Evolving Fertility Goals and Behaviors in Current U.S. Childbearing Cohorts

Abstract

In the post-Recession era, U.S. fertility rates have continued to fall. It is unclear if these declines are driven by shifts in fertility goals or growing difficulty in achieving goals. In this paper, we construct synthetic cohorts of men and women to examine both cross-cohort and within-cohort changes in fertility goals using multiple cycles of the National Survey of Family Growth. Although more recent cohorts exhibit lower achieved fertility at younger ages than earlier cohorts at the same age, intended parity remains around two children, and intentions to remain childless rarely exceed 15%. There is weak evidence of a growing fertility gap in the early 30s, suggesting more recent cohorts will need considerable childbearing in the 30s and early 40s to ‘catch up’ to earlier goals, yet low-parity women in their early 40s are decreasingly likely to have unfulfilled fertility desires or intentions to have children. Low-parity men in their early 40s, though, are increasingly likely to intend children. Declines in U.S. fertility thus seem to be largely driven not by changes in early-life fertility goals so much as either a decreasing likelihood of achieving earlier goals or, perhaps, shifts in the preferred timing of fertility that depress period measures.

During the Great Recession, birth rates in the United States fell – a typical demographic response to economic downturn (Cherlin, Cumberworth, Morgan, and Wimer 2013; Örsal and Goldstein 2018; Sobotka, Skirbekk, and Philipov 2011). After the Great Recession, birth rates did not fully rebound and, in fact, continued to decline. In 2019, the total fertility rate (TFR) in the United States was 1.71, the lowest level observed since the 1970s (Martin et al. 2021). These relatively low birth rates have sparked considerable concern in the public sphere (e.g., Carroll and Charles 2019; Cha 2017, 2018; Keshner 2019), and the lack of a full recovery after the Great Recession has also prompted curiosity in the scholarly community (Buckles, Guldi, and Schmidt 2020; Kearney, Levine, and Pardue 2022).

Some of the decline in observed period fertility likely reflects postponement of childbearing among younger women, including reductions in early unintended childbearing, that may eventually be recuperated at older ages (Buckles, Guldi, and Schmidt 2020; Finer and Zolna 2016). Thus, some research suggests that eventual cohort completed fertility will remain around two children per woman (Wu and Mark 2019). However, there is also some evidence that individuals of childbearing age intend slightly fewer children than they did just a decade earlier (Hartnett and Gemmill 2020). Further, delayed childbearing can evolve into foregone childbearing as individuals move through the life course and contend with competing obligations or sub-fecundity (Beaujouan and Berghammer 2019; Gemmill 2019; Hayford 2009, 2013). It is possible, then, that postponed childbearing will not be recuperated and that current cohorts will finish their childbearing years with fewer children than they (and demographers) anticipate.

Tracking the evolution of childbearing goals (a broad term encompassing preferences, desires, and intentions) and achieved childbearing over the life course is challenging empirically. Rich data from longstanding panel surveys in the U.S. and the U.K. have generated many comprehensive studies of these processes, but these data are only available for a few cohorts. There is a paucity of longitudinal data describing fertility among cohorts currently of childbearing age or how they compare to earlier cohorts. In this article, we take a different approach than prior work by leveraging multiple cycles of the National Survey of Family Growth (NSFG), the most comprehensive survey of U.S. fertility goals and behaviors. The NSFG is an ongoing repeated cross-sectional survey that has collected nationally representative data since the 1970s. Because the NSFG represents the U.S. population, the repeated cross-sections can be used to create and follow synthetic cohorts over time as they age and to track the aggregate-level fertility characteristics of these cohorts (cf. Parrado 2011).

We use NSFG data collected in 1988, 1995, 2002, 2006-10, 2011-15, and 2015-19 to trace fertility goals and behaviors for multiple cohorts of men and women, with particular attention to intended and actual parity as well as intended and actual childlessness. We compare cohorts currently in their childbearing years to earlier cohorts who have completed childbearing, and we follow cohorts as they pass through the reproductive life course. We examine the gap between early-life goals and later achieved fertility as well as unfulfilled fertility desires at the end of the reproductive years. We seek to understand post-Recession fertility trends in the United States but also to contextualize these trends in longer-term patterns and to contribute to knowledge on life-course patterns in fertility goals. We reflect on the implications of our results for understanding current low fertility and future fertility trends, particularly in light of the ongoing COVID-19 pandemic.

Long-term trends in fertility in the United States

Following World War II, the United States experienced a long and sustained Baby Boom, with elevated fertility rates spanning a period of nearly two decades. After the Baby Boom, fertility declined, returning to a level more consistent with the birth rates observed in the early 20^th century. Like the Baby Boom itself, the post-Boom decline was largely unexpected by population scientists at the time (Van Bavel and Reher 2013). The U.S. TFR reached its Baby Boom peak of 3.58 in 1958, then fell sharply through the 1970s, reaching a low of 1.77 in 1972. Birth rates rose slowly in the 1970s and 1980s, then stabilized (Morgan 1996); between 1990 and 2007, the TFR remained virtually unchanged at just over 2 children per woman (Hamilton and Kirmeyer 2017).

Following this period of aggregate stability in the TFR, the Great Recession ushered in another set of changes. For the period 2007-2011 (covering the two years marking the “official” recession and two recovery years), the period TFR fell 11% (2.12 to 1.89), and the general fertility rate, the number of births per 1000 women age 15-44, fell 9% (Cherlin et al., 2013). This demographic response to economic crisis was expected, given theory and previous research about the impact of economic conditions on childbearing, and was widely observed across the United States and Europe (Comolli 2017). In the U.S., recession-related declines in fertility were largest among younger women, particularly those in their teens and early 20s; declines were smaller for women in their 30s, and birth rates actually increased during the recession period for women in their 40s (Cherlin et al. 2013). These patterns represent an acceleration of ongoing trends in birth timing and are consistent with the idea that people were postponing births during the Recession that they intended to have at a later point.

After the Great Recession, although overall unemployment rates and other economic indicators returned to pre-Recession levels, birth rates did not fully recover and, in fact, have continued to fall (Martin et al. 2021; Seltzer 2019). The ongoing decline suggests that current low birth rates in the United States may be the product of sustained structural and social conditions, whether economic systems, institutional and policy characteristics related to (the lack of) support for families, or shifting values and preferences for childbearing. (It is worth noting, though, that the U.S. is not unique in failing to experience a post-Recession fertility recovery (Vignoli et al. 2020), and thus explanations for fertility trends need not be specific to the United States.) It is also possible that the post-Recession stagnation in birth rates results from a combination of delays in full economic recovery across sub-groups (particularly young adults) and demographic artifacts of changing birth timing.

Life course patterns of intended and achieved fertility

Declines in birth rates during and after the Great Recession may simply result from temporary postponement of childbearing. Period trends in fertility reflect short-term responses to economic and social fluctuations and may not correspond to trends in completed cohort fertility. Generally, completed cohort fertility is less responsive to short-term conditions than period TFR, and measures of cumulative childbearing over the life course typically show smoother trend lines than period measures (Sobotka 2017; see, e.g., Myrskylä, Goldstein, and Cheng (2013) for cohort trends vs. Rindfuss, Choe, and Brauner-Otto (2016) for period trends). If people postpone births during periods of economic downtown and then go on to have those births at later time points, period TFRs during the years when postponement is taking place will systematically underestimate cohort completed fertility (Bongaarts and Feeney 1998).

