Life Expectancy in the Interwar Period: Fixing Data Gaps for Lithuania and Explaining Early Demographic Transition in Estonia and Latvia

Abstract

Table 10.1 provides a compilation of the available data on life expectancy at birth, based on life tables for the Baltic countries and Finland in 1897–1938, which is included because of its role as a regional benchmark country in the perceptions of the indigenous populations of the Baltic countries. It also provides Gapminder (Lindgren 2020) estimates, which could not be neglected because of my interest (for application of the OOST) in the 1913 life expectancy values. They could only be interpolated or estimated from values in the years for which life tables could be constructed. I rely on my estimates, but those coming from Gapminder are useful as a control instance.

Table 10.1 provides a compilation of the available data on life expectancy at birth, based on life tables for the Baltic countries and Finland in 1897–1938, which is included because of its role as a regional benchmark country in the perceptions of the indigenous populations of the Baltic countries. It also provides Gapminder (Lindgren 2020) estimates, which could not be neglected because of my interest (for application of the OOST) in the 1913 life expectancy values. They could only be interpolated or estimated from values in the years for which life tables could be constructed. I rely on my estimates, but those coming from Gapminder are useful as a control instance.

Table 10.1 Life expectancy in the Baltic countries and Finland for 1897–1938

The first life tables on the Baltic countries were constructed by Ludvig Besser and Karlis Ballods (1897), covering the Baltic provinces (Estlandia, Livliandia and Kurland) of the Russian Empire. They were based on the findings of the regional census in 1881. According to these tables, in 1880–1883 life expectancy at birth (e°₀) of males was 39.1 years and 42.7 for females. This is far above the mean value of 31.32 years for males and 33.41 for females in 1896–1997 in 50 European provinces of the Russian Empire (Novosel‘skij 1916: 120, 125), being not far below the life expectancy in France in 1880–1883 (41.6 for males, 44.1 for females) and slightly above Prussia (37.6 for males, 40.7 for females) (Besser and Ballods 1897: 107–124).

The figures of Besser and Ballods refer to the three Baltic provinces as one entity. Hence, their data do not allow differentiating between the Baltic countries as they are known today. Moreover, 26% of the Latvian population lived in the Western part of the Vitebsk province, Latgale, which did not belong to the historical Baltic provinces. It had been under the Polish-Lithuanian rule since the sixteenth century and was annexed by Russia in 1772, while Estland and Livland came under Russian suzerainty in 1721. The Kurland province shared a common fate with Lithuania in 1795. Latgale differed from mainland Latvia culturally as its Latvian population was Catholic, different from the Protestant Lutheran Baltic provinces. It was also different economically, displaying lower productivity, and an inferior standard of living (see below in this chapter).

Historical changes in borders and administrative delimitations also complicate the interpretation of life expectancy estimates based on the All-Russian population census in 1897. Unfortunately, Sergei Novosel‘skij, who based his life tables on data from the 1897 Russian census, did not provide them for separate provinces or regions. Therefore, for Lithuania and Latvia in 1897, I use life expectancy values from abridged life tables for 1896–1897 by the Ukrainian demographer Mikhail Ptukha (1960), based on the 1897 census data and originally published in 1928.

Ptukha’s estimates include for the first time Lithuanians, whose life expectancy at birth was 41.12 for men and 42.40 years for women. Also, for the first time these data distinguish inhabitants of the Baltic provinces, providing life expectancy values for Estonians (males 41.61 and females 44.58 years) and Latvians (respectively 43.07 and 46.01). However, Ptukha was not interested in life expectancy of particular territories, but in its variation among the ethnic groups or “nationalities” of the Russian empire, including Russians, Ukrainians, Belarusians, Moldovans, Jews, Tatars, Bashkirs and Chuvashs.

According to Ptukha’s findings, the Baltic nations had the longest life expectancy, while Russians had the shortest, with 27.49 years for males and 29.82 for females. For Estonia, Ptukha’s tables are superseded by estimates on the territorial basis in the life tables constructed from the 1897 census and mortality data within the borders of contemporary Estonia by Kalev Katus and Allan Puur (1991: 2540). So they (45.5 years for males and 41.87 years for females) are used in Table 10.1 for Estonia in 1896–1897, while for Lithuania still there are no alternatives for the estimates made by Ptukha.

During the interwar independence period (1918–1940), two population censuses were held in Estonia, in 1922 and 1934. Using the data from the 1934 census, the Estonian statistical office constructed 3x1 format life tables for 1932–1934 (Riigi Statistika Keskbüroo 1937; Reiman 1936). According to Riigi Statistika Keskbüroo (1937: 38–39), life expectancy in 1932–1934 was 53.12 years for men and 59.60 years for women. Interwar Latvia held four censuses, in 1920, 1925, 1930, and 1935, with findings of the first census remaining partially unpublished. Based on data from the 1930 census, the Latvian statistical office published 4x1 format life tables for 1929–1932 (Valsts statistiskā pārvalde 1936). They used the 1935 population census for the production of 3x1 format life tables for 1934–1936 (Valsts statistiskā pārvalde 1938). According to these estimates, life expectancy at birth in Latvia increased from 54.56 years for males and 60.10 years for females in 1929–1932 to 55.59 years for males and 60.93 years for females in 1934–1936.

The research on life expectancy in interwar Estonia and Latvia came to a real breakthrough thanks to Katus (1955–2008), who constructed abridged (1 × 5 format) life tables for Estonia (1923–1938) and Latvia (1925–1938). Katus’ tables for Estonia are accessible in the Internet (Katus 2004). His tables for Latvia (Katus 2008) were produced working on the collaborative project on the health crisis in the Baltic countries coordinated by the French National Demographic Research Institute (INED). They were introduced by former collaborators in Vallin et al. (2017) in less detail (only life expectancy values at birth are presented) and published complete in Norkus, Jasilionis et al. (2022). Katus’ tables for Estonia and Latvia are my source for Estonia in 1923–1938 and Latvia in 1925–1938. The estimate for Estonia in 1922 is from Katus and Puur (1991: 2540). Katus only provided life tables for males and females, so the estimates for both sexes that are needed for comparison with the Gapminder estimates are derived simply as arithmetic means of both estimates.

