BENCHMARKS AND DISPUTED TRENDS IN BULGARIAN GDP

Though there continues to be room for disagreement about the appropriate benchmark figures for European per capita incomes in 1913, particularly as concerns Spain, Portugal, and Italy, there is a high degree of consensus amongst more recent publications in placing Bulgarian GDP just prior to World War I at between 1,456 and 1,521 Geary-Khamis dollars of 1990 and thus at the very bottom of the European league. The method through which this benchmark is arrived at varies according to the authority one chooses. In the case of Lains, the reference is to Gini et al.'s estimates of 1959, which provide a ratio relative to U.K. GDP per capita of 0.309.¹³

Maddison arrives at his figures through a different route, referring in the first instance to the 1957 edition of Colin Clark's Conditions of Economic Progress.¹⁴ However, a check on Clark's footnotes reveals that his tables for Bulgaria also relate back to an earlier publication by Gini, which may help to explain the agreement between the Maddison-Clark estimates and those derived by Lains directly from Gini.¹⁵ Bairoch was somewhat less optimistic, placing Bulgarian GDP relative to that of Britain at 26.6 percent.¹⁶ Until recently, the major exception to the consensus were Good and Ma's figures as published in 1996, which placed Bulgarian GDP per capita at only 0.23 of the British benchmark.¹⁷ However, in his more recent publication with Ma, Good too has converged on a ratio of 0.309.¹⁸ In light of this agreement, we feel relatively secure in taking ca. 1,500 Geary-Khamis dollars of 1990 as our best guess for Bulgarian GDP per capita in the period immediately prior to World War I.

Far more controversial is the question of growth rates. For lack of adequate nineteenth-century sources, the most common method for estimating growth rates before 1914 has been the use of proxy measures. The most technically sophisticated effort in this direction has certainly been that of Good and Ma.¹⁹

Using maximum likelihood methods they have constructed a proxy equation with which to estimate GDP growth in the eastern European countries for which no appropriate data are available. The basis of their estimates, however, can only be described as fragile. Good and Ma estimate movements in GDP between 1870 and 1910, on the basis of a constant term and a cocktail of only three proxies: the crude death rate, the number of letters posted per capita, and the share of nonagricultural employees in the workforce. As they candidly admit, their methodology assumes the stability of the structural relationship between these three proxy variables and GDP over time and the applicability of this same equation across the entire continent of Europe. And these claims are put under severe pressure by the startling nature of their results. Restricted by the availability of sources, Good and Ma limit themselves in the case of Bulgaria, to an estimate for 1890 and 1910. But this shows GDP per capita in postindependence Bulgaria growing at a vigorous 1.3 percent per annum, in line with the European average and faster than in Britain, France, and the Benelux countries. Their optimism with regard to Bulgaria is typical of their broader conclusions. Good and Ma are ardent advocates of the convergence hypothesis. The “territories of Central and Eastern Europe” they argue “… took full advantage of the growth potential inherent in their relatively low income levels in 1870.”²⁰ For this part of the European periphery at least, they claim that convergence was a reality. They qualify this conclusion only with regard to Spain, Greece, and Italy. The Mediterranean they conceded was indeed the “slow growing region” of Europe. Though Maddison expresses skepticism about Good and Ma's proxy methodology, he has incorporated their growth rates for Bulgaria, Poland, and Romania into the most recent edition of his database, on grounds of their general “plausibility,” pending the availability of direct estimates.²¹

However, serious doubts must be raised about the Good and Ma proxies at least as far as Bulgaria is concerned. Of their three key variables, only one in fact points consistently in an upward direction in Bulgaria prior to 1913. As will be discussed, the share of nonagricultural population, surely the most inherently significant indicator, was stuck in Bulgaria at a very low level for the 50 years following independence, with no decisive upward trend. The crude death rate fell between the early 1890s and the immediate prewar period, but the recorded figures after 1890 were consistently worse than those reported for the first decade after independence.²²

