Throughout the twentieth century, Italy was characterized by significant regional differences in fertility rates. In this article, a cohort approach is employed in order to further explore regional variation. More specifically, total cohort fertility rate is calculated for the various Italian regions for cohorts born between 1935 and 1968. In addition, recently published data from the Italian National Institute of Statistics (Istat) for the years 1999 2005 are examined. They allow us to extend our analysis to the first phase in the reproductive lives of cohorts born in the 1970s and in the early 1980s. This analysis shows that first, for the cohorts born up until the 1960s, fertility declines and regional differences remain relevant; and second, for the cohorts born from the 1970s on, the speed of the fertility decline slows in the north and even come to a halt in several regions, while continuing to drop in southern Italy. If this trend remains constant over time, cohort fertility levels in all Italian regions may eventually converge at similar levels.
Tout au long du xxe siècle, l'Italie a été marquée par de grandes différences régionales de fécondité. Adoptant une approche longitudinale pour approfondir l'analyse de ces variations, l'auteur a calculé, dans chaque région, la descendance finale des générations nées entre 1935 et 1968. Les données récemment publiées par l'Institut national italien de statistique (ISTAT), qui portent sur les années 1999-2005, permettent d'étendre l'analyse à la première partie de la vie féconde des générations nées dans les années 1970 et au début des années 1980. Les résultats montrent que : (1) pour les générations antérieures à 1970, la chute de la fécondité et les disparités régionales sont bien confirmées ; (2) pour les générations plus récentes, le rythme de la baisse de la fécondité se maintient en Italie du Sud, tandis qu'il ralentit en Italie du Nord, allant même jusqu'à la stagnation dans certaines régions. Si cette tendance perdure, les descendances finales des générations de toutes les régions italiennes pourraient, à terme, converger vers un même niveau.
A todo lo largo del siglo XX, Italia ha estado caracterizada por grandes diferencias regionales de fecundidad. Adoptando un enfoque longitudinal para profundizar el análisis de estas variaciones, el autor ha calculado, en cada región, la descendencia final de las generaciones nacidas 1935 y 1968. Los datos recientemente publicados por el Instituto nacional italiano de estadística (ISTAT), que se refieren a los años 1999-2005, permiten extender el análisis a la primera parte de la vida fecunda de las generaciones nacidas en los años 1970 y a principios de los años 1980. Los resultados muestran que : (1) para las generaciones anteriores a 1970, la caída de la fecundidad y las disparidades regionales están bien confirmadas ; (2) para las generaciones más recientes, el ritmo de la baja de la fecundidad se mantiene en Italia del Sur, mientras que disminuye en Italia del Norte, yendo incluso hasta el estancamiento en ciertas regiones. Si esta tendencia perdura, las descendencias finales de las generaciones de todas las regiones italianas podrían, dentro de algún tiempo, converger hacia un mismo nivel.
Since the early 1980s, Italy has been characterized by low fertility, and since the late 1990s, by lowest-low fertility. Indeed, the country has often been cited as an example of extreme fertility decline (Frejka and Calot, 2001; Kohler et al., 2002; Sobotka, 2003; Frejka and Sardon, 2005, 2006; Sardon, 2006). However, since the turn of the twenty-first century, new Italian fertility trends(1) have emerged, with radical changes in birth rates and in the timing of fertility (Dalla Zuanna and Micheli, 2004; Mencarini and Tanturri, 2006).
It is possible to study recent fertility behaviour using data released by the Italian National Institute of Statistics (Istituto nazionale di statistica, ISTAT) on age-specific fertility rates from 1999-2005. Although these data originate from a source different from that of an earlier series (1952-1996, see below) the two data sets are of a comparable nature (ISTAT, 2006a) and can therefore be used together.
A number of different approaches have been employed when studying fertility (Frejka and Sardon, 2005) such as using period measures, tempo-corrected period measures (Ryder 1964; Bongaarts and Feneey, 1998), cohort measures (Santini, 1974, 1995; Frejka and Sardon, 2005), or even a combination of these techniques (Livi Bacci, 1977; Festy, 1979). In this paper, a cohort approach (building on the work of Rallu, 1983 and Sorvillo and Terra Abrami, 1993) is employed in order to provide an updated picture of fertility at a cohort level across Italian regions.