In interwar Lithuania, only one (in 1923) population census took place, and its statistical office did not publish any life tables. Except for Ptukha’s life table for Lithuanians in 1897 (most probably referring to the ethnically Lithuanian part of the Kaunas/Kovno province), the only available life tables for the interwar period (1925–1926) were constructed by Antanas Merčaitis (1966) as part of his PhD dissertation, and were never completely published by the author himself.¹ According to Merčaitis, life expectancy at birth in 1925–1926 was 48.58 years for males, 51.89 for females, and an average of 50.25 for both sexes. Gapminder’s estimates for 1925 (50.5) and 1926 (51) are close but not identical to the last figure, suggesting that the compilers of this database did not know of or use Merčaitis’ work.

In order to provide the application of the OOST to Lithuania with a more solid basis than just the Gapminder (Lindgren 2020) guestimates, we constructed 30 life tables for the 1925–1934 period (Norkus, Jasilionis et al. 2022). Life expectancies for different age categories, extracted from these life table, are provided as supplements to this book together with Katus’ results on interwar Estonia and Latvia. Let us explain why this still was not done in research of historical demography of interwar Lithuania.²

Merčaitis’ pioneering estimates of life expectancy in 1925–1926 were only part of the historical introduction to the main part of his work, which, judging by its title, dealt with the issue of “reproduction of the population” in Soviet Lithuania, based on the 1959 census data. The limitations of the post-1959 period are likewise characteristic to the later work of Lithuanian demographers. With no census data to apply sophisticated quantitative demographic analysis tools, Lithuanian demographers generally avoid topics from the pre-1959 era.

On the other hand, Lithuanian historians continue to perceive quantitative research in historical demography as a subject outside of their field of professional competence. Thus, it escaped the attention of both demographers and historians that statistical publications of interwar Lithuania actually contain the data necessary for the construction of life tables, encompassing a broader time span than the 2 years covered by the pioneering contribution of Merčaitis.

The Lithuanian Statistical Office has published mortality data distributed by age since 1925. For the 1925–1927 period, this distribution included the categories of infants (0–1 years old), 0–4-year-old children, 5-year-long broad categories for the population aged 5–29, continuing with 10-year-long categories for the older population up to the 80+ age group. This distribution was refined from 1928, providing data for 5-year-long broad categories for the entire population aged from 5 to 100+. Due to the absence of such data for 1923–1924, Merčaitis constructed life tables for 1925–1926, assuming that in 1925–1926, the distribution of population by age remained unchanged.

However, the Lithuanian Statistical Office published estimates of population distribution by age for some later years as well. The last such publication is available for 1932 (Centralinis Statistikos Biuras 1933: 10). Besides this, mortality and population distribution data for 1928–1934 were published in annual surveys of the state of public health, published by the Health Department of the Lithuanian Ministry of Internal Affairs (Sveikatos departamentas 1928: 7, 10; 1929: 7, 10; 1930: 7, 10; 1932: 9, 12; 1934: 9, 12; 1936: 7, 10). This publication contains data on population distribution by age also for 1933 and 1934, which are missing in the annuals of the Lithuanian Statistical Office.

Neither Lithuanian statistical annuals nor public health surveys provide information on how estimates of the distribution of population by age were derived. It appears that the shares of population established by the 1923 census were used as benchmarks. However, there is no exact correspondence between the published figures and those which can be derived from population and subpopulation (males and females) totals, applying to them the exact shares of age categories established by the 1923 census. Either way, these data are authentic historical statistics, and are sufficient to construct life tables covering a 10-year- period (1925–1934).

Our tables also cover 1925–1926, which refer to Merčaitis’ tables. Different from his biannual tables, our tables are annual. Merčaitis used the 1925–1926 mortality means for each age category and population size on 1 January 1926, distributing the population into age groups according to proportions disclosed in the census on 17 September 1923. We used the midyear population, calculating it as a mean value of the population on the first of January over two contiguous years. Merčaitis applied the life table construction method of Russian Soviet demographer Vladimir Paevskij (1970: 17–46) to his data. We used the Excel template for life expectancy calculation, provided by the UK Office for National Statistics, which is based on the methodology of construction of abridged life tables developed by Ching Long Chiang (1979).³ To obtain age-specific death rates M_x up to the last open-ended interval 85+ and adjust for possible age-reporting problems at old ages, we apply the Gompertz-Makeham modeling (see Norkus, Jasilionis et al. 2022; Vallin and Berlinguer, 2006 for details).

Differences in methods may account for discrepancies between our and previously published estimates. To recall, according to Merčaitis, life expectancy at birth in 1925–1926 was 48.58 years for males 51.89 for females. According to our findings, life expectancy at birth in 1925 was substantially lower – 45.38 for males, 47.96 for females. We argue that our estimates are more realistic since mortality data used by Merčaitis are not adjusted for under-registration of infant deaths and most likely underestimate old age mortality. Similar problems may also have influenced previously published estimates by Katus (2008). Vallin et al. (2017: 196) proposed a correction factor for underregistration of infant deaths for the period since early 1950s to early 1990s. They are not large (less than 0.2 years). Different from Estonia and Latvia, which had civil registration offices, in Lithuania it was confessional communities that were entrusted with this task. Thus, in this country, only babies surviving until the baptism act were registered. We take a conservative approach assuming that the undercount of infant deaths in Lithuania in 1925–1934 was similar to that of the 1950s, that is, 9%.