The degree to which period TFRs underestimate cohort fertility – and the long-term impact of economic recessions on cohort fertility – depends on whether births postponed early in the life course are recuperated at later points in the childbearing years. For some people, delaying births may mean that they reach an age where it is no longer physically possible (or socially desirable) to have children. Other people may change their minds about wanting children after repeated delays in childbearing. The question of whether and when births delayed during the Great Recession will take place is central to understanding the impact of the Great Recession and the future of fertility in the U.S. Since 2010, birth rates to women in their teens and twenties have continued to decline, while birth rates to women age 35 and older increased (Martin et al. 2021). The combined impact of these trends for completed cohort fertility is not yet clear. For women who were in their 30s and 40s during the Great Recession, completed cohort fertility is, if anything, slightly higher than for women who completed childbearing before the Recession (Human Fertility Database n.d.). Women who were in their twenties or younger during the Recession – those who experienced the largest impact on period fertility – are still in their childbearing years. The Great Recession occurred during a period in which the transition to adulthood was already becoming longer and more difficult (Settersten and Ray 2010), and, during the Recession and recovery, economic pressures intensified as the social safety net weakened and labor markets became increasingly polarized (Cooper 2014; Seltzer, 2019; Silva 2013; Vespa 2017). This is the context in which more recent cohorts of young adults entered their childbearing years. Understanding their early life-course fertility goals, and the degree to which these goals evolve as they aged, is crucial for understanding future fertility trends.

A full understanding of how fertility goals and behaviors evolve over the life course requires longitudinal survey data. Surveys with adequate breadth and depth are rare, but the few existing datasets with the necessary longitudinal measures have been used to describe fertility goals (expectations, intentions, or desires, depending on the survey and the study) and achieved fertility, analyzing changes in fertility goals over the life course and the degree to which early-life goals are realized. In the United States, this research primarily draws on data from the National Longitudinal Survey of Youth-1979 (NLSY79) cohort, a longitudinal survey of people age 14-22 in 1979 who have been re-interviewed annually or at two-year intervals since the initial survey. A small but robust body of literature has identified key features of the childbearing career for these cohorts. Early in the life course, both the average desired family size and the average expected family size have consistently been between two and three children, for both men and women (Morgan and Rackin 2010; Musick et al. 2009; Nitsche and Hayford 2020; Quesnel-Vallée and Morgan 2003). The average completed fertility (measured in the early 40s) in the NLSY79 cohort is just under two children (Morgan and Rackin 2010). In the aggregate, achieved parity in the early 40s is slightly below expected parity for this cohort– by about one quarter of a child among women and about half a child among men (Morgan and Rackin 2010). This “fertility gap” between goals early in the life course and achieved fertility at the end of the reproductive years has been noted in other countries as well, at least for the cohorts born in the early 1970s (Beaujouan and Berghammer 2019).

At the individual level, there is substantial mismatch in both directions; by age 45, about 40% of NLSY79 respondents had fewer children than they expected at age 24, and about 20% had more than they expected (Morgan and Rackin 2010). In particular, among women who remain childless at the end of the childbearing years, most begin by expecting or desiring children earlier in their childbearing years (Gemmill 2019; Gray, Evans, and Reimondos 2013; Rybińska and Morgan 2019). Some of the gap between observed preferences early in the life course and achieved fertility at the end of the childbearing years is attributable to changes in preferences as people gain new knowledge and experiences. Still, some men and women who are approaching the end of their childbearing years continue to desire children, and some still intend to have children (Casterline and Han 2017; Guzzo 2018, 2021).

Most of what we know about the life course dynamics of fertility goals in the United States draws from a single cohort study – the NLSY79 – and, with a few exceptions, focuses on women. Respondents in this study were age 14-22 in 1979 and were in their peak childbearing ages in the 1980s and 1990s. Sociodemographic changes in the ensuing decades have altered the context for childbearing – for example, college enrollment rates have increased, cohabitation has become more widely accepted, and new contraceptive technologies have been developed and popularized. Marriage rates have declined, and female labor force participation has increased. Thus, some elements of the NLSY79 cohort’s childbearing experiences may not generalize to more recent cohorts. The Great Recession and its uneven recovery may also have changed the way individuals form fertility goals and the degree to which they can achieve these goals (Vignoli et al. 2020). Further, men’s fertility goals may differ – or may have differentially changed over time – relative to women. For instance, men were hit harder during the Great Recession (but also recovered more, e.g., Kalleberg and Von Wachter 2017), and women now outpace men in college completion (Pew 2021). In general, men and women in the same population appear to exhibit differences in fertility timing and quantum (Dudel and Klüsener 2021), suggesting that there may be differences in fertility goals as well.

The present study

Our overarching aim in this article is to describe trends in childbearing goals in the United States and in how fertility goals and behavior evolve over the life course. This aim is motivated in part by the continued declines in fertility rates since the Great Recession in the U.S. and the effort to understand the contribution of both changes in quantum goals and changes in the achievement of goals to these declines. We also add to the literature on life course fertility patterns by comparing multiple cohorts over time, moving beyond existing research that has largely focused on the NLSY79 cohort, and by analyzing both men and women. We use repeated cross-sectional surveys to observe synthetic cohorts in the absence of existing longitudinal data covering the most recent cohorts.

In our analysis and presentation of results, we begin by illustrating long-term patterns and then focus more specifically on the post-Recession period. We describe trends and patterns for both men and women, analyzing them separately because of differences in how data are collected for men and women and because previous research suggests that men and women may have different fertility goals and outcomes (Nitsche and Hayford 2020; Quesnel-Vallée and Morgan 2003; Rybińska 2021). Because of the differences in data collection, and because previous research has found limitations both in the sample frames for data on men’s fertility and on the reporting of births by men (Joyner et al. 2012), we exercise caution in drawing direct comparisons between men and women.

We address the following research questions.

RQ1a. Have early life-course goals for childbearing changed across cohorts?

Given the stability of the two-child norm, we expect limited changes in goals for childbearing measured early in the life course, with potentially modest declines in average intended parity and modest increases in the proportion of young adults intending to be childless but with most people intending to have children and average intended family sizes largely stable around two children.

RQ1b. Have there been larger changes in early life-course goals for childbearing in the post-Recession period?

It is possible that changing economic conditions and social norms have made the transition to adulthood and parenting more difficult since the Recession and that as a result young people have become less oriented toward childbearing. We examine both intended parity and intended childlessness to assess this possibility.

RQ2a. Have life-course patterns of evolution in childbearing goals changed across cohorts?

Here, previous research suggests multiple possible outcomes. On the one hand, growing social acceptance of childlessness and small families may mean that intended parity is declining earlier in the life course among more recent cohorts than among earlier cohorts. On the other hand, increased postponement of childbearing may imply that people retain their goals for childbearing at later ages more in recent cohorts than in earlier cohorts.

RQ2b: Have childbearing goals declined more with age for cohorts in the childbearing ages during and after the Great Recession than cohorts in the childbearing age before the Great Recession?

A possible response to unfavorable conditions for childbearing in the post-Recession period is to reduce goals for childbearing. This response might be particularly likely if people experience competing demands from other life course trajectories (e.g., more fulfilling career options), but could also result from a perception that conditions for childbearing are unlikely to improve. This change would result in different age schedules of fertility goals, with people potentially reducing goals earlier in the life course or reducing goals to lower levels relative to earlier cohorts.

RQ3a: Has the “fertility gap” at older ages changed over time?

