30 December, 2021; originally announced January 2022.

The wealth of nations and the health of populations: A quasi-experimental design of the impact of sovereign debt crises on child mortality

Authors: Adel Daoud

Abstract: The wealth of nations and the health of populations are intimately strongly associated, yet the extent to which economic prosperity (GDP per capita) causes improved health remains disputed. The purpose of this article is to analyze the impact of sovereign debt crises (SDC) on child mortality, using a sample of 57 low- and middle-income countries surveyed by the Demographic and Health Survey betwee… ▽ More The wealth of nations and the health of populations are intimately strongly associated, yet the extent to which economic prosperity (GDP per capita) causes improved health remains disputed. The purpose of this article is to analyze the impact of sovereign debt crises (SDC) on child mortality, using a sample of 57 low- and middle-income countries surveyed by the Demographic and Health Survey between the years 1990 and 2015. These surveys supply 229 household data and containing about 3 million childbirth history records. This focus on SDC instead of GDP provides a quasi-experimental moment in which the influence of unobserved confounding is less than a moment analyzing the normal fluctuations of GDP. This study measures child mortality at six thresholds: neonatal, under-one (infant), under-two, under-three, under-four, and under-five mortality. Using a machine-learning (ML) model for causal inference, this study finds that while an SDC causes an adverse yet statistically insignificant effect on neonatal mortality, all other child mortality group samples are adversely affected between a probability of 0.12 to 0.14 (all statistically significant at the 95-percent threshold). Through this ML, this study also finds that the most important treatment heterogeneity moderator, in the entire adjustment set, is whether a child is born in a low-income country. △ Less

Submitted 29 December, 2020; originally announced December 2020.

Extending Social Resource Exchange to Events of Abundance and Sufficiency

Authors: Jonas Bååth, Adel Daoud

Abstract: This article identifies how scarcity, abundance, and sufficiency influence exchange behavior. Analyzing the mechanisms governing exchange of resources constitutes the foundation of several social-science perspectives. Neoclassical economics provides one of the most well-known perspectives of how rational individuals allocate and exchange resources. Using Rational Choice Theory (RCT), neoclassical… ▽ More This article identifies how scarcity, abundance, and sufficiency influence exchange behavior. Analyzing the mechanisms governing exchange of resources constitutes the foundation of several social-science perspectives. Neoclassical economics provides one of the most well-known perspectives of how rational individuals allocate and exchange resources. Using Rational Choice Theory (RCT), neoclassical economics assumes that exchange between two individuals will occur when resources are scarce and that these individuals interact rationally to satisfy their requirements (i.e., preferences). While RCT is useful to characterize interaction in closed and stylized systems, it proves insufficient to capture social and psychological reality where culture, emotions, and habits play an integral part in resource exchange. Social Resource Theory (SRT) improves on RCT in several respects by making the social nature of resources the object of study. SRT shows how human interaction is driven by an array of psychological mechanisms, from emotions to heuristics. Thus, SRT provides a more realistic foundation for analyzing and explaining social exchange than the stylized instrumental rationality of RCT. Yet SRT has no clear place for events of abundance and sufficiency as additional motivations to exchange resources. This article synthesize and formalize a foundation for SRT using not only scarcity but also abundance and sufficiency. △ Less

Submitted 6 October, 2020; originally announced October 2020.

Combining distributive ethics and causal Inference to make trade-offs between austerity and population health

Authors: Adel Daoud, Anders Herlitz, SV Subramanian

Abstract: The International Monetary Fund (IMF) provides financial assistance to its member-countries in economic turmoil, but requires at the same time that these countries reform their public policies. In several contexts, these reforms are at odds with population health. While researchers have empirically analyzed the consequences of these reforms on health, no analysis exist on identifying fair tradeoff… ▽ More The International Monetary Fund (IMF) provides financial assistance to its member-countries in economic turmoil, but requires at the same time that these countries reform their public policies. In several contexts, these reforms are at odds with population health. While researchers have empirically analyzed the consequences of these reforms on health, no analysis exist on identifying fair tradeoffs between consequences on population health and economic outcomes. Our article analyzes and identifies the principles governing these tradeoffs. First, this article reviews existing policy-evaluation studies, which show, on balance, that IMF policies frequently cause adverse effects on child health and material standards in the pursuit of macroeconmic improvement. Second, this article discusses four theories in distributive ethics (maximization, egalitarianianism, prioritarianiasm, and sufficientarianism) to identify which is the most compatible with the core mission of the IMF, that is, improved macroeconomics (Articles of Agreement) while at the same time balancing consequences on health. Using a distributive-ethics analyses of IMF polices, we argue that sufficientarianism is the most compatible theory. Third, this article offer a qualitative rearticulation of the Articles of Agreement, and formalize sufficientarian principles in the language of causal inference. We also offer a framework on how to empirically measure, from observational data, the extent that IMF policies trade off fairly between population health and economic outcomes. We conclude with policy recommendations and suggestions for future research. △ Less

Submitted 10 August, 2020; v1 submitted 30 July, 2020; originally announced July 2020.

Comments: Working paper