Understanding Recent Fertility Trends in the UK and Improving Methodologies for Fertility Forecasting

过去二十年见证了生育率水平的剧烈波动，这是人口学家或政府统计学家没有预测到的：生育率在21世纪的头十年显著上升，而此后却有所下降。这些波动对国家和地方两级的规划和决策产生重大影响。例如，2001年至2012年间的生育率增长导致英国每年新增6万多名新生儿。最近生育率变化的原因尚不清楚。一些研究人员将最近生育率水平的波动归因于生育时间的变化，即推迟或加速生育。其他人则强调人口构成变化或生育行为变化的重要性，以应对过去的政策变化和2008年后的经济衰退。国家统计局（ONS）、苏格兰国家记录局（NRS）和北爱尔兰统计研究局（NISRA）的政府统计人员使用的出生登记数据告诉我们出生总数和综合生育率；然而，它们没有提供按胎次（出生顺序）计算的生育趋势信息，这对理解和预测生育趋势至关重要。高质量的大规模纵向数据提供了对特定胎次生育率进行详细分析的机会；例如，确定生育率最近下降是由于（进一步）推迟生育和妇女无子女人数增加，还是由于母亲家庭人数减少（例如第三胎减少）。生育自然是一个循序渐进的过程；是否要再要一个孩子的决定很可能是根据以前孩子的经验来评估的。因此，按胎次对生育率进行详细分析将大大提高我们预测未来生育率的能力。在这个项目中，我们将协调来自国家统计局纵向研究、苏格兰纵向研究和北爱尔兰纵向研究的与人口普查相关的行政数据，以及来自理解社会研究的调查数据。因此，该项目的新颖之处在于，它使用了来自英国所有四个组成国家的数据；重点按出生顺序分析生育趋势，并汇集人口学和统计预测方面的专家，以发展更好的生育率预测方法。首先，我们将计算英国国家的年度特定生育率，以确定生育率水平的变化在多大程度上归因于第一，第二，第三或第四胎的变化。然后，我们将根据该人口的特征（如出生地、教育水平）调整生育率，以确定英国生育率随时间的变化在多大程度上归因于人口构成的变化，以及在多大程度上归因于生育行为的变化（可能是政策和经济环境变化的结果）。最后，我们将使用贝叶斯方法来预测英国未来的生育水平。该项目将汇集来自圣安德鲁斯大学和南安普敦大学的研究人员，以及来自国家统计局、国家统计局和NISRA的政府统计人员，共同研究一个重要的政策相关主题。该项目将极大地提高我们对英国生育率动态变化相关因素的理解，并将展示现有的大规模纵向数据集如何用于按出生顺序进行生育率的跨国分析。它还将显著改进用于英国及其组成国家的生育率预测的方法。更好地了解目前的生育趋势和预测未来的发展，对规划各种公共服务（例如托儿所、学额和房屋）的需求至关重要。

The last two decades have witnessed dramatic fluctuations in fertility levels, which were not predicted by demographers or government statisticians: Fertility significantly increased in the first decade of the 21st century, whereas it has declined thereafter. These fluctuations have significant implications for planning and policy making, at both national and local levels. For example, the fertility increase between 2001 and 2012 led to more than 60 thousand additional births in the UK annually. The causes of the recent fertility dynamics are unclear. Some researchers attribute the recent fluctuations in fertility levels to changes in fertility timing - i.e. the postponement or acceleration of childbearing. Others emphasise the importance of changes in population composition or changes in childbearing behaviour in response to past policy changes and the post-2008 economic recession. Birth registration data used by government statisticians at the Office for National Statistics (ONS), National Records of Scotland (NRS) and Northern Ireland Statistical Research Agency (NISRA) inform us about the total number of births and aggregated fertility measures; however, they do not provide information about childbearing trends by parity (birth order), which is critical to understanding and predicting fertility trends. High-quality large-scale longitudinal data provide the opportunity to conduct a detailed analysis of parity-specific fertility; for example, to determine whether fertility has recently declined because of the (further) postponement of childbearing and increased childlessness among women or because of declining family size among mothers (e.g. fewer third births).Childbearing is naturally a sequential process; decisions on having an additional child are likely to be evaluated on the basis of experience with previous children. Detailed analysis of fertility by parity will thus significantly enhance our ability to forecast future fertility. In this project we will harmonise census-linked administrative data from the ONS Longitudinal Study, Scottish Longitudinal Study, and Northern Ireland Longitudinal Study, together with survey data from the Understanding Society study. The project is thus novel in that it uses data from the all four UK constituent countries; focuses on the analysis of childbearing trends by birth order, and brings together experts in demography and statistical forecasting to develop better methods for fertility forecasting. First, we will calculate annual parity-specific fertility rates by UK country to determine how much changes in fertility levels are attributable to the changes in first, second, third or fourth births. We will then adjust fertility rates for characteristics of that population (e.g. place of birth, educational level) to determine how much a change in fertility levels in the UK over time is attributable to changes in population composition, and how much to changes in childbearing behaviour, possibly as a result of changing policies and economic environment. Finally, we will use information on parity-specific fertility to forecast future fertility levels in the UK using Bayesian methodology.The project will bring together researchers from the Universities of St Andrews and Southampton, as well as government statisticians from ONS, NRS, and NISRA, to work on an important policy-relevant topic. The project will greatly improve our understanding of the factors associated with changing fertility dynamics in the UK and will show how existing large-scale longitudinal datasets can be used for cross-country analysis of fertility by birth order. It will also significantly improve the methodology used for fertility forecasts for the UK and its constituent countries. A better understanding of the present childbearing trends and forecast of the future developments will be critical to inform the planning of demand for various public services (e.g., nurseries, school places and housing).