The importance of research on reproductive intentions for fertility forecasting

Forecasting changes in demographic processes is one of the key tasks of demography as a science. Demographic forecasts support the validity of demographic policy targets, ensuring their feasibility or demonstrating inaccuracies in tactical and strategic planning. A number of demographic development indicators are included in the system of target indicators of national projects implemented in Russia and aimed at stopping depopulation and ensuring at least simple reproduction. In this regard, special attention is paid to stimulating fertility.

When developing demographic forecasts, it is important to take into account that the measures introduced by demographic, family, and social policies affect fertility indirectly, through demographic behavior of the population. Ideas about the future family, number of children (Kopeikina, 2006), their desired gender, the timing of pregnancies, family planning, etc. influence life trajectories, regulating, in fact, the length of intervals between marriage and first birth and between successive births, and the realized number of children. It is the behavioral nature of this fertility determinant that contributes to the lability of the population to implemented and new supportive measures for families with children, the effect of “getting used to” them and the need to monitor the characteristics of reproductive behavior for the permanent modernization of mechanisms to support and stimulate fertility, whereas the development of reproductive pro-family intentions is becoming a separate important objective on the way to saving people.

Special attention is paid to the formation of values, including family values, at the federal level. In 2022, the President of the Russian Federation signed Decree 809 “On the approval of basic principles of national policy on the preservation and strengthening of traditional Russian spiritual and moral values”, which considers traditional values as the foundation of Russian society, allowing protecting and strengthening Russia’s sovereignty. The national project “Family” includes a separate federal project “Family values and cultural infrastructure”, which is aimed, among other things, at the formation of traditional family values. Given the intention of the majority to have few children and pronounced signs of marriage devaluation (Shabunova, Kalachikova, 2024), such measures and a clear government position seem particularly relevant, and systematic monitoring of reproductive plans is one of the key tools for assessing the effectiveness of family and demographic policy aimed at an increase in fertility.

The aim of the study is to assess the possibility of using data on the reproductive intentions in predicting fertility.

Forecasting fertility in real generations in Russian and foreign studies

The fertility forecast is usually made on the basis of time series of calendar indicators (age-specific and total fertility rates) and an analysis of their determination. However, fertility rates in real generations can also be used for forecasting (in fact, they reflect the results of the reproductive behavior). This is most appropriate for long-term fertility forecasts, as well as in cases when the previous changes in fertility rates could have been significantly influenced by “timing shifts”, i.e. when due to one reason or another there are many early or, conversely, postponed births. The calendar fertility rates depend on them, whereas the realized fertility in real generations is not affected by “timing shifts”. The obtained predicted average numbers of children born in real generations are transformed into age-specific and total fertility rates.

In Russia, fertility forecasts in real generations are very rarely calculated. In 2007, specialists from the Institute of Demography of the Higher School of Economics made forecasts both on the number of children born in real generations and on the total fertility rate. However, there is no mention in the paper that one of these forecasts was calculated using the other (Zakharov et al., 2008). The forecast on the average number of children born in real generations was given in one of our works, and the forecast on the total fertility rate was calculated on its basis (Kozlova, Arkhangelskiy, 2021).

P.A. Kishenin, on the contrary, calculates forecasts on the average number of children born in real generations based on forecasts on total and agespecific fertility rates (Kishenin, 2023). Predictions for real generations based on the forecast of agespecific fertility rates for Russia and the Ozersk atomgrad (atom-city) were carried out by V.I. Telnov (Telnov, 2014; Telnov, 2021).

Among the foreign works on forecasting fertility rates in real generations, the report of T. Sobotka, K. Zeman, R. Lesthaeghe and T. Frejka (Sobotka et al., 2011) should be noted.

Unlike forecasts on total and age-specific fertility rates, if the fertility forecast is based on fertility rates of real generations, research data on reproductive intentions can be used.

The possibilities of use and degree of consistency of data on reproductive intentions for making fertility forecasts have not yet been studied sufficiently. First of all, of course, we note the paper by E.M. Andreev and G.A. Bondarskaya. The authors used data of generations of married and all women on the average expected number of children and the number of children born, according to the surveys conducted by the Demography Department of the Central Statistical Administration of the USSR since 1967, the 1985 and 1994 microcensuses, as well as on the number of children born, according to the 1979 and 1989 censuses. Findings of the study on married women show that “on average, women perform their reproductive intentions quite accurately”. However, the authors note that “for young cohorts, information about the expected number of children can be used in forecasting with great caution” (Andreev, Bondarskaya, 2000).

