Reforging the Wedding Ring:: Exploring a Semi-Artificial Model of Population for the United Kingdom with Gaussian process

Has demography witnessed a data revolution? Promises and pitfalls of a changing data ecosystem

Over the past 25 years, technological improvements that have made the collection, transmission, storage, and analysis of data significantly easier and more cost efficient have ushered in what has been described as the 'big data' era or the 'data revolution'. In the social sciences context, the data revolution has often been characterized in terms of increased volume and variety of data, and much excitement has focused on the growing opportunity to repurpose data that are the by-products of the digitalization of social life for research. However, many features of the data revolution are not new for demographers, who have long used large-scale population data and been accustomed to repurposing imperfect data not originally collected for research. Nevertheless, I argue that demography, too, has been affected by the data revolution, and the data ecosystem for demographic research has been significantly enriched. These developments have occurred across two dimensions. The first involves the augmented granularity, variety, and opportunities for linkage that have bolstered the capabilities of 'old' big population data sources, such as censuses, administrative data, and surveys. The second involves the growing interest in and use of 'new' big data sources, such as 'digital traces' generated through internet and mobile phone use, and related to this, the emergence of 'digital demography'. These developments have enabled new opportunities and offer much promise moving forward, but they also raise important ethical, technical, and conceptual challenges for the field.

Bridging implementation gaps to connect large ecological datasets and complex models

Merging robust statistical methods with complex simulation models is a frontier for improving ecological inference and forecasting. However, bringing these tools together is not always straightforward. Matching data with model output, determining starting conditions, and addressing high dimensionality are some of the complexities that arise when attempting to incorporate ecological field data with mechanistic models directly using sophisticated statistical methods. To illustrate these complexities and pragmatic paths forward, we present an analysis using tree-ring basal area reconstructions in Denali National Park (DNPP) to constrain successional trajectories of two spruce species (Picea mariana and Picea glauca) simulated by a forest gap model, University of Virginia Forest Model Enhanced-UVAFME. Through this process, we provide preliminary ecological inference about the long-term competitive dynamics between slow-growing P. mariana and relatively faster-growing P. glauca. Incorporating tree-ring data into UVAFME allowed us to estimate a bias correction for stand age with improved parameter estimates. We found that higher parameter values for P. mariana minimum growth under stress and P. glauca maximum growth rate were key to improving simulations of coexistence, agreeing with recent research that faster-growing P. glauca may outcompete P. mariana under climate change scenarios. The implementation challenges we highlight are a crucial part of the conversation for how to bring models together with data to improve ecological inference and forecasting.

The impact of demographic change in the balance between formal and informal old-age care in Spain. Results from a mixed microsimulation–agent-based model

Changes in population and family structures are altering the provision of care for dependent older people. In Southern European countries like Spain, such care is still largely provided by family, typically spouses and adult daughters. However, an increasing proportion of women have entered the labour force, thereby affecting their availability. To study the demand and supply balance of informal care and to quantify the need for formal care when there is a deficit, we have developed a mixed microsimulation–agent-based model (ABM). Based on nuptiality, fertility and mortality levels of cohorts born at ten-year intervals between 1908 and 1968, the model starts with a microsimulation of the lifecycle of individuals and their close relatives until death. The ABM then determines the amount of time available or needed to care for family members, starting from age 50. Estimates are derived from Spanish survey data on employment, disability and time of care received. Surprisingly, results show that the family care deficit was greater in the older cohorts due to higher mortality and thus a greater impact of widowhood. However, for future elderly persons, we foresee that persistent below-replacement fertility and, paradoxically, the prolongation of the lifespan of couples will increase the demand for formal care as there will be more couples with both members incapacitated but without children to take care of them.

The use of mixture density networks in the emulation of complex epidemiological individual-based models

Complex, highly-computational, individual-based models are abundant in epidemiology. For epidemics such as macro-parasitic diseases, detailed modelling of human behaviour and pathogen life-cycle are required in order to produce accurate results. This can often lead to models that are computationally-expensive to analyse and perform model fitting, and often require many simulation runs in order to build up sufficient statistics. Emulation can provide a more computationally-efficient output of the individual-based model, by approximating it using a statistical model. Previous work has used Gaussian processes (GPs) in order to achieve this, but these can not deal with multi-modal, heavy-tailed, or discrete distributions. Here, we introduce the concept of a mixture density network (MDN) in its application in the emulation of epidemiological models. MDNs incorporate both a mixture model and a neural network to provide a flexible tool for emulating a variety of models and outputs. We develop an MDN emulation methodology and demonstrate its use on a number of simple models incorporating both normal, gamma and beta distribution outputs. We then explore its use on the stochastic SIR model to predict the final size distribution and infection dynamics. MDNs have the potential to faithfully reproduce multiple outputs of an individual-based model and allow for rapid analysis from a range of users. As such, an open-access library of the method has been released alongside this manuscript.