The remarkably rapid growth rate claimed by Good and Ma must therefore be attributable to their third variable, which measures the rate of postal activity. The growth in postal services in Bulgaria was certainly spectacular. The number of post offices and postal agencies rose from 41 in 1879 to 374 post offices and 1,871 private postal agencies by 1911. The number of letters dispatched increased more than 40-fold between 1881 and 1911.²³ Nor can there be any doubt that this was indicative of a generalized process of modernization, including the rapid spread of literacy, a point to which we will return. But such spectacular and clearly disproportionate growth in one small sector is hardly the basis on which one would wish to make bold claims about convergence in per capita national incomes. The danger of doing so is further revealed if we examine not rates of growth but absolute levels of postal traffic. According to contemporary postal statistics, though the number of post offices per capita in Bulgaria and Britain in 1911 was remarkably similar, the volume of letters and postcards dispatched per capita in Bulgaria was only 7.5 percent of that in Britain. As we have already seen, this is far lower than any credible estimate of relative per capita income in 1913. In short, Bulgaria in 1878 suffered from a particularly severe developmental lag in postal traffic, which is hardly surprising given the very low levels of literacy prevailing at the time. It is this which accounts for the explosive growth in this proxy variable, which cannot therefore be taken as a reliable guide to general trends.

Good and Ma's is not the only exercise in proxy methodology for Europe's periphery. An earlier experiment was carried out by Bairoch in the 1970s. Though he did not make his methods entirely explicit, Bairoch relied on a wider bundle of 12 proxy indicators, of which two—the crude death rate and the percentage of the workforce in nonagricultural employment—match those employed by Good and Ma.²⁴

Bairoch, however, arrived at far less optimistic conclusions. Over the entire period between 1860 and 1913 he estimated that Bulgarian GNP per capita increased by only 25 percent, compared to a growth rate for the continent as a whole of 75 percent over the same 50 year period. These figures placed Bulgaria amongst the slowest growing economies in Europe. Only Spain and Portugal were more sluggish, and Spain, at least, enjoyed higher levels of GNP per capita than Bulgaria both in 1860 and 1913. Bairoch's Europe, in short, was emphatically nonconvergent. More recently, Foreman-Peck and Lains and Şevket Pamuk have followed him in this assessment.²⁵ As Foreman-Peck and Lains note in a recent paper, they can find no evidence that “the poorest regions of Europe showed” any “tendency to catch up with the most prosperous economies” in the period before 1914.²⁶

Nor are Pamuk, Foreman-Peck and Lains the most pessimistic voices. Palairet and the Sundhaussen agree in seeing both the Bulgarian and Serbian economies as locked not into a cycle of slow growth, but of actual decline. To make this case for Bulgaria, Palairet draws on a wide variety of indicators, one of which—his agricultural output index—we will return to below. But in making his case for decline, Palairet also invokes Bulgaria's little appreciated tradition of national income estimation. In the late nineteenth century newly independent Bulgaria was amongst a cluster of countries to produce their first estimates of national income.²⁷

In 1915 Kiril Popov, the founder and first Director of the Statistical Office of Bulgaria, published spot estimates for Bulgarian national income for 1892 and 1911.²⁸ For 1892 Popov claimed a total national income of 1109 million leva. For 1911 he put the figure at 1647 million leva. This gives nominal income growth of just short of 50 percent. Popov did not say so, but as Palairet points out, the implication of these numbers, allowing for retail price inflation of 41.3 percent and population growth of 32 percent, is that real per capita incomes in Bulgaria in the generation prior to World War I may have fallen, perhaps by as much as 20 percent.²⁹

Given the startling implications of this claim, the first aim of this article is to reconstruct Popov's estimates and to explore the interpretative problems they raise. As our template for the construction of Bulgarian GNP estimates for the pre-1914 period, we draw on the interwar work of Asen Chakalov. In the 1930s Chakalov was associated with the Statistical Institute for Economic Studies at the Sofia State University headed by Oscar Anderson.³⁰