Foreign works containing the findings on reproductive intentions themselves and the influence of “uncertainty” factors on them are quite numerous. Depending on the field of research, reproductive intentions are measured considering either plans to have a certain number of children, or the ideal parity. Among foreign studies on the predictive validity of reproductive intentions, we primarily underline the work (Hendershot, Plaek, 1985), as well as the papers (Westoff, Ryder, 1977; Toulemon, Testa, 2005; Beaujouan, Toulemon, 2013).

Scholars actively discuss the role of socioeconomic and demographic factors in shaping reproductive intentions, which should be taken into account when predicting fertility. The most common studies are those assessing the impact of education, financial security and the economic status of both women and men on reproductive intentions (Norling, 2022). The role of contraception availability, on the one hand, and reproductive technologies, on the other, is also analyzed (Alazbih et al., 2017).

There are works based on a multidimensional analysis of genealogical data, the authors of which use intergenerational approach to identifying the desired or ideal parity. The results of these studies confirm the high probability of transmission of reproductive behavior, as well as fertility-related life-course events, across generations (Anderton et al., 1987).

Modern publications suggest a new (it can be called psychological) approach to the study of the relationship between economic uncertainty and fertility using the Narrative Framework concept, suggesting that people act “according to or despite uncertainty”, linking reproductive intentions with their imagined futures created under the influence of their social environment (parents, other relatives, friends), which is to some extent related to an intergenerational approach (Vignoli et al., 2020).

Here we can also mention the results of a study based on binary longitudinal data on couples who participated in a panel analysis of the changes in their relationship over a year. Given a favorable emotional background in the couple’s relationship, the authors assessed the consistency between fertility expectations and intentions as high (Heiland et al., 2005).

To conclude, we note that, despite all the diversity, most foreign studies analyzing the predictive validity of reproductive intentions do not contain reliable methods for predicting fertility. As researchers admit, currently available prediction models are far from ideal and there is an urgent need to develop the “design” of such studies (Maheshwari et al., 2008). In this regard, we decided to evaluate the predictive validity of data on the expected number of children by comparing the intended and actual parity of women in Russia based on representative and relevant information sources.

Methodology and information base of the study

Our approach is to assess the degree of realization of reproductive intentions, recorded by statistics on births. The identification of a gap between intentions and performance (realized number of children) will become the basis for calculating correction factors that will allow assuming realized fertility in real generations of women. To do this, we will determine for each generation the average number of children born, the average expected number of children, and calculate their difference. It will show the degree of realization of reproductive intentions, and its value will measure the predictive validity of this empirical indicator. In addition, we will see intergenerational differences in reproductive intentions and the degree of their realization. Next, we will determine the average number of children born in a specific time period (2013–2023) and compare it with the reproductive intentions for this period, taking into account the data on the intention to have more children. In the sample survey of reproductive plans in 2012, a question was asked about the expected number of children (“How many children (including those you have) do you intend to have?”) and the number of children born. The difference between them can be interpreted as the “expected number of more children”. The data from population surveys allow assessing each respondent and calculating the average aggregate value for women of one generation.

There are changes in reproductive behavior at different parities which influence fertility rates and make it necessary to predict fertility rates differentially by parity. This is substantiated by the objective “annual growth in the total fertility rate in terms of the third and subsequent children” included in Presidential Decree 309, dated July 5, 2024 “On the national development goals of the Russian Federation for the period up to 2030 and for the future up to 2036”. To this end, the answers to the question of the intention to have another child can be used differentially by parity. The corresponding question was not included in the micro-census in 2015, but it is used in the selective monitoring of reproductive intentions by Rosstat. Such a study was carried out in 2012, 2017 and 2022. The necessary microdata of the 2017 survey are not available, so we will appeal to the 2012 survey, which was conducted in only 30 regions. In this regard, a correct comparison of intentions to have another child by women cohorts based on the 2012 survey with the average number of children born in 2013– 2023 in real generations by parity is possible only for individual regions or for a group of regions.