As people increasingly postpone childbearing, it may become more difficult to meet early-life goals for childbearing because of biological challenges to fecundity. People may also find it more difficult to achieve conditions they see as necessary for parenting, such as financial security or stable partnerships. In response to these difficulties, people may change their goals for childbearing, as suggested in the question above, or they may reach the end of their childbearing years with unfulfilled fertility goals. This “fertility gap” can be conceptualized and measured in different ways. One way is to consider the gap at the macro level by comparing aggregate completed cohort fertility at later points of the childbearing years to aggregate early-life intentions. This aggregate measure captures the combined impact of both changing goals and cumulative behavior. Another way is to consider fertility desires relative to intentions at the micro level. When focusing on low-parity individuals at the end of the childbearing years, looking at the difference between whether men and women would have a child if they could vs whether they actually plan on doing so gives some insight into the extent to which people have downgraded intended parity vs are continuing to delay (i.e., whether delayed fertility/delayed fertility preferences are underlying factors contributing to cohort shifts).

RQ3b: Is the fertility gap at older ages larger after the Great Recession than before the Great Recession?

Cohorts who experienced the Great Recession during their childbearing years may have begun their lives with similar goals as previous cohorts but may have more difficulty fulfilling these goals due to unfavorable economic contexts, limited policy support, or other challenges. If this is the case, low-parity men and women cohorts will exhibit growth in the share wanting but not intending to have a child in the future. Conversely, if low-parity men and women are increasingly likely to have changed both desires and intentions with more exposure to post-Recession conditions (i.e., across cohorts), then a greater share of these individuals would have fertility desires and goals that are aligned – either neither wanting nor intending a child or both wanting and intending to have a child.

Data and approach

The National Surveys of Family Growth (NSFG) collect full birth histories and detailed measures of prospective fertility goals, including fertility desires and fertility intentions, from a nationally representative sample. From 1973 through 1995, the survey included female respondents age 15-44, with male respondents added in the 2002 cycle and individuals up to age 49 added in 2015; the NSFG refers to female respondents/women and male respondents/men without other measures or elaboration of gender identity; we follow that convention. Beginning in 2006, the NSFG moved to continuous interviewing, with data released in two-year and four-year segments. In these multi-year releases, the data are nationally representative for the short span of years, but not for individual years within the span; survey weights are constructed to refer to the middle year of the data range. For example, the 2006-2010 data release is representative of the population in 2008. Because the NSFG were designed to track change over time, question wording and universes have been largely stable across cycles.

Response rates for the NSFG, as for other large surveys, have declined over the past decades (National Center for Health Statistics 2020). Although sample weights are designed to address this to some extent, comparisons of birth reports from the NSFG to Vital Statistics data indicate that survey coverage of births has gotten slightly worse over time (National Center for Health Statistics 2020b). Much of our analysis combines data from multiple NSFG surveys, which mitigates somewhat the impact of data quality in any single survey. Still, given data issues, it is possible that our analyses underestimate parity among young adults in the more recent cohorts, thus making our estimates likely to be outer bounds of the changes over time. We discuss these issues more fully in the limitations section and in the Appendix.

Since our interest lies in situating current patterns with the longer arc of fertility trends, the 2015-2019 data release was our starting point. We worked backwards to trace the birth cohorts represented in this cycle earlier in the life course. Because we want to both look within cohorts over time and compare across cohorts at specific ages, we exclude those age 45-49 in 2015-2019, as this age range has no comparison in earlier NSFG cycles. Of those age 15-44 in 2015-2019, the oldest cohort was first included in an NSFG sample frame in 1988. Thus, the first NSFG cycle we use is the 1988 survey. We constructed five-year birth cohorts, beginning with the 1940-44 cohort (representing the oldest age group included in the 1988 NSFG) and ending with the 2000-04 cohort (representing the youngest group in the 2015-2019 NSFG), for a total of 13 birth cohorts.¹ This approach produced a preliminary sample size of 93,990 men and women. Due to shifts in immigration and inconsistent measurement of migration timing across cycles, we exclude foreign-born respondents (n = 13,980) and those with missing information on nativity (n = 64), for a final analytical sample of 79,946 individuals (52,221 women and 27,725 men; recall that men were not included in the NSFG until 2002). Table 1 shows the cohort sizes by cycle for reference (Panel 1a), although no analyses are conducted by NSFG cycle. Rather, we focus on birth cohorts and age groups; sample sizes for these units of analysis are shown in Panel 1a of Table 1. The shading in Panel 1b indicates each cohort’s experience with the Great Recession. Because the NSFG surveys occurred at irregular intervals prior to 2006, and because the data releases for the continuous cycles include multiple years of data, the age groups do not map exactly on to NSFG cycles, and the five-year age groups shown in the table are, for some cohorts, an approximation. For instance, the 309 women in the 40-44 age group from the 1940-44 birth cohort are actually age 43-44 since this cohort was interviewed in 1988.

TABLE 1.

Sample Size by Birth Cohort

	1988	1995	2002	2006-10	2011-15	2015-19
Panel 1a By NSFG Cycle
1940-44	309
1945-49	1,162
1950-54	1,407	1,559
1955-59	1,567	1,909	720
1960-64	1,405	1,902	1,568	820
1965-69	1,202	1,540	1,654	2,459	631
1970-74	743	1,389	1,597	2,514	2,199	1,028
1975-79		1,357	1,716	2,992	2,381	2,098
1980-84		47	2,045	3,385	2,988	2,572
1985-89			1,071	3,574	3,120	2,925
1990-94				2,979	3,284	2,892
1995-99				24	2,666	2,923
2000-04					32	1,591
Panel 1b By Five-Year Age Group
	15-19	20-24	25-29	30-34	35-39	40-44	Age Range in 2007
1940-44						309	63-67
1945-49					419	743	58-62
1950-54				516	891	1,559	53-57
1955-59			525	1,042	1,805	824	48-52
1960-64		457	948	1,809	811	1,670	43-47
1965-69	428	774	1,450	845	1,664	2,325	38-42
1970-74	743	1,300	794	1,613	2,372	2,648	33-37
1975-79	1,276	959	1,801	2,715	2,827	966	28-32
1980-84	1,022	2,064	3,221	3,554	1,176		23-27
1985-89	2,331	3,129	3,843	1,387			18-22
1990-94	4,022	3,745	1,388				13-17
1995-99	4,574	1,039					8-12
2000-04	1,623						3-7

No shading in the cells in Panel B means that all members of this cohort passed through this age group prior to the start of the Great Recession in 2007. Lighter shading means that the youngest members of this cohort were within this age group at the start of the Great Recession. Darker shading reflects that all members of this cohort experienced this age group during or after the start of the Great Recession.

To illustrate how different cohorts are covered by the NSFG and how different cohorts experienced more or less of their childbearing years in the Great Recession and its recovery, Figure 1 displays a Lexis diagram, highlighting three cohorts as examples (1955-59, 1975-79, and 1995-99). The dashed lines represent either single-year surveys or continuous survey cycles (dashed lines with connecting arrows). Because the NSFG has traditionally only included those age 15-44, the axis for age begins and ends at those ages. The 1955-59 birth cohort experienced all of its childbearing years prior to the Great Recession, though we can only observe them beginning with the 25-29 age range in the 1988 cycle (and only for women; for men, we do not observe this cohort until age 40-44 in the 2002 cycle since men were not included in prior cycles). We can observe the 1975-79 cohort for women from the 15-19 age range through the 40-44 age range across multiple cycles; for men, we begin observing them in 2002 at age 23-27. The Great Recession hit this cohort in its late 20s and early 30s, and so many had likely begun (and possibly completed) childbearing already. For the 1995-99 cohort, who are only observed through their early 20s, the entirety of their observed childbearing years were spent in the post-Recession recovery years. (It should also be noted that, for this latter cohort, their mid-20s and early 30s – prime childbearing years – will occur after the onset of the pandemic. We return to this point in the Discussion section.)