An analysis at the regional level also allows for a more detailed depiction of Italian fertility. Although researchers have investigated Italian fertility at the national level and have observed the country's general pattern of very low fertility, a number of scholars have demonstrated that this approach is somewhat misleading (Livi Bacci, 1977; Rallu, 1983; Del Panta, 1996 and most recently, Kertzer et al., 2006 and Santini, 2006). [End Page 157]
Not only are there considerable differences in fertility patterns between northern and southern Italy (Santini, 1995; De Sandre, 2000), but divergences also exist within these two macro-regions (Santini, 2006). In fact, the range of fertility levels across the country recently motivated ISTAT to employ, for more accuracy, a multiregional model for its population forecast (Terra Abrami, 1998; ISTAT, 2002).
This short paper has four main aims: i) to describe changes in total cohort fertility rate (TCFR) for female cohorts who have already completed their reproductive period (1935-1956); ii) to estimate TCFR for those cohorts (1957 1968) who, even if they have not yet completed their reproductive period, have only a small fraction (no more than 10%) yet to conclude; iii) to verify whether Italy's long-standing decline in cohort fertility continues to hold true for those cohorts born after 1969, or whether there are signs that this process is slowing or even coming to a halt; iv) to explore whether women born in the late 1960s and early 1970s tend to recover fertility levels after the age of 29, and to compare their fertility behaviour with that of women born in the 1950s and early 1960s.
Data from two of the most representative regions of Italy have been chosen in order to depict changes in fertility behaviour(2): Lombardy, the most industrialized and populated region of northern Italy, whose major city is Milan, and Campania, the most inhabited southern region of Italy, whose largest urban centre is Naples.
1. Data and methods
Regional age-specific fertility rates analysed here come from two different sources: the Stato Civile registers, which record births (and other vital events) from the present (de facto) population (ISTAT, 1997, 1998a, 1998b, 2000; Terra Abrami and Sorvillo, 1993) for the years 1952-1996, and the Anagrafe registers, which record births (and other vital events) from the resident (de jure) population, for the years 1999-2005(3) .
Until now, no recent studies have validated the comparability of these two sources. However, Istat's (1998) analysis of fertility in Italy from 1952-1993 did evaluate the number of births per year given in the two different sources, and found only negligible differences. ISTAT (2006a) has also used data from these two sources to study changes in Italian fertility over the last ten years (1995-2004). Moreover, the data series available from the Anagrafe have excellent coverage, which increased from 95.8% in 1999 to 98.9% in 2001 (the latest coverage available). For these reasons, it should be acceptable to use the two series together without introducing significant bias into the analysis.
Age-specific fertility rates for the years 1997 and 1998 are not available due to the sudden termination of the Stato Civile data collection (Prati et al., 2004). [End Page 158] These rates were thus estimated by means of a least squares procedure, using data from 1992 to 1996 and from 1999 to 2003. Two interpolative estimates were produced: one linear, using data by calendar years, the other parabolic, through use of cohort data. The final estimation of each age-specific fertility rate at a regional level for the years 1997 and 1998 was given by the unweighted average of the two(4).
The aggregation of cross-sectional data at a cohort level provides the complete age schedule of fertility for the 1937-1956 cohorts. For the 1935 and 1936 cohorts, fertility rates at ages 15 and 16 were estimated, assuming them to be the same as the 1937 cohort.
For the 1957-1968 cohorts, TCFR was partially estimated, hypothesizing that age-specific fertility rates will not change between younger and older cohorts(5). Indeed, this portion is never more than 10% of the TCFR.
Finally, due to the intrinsic characteristics of the data sources, the age-specific fertility rates analysed here refer to two overlapping cohorts. Consequently, the birth year of the younger cohort is used for the sake of simplicity (as seen in Rallu, 1983).