For individual regions, the number of respondents surveyed in 2012 is relatively small and random fluctuations in indicators may occur, so it is advisable to use data for a group of regions. The choice of regions for the group is determined by the following criteria:

– participation in the selective monitoring of reproductive intentions in 2012;

– accuracy of calculated fertility rates in real generations based on one-year age-specific fertility rates for a given region (accuracy degree was estimated earlier in one of our studies by comparing estimated and actual (according to the 2002 and 2010 censuses) average number of children born in real generations of women) (Arkhangelskiy, 2016);

– availability of annual age-specific fertility rates by parity for 2013–2023.

Taking into account these criteria, a comparative assessment of reproductive intentions of real generations by parity (according to the data of the 2012 sample survey of reproductive intentions) and the average number of children born over the period 2013–2023 can be made for a group that includes 20 regions: republics of Bashkortostan, Buryatia, Komi and Khakassia; Amur, Astrakhan, Belgorod, Vologda, Kaliningrad, Kaluga, Kostroma, Orenburg, Rostov, Ryazan, Samara, Sverdlovsk, Smolensk, Tver, Ulyanovsk and Chelyabinsk regions.

There are 3179 women surveyed in 20 regions born in 1967–1994. Based on the objectives of the study, it is advisable to consider the generations of women born in 1970–1994 (2911 people).

In 2023, women born in 1994 turned 29, women born in 1967 turned 56, which is significantly higher than the limit of reproductive age, while women born in 1970 turned 53 (49 years – born in 1974).

In other words, the survey allows assessing the similarity of reproductive intentions and realized fertility of generations older than 1974, identifying relevant correction factors and, on their basis at the next stage of the study, assuming realized fertility for childbearing-aged women over 29.

Findings

Let us consider the average number of children born at the time of the survey (average parity), the average expected number of children and the difference between them (“total” fertility potential), the average number of children born in the period after the survey (2013–2023) and the difference between the fertility potential (the “expected number of more children”) and fertility realized in 2013–2023. The calculation was performed for women of each year of birth (from 1970 to 1994) and for five five-year groups (Tab. 1).

It is advisable to compare the average expected number of more children with the number of children born in the period after this survey. The average expected number of more children, according to the 2012 sample survey of reproductive intentions, is about 0.3 more than the actual number of children born in the period 2013–2023.

Of course, it should be mentioned that, unlike the 2015 micro-census, the sample of this survey is small and this may affect data representativeness. But if it caused inconsistency in the results, the ratio of the average expected number of more children and the average number of children born in 2013– 2023 could differ significantly across women cohorts.

Таблица 1. Reproductive intentions of women, according to the 2012 sample survey of reproductive intentions, and the average number of children born among 1970–1994 women cohorts in 2013–2023

Women’s year of birth	Number of respondents	Average parity	Average expected number of children	Difference between the average expected number of children and average parity (“average expected number of more children”)	Average number of children born in 2013–2023	Difference between the “average expected number of more children” and the average number of children born in 2013–2023
1970	108	1.41	1.62	0.21	0.01	0.20
1971	114	1.61	1.85	0.24	0.01	0.23
1972	118	1.59	1.72	0.13	0.02	0.11
1973	115	1.69	1.90	0.21	0.04	0.17
1974	117	1.62	1.90	0.28	0.06	0.22
1975	150	1.51	1.87	0.36	0.09	0.27
1976	155	1.48	1.89	0.41	0.13	0.28
1977	139	1.53	1.98	0.45	0.18	0.27
1978	115	1.57	2.03	0.46	0.24	0.22
1979	122	1.42	1.92	0.50	0.30	0.20
1980	123	1.35	2.00	0.65	0.36	0.29
1981	111	1.40	2.07	0.67	0.44	0.23
1982	142	1.18	2.04	0.86	0.52	0.34
1983	142	1.15	2.02	0.87	0.62	0.25
1984	118	0.93	2.01	1.08	0.70	0.38
1985	131	0.94	1.92	0.98	0.76	0.22
1986	130	0.98	1.97	0.99	0.83	0.16
1987	160	0.74	1.93	1.19	0.93	0.26
1988	137	0.66	2.02	1.36	0.97	0.39
1989	101	0.55	2.00	1.45	0.99	0.46
1990	99	0.37	1.91	1.54	0.99	0.55
1991	89	0.24	1.99	1.75	1.01	0.74
1992	81	0.17	1.86	1.69	1.01	0.68
1993	52	0.06	1.96	1.90	0.98	0.92
1994	42	0.07	1.89	1.82	0.96	0.86
1970–1974	572	1.58	1.80	0.22	0.03	0.19
1975–1979	681	1.50	1.93	0.43	0.18	0.25
1980–1984	636	1.20	2.03	0.83	0.53	0.30
1985–1989	659	0.78	1.96	1.18	0.89	0.29
1990–1994	363	0.21	1.92	1.71	0.99	0.72