FIGURE 1.

NSFG Cohorts by NSFG Survey

Fertility measures

The NSFG includes several measures of fertility goals and behaviors. In addition to an indicator of parity at the time of interview (children ever born), respondents were also asked if they intended to have any children in the future. Respondents who reported either intending more children or being unsure were then asked how many children they intended to have. Respondents who were unsure of how many children they intended were asked to estimate the fewest and largest number of possible children, and the midpoint range was used. NSFG imputed information on those who still did provide a valid numeric answer or range, but this occurred for less than 0.3% of the analytical sample and did not affect results (not shown). Individuals who were not fecund, or who were partnered with someone who was not fecund, were not asked the question and were coded as not intending any children. Combining information on parity at the time of the interview with intended parity, we create a measure of total intended parity, similar to that used by Hartnett and Gemmill (2020).

For some analyses, we also use information on fertility desires, asked of all respondents regardless of fecundity: if it were possible, would the respondents want to have a child? This measure was only asked in the 2002 and later cycles and is missing for 0.62% of the analytical sample (386 cases) in cycles when it was asked. We combine this information with intentions to create a measure of desires relative to intentions that gives a more nuanced measure of future plans. The measure initially contained four categories: both wants and intends (more) children, wants but does not intend, neither intends nor wants, and does not want but intends. The last category accounted for less than 2% of the valid cases; when disaggregating by cohort and age group, there were too few cases to meaningfully interpret, and so we grouped these individuals with those both wanting and intending to have a child (results do not differ when they are excluded or categorized differently, not shown).

Analytical approach

We conduct a number of descriptive analyses to answer our research questions (RQs) about both long-term patterns and post-Recession changes. All analyses present measures by birth cohort and age group, pooling NSFG surveys as described in the data section above, and are conducted separately for men and women. Analyses are weighted using Stata’s svy commands; we use the four-year weights for the continuous cycles, but the two-year weights yielded virtually identical results.

We begin by describing achieved fertility over the life course and across cohorts. We show the average number of children and the proportion childless by age and birth cohort. These figures provide the background context of how fertility behavior has evolved, both in terms of timing and in terms of number of births. Although overall fertility trends are reasonably well known, this description showing both timing and quantum of childbearing is useful for motivating our analysis of childbearing goals. We then present changes in fertility goals to more directly answer our research questions.

First, to address whether early life-course childbearing goals have changed (RQ1a), we consider intended parity and intended childlessness among 15-19 year olds and 20-24 year olds across cohorts. Vertical black lines indicate which cohorts were entirely in those age groups after the start of the Great Recession in 2007, allowing us to address RQ1b.

Second, to address RQ2a about the evolution of childbearing goals across the life course and across cohorts, we graphically display total intended parity and intended childlessness by age and cohort. We also present figures that focus just on the age group-cohort combinations that fell during or after the Great Recession to address RQ2b.

Third, to address RQ3 about the fertility gap, we present multiple comparisons of fertility goals and behavior at both the macro (cohort) and micro level. At the macro level, we compared average intended total parity for cohorts early in the life course (age 20-24) to achieved parity in the middle of the reproductive years (age 30-34) and toward the end of the reproductive years (age 40-44) for the same cohort. We present this comparison both as a ratio (share of intended total parity achieved, achieved parity / intended parity) and as an absolute difference (fertility gap: achieved parity – intended parity; cf. Beaujouan and Berghammer 2019). At the micro level, we compare fertility desires relative to intentions at age 40-44 for men and women at parities one and below (parities below the mean intended family size earlier in the life course for these cohorts).

Results

We first describe childbearing behavior among recent cohorts. Figures 2 and 3 show mean achieved parity and childlessness, respectively, for women, by cohort and age group, and Figures 4 and 5 show the corresponding information for men. (The estimates used to produce these figures are in Appendix Table A1, along with confidence intervals). We present childlessness in addition to mean parity because it more directly illustrates changes in the timing of first birth.

FIGURE 2.

Achieved Parity among Women by Birth Cohort & Age Group

FIGURE 3.

Proportion Childless among Women by Birth Cohort & Age Group

FIGURE 4.

Achieved Parity among Men by Birth Cohort & Age Group

FIGURE 5.

Proportion Childless among Men by Birth Cohort & Age Group

Looking first at achieved parity for women (Figure 2), the overall story is what one would expect – parity is low at young ages and increases over time. This pattern is true for all observed (synthetic) cohorts. Further, the number of children ever born, for those cohorts of women observed at age 40-44, hovers around two, ranging from a low of 1.91 for the 1950-54 birth cohort to 2.31 for the 1940-44 birth cohort (though this latter age group-birth cohort combination was actually age 43-44 at the only observation, which may have elevated completed parity slightly). Notably, the most recent of the observed cohorts completely followed through the childbearing years (from ages 15-19 to 40-44) – those born 1975-79 – had among the highest number of children ever born, at 2.11. This cohort, however, did not experience a substantial portion of its childbearing years in the post-Recession period. For more recent cohorts at younger ages, we see that achieved parity was lower than it was for earlier cohorts at the same age. Take the 30-34 age group: the average number of children ever born hovered between 1.54 and 1.70 for the 1950-54 through 1980-84 cohorts, before exhibiting a much lower mean (1.36) for the 1985-89 cohort. For the 20-24 age group, mean achieved parity was halved between the 1990-94 and 1995-99 cohorts. (As noted above and in the Appendix, the most recent NSFG may slightly underestimate fertility; if this is the case, true declines in achieved parity could be even smaller.)

Much of these declines in achieved parity are likely driven by delays in the timing of first births (Figure 3) for more recent cohorts at younger ages. The solid horizontal line in the figure indicates 50%, the point where half of the cohort had at least one child. Among the more recent cohorts of women, those born before 1985 reached this point in the 25-29 age group, but the 1985-89 and the 1990-94 cohorts did not reach this point until age 30-34 (though the former cohort just missed this, at 51% childless). For the 35-39 and 40-44 age groups, there was little variation in the proportion childless across cohorts for women, but at younger ages, the cohorts who experienced more of their prime childbearing years in the post-Recession period had higher levels of childlessness than earlier cohorts. For instance, among the 1980-84, 1985-89, and 1990-94 birth cohorts, between 70% and 74% of women were childless at age 20-24, but for the 1995-99 birth cohort, 86% were childless.

For men, the life course pattern was essentially the same as women’s – low mean parity at younger ages but rising as the cohorts age (Figure 4). However, completed parity is below two for all five male cohorts observed at 40-44, ranging from a low of 1.66 for the 1960-64 cohort to a high of 1.94 for the 1955-59 cohort. The most recent cohort to be observed at age 40-44, the 1975-79 cohort, had the highest mean parity (1.80) since the 1955-59 cohort. As with women, the most recent cohorts of young men tended to have considerably lower mean parity than earlier cohorts at the same age. For instance, among the 20-24 age group, the 1980-84, 1985-89, and 1990-94 birth cohorts reported having an average between 0.17 and 0.21 children ever born, but for the 1995-99 cohort, mean parity was more than halved, falling to 0.07. Again, these patterns can be partially explained by delayed first birth timing, as represented by childlessness (Figure 5). For most cohorts, more than half of men were fathers (solid horizontal line) by the 30-34 age group, but, among the last observed cohort for this age group (1985-89), only 46% were fathers (i.e., 54% of men were still childless). For the oldest age group (40-44), childlessness varied little, between 21% and 26% between the 1955-59 and 1975-79 cohorts. But at younger ages, the most recent cohort often exhibited much higher levels of childlessness as more of men’s childbearing years occurred in the post-Recession period. For the 25-29, 30-34, and 35-39 age groups, the most recently observed cohorts had proportions childless about ten percentage points higher than the immediately prior cohort.