As the official population estimates did not completely account for international migration, further refinement is needed to obtain more precise population denominators, which are necessary to compute death rates. With no noticeable immigration to Lithuania, the total emigration during 1925–1934 was 70,886 persons, which is 23.03% of the natural population increase (Vaskela 2011). Emigration did not affect all age groups equally, as most emigrants were young males. Most probably, corrected life expectancy figures should be slightly lower than our estimates. However, we leave these refinements for further research, which needs more detailed data on the distribution of emigrants by age.

Although still considerable, interwar emigration from Lithuania was lesser in comparison with the previous period. According to estimates by Liudas Truska (1961) (which remain undisputed), the total scope of emigration from “ethnographic Lithuania” between 1868 and 1914 amounted to 635,000 persons. This number includes 215,000 persons, born in Lithuania but living in other provinces of Russia (calculated from census data) as of 28 January 1897, and 74,000 ethnic Lithuanians who moved to these provinces between 1897 and 1914. The main destination for Lithuanian emigration was theUSA, with 55,000 ethnic Lithuanians immigrating between 1868 and 1998, and 252,594 between 1899 and 1914. Emigration to the USA involved most healthy members of the population—mainly 20–25-year-old males escaping conscription to the Russian army.⁴

While the first figure (55,000) about Lithuanian immigration to the USA is an estimate (from Račkauskas-Vairas 1915), the second one (252,594) is reported by US immigration statistics (Eidintas 2005: 61). Lithuanian emigration peaked in 1907–1913, almost equalling the natural population increase levels in the Lithuanian governorates of the Russian Empire. Huge emigration could not remain without consequences on the changes in life expectancy in Lithuania: healthy (and, as expected, long-living) males were removed from the population, and those which could expect not to be drafted because of health problems remained.1F.⁵

For my aims (application of the OOST guided by the CRHPS), the real problem is that data series are still too short. Regarding life expectancy in 1913, there is no other data apart from Gapminder (Lindgren 2020) guestimates for the total (both sexes) population. Besides that, Lithuanian data for the years 1935–1938 is missing. So male and female life expectancy values at birth for Estonia in 1913 are linearly retropolated from Katus and Puur (1991) for Estonia in 1897 and 1922 and Katus (2004) for Estonia in 1923–1938. For Latvia, 1913 values are estimated from Ptukha (1960) for 1897 and Katus (2008) for 1925–1938. For Lithuania, 1913 values are estimated from Ptukha (1960) and Norkus, Jasilionis et al. (2022) for 1925–1934. The same data is used in linear extrapolation to estimate life expectancies in 1938.

To make our data comparable with the Gapminder (Lindgren 2020) dataset, providing only estimates for total population, Table 10.1 presents also life expectancy figures for both sexes, calculated as means of male and female life expectancy values.⁶ We find that Gapminder’s estimates for 1938 are quite close to figures for Estonia and Latvia, implied by the Katus (2004) findings, but strongly (by 4–5 years) overshoot the Lithuanian values, implied by our data and calculations. As there is close agreement between Gapminder’s and our estimates for 1913, the disparity should be accounted by the exaggeration of interwar Lithuania’s health progress in the Gapminder database.

All available data on the male and female life expectancy at birth in the interwar Baltic countries is visualised in Figs. 10.1 and 10.2, including also Finland as a comparator country. Its inclusion helps to detect the impact on the health progress of the change in the socioeconomic system during the later period, which Finland successfully resisted in 1939–1940, while the Baltic countries failed.

Fig. 10.1

Female life expectancy at birth in Estonia (1922–1938), Latvia (1925–1938), Lithuania (1925–1934) and Finland (1922–1938). Author’s own production. Data sources: Estonia 1922—Katus and Puur (1991: 2540); Estonia 1923–1938—Katus (2004); Latvia 1925–1938—Katus (2008); Lithuania: Norkus, Jasilionis et al. (2022); Finland—HMD (2022). See also: Vallin et al. (2017)

Fig. 10.2

Male life expectancy at birth in Estonia (1922–1938), Latvia (1925–1938), Lithuania (1925–1934) and Finland (1922–1938). Author’s own production. Data sources: Estonia 1922—Katus and Puur (1991: 2540); Estonia 1923–1938—Katus (2004); Latvia 1925–1938—Katus (2008); Lithuania: Norkus, Jasilionis et al. (2022); Finland—HMD (2022). See also: Vallin et al. (2017)

So we find, firstly, in 1925–1938 Latvia was the best performing country among the three Baltic countries in terms of life expectancy. Estonia’s decadal life expectancy increase rate was slightly higher, but it was not sufficient to close the lag behind Latvia. Actually, in 1925 Latvian male life expectancy was slightly below (by 0.92 years) Estonia’s level, while the Estonian edge over the Latvian female life expectancy lag (2.23 years) was considerable. This could have been due to the legacy of Latvia’s greater destruction (including the dislocation of its population) during World War I, making recovery to the prewar economic level a much more formidable task.

However, by 1938, both Latvian males and females displayed superior life expectancy, Latvian males surpassing their Estonian peers by 1.42 years and Latvian females by 0.77 years. This restored the running order between the Baltic countries, which is disclosed by Ptukha’s findings for 1896–1897. Lithuania remained the worst performer for the complete period with available data. Importantly, despite its superior economic performance, interwar Finland did lag behind Latvia (see Table 10.1), but there was near parity with Estonia. So in this case, Latvia’s first rank corresponds to its top ranking among the Baltic countries according to output per capita (cp. Chap. 8).