Infectious disease modellers have a growing need to expose their models to a variety of stakeholders in interactive, engaging ways that allow them to explore different scenarios. This approach can come with a considerable computational cost that motivates providing a simpler representation of the complex model. We propose the use of mixture density networks as a solution to this problem. MDNs are highly flexible, deep neural network-based models that can emulate a variety of data, including counts and over-dispersion. We explore their use firstly through emulating a negative binomial distribution, which arises in many places in ecology and parasite epidemiology. Then, we explore the approach using a stochastic SIR model. We also provide an accompanying Python library with code for all examples given in the manuscript. We believe that the use of emulation will provide a method to package an infectious disease model such that it can be disseminated to the widest audience possible.

Modelling social care provision in an agent-based framework with kinship networks

Current demographic trends in the UK include a fast-growing elderly population and dropping birth rates, and demand for social care among the aged is rising. The UK depends on informal social care-family members or friends providing care-for some 50% of care provision. However, lower birth rates and a greying population mean that care availability is becoming a significant problem, causing concern among policy-makers that substantial public investment in formal care will be required in decades to come. In this paper, we present an agent-based simulation of care provision in the UK, in which individual agents can decide to provide informal care, or pay for private care, for their loved ones. Agents base these decisions on factors including their own health, employment status, financial resources, relationship to the individual in need and geographical location. Results demonstrate that the model can produce similar patterns of care need and availability as are observed in the real world, despite the model containing minimal empirical data. We propose that our model better captures the complexities of social care provision than other methods, due to the socioeconomic details present and the use of kinship networks to distribute care among family members.

The Dynamics of Son Preference, Technology Diffusion, and Fertility Decline Underlying Distorted Sex Ratios at Birth: A Simulation Approach

We present a micro-founded simulation model that formalizes the "ready, willing, and able" framework, originally used to explain historical fertility decline, to the practice of prenatal sex selection. The model generates sex ratio at birth (SRB) distortions from the bottom up and attempts to quantify plausible levels, trends, and interactions of son preference, technology diffusion, and fertility decline that underpin SRB trajectories at the macro level. Calibrating our model for South Korea, we show how even as the proportion with a preference for sons was declining, SRB distortions emerged due to rapid diffusion of prenatal sex determination technology combined with small but growing propensities to abort at low birth parities. Simulations reveal that relatively low levels of son preference (about 20 % to 30 % wanting one son) can result in skewed SRB levels if technology diffuses early and steadily, and if fertility falls rapidly to encourage sex-selective abortion at low parities. Model sensitivity analysis highlights how the shape of sex ratio trajectories is particularly sensitive to the timing and speed of prenatal sex-determination technology diffusion. The maximum SRB levels reached in a population are influenced by how the readiness to abort rises as a function of the fertility decline.

Multistate modelling extended by behavioural rules: An application to migration

We propose to extend demographic multistate models by adding a behavioural element: behavioural rules explain intentions and thus transitions. Our framework is inspired by the Theory of Planned Behaviour. We exemplify our approach with a model of migration from Senegal to France. Model parameters are determined using empirical data where available. Parameters for which no empirical correspondence exists are determined by calibration. Age- and period-specific migration rates are used for model validation. Our approach adds to the toolkit of demographic projection by allowing for shocks and social influence, which alter behaviour in non-linear ways, while sticking to the general framework of multistate modelling. Our simulations yield that higher income growth in Senegal leads to higher emigration rates in the medium term, while a decrease in fertility yields lower emigration rates.

Conceptualizing the Dynamics between Bicultural Identification and Personal Social Networks

An adequate understanding of the acculturation processes affecting immigrants and their descendants involves ascertaining the dynamic interplay between the way these individuals manage their multiple (and sometimes conflictual) cultural value systems and identifications and possible changes in their social networks. To fill this gap, the present research examines how key acculturation variables (e.g., strength of ethnic/host cultural identifications, bicultural identity integration or BII) relate to the composition and structure of bicultural individuals' personal social networks. In Study 1, we relied on a generationally and culturally diverse community sample of 123 Latinos residing in the US. Participants nominated eight individuals (i.e., alters) from their habitual social networks and across two relational domains: friendships and colleagues. Results indicated that the interconnection of same ethnicity alters across different relationship domains is linked to cultural identifications, while the amount of coethnic and host individuals in the network is not. In particular, higher interconnection between Latino friends and colleagues was linked to lower levels of U.S.