Have early life-course goals for childbearing changed across cohorts (RQ1a), especially since the Great Recession (RQ1b)?

Figures 6 and 7 show intended parity and intended childless, respectively, among 15-19 and 20-24 year old men and women. (The estimates used to produce these figures are in Appendix Table A2, along with confidence intervals (CIs)). Moving from left to right, the shading of the columns lightens successively with each more recent cohort, and the columns to the right of the vertical bar represent cohorts who were in that age group after the onset of the Great Recession. Looking first at intended parity (Figure 6), for women 15-19, intended parity is between 2.0 and 2.25 births for all observed cohorts. There is some fluctuation in between the 1965-69 and 1990-94 cohorts before declines with the two most recent cohorts. There are fewer observations for men 15-19, but the available data suggests a similar pattern. For 20-24 year olds, women’s intended parity is generally higher than observed for 15-19 year olds, often exceeding 2.25 births. Again, there is fluctuation for the earlier cohorts (1960-64 through 1985-89), though perhaps with an overall pattern of increase across cohorts, before a decline across the later cohorts. For men ages 20-24, there is a hint of a rise in intended parity across the two earliest observed cohorts (1975-79 and 1980-84) before the decline observed for age-gender groups for recent cohorts.

FIGURE 6.

Intended Parity among Young Adults

FIGURE 7.

Intended Childlessness among Young Adults

Trends in intended childlessness (Figure 7) correspond with trends in intended parity. For 15-19 year old women, the share intending no children doubled from about 6% in the oldest observed cohort (1965-69) to 12% for those born in the more recent cohorts (1995-99 and 2000-04). For women ages 20-24, the share who intend to have no children fluctuates a bit for those in 1975-79 cohort and earlier before increasing for later cohorts, with a large uptick in the last observed cohort with a peak of 16% (the CI for this estimate is particularly large and should be interpreted with caution). Among men ages 15-19, intended childlessness declined very slightly – by a little over a percentage point – between the 1980-84 and 1990-94 cohorts before rising to just under 11% for the most recent cohort (2000-04). Among men ages 20-24, there is very minor fluctuation for the earliest cohorts around 8-9% before increasing to around 13%-14% for the 1990-94 and 1995-99 cohorts. In sum, then, with few exceptions, young adults across all cohorts intended to have at least two children on average, and less than 15% intended to have no children, suggesting that early life-course goals have remained in general fairly stable (RQ1a). That said, there is some evidence of a slight decline in intended parity and rise in intended childlessness in the post-Recession years (RQ1b), though the fairly wide confidence intervals (Appendix Table A2) and data quality issues for later cohorts suggest caution in this interpretation. In any case, there is not strong evidence of a longer-term decline (RQ1a).

Have life-course patterns of evolution in childbearing goals changed across cohorts (RQ2a), and are the changes greater post-Recession (RQ2b)?

Figure 8 shows mean total intended parity for women, with Figure 8a showing this for all age groups and cohorts (extending the figures above to show a wider age range) and Figure 8b focusing in on the age group-cohort combinations whose childbearing years fell during or after the Great Recession (i.e., the dark shaded cells in Table 1, Panel 1b). There are several overall trends to observe that correspond with RQ2a. First, mean intended parity was nearly always at or above two children (the solid horizontal line) both across (synthetic) cohorts and within each cohort’s reproductive life course. The only exception was for the 1950-54 cohort at age 40-44 (just below two, at 1.98). Second, within the reproductive life course, intended parity generally peaked in the early or late twenties. The largest of these peaks was for the 1975-79 cohort, rising from 2.25 for the 15-19 age group to 2.47 for the 20-24 age group and then declining to 2.28 for the 25-29 age group. Third, after the peak, nearly all cohorts had a lower total intended parity at age 40-44 than that cohort had at an earlier age. Thus, these general trends suggest little change overall (RQ2b) yet may mask post-Recession trends, so we turn to Figure 8b. Here we see more evidence of downward shifts within an age group across cohorts, with the largest differences across cohorts occurring for those in the 20s and early 30s – the prime childbearing years. For instance, among women 30-34 years old, the oldest cohort to experience this age during or after the Recession (meaning that their teens and 20s were pre-Recession) were those born between 1975-79. This cohort intended an average of 2.31 children. But for the 1985-89 cohort, who had experienced all of their 20s in the Recession and post-Recession period, their intended parity at 30-34 was 2.10. Despite this general pattern of declining fertility intentions across more recent cohorts during and after the Recession, intended parity across these cohorts is at or about two children.

Figure 8.

Total Intended Parity among Women by Birth Cohort & Age Group

Intended childlessness is shown for women in Figure 9, with Figure 9a again showing the full set of age groups and cohorts and Figure 9b zeroing in on those age group-cohort combinations during and after the Great Recession. Intended childlessness is relatively rare, but within each cohort, the share of women who intended to be childless generally increases with age. Although intended childlessness is often somewhat high in the 15-19 age group, it is lower in the 20-24 age group (perhaps representing the increased union formation that occurs in these ages relative to the teen years and the link between union formation and fertility) and then increases with age. Most cohorts reach a peak level of intended childlessness at ages 40-44, with the highest level at this age reaching 15.6% for the 1950-54 cohort. The overall patterns are consistent across cohorts, suggesting little overall change (RQ2a). However, as shown in Figure 9b, there is some evidence that cohorts who experienced more of their early childbearing years in the post-Recession period had dampened intentions to have any children (RQ2b). For instance, the share of 20-24 year olds who intended to have no children rose from 9.2% for the 1990-94 cohort to 16.2% in the 1995-1999 cohort, while the corresponding share among 25-29 year olds rose from 9.5% to 14.8%. Nonetheless, while there may be some increases in intended childlessness among recent cohorts at younger ages, intended childlessness rarely exceeded about 15%.

Figure 9.

Proportion Intending Childlessness among Women by Birth Cohort & Age Group

Figure 10, Figures 10a and 10b, show total intended parity for men, paralleling Figure 8 for women. For men, total intended parity hovered around two but frequently fell below two. As shown in Figure 10a, two of the five cohorts observed at age 35-39 and 40-44 had a mean total intended parity below two (represented by the solid horizontal line), as did three of the five cohorts observed at age 30-34. There was no consistent evidence of a life course peak across the synthetic cohorts, and, of the four cohorts that have both observations for more than one age group and an observation for age 40-44, only two exhibited a pattern where intended parity at age 40-44 was lower than earlier in that synthetic cohort’s reproductive life course. For all but the oldest age group, there was also some evidence that intended parity declined across cohorts at each age group. While it is less clear what the patterns are overall for the full set of cohorts observed here (RQ2a), the patterns in Figure 10b suggest that – like women – there has been a downward shift in intended parity after the Recession (RQ2b). Among the 20-24 age group, for example, total intended parity fell from 2.32 for the 1980-84 cohort to 1.96 for the 1995-99 cohort. Still, all of these cohorts whose childbearing years fell during or after the Great Recession reported an average intended parity of around two children.

Figure 10.