During the complete period with data available for all three countries (1897–1934), Lithuania was the worst performing country. Importantly, Lithuania’s lag both behind Estonia and Latvia increased. For females, it enlarged from 3–4 for years in 1897 to 7–8 years in 1934. Life expectancy of Latvian males was by 2 years larger in 1897, and in 1934 it was by more than 4 years larger than in Lithuania. The life expectancy lag of Lithuanian males behind their Estonian peers was less considerable, in 1928 and 1933 the Lithuanian life expectancy value nearly reaching Estonian level. However, the life expectancy gap between Estonian and Lithuanian women was never lower than 3.68 years, as in 1933.

Secondly, extending the cross-country comparison to other age categories (see Tables S1–S6 in the supplement to this book), we find that differences across the Baltic countries were largest for life expectancy at birth. They are related to the much larger infant mortality (during first year of life) in Lithuania. Indeed, during the whole interwar period, infant mortality in Lithuania was almost two times higher than in Latvia (see Table 10.2). Differences in remaining life expectancy levels are quite conspicuous also for children (until the age of 10), fading in the older age categories. While for Estonia and Latvia the end of the first year was the year of top remaining life expectancy, this was not the case of Lithuania. As for the male population, five-year-old boys recorded the top remaining life expectancy. Again, this indicates significantly higher mortality levels for Lithuanian infants and children.

Table 10.2 Infant mortality in the Baltic countries, 1920–1939 (Promilles)

The fading of differences in the remaining life expectancy in adult age is most characteristic for males. Older Lithuanian men generally recorded a higher remaining life expectancy than their Estonian and even in some cases their Latvian peers. Again, this can be explained by higher infant and child mortality, which normally represents an early selection for survival. However, it did not apply to Lithuanian women, whose remaining life expectancy was shorter than that of their Estonian and especially Latvian peers.

Thirdly, among all age categories, life expectancy for infants and children increased most markedly in all three Baltic countries, while that for the adult population increased either insignificantly or stagnated. The decomposition analysis (see Norkus, Jasilionis et al. 2022) shows that the main driver of life expectancy progress was declining infant and child mortality in the age group 1–4. Reduction of infant mortality alone explains more than 50% of the total increase in life expectancy at birth. In fact, decreases in both infant and child mortality at age 1–4 years accounted for around 80% of the total longevity improvement. However, although infant mortality markedly decreased in Lithuania in 1925–1939, by 1938 it still lagged very considerably behind the other two Baltic countries and Latvia preserved its edge over Estonia (see Table 10.2).

Fourthly, the life expectancy gap between both sexes in Lithuania was smaller than in the other two Baltic countries. According to Ptukha’s estimates, life expectancy at birth for males and females in the ethnically Lithuanian population differed by only 1.28 years, while for Latvians this difference was 2.94 years in 1896–1897. For Estonians, this difference was 3.63 years according to Katus and Puur (1991), growing to 6.27 years in 1922 according to the same source. In 1925, it reached 5.65 years in Latvia according to Katus (2008), and 2.58 years in Lithuania according to our findings. In 1934, which is the last year when data based on life tables are available for all three countries, the life expectancy gap between males and females was 5.38 years in Latvia, 7.29 years in Estonia, but remained nearly at the same level as in 1925 (2.62 years) in Lithuania.

The increase of the gender gap in Estonia and Latvia is most probably related to fertility decrease in these countries, where birthrates dropped to a stable population reproduction level already by 1913. Birth complications were a major death factor for adult women before the rise of modern medicine and the advancement of the welfare state making obstetric services available for everyone. According to Neniškis (1998: 697), up to 7 per 1000 deliveries in Lithuania annually ended with lethal outcome for mothers in 1924–1939.⁷ Thus, a lesser frequency of births did increase the remaining life expectancy for women of childbearing age.

It can be argued that being raised in smaller households also increases the chances of survival for infants and children. This is an outcome of optimising in the context of ‘quality-quantity trade-off’ (Becker 1960; Becker and Lewis 1973). Children in small families may have tended to receive better care, nourishment and a more hygienic home environment. A larger family could have increased physiological stress due to overcrowded homes. It is more difficult to keep overcrowded spaces clean, and it may be a harder challenge to adhere strictly to personal hygiene. Even if parents could afford to expand their living space, children in larger families were at higher risk of infectious diseases (Hatton 2017: 183–184).

This argument may seem contradicted by the slightly smaller gender gap, higher birth and natural increase rates in Latvia than in Estonia (see Katus 1994: 94–95), despite the slightly higher life expectancy in Latvia. However, in the remaining part of this chapter I will argue that higher birth and natural increase rates in interwar Latvia are the result of the socio-economic territorial heterogeneity of Latvia. I will also argue that the most important factor accounting both for differences in the life expectancy gender gap and in infant mortality between Lithuania and the other two Baltic countries is the early start to fertility transition in Estonia and mainland Latvia (the Baltic provinces).

The Baltic provinces ‘may have been among those scattered regions of Europe where the so-called demographic transition—a sustained decrease in fertility and mortality rates—had already begun in the mid-decades of the nineteenth century, a generation earlier than in surrounding areas’ (Plakans 1995: 88). Earlier timing of the demographic transition in Estonia and mainland Latvia is indicated by the time series of crude birth rates (CBR) in the provinces of the Russian Empire (Table 10.3) and the then independent Baltic States (Table 10.4).