Considering the generations of women born in 1970–1987, only in three of them the discrepancy between these indicators was less than 0.2 and only in two of them it was more than 0.3 (Tab. 1). In other words, we can say that this difference between generations of women is sufficiently stable. This is very important when evaluating the predictive validity of data on reproductive intentions. The discrepancy is greater in younger generations, as they are more likely to have more children and, consequently, to reduce the gap between the expected number of children (according to the 2012 survey) and the realized number of children.

Aggregating the data for five-year groups of women, we see that in the generations that are above the childbearing age, the discrepancy between the actual number of children born in 2013–2023 and the expected number was less than 0.19, so we note that the differences are relatively low for these generations of women. For generations aged 44–49 for the analysis period (the last 5 years of reproductive age), the discrepancy is higher – 0.25, for women aged 39–44 (born in 1980–1984) – 0.3, for women aged 34–39 – 0.29, and for women aged 29–34 – 0.72.

As already noted, the data from the 2012 sample survey of reproductive intentions allow comparing reproductive intentions and the realized number of children, differentiated by parity. To do this, the answers to the question “Do you intend to have a child (the first one if you do not have children, or another one)?” are used. As evidence of the intention to have another baby we considered the answers: “I am already pregnant”, “yes, in the near future”, “yes, but a little later, we are postponing for now”.

The number of such responses was differentiated by parity: for those at zero parity – as an intention to have the first child, for those at one parity – as an intention to have the second child, for those at two parity – as an intention to have the third child. However, when comparing with the average number of children born by corresponding parity in 2013–2023, it is correct to consider the share of such responses among all surveyed women of a given generation, rather than among those at particular parity, since the one-year age-specific fertility rates by parity, which is used to calculate the average number of births by corresponding parity, are calculated aggregately for all women of a given age, and not for particular parity.

The share of women intending to have their first child among all surveyed women of this generation is slightly higher than the average number of first births in 2013–2023. In the 1970– 1974, 1975–1979, and 1980–1984 generations of women, the discrepancy is 0.02–0.04, so we note answering the question about the intention to have the first child as having fairly good predictive validity ( Tab. 2 ). For women born in 1990–1994, the discrepancy is greater (0.08), but in 2023 they were 29–33 years old, and some of them may still have their first child.

The situation is somewhat different for second births ( Tab. 3 ). For the 1970–1974, 1975–1979, and 1980–1984 generations, the number of women intending to have a second child is slightly (by 0.04– 0.07) higher than the average number of second births in 2013–2023, according to the 2012 sample survey of reproductive intentions. We think that it is the evidence of certain predictive validity of answers to the question about the intention to have a second child. In the 1985–1989 and 1990–1994 generations, the average number of second births in 2013–2023 is higher than the share of those planning to have a second child, revealed by the data of the 2012 sample survey of reproductive intentions (by 0.09 and 0.25, respectively). But, apparently, it would be wrong to talk about the “overfulfilling” of reproductive plans in these generations. Given their age in 2013–2023 and the length of this period, it can be assumed that some of the women, who had

Table 2. The share of women intending to have their first child, according to the 2012 sample survey of reproductive intentions, and the average number of first births among 1970–1994 women cohorts in 2013–2023

Women’s years of birth	Number of respondents	Number of childless women	Number of women intending to have a first child (childless, whose answers to the question about the intention to have a child was: “I am already pregnant”, “yes, in the near future”, “yes, but a little later, we are postponing for now”)	The share of women intending to have their first child		Average number of first births in 2013– 2023	Gap between the share of women intending to have their first child among respondents and the average number of first births in 2013–2023
				among childless women	among respondents
1970–1974	572	48	15	0.31	0.03	0.00	0.03
1975–1979	681	73	44	0.60	0.06	0.02	0.04
1980–1984	636	116	88	0.76	0.14	0.10	0.04
1985–1989	659	260	199	0.77	0.30	0.28	0.02
1990–1994	363	296	205	0.69	0.56	0.48	0.08