Total Intended Parity among Men by Birth Cohort & Age Group

Intended childlessness for men is shown in Figure 11, corresponding with Figure 9 for women, with Figure 11a covering the full set of age groups and cohorts and Figure 11b focusing in on the Recession and afterwards. In general, intended childlessness increased with age within cohorts (RQ2a). Intended childlessness among men remained relatively uncommon across at ages under 35, hovering at or below 15%. By ages 35-39 and 40-44, some cohorts had levels higher than 15%, with the 1980-84 cohort reaching a peak of 20.1% at ages 35-39. As with women, there is some weak indication, as seen in Figure 11b, that more recent cohorts experiencing their early childbearing years in the post-Recession period had higher levels of intended childlessness, though remaining around 15% or lower (RQ2b). And although gender comparisons are not the focus of this paper, it is worth noting that mean total intended parity was generally lower, and intended childlessness higher, among men than women for each cohort-age group combination.

Figure 11.

Proportion Intending Childlessness among Men by Birth Cohort & Age Group

Has the “fertility gap” between fertility goals and achieved fertility changed over time (RQ3a), especially since the Great Recession (RQ3b)?

Finally, we take two different approaches to consider fertility gaps, beginning with a macro-level approach that compares synthetic cohorts’ early life intentions (20-24) to achieved fertility at ages 30-34 and 40-44. Then we consider the extent to which men and women have (as-yet) unfulfilled intentions and desires toward the end of the reproductive years at the micro level. Looking first at the aggregate patterns, Table 2 shows mean intended parity at 20-24 and achieved parity at ages 30-34 and 40-44. For women, by age 30-34, those in the 1965-69 through 1980-84 cohorts had achieved between 67% and 75% of the parity intended at age 20-24. But for the most recent cohort observed at age 30-34 (the 1985-89 cohort), the share of 20-24 intended parity achieved by age 30-34 had fallen to 56%. Further, for this cohort, women had a fertility gap of just over one child, whereas earlier birth cohorts had a fertility gap of around 0.80 or less. At age 40-44, the share of parity intended at age 20-24 achieved was between 85% and 91%; the cohorts observed in the NSFG cycles for this age range (1960-64 through 1975-79) experienced fairly little of their childbearing years during the Great Recession and its aftermath. Still, they all exhibited a fertility gap, ranging from 0.20 to 0.36. For women, the results in Table 2 are consistent with continued postponement of births (more unfulfilled intentions at age 30-34) that may be eventually made up, as there is no substantial increase in unfulfilled intentions at age 40-44.

TABLE 2.

Underachievement of Intentions at 20-24 at 30-34 and 40-44 by Cohort, Age Group, and Gender (95% CIs)

	Intended parity at 20- 24	Actual parity 30-34	% of intended parity at 30-34	Fertility gap at 30-34	Actual parity 40- 44	% of intended parity at 40-44	Fertility gap at 40-44
	Women
1960-64	2.41 (2.29-2.53)	1.56 (1.49-1.64)	65.0% (55.7-72.5%)	0.84 (0.41-1.28)	2.08 (1.92-2.23)	86.3% (74.1-95.8%)	0.33 (−0.05-0.70)
1965-69	2.25 (2.16-2.35)	1.70 (1.58-1.82)	75.4% (61.0-89.9%)	0.56 (0.28-0.83)	2.05 (1.92-2.18)	90.9% (80.1-1.02%)	0.20 (−0.10-0.51)
1970-74	2.29 (2.22-2.36)	1.58 (1.46-1.70)	69.0% (55.6-83.3%)	0.71 (0.38-1.04)	1.94 (1.84-2.04)	84.7% (73.4-97.3%)	0.35 (0.05-0.64)
1975-79	2.47 (2.29-2.64)	1.66 (1.55-1.76)	67.1% (54.0-83.0%)	0.81 (0.34-1.28)	2.11 (1.94-2.28)	85.6% (66.8-1.08%)	0.36 (−0.18-0.90)
1980-84	2.32 (2.20-2.45)	1.60 (1.51-1.70)	69.0% (58.3-81.0%)	0.72 (0.41-1.03)
1985-89	2.35 (2.22-2.48)	1.32 (1.20-1.45)	56.3% (37.8-76.5%)	1.03 (0.63-1.42)
	Men
1975-79	2.29 (2.06-2.52)	1.18 (1.08-1.29)	51.7% (39.6-66.1%)	1.11 (0.63-1.58)	1.8 (1.63-1.97)	78.6% (60.8-1.00%)	0.49 (−0.02-0.99)
1980-84	2.32 (2.06-2.58)	1.06 (0.98-1.15)	45.8% (36.6-56.1)	1.26 (0.94-1.58)
1985-89	2.18 (2.09-2.26)	0.87 (0.71-1.03)	39.8% (22.3-58.4%)	1.31 (0.96-1.66)

No shading in the cells means that all members of this cohort passed through this age group prior to the start of the Great Recession in 2007. Lighter shading means that the youngest members of this cohort were within this age group at the start of the Great Recession. Darker shading reflects that all members of this cohort experienced this age group during or after the start of the Great Recession.

For men (of whom there are fewer observed cohorts), the share of intended fertility (at age 20-24) achieved by age 30-34 was low, consistent with men’s later childbearing age schedule, but also declining across cohorts. For the 1975-79 cohort, mean achieved parity at 30-34 was about half of parity intended at age 20-24, falling to about 40% for the 1985-89 cohort; however, all three observed cohorts among men had a fertility gap of more than one child at 30-34. By age 40-44, just over three-fourths of men’s earlier intended parity had been achieved for the 1975-79 cohort (the only cohort observed at this age for men), and the fertility gap was about half a child. Thus, in answer to RQ3a, for both men and women, we do not find strong evidence of a growing fertility gap at older ages, though the youngest cohorts have yet to reach the oldest ages, and there is evidence that the fertility gap in the early 30s is growing for those who experienced more of their childbearing years during and after the Great Recession (RQ3b). However, to the extent that the more recent cohorts’ parity data is taken from the latter NSFG cycles that may have less accurate fertility data, the size of the fertility gap may be overestimated.

Finally, to consider the fertility gap at the micro level, Table 3 shows unfulfilled fertility desires and intentions toward the end of the reproductive years (age 40-44) for those at parities zero and one. This table uses a three-category measure: does not want or intend any children, wants but does not intend children, intends children (regardless of desire). Essentially, this measure shows the extent to which men and women with fewer children at the end of their reproductive years than the cohort’s average intended parity earlier in the life course have unfulfilled fertility desires (wanting but not intending) or are continuing to delay fertility (still intending to have at least one child). All of these cohorts reached their early 40s during or after the Great Recession, with the latter cohorts experiencing more exposure than others.

TABLE 3.

Fertility Desires and Intentions at 40-44 at Parities ≤ 1, by Gender

	Women 40-44 Parity≤1 (N=1757)	Men 40-44 Parity≤1 (N=1102)
Doesn't want or intend
1960-64	66.4% (60.3-72.5%)	57.5% (49.2-65.7%)
1965-69	68.5% (62.2-74.8%)	55.1% (47.7-62.6%)
1970-74	71.5% (66.6-76.4%)	57.8% (51.8-63.7%)
1975-79	78.3% (71.5-85.1%)	47.2% (38.5-55.9%)
Wants, doesn't intend
1960-64	16.8% (12.7-20.9%)	15.1% (9.6-20.7%)
1965-69	15.2% (10.5-20.0%)	13.7% (9.5-18.0%)
1970-74	17.4% (13.1-21.8%)	18.5% (13.5-23.5%)
1975-79	9.2% (4.6-13.8%)	13.8% (6.8-20.7%)
Intends (regardless of want)
1960-64	16.8% (11.8-21.7%)	27.4% (20.8-34.1%)
1965-69	16.2% (11.5-21.0%)	31.1% (24.3%-37.9%)
1970-74	11.1% (7.9-14.2%)	23.7% (18.7-28.7%)
1975-79	12.5% (7.4-17.6%)	39.0% (30.3-47.8%)

Lighter shading means that the youngest members of this cohort were within this age group at the start of the Great Recession in 2007. Darker shading reflects that all members of this cohort experienced this age group during or after the start of the Great Recession.