Table 10.3 Births (rates per 1000 population) in the Baltic provinces and the Vitebsk, Vilna and Kovno provinces in 1861–1913

Table 10.4 Births (rates per 1000 population) in Estonia, Latvia and Lithuania in 1915–1940

So as to understand the relationship between both tables, it is important to note that the Estland Province and the northern part of the Livland Province (together with parts of Saint Petersburg and the Pskov provinces) made up the territory of independent Estonia. Latvia consisted of the southern part of the Livland Province, almost all of the Kurland Province (except for small pieces, which were exchanged with Lithuania), and three counties (uezd) of the Vitebsk Province, also known as Latgale, populated by Catholic Latvians, while Lutheran Protestantism is the dominant confession in mainland Latvia. The main part of the territory of Lithuania was formed from the Kovno (Kaunas) Province and the western part of Wilno/Vilna (Vilnius) Province, augmented by the Suwałki (Suvalkai) Province, which was part of the Kingdom of Poland under Russian rule (Map 10.1).

Map 10.1

Boundaries of the tsarist Russian provinces (governorates) (1913; dotted lines) and contemporary borders of the Baltic States. Source: author’s own production, credits to Vaidas Morkevičius for his assistance

As was explained in the introduction to this part of the book, demographic transition refers to the historical shift from a Malthusian equilibrium between mortality and fertility, defined by high birth rates and high death rates, to a modern equilibrium, defined by low fertility and low death rates. Demographic transition is often associated with accelerating population growth, while both Malthusian and modern equilibria are described by very slow or zero natural population increase. It includes fertility and mortality transitions. Fertility transition ends with the total fertility rate (TFR; expected number of children women would have during their lifetime) being close to or below 2.

Mortality transition involves change in the structure of morbidity and causes of mortality. It starts with a receding of mortal pandemics killing both adult and child populations and continues with the replacement of infectious diseases as the dominant causes of mortality by man-made and degenerative diseases, represented mainly by cardiovascular diseases and cancer (therefore, it is also called epidemiological transition). Infectious diseases kill mainly children, infants and childbearing mothers. Their recession leads to decreasing infant and children mortality, and most people living to an older age and dying from degenerative diseases.

The increase of their relative proportion among the causes of death is promoted also by changes in the way of life, with the majority of the population moving to cities to exchange work in agriculture for that in industry and services, where work in many occupations is associated with low physical activity but high levels of stress. Closing of mortality transitions is indicated with the life expectancy at birth approaching the 60 years mark for both genders (Omran 1971, 1983, 1998; Olshansky and Ault 1986).

The mortality transition in Russia proper took place only under the intermediate socialist system, with life expectancy (both sexes) increasing from 33.34 to 67.65 (by 33.34 years) in 1913–1959, from 37.4 to 67.65 (by 30.25 years) in 1929–1959, from 41.4 to 67.65 (by 26.25 years) in 1938–1959 (Lindgren 2020). This makes Russia a very conspicuous instance of the accelerated epidemiological transition model, which experts in historical demography oppose to the classical or Western model.

In the last model, mortality started to decrease from the late eighteenth century due to early successes in the control of epidemics, improvement of nutrition and public hygiene, as well as expansion of the supply of medical services along with progress in biomedicine, making some formerly mortal illnesses curable diseases (Omran 1971; McKeown 1988; Fogel 2004). Our life expectancy data (see Table 10.1) indicate that the demographic shift in the three Baltic countries (with some reservations for Lithuania) can be described with the Western epidemiological model, with life expectancy levels surpassing those of ethnic Russia by more than 10 years already in 1897.

There is also a third—delayed mortality transition—model, which is exemplified by countries (mainly in Africa, Latin America and South Asia) where infectious diseases killing children and mothers giving birth were controlled only after World War II, due to the spread of new innovations in biomedicine (antibiotics). However, this was not accompanied by a decrease in birth rates, improvement in nutrition or in living standards. Among the former communist countries, Mongolia and Tajikistan exemplify this model (see Norkus 2023).

In addition to Russia proper, some Yugoslavian republics, Bulgaria, Romania and Moldova also seem to exemplify the accelerated epidemiological transition model. They share unchanging or even deteriorating life expectancy before World War I, followed by a very rapid improvement during the interwar period. However, by 1938 accelerated epidemiological transition was not yet complete, ending only in the 1960s under the socialist system. Most fSU republics also seem to represent this model, although improvement of life expectancy began here before World War I. A distinctive feature of belonging to this cluster of countries is life expectancy of the total population below or not much above 50 years of age by 1938.

The slower pace of life expectancy increase in Lithuania in 1913–1938 and its acceleration after World War II (see Table 11.1 in the next chapter) suggests that Lithuania may display similarities to these countries and belong to the Western epidemiological model only as its peripheral member. However, Estonia and Latvia (together with the Czech Republic, Hungary and Slovakia) may be included into this model without reservations, except that mortality transition was still incomplete also in these countries by 1938 because the replacement of infectious diseases as the dominant cause of mortality by man-made and degenerative diseases (and so the epidemiological transition itself) was still unfinished by this time. This is what the newly collected historical data on the causes of death indicate (see Table 10.5).

Table 10.5 Causes of death in the Baltic countries in 1925–1939

Although mortality from infectious and parasitic diseases declined over the 15 years documented by data, in the late 1930s, 20–25% of all deaths in Lithuania and Estonia were still caused by these diseases and closely related respiratory system diseases, including influenza and pneumonia. Latvia appears to be an exception because 45.9% of all deaths in 1938 were caused by degenerative diseases, such as neoplasms and circulatory system diseases. It may be tempting to consider this as another piece of evidence of Latvia’s leadership in the mortality transition and claim that it was almost complete by World War II in this country, defining completion as the absolute dominance (at least 50%) of degenerative diseases as all causes of death. However, in the case of Latvia, data on causes of death are available only for Riga, the only metropolitan city in the interwar Baltics. At the time of the last Latvian interwar census in 1935, only 34.62% of the total population lived in cities and towns (Norkus et al. 2021d).