Table 3. The share of women intending to have a second child, according to the 2012 sample survey of reproductive intentions, and the average number of second births in the 1970–1994 women cohorts in 2013–2023

Women’s years of birth Number of respondents Number of women at one parity Number of women intending to have a second child (childless, whose answers to the question about the intention to have a child was: “I am already pregnant”, “yes, in the near future”, “yes, but a little later, we are postponing for now”) The share of women intending to have their second child Average number of second births in 2013–2023 Gap between the share of women intending to have their second child among respondents and the average number of second births in 2013–2023 among women at one parity among respondents 1970–1974 572 223 30 0.13 0.05 0.01 0.04 1975–1979 681 279 96 0.34 0.14 0.07 0.07 1980–1984 636 319 198 0.62 0.31 0.25 0.06 1985–1989 659 291 202 0.69 0.31 0.40 -0.09 1990–1994 363 57 41 0.72 0.11 0.36 -0.25 not yet had a single child at the time of the 2012 survey, had their first and second children in 2013– 2023. Therefore, they could not get into the share of those who were intending to have a second child, since the question was asked about the intention to have only one more child.

In this regard, it may be advisable to reduce the time period for calculation of second births from 11 to 5 years, namely to use the period 2013–2017 to consider second births only (Tab. 4). In this case, the share of those intending to have a second child is less than the average number of second births only in the 1990–1994 generation. It should be noted that this small generation was born in difficult socio-economic conditions and has its own sociocultural peculiarities.

The situation is similar for third births. The average number of third births in 2013–2023 is 0.04 less than the share of those intending to have a third child in the 1970–1974 generation.

For women born in 1975–1979, these indicators coincide. In younger generations, the average number of third births in 2013–2023 is higher than the share of those intending to have a third child, as revealed by the 2012 sample survey of reproductive intentions. Moreover, the younger the generation, the higher the discrepancy between the indicators: for 1980–1984 cohort it is 0.03; for 1985–1989 cohort – 0.08; for 1990–1994 cohort – 0.10 (Tab. 5).

Table 4. The share of women intending to have a second child, according to the 2012 sample survey of reproductive intentions, and the average number of second births in the 1970–1994 women cohorts in 2013–2017

Women’s years of birth	Number of respondents	Number of women at one parity	Number of women intending to have a second child (childless, whose answers to the question about the intention to have a child was: “I am already pregnant”, “yes, in the near future”, “yes, but a little later, we are postponing for now”)	The share of women intending to have their second child		Average number of second births in 2013–2017	Gap between the share of women intending to have their second child among respondents and the average number of second births in 2013– 2017
				among women at one parity	among respondents
1970–1974	572	223	30	0.13	0.05	0.01	0.04
1975–1979	681	279	96	0.34	0.14	0.06	0.08
1980–1984	636	319	198	0.62	0.31	0.18	0.13
1985–1989	659	291	202	0.69	0.31	0.25	0.06
1990–1994	363	57	41	0.72	0.11	0.15	-0.04

Table 5. The share of women intending to have a third child, according to the 2012 sample survey of reproductive intentions, and the average number of third births in the 1970–1994 women cohorts in 2013–2023

Women’s years of birth	Number of respondents	Number of women at two parity	Number of women intending to have a third child (childless, whose answers to the question about the intention to have a child was: “I am already pregnant”, “yes, in the near future”, “yes, but a little later, we are postponing for now”)	The share of women intending to have their third child		Average number of third births in 2013–2023	Gap between the share of women intending to have their third child among respondents and the average number of third births in 2013–2023
				among women at two parity	among respondents
1970–1974	572	239	28	0.12	0.05	0.01	0.04
1975–1979	681	260	34	0.13	0.05	0.05	0.00
1980–1984	636	162	65	0.40	0.10	0.13	-0.03
1985–1989	659	100	44	0.44	0.07	0.15	-0.08
1990–1994	363	9	4	0.44	0.01	0.11	-0.10

Table 6. The share of women intending to have a third child, according to the 2012 sample survey of reproductive intentions, and the average number of third births in the 1970–1994 women cohorts in 2013–2017

