Incorporating big microdata in life table construction: A hypothesis-free estimator

Estimation of the Combined Effects of Ageing and Seasonality on Mortality Risk: An Application to Spain

“…In this paper, we look more closely at another topic that has received less attention in the literature: the analysis of probabilities of surviving from fractional ages (or for fractional durations) and its interaction with seasonal mortality patterns. As Lledó et al (2019) stated, death statistics show clear intra-age and calendar patterns. Although the existence of patterns between consecutive integer ages was quickly acknowledged and internalised by statistics and the actuarial literature, considering fractional age assumptions and/or continuous survival models (Hoem, 1984;Pascariu, 2018), the prevalence of seasonality patterns in death statistics has been almost forgotten (avoided) in this literature (Richards et al, 2020, is an exception), despite it being well-documented in the demographic, epidemiological and sociological literatures (e.g.…”

“…The question is to what extend the death seasonality patterns impact on the sub‐annual distributions of mortality risks after taking into account the seasonality of other demographic vital events (birth dates and migration flows) and ageing effects. In this research, we answer a research query posed by Lledó et al (2019, pp. 144–145) who point to the importance of studying ‘the appropriateness of decomposing mortality rates by quarters or months’, taking into account the ‘age/calendar distribution of deaths’.…”

“…The more popular (annual) period‐based estimators used nowadays approximate central death rates, , after implicitly assuming (Lledó et al, 2019) hypotheses (H1)–(H3), while the less hypothesis‐demanding, but more data‐demanding, (annual) cohort‐based estimators obtain death probability estimates, , by implicitly imposing (Lledó et al, 2017; Pavía et al, 2012) hypotheses (H1) and (H2). Whatever the approach, all previous studies on mortality had one thing in common: the biometric variables were computed based on annual figures and the life tables of dying and surviving probabilities were only offered for integer ages.…”

“…The information technology revolution has made demographic microdata (births, migrations and deaths) more accessible, allowing researchers to study and incorporate the exact timing of each demographic event in the construction of death estimators. As a result of this, Lledó et al (2019) propose a new annual period-based estimator for general populations that includes the exact time of every demographic event. Their analysis tackles an interesting debate about the three assumptions mentioned above.…”

Data granularity in mid-year life table construction

“…For death rates, the numerator refers to the number of deaths with age x in year t, , while the denominator accounts for either the total number of 'person-years' at risk (e.g., Wilmoth et al 2007;Arias, 2015;INE, 2016) or the average population at risk of dying, known as mid-year population estimates (e.g., ONS, 2012). In the era of the IT revolution and the boom of big data, the approach for computing the total number of 'person-years' at risk has been extensively studied in Lledó et al (2019), while the approach for estimating mid-year figures remains under-researched.…”