He was fully conversant with the new methods of national income accounting being developed in Europe and the United States at the time. At the end of World War II Chakalov published a study of Bulgarian GNP covering the period 1924–1944. His results are fully documented and conform to international conventions of national income estimation. They have become a standard reference in the literature amongst others for Maddison.³¹ Given the sophistication of Chakalov's estimates and the desirability of linking new data to his estimate for 1924, we proceed by replicating Chakalov's figure for 1924 on the basis of original sources. Then, using these same sources, we create a series of fully documented figures for the years between 1892 and 1921. These direct estimates provide an important check on the proxy measures hitherto cited in the literature. Furthermore, they address a second key question, which has been overshadowed in recent years by the focus on the pre–World War I period: the question of how the pre-1914 period ought to be linked to the experience of the 1920s and thus how we narrate the economic history of Bulgaria in the interwar period.

COMPILING A NEW GNP ESTIMATE

As Rositsa Rangelova has recently shown, Chakalov's national accounting methodology corresponds closely to modern definitions of Gross National Product (GNP), i.e., the total income of Bulgarian individuals and legal entities regardless of their location within or outside the national borders.³²

In contemporary terms Chakalov agreed with Alfred Marshall that only labor valued in money terms by the market could form GNP. Domestic labor such as cooking, washing, or cleaning unless done by servants was not included. On the other hand, given the predominance of subsistence agriculture in Bulgaria, a national income estimate would be meaningless unless it made ample allowance for farmers' own consumption of agricultural output and subsistence manufacturing.

As in the original calculations by Popov and Chakalov, arable agriculture emerges as the largest component of our estimate, constituting 30.5 percent of the total in 1911.³³

These figures were calculated by multiplying output volumes with farmgate prices as given in the national statistics. Feed is counted as part of arable output and deducted as a cost from the output of livestock farming. We follow Chakalov in making a deduction for seed. Strictly speaking, because the stock of seed is capital, this means that our measure of arable output is adjusted for at least this element of depreciation. Such a procedure is not ideal. However, it was necessary to make this adjustment to preserve continuity with Chakalov's figures. Fortunately, because the percentage used as seed is fixed, it does not distort the movement of our series over time.

The calculation of value added in livestock farming (16.2 percent) rests on the number of beasts in key census years and average milk and meat yields minus the costs of feed supplied from the arable sector. For sake of consistency we have applied the milk yields used by Chakalov with a downward adjustment for the earlier period. For 1924 Chakalov reckoned with 1,000 liters of milk per cow per annum. For 1892 we assume 900 liters. The same technique is also applied for wool and other animal hair production. The output of eggs (0.8 percent of the total in 1911) is estimated from the transport statistics.

The production of forestry and fishing (3.4 percent) is derived from the official figures published by General Directorate of Statistics. The output of home industry (8.5 percent) is calculated by multiplying the number of females, aged between 16 and 60 years by the average number of days spent for home work based on peasant account books and contemporary surveys. A similar procedure is applied for servants (0.2 percent). Implicit rental income of both rural and urban dwellings (13 percent) is based on census information about the number of buildings and the yearly data on rents. Interest on agricultural credits is subtracted, following Chakalov's methodology. To conform to the definition of GNP, we also add to the total of Bulgarian rural incomes the remittances of seasonal migrants.

About the production of large-scale industry and mining (3 percent of the total in 1911), we possess fairly good statistical information, thanks to the protective policy of the Bulgarian government. Value added is arrived at in the normal fashion by deducting input costs from sales figures. To conform to the definition of GNP we also deduct interest and dividend payments to foreigners. Only for 1892 and 1899 were we compelled to use coal consumption and the number of factories to arrive at a reasonable estimate of the industrial output.

For construction and the crafts sectors the source situation is less good. Figures for public construction were taken from central and local government budgets. For private construction and the craft sector (6.2 percent), we follow Chakalov in relying on census figures for the number of craftsmen and builders, multiplied by official statistics of the annual earnings of masons, masters, journeymen, and apprentices.