IDENTIFICATION

Conversely, the interconnection of non-Latino friends and colleagues was associated with lower levels of Latino identification. This pattern of results suggests that the relational context for each type of cultural identification works in a subtractive and inverse manner. Further, time spent in the US was linked to both Latino and U.S. cultural identifications, but this relationship was moderated by the level of BII. Specifically, the association between time in the US and strength of both cultural identities was stronger for individuals reporting low levels of BII. Taking the findings from Study 1 as departure point, Study 2 used an agent-based model data simulation approach to explore the dynamic ways in which the content and the structure of an immigrant's social network might matter over time in predicting three possible identity patterns: coexisting cultural identifications, conflicting cultural identifications, and a mixture of the two. These simulations allowed us to detect network constellations, which lead to identification or disidentification with both cultures. We showed that distinct patterns of social relations do not lead to identity outcomes in a deterministic fashion, but that often many different outcomes are probable.

Integrating macro- and micro-level approaches in the explanation of population change

Demographers study population change across time and place, and traditionally they place a strong emphasis on a long-range view of population change. This paper builds on current reflections on how to structure the study of population change and proposes a two-stage perspective. The first stage, discovery, focuses on the production of novel evidence at the population level. The second stage, explanation, develops accounts of demographic change and tests how the action and interaction of individuals generate what is discovered in the first stage. This explanatory stage also provides the foundation for the prediction of demographic change. The transformation of micro-level actions and interactions into macro-level population outcomes is identified as a key challenge for the second stage. Specific instances of research are discussed.

Bayesian demography 250 years after Bayes

Bayesian statistics offers an alternative to classical (frequentist) statistics. It is distinguished by its use of probability distributions to describe uncertain quantities, which leads to elegant solutions to many difficult statistical problems. Although Bayesian demography, like Bayesian statistics more generally, is around 250 years old, only recently has it begun to flourish. The aim of this paper is to review the achievements of Bayesian demography, address some misconceptions, and make the case for wider use of Bayesian methods in population studies. We focus on three applications: demographic forecasts, limited data, and highly structured or complex models. The key advantages of Bayesian methods are the ability to integrate information from multiple sources and to describe uncertainty coherently. Bayesian methods also allow for including additional (prior) information next to the data sample. As such, Bayesian approaches are complementary to many traditional methods, which can be productively re-expressed in Bayesian terms.

Decision-Making in Agent-Based Models of Migration: State of the Art and Challenges

We review agent-based models (ABM) of human migration with respect to their decision-making rules. The most prominent behavioural theories used as decision rules are the random utility theory, as implemented in the discrete choice model, and the theory of planned behaviour. We identify the critical choices that must be made in developing an ABM, namely the modelling of decision processes and social networks. We also discuss two challenges that hamper the widespread use of ABM in the study of migration and, more broadly, demography and the social sciences: (a) the choice and the operationalisation of a behavioural theory (decision-making and social interaction) and (b) the selection of empirical evidence to validate the model. We offer advice on how these challenges might be overcome.

Assortative mating and the reversal of gender inequality in education in europe: an agent-based model

While men have always received more education than women in the past, this gender imbalance in education has turned around in large parts of the world. In many countries, women now excel men in terms of participation and success in higher education. This implies that, for the first time in history, there are more highly educated women than men reaching the reproductive ages and looking for a partner. We develop an agent-based computational model that explicates the mechanisms that may have linked the reversal of gender inequality in education with observed changes in educational assortative mating. Our model builds on the notion that individuals search for spouses in a marriage market and evaluate potential candidates based on preferences. Based on insights from earlier research, we assume that men and women prefer partners with similar educational attainment and high earnings prospects, that women tend to prefer men who are somewhat older than themselves, and that men prefer women who are in their mid-twenties. We also incorporate the insight that the educational system structures meeting opportunities on the marriage market. We assess the explanatory power of our model with systematic computational experiments, in which we simulate marriage market dynamics in 12 European countries among individuals born between 1921 and 2012. In these experiments, we make use of realistic agent populations in terms of educational attainment and earnings prospects and validate model outcomes with data from the European Social Survey. We demonstrate that the observed changes in educational assortative mating can be explained without any change in male or female preferences. We argue that our model provides a useful computational laboratory to explore and quantify the implications of scenarios for the future.