Thus, we may guess that at the national level, the Latvian causes of death pattern was close to Estonia, with mortality from degenerative diseases amounting to 25–30% of deaths. This is in sharp contrast to the Lithuanian pattern (approximately 15%), but is still far from the 50% mark. We should be cautious, however, taking these data at face value because of the huge cross-country differences in the share of unknown death causes, which itself provides another important piece of evidence about disparities in the accessibility of medical services. In Lithuania, the share of deaths from unknown causes was larger by a factor of two or more than in Estonia, while in Riga the share of such deaths was less than 1%. This indicates that under urban conditions, nearly all cases of death were examined by medical professionals, but this was not the case under rural conditions in the interwar Baltic states (most probably, also in Latvia).

However, fertility transition in Estonia and Latvia was already over in the interwar years, when fertility in these countries fell below the replacement level, although they were still far behind the West European and Scandinavian countries in terms of GDP per capita. ‘Formation of the modern population in Estonia was completed prior to the Second World War, simultaneously with the pioneering countries of demographic transition in Northern and Western Europe’ (Puur 2011: 74; see also Katus 1990; Plakans 1984; Krūmiņš 1993; Krūmiņš and Zvidriņš 1976). That is why these countries were exposed to massive immigration from other Soviet republics during the period of Soviet rule, dramatically changing the ethnic composition of the population in these countries. There was no such early and rapid change in Lithuania, where demographic transition started and ended later.

As Plakans has noted, ‘because the demographic transition is usually associated with socio-economic modernization (industrial and technological growth, urbanization, high literacy rates), the Baltic area, by showing transition characteristics earlier, may have been an anomaly’ (Plakans 1995: 88). How can this “anomaly” be explained? The availability of reliable demographic data imposes the middle of the nineteenth century as the starting point of our time series. However, the origins of the disparities in the birth levels at this time can be traced back to much earlier times. Specifically, they can be related to the famous Hajnal line (Trieste-Saint Petersburg) crossing the Baltic countries, and separating two areas characterised by different levels of nuptiality (Hajnal 1965).

Already in the early modern times, a significant minority of women married late or remained single to the west of this line. Therefore, marriage rates and thus fertility were comparatively low here. To the east of the line and in select pockets of Northern Europe, early marriage was the norm, while high fertility was offset by high mortality. Although evidence may be too scarce for such conclusions, many experts place the Baltic provinces to the west of this line, while Lithuania is considered as a mixed case or a broad frontier area (Leinarte 2017: 62–3). However, the problem is why birth rates started to decrease in the Baltic provinces at least one generation earlier than in the neighbouring regions? Why did they decrease here more rapidly than in some countries to the west of the Hajnal line, which were far ahead of them in terms of economic advancement?

In the original West European version of demographic transition, it was preceded by an increase in GDPpc, followed by the decrease of mortality. With fertility remaining at the former level or even increasing, the outcome was a rapid increase in the total population. The fertility decrease did lag for some time behind the mortality decrease, with mortality and fertility reaching a new equilibrium at a much higher level in comparison with the original Malthusian regime, defined by high fertility, high mortality and (therefore) short life expectancy at birth (around 30 years of age). Together with France, the Baltic countries belonged to the group of exceptions or deviations from this baseline demographic transition model. In these exceptions, fertility decrease preceded mortality decrease and marked GDP increase (on the last point, further research in the quantitative economic history of the Baltic provinces in the nineteenth–early twentieth century is needed). Therefore, demographic transition in the Baltic countries did not double or triple the population there in only a few decades, which was the case in countries exemplifying the baseline version of the demographic transition model.

I will argue that differences in the ways the agrarian evolution occurred provide a clue explaining the variation in the timing of the demographic transition in Eastern Europe. By the early nineteenth century, the agricultural political economy of the Baltic provinces was not very different from that in the neighbouring East European countries. Folwarks were its backbone—landed estates, exploiting the work of serfs. Unlike medieval manors, folwarks were commercial enterprises, producing their output for market (mainly foreign) sale (Strods 1972; Žiemelis 2013). Serfs lived on the farmsteads, which were mainly subsistence production units on land owned by the landlord. Heads of farmsteads had to provide working hands to labour in the landlord’s fields. The number of hands was dependent on the size and quality of the land allotted to a farmstead. The welfare of a farmstead was conditional on the size of its internal workforce.

Farmstead heads with many sons (including those already married) and daughters of working age had a sufficient workforce not only to fulfil their labour dues, but to also properly cultivate the allotted parcel, which usually was the only source of the farmstead’s revenue. The heads of farmsteads with no sufficient internal workforce were expected to hire working hands and send them to work on the landlord’s fields, with no sufficient workforce left to meet the tight production deadlines imposed by weather conditions on their “own” landholdings. Therefore, farmsteads usually hosted large, complex, extended households, including several conjugal families and unmarried adult relatives (Plakans 1975: 4–5; cp. Vyšniauskaitė et al. 1995: 109–115).

In 1816–1819, serfdom was abolished in the Baltic provinces, inaugurating the capitalist era. The liberation of serfs in Russia (in 1861) was accompanied by land reform, involving the state, the landlords and peasants. Peasants could keep their plots, while landlords were compensated by the state, which collected a repurchase tax from the former serfs. Until the 1905–1907 revolution, when the Russian government abolished this tax and enacted a new agrarian reform to create a farmer-owner class, very few peasants were able to pay the full sum of the repurchase tax.

In Russia, the peasant community was the taxation unit, considered also as the collective owner of the allotted land to be repurchased. In Lithuania, parts of Belarus and Eastern Latvia, this unit was an extended family, holding the allotment. Until the repurchase tax was paid, allotments could not be sold, divided or transferred by other means. Although de facto divisions among adult children did happen, they were considered illegal. This particular landholding regimen preserved extended families, consisting of several nuclear families, who had no strong incentives to control fertility.