Due to the significant presence of the state in the transport and communication sector we have at our disposal very detailed data on their income net of working expenses, accounting for 2.1 percent of the total. The same applies to a large extent to the financial services sector (4 percent). Only for private banking or insurance institutions (1.9 percent) are we compelled to resort to profit and loss accounts published in the State Gazette or preserved in the archives.

The relatively low level of marketization of the economy creates a serious problem for the calculation of commercial incomes (6.5 percent of the total in 1911). Following Chakalov we construct a proxy by assuming a fixed markup on the sales of agricultural goods and the value added of industry and the import trades. The flow of traded agricultural goods is approximated by summing the value of exports and the value of commodities transported by the railways. The industrial and craft figure is arrived at as for agricultural production. To avoid double counting, industrial output is netted of local raw material consumption and of excises and taxes paid. The income of commerce is taken to be a fixed 15 percent share of this total sales figure for agriculture, industry, and crafts.

Aggregate incomes of freelance professionals and services (2 percent) were estimated by applying estimated salary levels to census data. For lack of alternative information, the salary levels were derived from masons' wages, adjusted for salary differentials and the longer working year of white-collar employees.

Earnings from state employment (4.5 percent of total in 1911) were taken directly from budget data for wage bills and the public works expenditure of national and local government as well as autonomous state agencies. Again, to conform to the definition of GNP, interest owed on Bulgaria's foreign debt is subtracted.

To summarize: For 47.6 percent of the estimate (in 1911) we used physical output data multiplied by prices, minus input costs; for 9.3 percent we can draw information on salaries and wage bills directly from state budgets and business accounts; for 36.5 percent of the estimate we are relying on employment census data multiplied by estimates for annual earnings; 6.6 percent of the estimate, covering the commercial sector, is derived by more approximate methods.

GNP OVER THE LONG RUN: THE PROBLEM OF DEFLATION

Applying Chakalov's methodology to the obvious official sources we achieve a reassuring match with his estimate for 1924. In addition and perhaps more surprisingly we are also able, using Chakalov's methodology, to replicate Popov's nominal estimates both for 1892 and 1911 as well as adding new data points for 1899, 1905, and 1921. (See the estimates compared in Table 1.)

To provide a meaningful comparison over the long run between 1892 and 1945, we must allow for price changes, population change, and the impact of territorial redistribution in the course of the Balkan wars and World War I. The latter two points will be discussed below and present few areas of disagreement. The choice of proper deflators, however, is a potentially controversial one.

Instead of deflating the nominal national income figures using the retail price index, we follow Chakalov's procedure in applying specific price series to each component of our GNP estimate. We prefer this methodology, because in an economy in which the vast majority of the population were self-provisioning in food and in which heavy tariff protection distorted the urban price level upwards, deflating total output by an urban cost of living index will result in serious biases. To arrive at a constant price series for agricultural output, we apply farm gate prices of 1911 to the physical output series. For lack of subsectoral data, industrial output is deflated by an unweighted average of the available wholesale prices for the different lines of industrial production. Employment data taken from the census were multiplied by wage and salary figures for 1911. In our constant price series, deflation by means of the retail price index is thus confined to the incomes of civil servants, financial services, and commerce, or just 13.6 percent of the total estimate. Because these were the sectors of Bulgarian society most dependent on the market, this seems a reasonable procedure.

Adopting this more sophisticated deflation procedure turns out to have major implications. Whereas Popov's retail price index rises by a surprising 51 percent between 1900 and 1911, producing the highly pessimistic story suggested by Palairet, our implicit “GNP deflator” increases by only 27 percent between 1899 and 1911. This discrepancy is largely accounted for by the heavy weight we accord to farm gate prices for agricultural output, which were under the influence of global price trends.