For women age 40-44 at parities zero and one, the share who do not want or intend to have children has risen across cohorts. About two thirds of low-parity women age 40-44 in the 1960-64 birth cohort (66.4%) did not want or intend to have children at the end of the reproductive years, a share that increased to 78.3% for the 1975-79 cohort. For low-parity men age 40-44, conversely, the proportion of those who did not want or intend a child decreased over time, falling from 57.5% in the 1960-64 cohort to 47.2% for the 1975-79 birth cohort.

Turning to the category that captures unfulfilled fertility desires at the end of the reproductive life course, there is a weak pattern of declining shares who want but do not intend children among low-parity women 40-44. Among women with zero or one child, 16.8% of the 1960-64 birth cohort wanted but did not intend to have any children. The share was lower in the 1965-69 cohort (15.2%), rose for the 1970-74 cohort (17.4%), and then fell again for the 1975-79 cohort (9.2%). For men age 40-44, the share of low-parity men with unfulfilled fertility desires shows a similar pattern, starting at 15.1% for the 1960-64 cohort, falling to 13.7% for the 1965-69 cohort, rising to 18.5% for the 1970-74 cohort, and then falling again to 13.8%.

Finally, delayed fertility intentions are shown in the bottom third of the table. Among low-parity women, fewer women ages 40-44 intend to have a child in the future across cohorts – the share fell from 16.8% for the 1960-64 cohort to 12.5% for the 1975-79 cohort. Still, about one in six low-parity women in the most recent birth cohort observed at age 40-44 intended to have a child. For men, the pattern appears to be in the opposite direction, with a rise from 27.4% in the 1960-64 cohort to 39.0% in the 1975-79 cohort.

In sum, although there is evidence of change in the fertility desires and intentions of men and women at low parities across cohorts, there is no evidence of increases in unfulfilled fertility desires (RQ3a). Instead, among the cohorts observed to age 40-44, women’s desires and intentions have become increasingly aligned as each successive cohort experienced more of its childbearing years during the Recession and recovery period (RQ3b), with more low-parity women neither wanting nor intending to have any children. Similarly, though the direction of the pattern differs, men, too, are not exhibiting evidence of rising unfulfilled fertility desires. For low-parity men, delayed fertility seems to be rising – in the latest cohort, nearly four in ten intend to have a child in the future. While low-parity women may increasingly have revised earlier life course goals downward with longer exposure to the post-Recession period, men seem to have shifted their fertility timing to older ages.

Discussion

In this paper, we took a cohort approach to better understand how the low U.S. fertility rates of the post-Recession years fit into larger trends. Fertility fell during the Great Recession and its immediate aftermath, as expected given theory and past recessions, but it did not rebound as expected as the economy recovered. Instead, it has continued to decline, with the U.S. TFR reaching a historic low of 1.71 in 2019 (Martin et al. 2021). There is a considerable literature documenting both the aggregate and individual conditions and factors that could be affecting fertility. In the U.S., for instance, there is work on the link between aggregate economic indicators and aggregate fertility rates (e.g., Cherlin et al. 2013; Schneider 2015), aggregate economic indicators and individual fertility behaviors (e.g., Schneider 2017; Schneider and Gemmill 2016; Schneider and Hastings 2015; Seltzer, 2019), and individual financial circumstances and individual fertility behaviors (e.g., Brauner-Otto and Geist 2018; Min and Taylor 2018; Nau, Dwyer, and Hodson 2015). This work has made important contributions, but most of it has one key drawback: it cannot tell us whether these factors inhibited the ability to achieve fertility goals or whether these factors affected the formation of fertility goals in the first place. To the extent that raising fertility levels is a target for U.S. policy – and whether it should be is the subject of considerable debate – the distinction between these two pathways has different implications for policy responses.

The data requirements to determine whether men and women of reproductive age have increasingly begun to limit their goals for childbearing or are having more difficulty achieving their fertility goals are steep. Longitudinal data that includes information about both fertility goals and fertility behaviors is needed for multiple cohorts covering the entire reproductive life course, and the data would need to include a host of both subjective and objective economic and social measures. No such data exists in the U.S. But, as a first step, we leveraged several cross-sectional cycles of the National Survey of Family Growth to create synthetic cohorts of men and women of childbearing age and examine fertility goals and behaviors within and across cohorts. Although the NSFG does not include sufficient economic and objective measures to test individual fertility responses to economic conditions, we can leverage the differential exposure these cohorts faced to the Great Recession to draw some tentative conclusions about post-Recession fertility trends.

We first showed that, among women, observed cohorts who have largely completed their childbearing years (age 40-44) typically have a completed parity around two, and childlessness for these cohorts does not exceed 20%. The variation is fairly minimal for these cohorts (those born between 1950 and 1975). Delays in the transition to parenthood are evident across cohorts, but the delays are more dramatic for the cohorts born after 1990, who experienced much of their childbearing years after the Great Recession. For men, the pattern is similar, but all the birth cohorts observed at age 40-44 have fertility below two, and childlessness is above 20%.

Might we expect different patterns in the future, if we were to look at fertility preferences at early ages across cohorts and especially those who experienced more of their childbearing years during the Great Recession and its recovery (RQ1)? The results show that while there is a noticeable decline in intended parity and an uptick in intended childlessness among men and women under 24 among post-Recession cohorts, intended parity remains around two and intended childlessness generally under 15% (with one exception). Further, the confidence intervals often overlap, suggesting that the absolute size of the decline should be taken with caution. Next, we considered patterns over the life course within cohorts and across cohorts at the same age (RQ2a). We found that, among women, most cohorts – even those who entered their childbearing years during the Great Recession and its recovery – intend to have slightly more than two children throughout the reproductive life course. The life course pattern in which intended parity peaks in the early 20s is evident across cohorts of women. This is true despite small declines in total intended parity and increases in intended childlessness at younger ages in the most recent cohorts (RQ2b). For men, at younger ages, intended parity remains at or above two, though intended parity at ages 30 and older more often falls below two. Thus, the Great Recession does not seem to have induced (or coincided with) a substantial shift in how individuals envision their future family size the role of childbearing role in their lives.