In the Baltic provinces, the serfs were freed without land, and all estate land remained the landlord’s property (Plakans 2001). Peasants were now considered the landlord’s tenants, renting his land. On the ground, nothing changed in terms of relations between landlords and farmstead heads, as nearly all of them paid their rents through labour dues. However, in 1849–1868, serf emancipation was supplemented by a protracted land reform, which was very different from its Russian pendant (Kozin 1976; Niedre 1986; Raun 2001 (1987): 43–49, 68–70).

First, some of the tenants received the right to purchase their farmstead and the surrounding lands. Second, their labour dues were replaced by monetary rents. The total annual payment of a farm’s tenant consisted of rent and land repurchase annuity. Heads of farmsteads, who were invited to make such land lease-to-own deals, had to assume complete, individual responsibility, risking losing the farmstead and their movable property to be sold at auctions if they failed to meet the payment deadlines. Under the 1861 Russian regimen, holders of the allotted land were collectively liable for paying ordinary and repurchase taxes to the state. They risked having to sell their movable property at auctions to pay for tax arrears, but they could not lose the land.

The land property rights settlement in the Baltic provinces removed the incentive to have a large number of children, characteristic for the folwark system and preserved by the 1861 system in other provinces of Russia. In the Baltic provinces, the rural population was divided into two classes. A small portion (20–30%) of the rural population were tenants of large farms (Vaskela 1998: 77–78). Many of them succeeded in repurchasing their homesteads and becoming well-to-do farmers, referred to as “grey barons” by their less successful compatriots. Only large tenant farms, run as efficient capitalist enterprises, could meet the conditions imposed by the lease-to-own deals. One could not even consider the partitioning of such farms. Therefore, only one son could expect to inherit the farm, while the other children were expected to look for employment elsewhere once they grew up. Instead of drawing on internal labour, such farms relied on hired farm hands, who (the system very different from that of other regions of tsarist Russia, including Lithuania) comprised the absolute majority (more than 60%) of the rural population (Siilivask 1990: 237–238).

Farmhands (both men and women) usually had only a 1-year-long work contract. Most of them were highly mobile, and moved from estate to estate, and from farm to farm every year in search of better employment conditions. This lifestyle discouraged both marriages and bearing children. However, the few late-born children in small households of the farm heads had a better chance to survive than their parents or grandparents, who were raised in extended families that prevailed before the middle of the nineteenth century. As a result, ‘rural marital fertility had already been substantially reduced before 1900 in the provinces inhabited primarily by Latvians and Estonians, and very little or not at all in other provinces in European Russia’ (Coale 1979: 113).

Finland and Lithuania provide two contrasting cases which help to substantiate the thesis of the decisive role of the Baltic mode of agrarian reforms for imparting demographic transition in Estonia and mainland Latvia, with its distinctive features. On account of the late start of its demographic transition, Finland’s population increased from around 1.8 million in 1870 to 3.1 million in 1917 and continued to grow during the interwar period to 3.7 million (Hjerppe 1989: 192). Between the census of 1881 and 1922 in the Baltic provinces, the population of Estonia increased from 881,000 to 1.1 million and then stagnated during the interwar period (Raun 2001 (1987): 246). To understand such a sharp contrast, we must consider that in Finland there were no folwarks or serfdom. Therefore, there were no agrarian reforms in the nineteenth century in Finland, with its agrarian and demographic development moving along the traditional path, marked by the predominance of small farmer landholdings.

The main class division in rural Finland was not between landlords and farmers, but among farmers themselves, differentiating them into landowners and tenants (crofters and cottagers) with long-term lease contracts. Due to the rapid population growth, many regions in Finland suffered from agrarian overpopulation by 1914, until it was eased by urbanisation and industrialisation after the interwar period, also changing the procreation behaviour of the population according to the standard scenario of demographic transition.

In the Lithuanian provinces, survival of extended households did not create very strong incentives to postpone marriages and control fertility for both the advantaged (minority) and disadvantaged (majority) parts of formerly enserfed populations, allowing for fragmentation of landed property and perpetuating the Malthusian dynamics of agrarian overpopulation. Actually, according to legislation from 1861, until the repurchase tax was paid, allotments could not be sold, divided or transferred by other means. Although de facto divisions among adult children did happen, they were considered illegal.

After the Stolypin reform in 1906 abolished legal obstacles for the legal division of allotments, the surviving extended households were divided, and Lithuania became a country of small landholdings. In 1913, the mean size of a farmer landholding was 15.2 ha in Lithuania, 29.4 ha in Estonia (Vaskela 1998: 57–58), 47.0 ha in the Latvian part of the Livland (Vidzeme) province and 41.5 ha in Kurland (Skujeneeks 1927: 402). However, the mass emigration to the USA that had been taking place since the late 1860s eased population pressure in Lithuania, preventing rapid population growth (Norkus et al. 2020: 599–601).

The fate of rural proletarians and the nutrition conditions of their children in the Baltic provinces improved significantly only after World War I. During the very first years after the establishing of independent states in 1918, radical agrarian reforms were implemented in all three Baltic countries. In Estonia and Latvia, this reform dispossessed German landowners. The land was distributed among landless farmhands, with the best allotments going to those who volunteered for military service during the independence wars in 1918–1920. However, except for a brief post-war “baby boom”, birthrates continued to decrease until the very end of the interwar period (see Table 10.4).

On the other hand, by transforming most of the former landless farmhands into subsistence farmers, the land reform created a permanent labour force shortage problem for the “grey barons”, who could not run their large commercially oriented farms without hired workers. This problem could only be alleviated by significant seasonal migration of agricultural workers. These workers came from Lithuania and Poland—mainly from those regions which the interwar Republic of Lithuania considered as “Polish-occupied Eastern Lithuania”. In Latvia, another supplier of farm hands was Latgale, which in terms of its culture and socio-economic conditions was closer to Lithuania than to regions of Latvia, which formerly belonged to the Baltic provinces.