Women’s years of birth	Number of respondents	Number of women at two parity	Number of women intending to have a third child (childless, whose answers to the question about the intention to have a child was: “I am already pregnant”, “yes, in the near future”, “yes, but a little later, we are postponing for now”)	The share of women intending to have their third child		Average number of third births in 2013– 2017	Gap between the share of women intending to have their third child among respondents and the average number of third births in 2013–2017
				among women at two parity	among respondents
1970–1974	572	239	28	0.12	0.05	0.01	0.04
1975–1979	681	260	34	0.13	0.05	0.04	0.01
1980–1984	636	162	65	0.40	0.10	0.07	0.03
1985–1989	659	100	44	0.44	0.07	0.06	0.01
1990–1994	363	9	4	0.44	0.01	0.03	-0.02

If we compare the average number of third births with the share of those intending to have a third child not for 2013–2023, but for the five-year period (2013–2017), then this share is slightly higher than the average number of third births in all generations, except for women of the 1990– 1994 cohort ( Tab. 6 ).

Therefore, the duration of the analysis period used for forecasting is important for second and subsequent births. However, the predictive validity of reproductive intentions of women aged over 35 is fairly high.

Conclusion

Finding relevant options for predicting fertility inevitably leads to the need to understand the mechanisms of shaping reproductive behavior, family size preferences and the conditions for their realization. In turn, this requires monitoring of reproductive intentions using sociological methods, which is implemented by the Federal State Statistics Service through regular surveys of reproductive intentions. A number of organizations carry out similar research using similar and our methodology. Earlier, the co-authors of this paper analyzed estimates of Russians’ reproductive intentions (Arkhangelskiy, Kalachikova, 2021). This indicates that there is a certain sociological database on reproductive behavior, which can be used to predict fertility. Regional studies are especially relevant in constituent entities with a pronounced ethno-cultural identity, such as the republics of the North Caucasus. However, in order to obtain up-to-date and complete information, it is necessary to continue regular monitoring of reproductive intentions.

One of the previous studies has already included the comparison between the average expected number of children among women aged 35–39 and 40–44 according to the 2015 micro-census and the average number of children born among women aged 40–44 and 45–49, respectively, according to the 2020 census (Arkhangelskiy et al., 2024). The discrepancy between the average expected number of children among women aged 40–44, according to the 2015 micro-census, and the average number of children born among women aged 45–49, according to the 2020 census, is 0.11, and the difference between these indicators among women aged 35– 39 and 40–44, respectively, is 0.13. These results already suggest that, with appropriate correction, data on the expected number of children can be used in predicting fertility.

The results of this paper indicate that, considering the correction factors, data on reproductive intentions reflect fertility in real generations. An analysis of data on the realization of expected parity in real generations of women showed that in the generations above the childbearing age, the gap between the actual fertility in 2013–2023 and the “expected number of more children” was less than 0.19, so for these women cohorts, the discrepancies are stable and relatively low. For generations of women aged 44–49 during the analysis period (the last 5 years of childbearing age), the difference is higher – 0.25, for women aged 39–44 (born in 1980–1984) – 0.3, for women aged 34–39 – 0.29, and for women aged 29–34 it is 0.72. This fact alone suggests that the reproductive intentions of the youngest group of women are still far from being realized (at least judging by the observed difference between 0.72 and 0.30). We understand that reproductive intentions of different generations were developed in different conditions, especially for those who were socialized in different countries and under different political systems, but in the analyzed period they are realized in one social, political, and economic environment.

Predictive validity of intentions to have second and third children shows even higher consistency, considering that the period for fertility realization is cut to exclude the interval between marriage and first birth.

The novelty of the paper lies in the empirical assessment of the degree of realization of reproductive intentions in real generations of women, including by parity. Correction factors have been determined, which will allow predicting cohort fertility and calculating the total fertility rate. It is assumed that the influence of living conditions on fertility is reflected in the discrepancy between intentions and realized family size. Accordingly, the higher it is, the stronger the impact of external conditions on fertility. On the other hand, a lower gap between intended and actual fertility indicates subjectively more favorable conditions for having family, including due to certain demographic policy measures.

At the next stage of the study, we plan to test the revealed patterns, develop correction factors and calculate fertility forecasts, considering the reproductive intentions of women in Russia.