As Figure 1 vividly illustrates, our reconstruction of the national income data suggests a substantial revision of existing views of Bulgaria's economic development between the late nineteenth century and the end of World War II. The following discussion will focus on the period before 1914. We will return to the interwar period thereafter.

By confirming Popov's nominal data and applying to them a more sophisticated deflation procedure than that suggested by Palairet, we steer a middle course between his pessimistic story of economic decline between 1890 and 1911 and the standard growth narrative peddled by Good and Ma. In spirit, our data are closest to the slow growth scenario first suggested by Bairoch and since adopted by Foreman-Peck and Lains. We qualify their story, however, to the extent that whatever growth may have occurred between the 1860s and 1913 must have taken place prior to 1890. Between the early 1890s and the mid 1930s, the best available evidence suggests that Bulgaria experienced very little growth in per capita incomes.

In this connection it is also worth noting what appears to be an inconsistency between Palairet's treatment of the Popov national income data, and his main index of Bulgarian economic development, which is a constant price series for agricultural output stretching from 1865 to 1914.³⁴

As we have seen, if we follow Palairet's suggestion of applying the retail price index to Popov's nominal income figures, then real per capita incomes drop sharply, particularly in the period after 1904 when inflation in retail prices was at its most rapid. Such a fall, however, is at odds with the main thrust of Palairet's declinist story, which rests not on the Popov data, but on Palairet's calculation of per capita farm output. In calculating his series for real agricultural output, Palairet proceeds much as we do, by multiplying the available data for physical output by prices recorded in the official statistics for 1910. Reassuringly, though we differ with Palairet in the relative importance we attach to arable and pastoral farming, the overall movement in our two series for agricultural output over the period 1892–1911 is very similar. Palairet's data, like ours, show agricultural output per head of farm population as being virtually identical in 1911 and 1892. However, given that agriculture was clearly the largest component of national income, this result is incompatible with Palairet's highly pessimistic reading of the Popov data. It also highlights the fact that the declinist case must rest not on the period under discussion here, i.e., 1890–1910, or even on the period between independence in 1878 and 1890, for which Palairet's agricultural index shows a moderate increase, but on a long-run comparison between the 1900s and the output of Bulgarian agriculture under Ottoman rule in the 1860s.³⁵ Between 1879 and the 1890s Palairet's index shows agricultural output per head of farm population rising from 163 gold francs to around 200 gold francs per capita, followed by a slight dip and a recovery to 190 gold francs per capita around 1910. If we confined ourselves to this period therefore we would have a story much like that suggested by Bairoch and Lains. It is only Palairet's figures for the 1860s, which show agricultural incomes of in excess of 220 gold francs per capita that create the impression of severe decline. In our view, however, the database on Bulgarian agriculture in the Ottoman period is at this point still too fragile to allow any far-reaching conclusions to be drawn.

If we accept that, at least in the period between the 1890s and the 1930s, per capita incomes in Bulgaria fluctuated around a plateau of roughly 1,500 Geary-Khamis dollars (1990), this stasis was conditioned by a no less striking continuity in sectoral shares. Up to World War II agriculture remained overwhelmingly the dominant sector in the Bulgarian economy, with industry contributing less than 15 percent, and the rest being made up of service sector and public sector incomes (see Table 2).

EXTENSIVE GROWTH

To explore the dynamics, which held this equilibrium in place for the best part of two generations, we must look in greater detail at developments at a sectoral level, particularly in agriculture. And we must focus in particular on the interaction between output, population, and land. For behind the stasis in per capita income figures in the generation before World War I, lies a dramatic story of what is perhaps best termed “extensive growth.”

In less than 20 years between the end of 1892 and the end of 1910, Bulgaria's population expanded by one-third, from 3.31 million to 4.34 million. At the same time, per capita incomes neither declined nor increased substantially. Instead, national output increased roughly in proportion to population, leaving per capita incomes unchanged. What were the motivating forces of this “extensive growth”?