Life expectancy differences between Estonian and Latvian males and females were noticeable already in 1897, indicating that Estonian women reaped life expectancy increase dividends due to the fertility decrease already before World War I. Although most Estonians born in 1890–1899 grew up in poorer (in comparison with the UK, Sweden or Germany) families, most of these families were small, providing better care, nourishment and a more hygienic home environment for their few children. This accounts for a lower infant and child mortality in comparison with Lithuania.

Another factor was the weaker supply of medical services in Lithuania. In Lithuania, there was one physician per 2900 of population, and 1.7 hospital beds per 1000 of population in 1938. In Latvia, there was one physician per 1247 of population, and 6.7 hospital beds per 1000 of population. In Estonia, in the same year there was one physician per 1183 of population and 4.3 hospital beds per 1000 of population (Reichskommissar für das Ostland 1942: 73; 112; 151).

As former parts of the Russian Empire before World War I, all three Baltic countries inherited its social legislation and built welfare institutions on the basis of this received infrastructure (Mančinskas 1971: 10–11; Mucinieks 1934: 18; Peep 2005: 20–34). This common legacy included legislation on accident (1903) and sickness insurance (1912) for workers in large industry enterprises, following the example of Bismarckian legislation in Germany. Lithuania started to build up or transform this legacy only by the end of the first decade of independence. In 1926, the law on the establishment of territorial sickness funds, co-financed by compulsory contributions by employers and employees, was passed (see Norkus, Morkevičius et al. 2021).

However, its implementation only started in 1929 (Mančinskas 1971: 57), with the government’s contribution limited to funding administrative costs and small maternity benefits for family members of the insured. It was not more significant in funding the implementation of the law on compulsory insurance of urban workers and employees against accidents, passed in 1936. It somewhat increased only when in 1939, insurance against accidents was extended to the agricultural working population (Mančinskas 1971: 129–130). However, the agricultural population remained excluded from sickness insurance schemes until the end of interwar independence.

In Latvia, the received Russian sickness insurance legislation was upgraded already in 1920, extending it to the entire employed urban population. It was co-financed by compulsory contributions by employers and employees, but was heavily subsidised by the government, its contribution adding up to 25% of the total revenue of sickness funds by 1930 (Aizsilnieks 1968: 576). The government mainly paid the cost of sickness benefits of the agricultural population, when after 1928, the sickness benefits scheme was extended to agricultural workers and farmers. In 1927, Latvia’s parliament passed the “Law on insurance of employees against accidents at work and sickness” to upgrade the Russian 1903 and 1912 legislation. By 1930, it covered 350,000 persons (18.4%) out of a total population of 1.9 million in Latvia (Zālīte 1999: 108), with accident insurance of the self-employed (still a majority in the population of the country) remaining voluntary.

In Estonia, the main novelty in addition to social insurance schemes received from the Russian period was introduced in 1924 with the passing of the pension law, which entitled state employees to state pensions. In 1926, it was extended to workers of state-owned enterprises (Ahelik 1964: 194). However, by the end of interwar independence, only some 3% of the 1.1 million population of Estonia was entitled to old age pensions (Ahelik 1964: 195). The received Russian legislation on insurance against accidents was modified several times, extending this insurance to agricultural workers in 1936. Otherwise, the agricultural population remained excluded from health insurance schemes because compulsory health insurance co-financed by contributions from employers and employees extended only to the employed urban population.

The differences in the development of the welfare state between the Baltic countries broadly correspond to those in economic development: Latvia ranks first, followed by Estonia and Lithuania closing the running order (Norkus, Morkevičius et al. 2021). The early establishment of the authoritarian regime in Lithuania (1926) did not favour the build-up of a welfare state, which had been the priority of socialist parties. The agenda of a ‘democratic class struggle’ (Korpi 1983) in 1920–1934 was very similar in Estonia and Latvia, with the main cleavage separating the agrarian parties, who promoted the interests of farmers, from the socialist parties, who conceived themselves as representatives of the urban working class and rural proletariat (see Graf 2000; Stranga 1998). The “historical alliance” between agrarian and socialist parties in the Nordic countries in the 1930s accelerated the growth of the welfare state in these countries, while the inability of the “Baltic sisters” of these parties in Estonia and Latvia to make compromises necessary for such alliances may explain why democracy lost out here in 1934 (Luebbert 1992: 259–266; Stranga,1998: 36–72).

A critical juncture in the parting of ways between Latvia and Estonia was the failed attempt at a communist coup in Tallinn on 1 December 1924. Legal socialist parties were staunchly anti-communist, and were not implicated in the preparation coup, continuing to participate in the coalition governments. However, after the coup’s failure, public opinion about leftist ideas had changed in Estonia, putting the socialists in a defensive position where they had to spend all their energy on defending themselves against the demagogic allegations of their political opponents that they were just “communists in camouflage”.

Tellingly, Estonia became one the few interwar European countries with a grassroots mass proto-fascist movement Eesti Vabadussõjalaste Liit (Union of Participants in the Estonian War of Independence; colloquially vapsid), which was prevented from gaining victory at the election by the autogolpe coup launched in 1934 by the incumbent government under Konstantin Päts, which had initially also been supported by the socialists, as they were the prime target of vapsid’s demagogic incriminations of crypto-communism (Kasekamp 2000).

Sitting in their strategic defence position, the Estonian socialists could not follow the example of their Latvian colleagues, even when they participated in governing coalitions. Very differently, both in the government and the opposition, the Latvian Social Democratic Worker Party successfully promoted several important reforms, making Latvia the leader in building a welfare state among the Baltic countries.