2. A comparison of period (1952-2006) and cohort (1935-1968) fertility rates across Italian regions
In the northern regions of Italy, the period total fertility rate (TFR) shows an irregular pattern, with a considerable increase in the early 1960s during the baby-boom years (Terra Abrami and Sorvillo, 1993), followed by a decline which then accelerated during the late 1970s. TFR reached its minimum in 1995 and then began to slowly increase. In the regions of southern Italy, no baby boom occurred and the decline in TFR was comparatively slower in the 1960s. This was followed by an accelerated decline in the 1970s. At present, the decline in TFR in southern Italy has not reached a minimum level, but has slowed, and almost stabilized at low levels in several regions (for further details on recent regional fertility trends in Italy see Livi Bacci and Salvini, 2000; Santini, 2006). [End Page 159]
Levels of TCFR and mean age at childbirth in the Italian regions, 1935-1965 cohorts
|Region||TCFR (children per woman)||Mean age at childbirth (years)|
|Source: Author's calculations based on ISTAT data (Stato Civile for the years 1952-1996; Anagrafe for the years 1999-2005; estimate for the years 1997 and 1998).|
[End Page 160]
[End Page 161]
The changes in TCFR, on the other hand, are less irregular than those seen in the TFR: there is no baby boom and the decrease in cohort fertility is slow but continuous, at least from the cohorts born in the late 1930s (Terra Abrami and Sorvillo, 1993). Table 1 and Figure 1 present these differences and the variations in regional TCFRs.
The decline in TCFR was slower and more regular in the regions of northern Italy, where the cohorts born in the 1930s already had low fertility levels. In the southern regions, on the other hand, the decline occurred more quickly, as illustrated by the case of Campania (Figure 2). Thus, changes in regional cohort fertility have followed the same general pattern of diffusion as that seen in the demographic transition in Italy: starting in the north and the west and then moving to the south and the east (Livi Bacci, 1977).
The decline in cohort fertility in the northern regions slows or comes to a halt for those cohorts born after 1970, while in southern Italy this decline continues. Women born in 1970 and in 1980 in the northern regions of Italy have similar characteristics with regard to the first part of their reproductive lives, as shown by the case of Lombardy (Figure 3). Women born in 1960, on the other hand, are comparatively quite different in that they began childbearing at an earlier age. However, overall TCFR for these cohorts may ultimately be rather similar, given that a fertility recovery appears to occur among younger cohorts after age 30 (beginning childbearing later). [End Page 162]
In the southern regions, such as Campania (Figure 4), the fertility of the 1970 cohort is much lower than that of the 1960 cohort. For the 1980 cohort, this decline continues (although less dramatically) and the timing of fertility begins slightly later (GCD-SIS, 2007).
[End Page 163]
This pattern of fertility recovery becomes evident when comparing cumulated cohort fertility rates (CCFR) up to age x for the 1950-1980 cohorts (Figures 5 and 6). We take the cohort born in 1950 as our base, and then compare their cumulated fertility with that of the younger cohorts(6) (Frejka and Calot, 2001; Sardon, 2006). Data are fully observed for the 1950 and 1955 cohorts, and partly estimated for the 1960 and 1965 cohorts. CCFRs are also shown up to the age of 35 for the 1970 cohort, up to the age of 30 for the 1975 cohort, and up to age 25 for the 1980 cohort.
That fertility decline has come to an end in the northern regions is evident when comparing the 1980 cohort with the 1975 cohort. Moreover, a comparison of the 1975 cohort with the cohort of 1970 reveals only a very slight decline in early fertility. In addition, these data demonstrate fertility recuperation on the part of the 1960 and 1965 cohorts when compared with the 1950 cohort. For example, in Lombardy (Figure 5) the difference in the CCFR between the 1950 and the 1965 cohorts decreases from a maximum of –0.602 child per women at age 29 to –0.331 at age 45. Between the 1950 and 1960 cohort, there is a decrease from –0.362 child per women at age 29 to –0.205 at age 45.
In southern Italy, the fertility recovery is almost non-existent for the 1960 and 1965 cohorts when compared with that of the 1950 cohort, as illustrated by the case of Campania. Fertility declines from the 1970 to the 1975 cohort. [End Page 164] It is only beginning with the 1980 cohort that the decline in fertility seems to slow when compared to the cohort born five years earlier (Figure 6).