In the Bulgarian parliament and amongst the middle-class elite in the 1890s there was much talk about industrialization.³⁶

However to judge by the statistics, the program of protectionism adopted in the 1894 was not a major success. It was not before the early 1900s that Bulgarian industry embarked upon positive growth and even then the results were far from impressive. The reality was that the share of agriculture in Bulgarian national economic life was nearly as large in 1911 as it had been in 1892. Over 70 percent of national output and more than 70 percent of employment was on the land. It was agriculture therefore that sustained two-thirds of the increase in total output in the 20 years before World War I. And it did so in the face of two serious setbacks.

The phylloxera outbreak of the 1890s, according to our calculations, reduced the value of grape production, which in 1892 accounted for 10 percent of rural incomes, by 80 percent. Then in the late 1890s the over-commitment of Bulgarian peasants to wheat farming was punished by a succession of bad harvests brought on by a serious drought. As a result, in 1899 per capita earnings from wheat were down by 39 percent relative to 1892. Over this seven year period wheat alone accounted for a 13 percent fall in total per capita agricultural incomes. How then did Bulgarian agriculture bounce back from these shocks, feeding a much-increased population at a roughly constant level?

Certainly we can rule out the narrative of intensification that lay behind the “green revolution” of Western and Central Europe. According to Van Zanden's analysis, the level of productivity in European agriculture by the 1870s depended first and foremost on a country's general level of modernization.³⁷

The best predictor of agricultural productivity was the share of population in nonagricultural employment. Given its overwhelmingly rural population, Bulgaria was thus at an enormous initial disadvantage. From this base level, Van Zanden then found that productivity growth after 1870 was driven by a variety of land-saving intensification strategies both in livestock and arable farming. One means of escaping the pressure of overseas grain import was to diversify out of grain into the production of milk, butter, and pork. With a climate that was unsuitable for pasturing, a small population of urban consumers and little access to capital, intensification was not an option for the Bulgarian peasantry.

Not that Bulgaria's farmers were without livestock. Both in absolute size and in per capita terms, Bulgaria had one of the largest sheep herds in Europe.³⁸

But this did not keep pace with Bulgaria's rapidly growing rural population between 1886 and 1910. In terms of horses and cattle per capita of its rural population, Bulgaria was close to the European average. Here the problem was one of quality not quantity. Unlike in the core countries of Western Europe, Bulgaria's cattle did not consist of well-bred animals, fed on high-energy animal feeds and dedicated to the efficient production of meat and milk. The majority of Bulgaria's cattle were in fact used primarily for plowing and tilling. Out of Bulgaria's herd of 1.606 million animals, only 367,723 or 23 percent were dedicated milk-producers and their milk yields were amongst the lowest in Europe.³⁹ Another striking deficit was in relation to swine, which served as an ideal “food bank” for the peasantry of North Western and Central Europe. Relative to the West European per capita average, Bulgarian peasants kept only one-third as many pigs. This was both a cultural legacy of the Ottoman occupation and a reflection of the shortage in Bulgaria of suitable feedstuffs.

Nor, despite its relative over-abundance of labor, was Bulgaria prior to 1914 able to significantly shift the basis of its arable farming towards more labor-intensive lines. With the vineyards blighted, Bulgaria remained trapped in grain farming, in which there was little evidence of intensification. Here perhaps the most relevant comparison is with Italy, which managed to achieve a considerable increase in agricultural output per capita from the 1890s onwards. Italy's advance rested on a shift into labor-intensive, high value fruit and vegetable crops and a dramatic increase in the application of artificial fertilizers.⁴⁰

By contrast, in Bulgaria even as late as the 1940s the application of artificial fertilizer remained at minimal levels.