This, of course, does not mean that goals will be achieved, and it is possible that there is a growing fertility gap or rising levels of unfulfilled fertility desires at later ages (RQ3a), especially for cohorts experiencing more of their childbearing years during or after the Recession (RQ3b). The fertility gap itself is not new; among women observed at age 40-44, the gap between their intended parity at age 20-24 and their achieved parity at age 40-44 has most frequently been about a third of a child, with the gap larger for men at about half a child. But looking at age 30-34, where more cohorts are observed, there is consistently a sizeable gap, and it is particularly large for the last cohort observed reaching this age, a cohort whose childbearing years overlapped more with the Recession and its recovery (RQ3b). This suggests that fertility will need to be considerable during the 30s and early 40s if more recent cohorts are to achieve their earlier life course fertility goals. Yet among cohorts observed at age 40-44, there is little evidence that the proportion of low-parity women with unfulfilled fertility desires has increased (RQ3a). By the end of the life course, many childless women and women with only one child appear to have adjusted their fertility preferences. The share of women who neither want nor intend children has increased, and the share who want but do not intend children has decreased across cohorts (RQ3b). Still, among the most recent cohort (1975-79), some low-parity women age 40-44 still intend to have a child – about 17% of childless women and 9% of mothers with one child. For men, the story is different to some extent. Although there is no growth in the share with unfulfilled fertility desires, there is an increase in the share of low-parity men who intend to have children in the future, with about four in ten childless men and one in three fathers with one child intending to have at least one birth in the future. Given the biological and social realities of men’s childbearing and partnership behaviors, such intentions may be likely to be fulfilled.

In sum, of the possible explanations for low fertility in the post-Recession period, there is more support for the interpretation that men and women are either facing difficulty in achieving their fertility goals at younger ages or increasingly shifting preferred timing to older ages in a way that depresses period fertility, rather than a wholesale shift to substantially lower fertility intentions and increased preferences for childlessness. Although it is possible that the small and gradual change across recent cohorts in intended parity may eventually result in average intended parities well below two and a growth in intended childlessness, it seems the larger issue is that more recent cohorts are less likely to achieve their goals at earlier stages of the life course, perhaps because they increasingly prefer to delay childbearing to later ages. This may ultimately reduce cohort completed fertility. If fertility is compressed into a shorter period towards the end of the reproductive life course, one that may correspond with lower fecundity (more so for women than men), the fertility gap between early life course fertility goals and achieved fertility later in the life course may widen. At the individual level, though, this gap may not be problematic, if people adjust their plans to what is possible, consistent with our findings that unfulfilled fertility desires are not increasing among low-parity individuals in recent cohorts. Some people, especially women, may downgrade their fertility goals over time (Gemmill 2019; Rybińska and Morgan 2019).

Certainly, today’s young adults who have yet to have children by their late 20s still have time to have children and fulfill their intentions during their 30s and early 40s, especially when intentions remain around two children. And it is good news that unintended births, which have always comprised a substantial portion of U.S. births at younger ages, are down (Finer and Zolna 2016). Further, advances in reproductive technology may extend the fecundity window, though access and affordability will likely remain a barrier for all but the most advantaged (Quinn and Fujimoto 2016). Together, such changes might mean that children will increasingly be raised by parents who have deliberately entered into parenthood and feel equipped for childbearing, perhaps leading away from the “diverging destinies” that has worried scholars over the past two decades (McLanahan 2004). It is also possible, of course, that some men and women will change their minds about wanting and/or intending children as social acceptance of childlessness grows (see the ‘childfree’ movement, e.g., Blackstone 2019) or as other competing interests or impediments arise. Still, it seems that the majority of women (and men, to a lesser extent) intend to have two children, and policies and programs that reduce the social and economic costs and pressures of childrearing earlier in the life course may make it easier for men and women to achieve their fertility goals.

Limitations

Although we made creative use of available data, we were unable to follow specific individuals over time. Moreover, the NSFG contains only very limited data on individual economic circumstances, and it does not collect information on individual motivations and reasonings for fertility goals. As a first step in situating current fertility trends within a cohort framework, our analysis was purely descriptive and did not incorporate factors known to be linked to fertility, such as changes in marriage (Hayford 2013) or educational attainment (Adserà 2017). We did not analyze other aspects of reproductive behavior, such as the strength of intentions, intended fertility timing, contraceptive behavior, or the use of assisted reproductive technology. It is also possible that the measure of fertility desires, when asked of 40-44 year olds, may under-represent those who would have liked to have had a child earlier in the life course but would not wish to be parents of young children at older ages. And in focusing on those at low parities towards the end of the reproductive years, we did not examine those at parities two or higher who may also have unfulfilled fertility desires.

There are data issues that raise concerns. Given declines in NSFG response rates and potential undercounts of fertility in the NSFG relative to Vital Statistics in recent cycles, we advocate caution in interpreting the magnitude of shifts in later years. If anything, these issues suggest that the extent to which fertility is falling in recent cohorts – and potentially also whether fertility preferences are changing – may be overestimated. Further, we should be cautious interpreting the shifts among men and especially in comparisons of men and women. Not only do we have a shorter timeframe to observe men, as they were not included in the NSFG until 2002, but there are reasons to be concerned about the accuracy and completeness of men’s fertility data (Joyner et al. 2012). Men may under-report births more than women. They are also differentially under-represented in the data given the sampling frame of non-institutionalized, civilian persons; men who are not attached to households (typically disadvantaged, young, and/or members of racialized minorities) may differ in ways that would bias estimates of intended parity either upwards or downwards. Men also have a different fertility schedule than women, with their achieved fertility at age 40-44 perhaps not representing their completed fertility to the same extent as women; paternal age has been increasing, though births fathered by men over 40 accounted for less than 10% of all births as recently as 2015 (Khandwala, Zhang, Lu, and Eisenberg 2017). Still, to the extent that men seem to have lower parity goals and higher shares who intend to be childless than women, the not-infrequent portrayals in the popular press of women’s selfishness and career ambitions as driving low fertility levels (Sobotka and Beaujouan 2018) are most certainly misdirected.

Conclusion

As the U.S. and other nations grapple with sustained fertility declines in the aftermath of the Great Recession – and the additional shock to fertility levels introduced by the pandemic – understanding whether changes in fertility behaviors are driven by growing mismatches between goals and behaviors or downward shifts in parity goals and plans to have any children is paramount importance for the success of any interventions. Raising fertility goals – encouraging people to want and intend to have more children – seems more formidable (and less desirable, from a human rights perspective) than identifying and addressing the factors that influence whether individuals achieve their earlier fertility intentions. These factors are most certainly multifaceted and numerous, inclusive of not only explicit family policies (such as parental leave and childcare access and affordability) but also housing costs, the costs of education, job security, and health care costs and access. Policies that merely provide financial transfers to parents seem to be less effective in affecting fertility behaviors than those that address the larger conditions of parenting (Bergsvik, Fauske, and Hart 2021). And in addition to factors that may depress fertility, there are those that support non-parental identities, such as the ‘childfree’ movement (Blackstone 2019).

The time period analyzed in our descriptive analysis ended right before the pandemic begin, and there is growing evidence that the pandemic has served as a major fertility shock affecting birth rates (Aassve, Cavalli, Mencarini, Plach, and Sanders 2021; Luppi, Arpino, and Rosina 2020). For men and women who delayed fertility in the aftermath of the Great Recession, the additional stressors of the pandemic may have a permanent and negative impact on their ability to make up those postponed births. Young cohorts in the earliest stages of the reproductive life course have much more time to have births in the future, though these are the cohorts that perhaps were already on the pathway to lower fertility goals. The pandemic’s disruption of education, career, and relationship trajectories may permanently depress fertility goals, especially given the very public nature of the pressures placed on parents to take on more of the responsibility for childrearing as schools and childcare centers shut down even while expectations to remain employed and productive remained. Alternatively, the pandemic’s impact may only be a small period shock with few long-term effects (Sobotka et al. 2022). Thus, it remains to be seen whether the fertility will be able to rebound from not just the Great Recession but the pandemic as well.

Data Availability

This paper uses the National Survey of Family Growth. These data are publicly available for download from the National Center for Health Statistics at https://www.cdc.gov/nchs/nsfg/index.htm