More generally, the overall recovery of fertility in Italy has exceeded the expectations of a number of scholars. Frejka and Calot (2001), Sobotka (2004), and Frejka and Sardon (2005), respectively estimated TCFR for the 1960 cohort in Italy at 1.61, 1.67, and 1.60 children per woman, while our estimation is 1.71 (see Table 1)(7).
Finally, the recuperation in cohort fertility is related to changes in women's fertility schedules: the CCFR in the northern regions of Italy after the age of 29 for the 1965 cohort is about 50% of the TCFR, while in southern Italy the CCFR is still around 35% with a minimum of 32.9% in Calabria and Sicily (Table 2). Indeed, the northern and southern regions of Italy have followed distinctly different patterns over time. In the northern regions, fertility rates [End Page 165]
Percentage of TCFR after age 29 for the 1950 and the 1965 cohort by Italian region
|Source: Author's calculations based on ISTAT data (Stato Civile for the years 1952-1996; Anagrafe for the years 1999-2005; estimate for the years 1997 and 1998).|
|Northern Italy||Southern Italy|
|Trentino Alto Adige||31.3||49.0||Apulia||27.2||36.6|
|Friuli Venezia Giulia||22.9||50.9||Calabria||27.8||32.9|
initially declined at older ages, while at younger ages fertility slightly increased. This pattern ended with the cohorts born in the late 1940s, and was followed by a decline in fertility at ages 15-29 and a simultaneous increase in fertility
[End Page 166]
at ages 30-49. The southern regions, on the other hand, were characterized by a first phase of sharp fertility decline at the older ages (due to a reduction in higher order births) and by relative stability at the younger ages. A second phase, beginning with the cohorts born in late 1940s, saw fertility at ages 30 49 decline slowly or stabilize, while fertility at ages 15-29 began to decline quite gradually up until the most recent data available (see Figure 7 – curves for all other northern and southern regions do not diverge significantly from those shown above).
An examination of Italian fertility on a regional level reveals two distinct patterns in reproductive cohort behaviour. In northern Italy, the decline in fertility slows or even stops beginning with the cohorts born in the 1970s. In several regions, such as in Emilia Romagna, fertility levels have even begun to increase, starting with the cohorts born in the early 1980s (Caltabiano, 2006). Although this increase has not been significant enough to reach replacement level, a continual rise in fertility over time could eventually stabilize at a TCFR of 1.6-1.7 children per woman. This level of fertility, accompanied by a sustainable flow of immigrants(8), may bring to an end the decline in the number of working-age individuals residing in the North.
In the South, fertility continues to decline for the cohorts born in the 1970s, while a clear trend has not yet emerged for the cohorts born in the early 1980s. There are also significant differences among the southern regions. For example, the fertility schedules in the regions of Abruzzo and Molise are becoming more and more similar to that of the northern regions. In Campania and in Sicily, fertility has decreased slowly, while in the regions of Apulia, Basilicata, and Calabria, the decline has been much more dramatic. This negative fertility trend will most likely continue over the next few years such that southern Italy will increasingly be characterized by a rapidly ageing population and by a considerable decrease in the number of overall inhabitants (ISTAT, 2002). The causes of such decline include smaller and smaller cohorts of newborns (made more apparent by the significant outflow of individuals towards the north and abroad – ISTAT, 2006b; GCD-SIS, 2007) and the limited appeal of the south for immigrants from less developed countries (Dalla Zuanna, 2005). [End Page 167]
Thus, the observations made by Rallu (1983) and Terra Abrami and Sorvillo (1993) continue to hold significance for the cohorts born up until the 1960s. The reproductive behaviour of these women certainly diverges across the Italian regions, as shown by the remarkable differences in cohort parity distribution between northern and southern Italy. We can therefore confirm the presence of two diverse reproductive regimes, deeply rooted in the social and economic disparities between regions (Santini, 1995; 2006; White et al., 2007).
Nevertheless, recent patterns in Italian fertility reveal a diminished variance between regions, beginning with the cohorts born in the 1970s(9). If this trend remains constant over time, regional differences may eventually become less marked, or even vanish altogether (see GCD-SIS, 2007 for further discussion).
The data presented in this short paper depict the general pattern of fertility in Italy and its different regions at the beginning of the twenty-first century. It would be possible to obtain a more complete picture if detailed cohort birth order data were available after 1997. This lack of information may partially be remedied by data from surveys such as the Famiglia and Soggetti Sociali Survey, conducted by ISTAT in 2003. This survey shows that for younger cohorts (born between 1960 and the early 1970s), the number of childless women is increasing, to as much as 17% – a level already experienced by women born in 1920 (Santini, 1995). Moreover, the proportion of first-order births is increasing slightly. Finally, there is a rise in second order births, while third and higher order births are decreasing (GCD-SIS, 2007).
Acknowledgements. The author would like to thank Franco Bonarini and Gianpiero Dalla Zuanna, as well as the anonymous referees of Population for their insightful comments. [End Page 168]
2. Complete data and figures for all Italian regions can be found in Caltabiano (2006).
4. Although ISTAT recently published an estimation of TFR and mean age at childbearing for 1997 and 1998 at a regional level using a complex ARIMA procedure (Battaglini and Iaccarino, 2006), they did not provide estimations of age-specific fertility rates. A comparison between Istat's results and the estimations in this paper reveal only minimal differences.
5. For example, for the 1957 cohort, the fertility rate at age 49 is estimated to be the same as that for the 1956 cohort at age 49; for the 1958 cohort, the fertility rate at age 49 is estimated to be the same as that for the 1956 cohort at age 49; at age 48 it is estimated to be the same as that of the 1957 cohort at age 48, and so on. The last TCFR estimated refers to that of the 1968 cohort, whose fertility rates between the ages of 38 and 49 are estimated to be respectively the same as those of the 1967 and the 1956 cohorts. This procedure may underestimate fertility, if the postponement of births continues over the years that follow, or if the recuperation is more significant than that of the recent past. The underestimation, however, is never higher than 0.02 children per woman, compared to a model with a constant increase in rates.
6. That is, 1950CTFR x – iCTFR x, where i is the cohort and x the age up to which age-specific fertility rates are cumulated.
7. Many scholars who, at the beginning of the present decade, attempted to estimate the TCFR for Italian female cohorts born in the 1960s often wound up underestimating fertility recuperation for these cohorts at older ages, due to a significant postponement of births. While the CCFR after age 29 for women born in the 1950s is less than a third of the TCFR, for women born in 1960s this same percentage is about half of the TCFR (see also Table 2 and Figure 7). More generally, changes in fertility schedule between one cohort and the next have occurred more quickly than scholars expected. Furthermore, studies at the turn of the century attributed a great deal of importance to the long-term structural components of fertility decline in Italy, which instead have played a lesser role than anticipated. However, note also that our estimations of the TCFR for the Italian cohorts of 1966 and 1967 (1.52 and 1.50 respectively) are consistent with those proposed by Sardon (2006) (1.49 and 1.47 respectively).
8. The role of immigration in the recent rise in period fertility in northern Italy is a controversial subject. Some scholars argue that the ever larger presence of immigrants in northern Italy has been a significant factor in the increase in fertility rates (GCD-SIS, 2007; ISTAT, 2006a). Others, however, believe that migrants have been less influential in changing fertility patterns. They suggest that the contribution of immigrant women to overall fertility rates in the last few years has fluctuated significantly and is so erratic that it is difficult to clearly define the role migration has played in changing fertility trends. Instead, these scholars often emphasise the recovery of fertility via postponed births on the part of Italian women who are over the age of 30 (Caltabiano et al., 2007).
9. In February 2007, ISTAT published TFR data by parity at a regional level (http://demo.istat. it/fecondita/index.html) demonstrating that birth distribution by parity remained unchanged in northern Italy between 1995 and 2004. In the regions of southern Italy (especially in Basilicata, Calabria, Sicily and Sardinia), the ratio of first births to total births increased while the ratio of 3+ order births decreased, suggesting at least a partial convergence toward the northern model.