The one important sign of substantial modernization in Bulgarian agriculture in the early twentieth century was the import of considerable numbers of iron ploughs to replace the wooden implements with which the peasants barely scraped the topsoil. This was a very promising development, as Bulgaria's dry summers put a premium on deep working of the soil during the wet autumn months.⁴¹

Progress however was painfully slow. It was not until the 1930s that even half of Bulgaria's ploughs were of a modern iron type.⁴²

The conclusion is inescapable: lacking the means for intensification along Western or Central European lines, the principal factor driving the expansion of output in Bulgarian agriculture was land. If we are to believe Popov's figures, between 1889 and 1911, the area under cereals increased by 157 percent.⁴³

This would imply that despite rapid population growth, the endowment of cultivated land per head of rural population actually doubled from the late 1880s, to reach its peak in 1911, at just over 0.8 hectare per capita. Such an increase may seem implausible. But even if we take the more credible census figures for 1897 as our starting point, it seems that up to the beginning of the first Balkan war in 1913, sufficient new land could be brought into cultivation every year to stave off any decline in land per capita. Furthermore the resulting distribution of land in Bulgaria was unusually even. By 1914 Bulgaria was truly a “peasant nation,” with 70 percent or more of its population settled on a mass of small independent farms.

As is clear from Table 3, we are not arguing that the Bulgarian peasantry were overgenerously endowed with land. Far from it: average farm sizes were very small by northwest European standards. By the interwar period, when the limits of easy extensive growth had been reached, the ratio of rural population to cultivated land was higher in Bulgaria than in even the most overcrowded regions of Italy and exceeded the national average for Poland or Rumania. Only in the Bukovina and Galicia and in the Serbian regions of Yugoslavia was land scarcity significantly worse. The point we are making is not that agricultural land was particularly abundant in Bulgaria, but that up to 1913 at least the land situation was not getting worse. Furthermore the prospect of land ownership was a realistic one for the vast majority of Bulgarians.

We stress this flexibility of the land constraint, because it seems essential to understanding three interconnected features of Bulgaria's social and economic development prior to 1914: the growth of total agricultural output in line with population, the striking lack of international emigration from Bulgaria, and the population boom itself.

Bulgaria's exceptionally dramatic demographic transition in the 70 years following independence dominates its economic and social history. From the mid nineteenth century onwards, population growth accelerated dramatically, reaching its peak in the early 1920s at just over 2 percent per annum. Prior to 1913 it was this explosive growth in population that ate up any advance in national income, reducing per capita income growth to zero.

The population explosion was due both to an increase in fertility from 33.7 births per thousand in 1881 to a maximum of 45.1 in 1914, amongst the highest rates ever recorded in Europe West of Russia, and, from the early 1890s onwards, to a sustained decline in mortality. So much is clear from the earliest statistics produced by the new Bulgarian nation state. The most likely cause for the decline in mortality is the low-cost imitation of Western public health measures, in particular the provision of clean water. Over the ultimate causes of the increase in fertility there remains a large measure of uncertainty. One popular theory linked the sudden rise in the birth rate to the breakdown in the traditional form of extended patriarchal households (zadruga) supposedly dominant throughout the Balkans.⁴⁴

In fact, however, zadruga households were only found on the Western fringe of Bulgaria. The disintegration of this particular “Balkan” family structure cannot therefore have been the primary determinant of demographic change.⁴⁵ On the other hand it is undeniable that some change in patterns of marriage and fertility must have occurred in the postindependence period, to explain the sudden increase in the birth rate. And both the rising fertility rates and falling mortality rates must clearly have rested on economic foundations. This raises the question of how one can square the declinist view of Bulgaria's economic development between 1878 and 1914 with the demographic data. Would population growth really have accelerated as rapidly as it did, if per capita incomes had actually been declining sharply in the late nineteenth century? A more plausible interpretation surely is that Bulgaria's population grew so rapidly because social constraints on marriage and fertility were loosened as hundreds of thousands of new hectares were opened up for settlement by young families.

And a similar logic applies to the question of emigration. Emigration provided one obvious escape from the overpopulated periphery of Europe. The departure of more than 14 million Italians between the 1880s and 1914, both overseas and within Europe, was one of the most spectacular voluntary population movements in history. But Bulgaria, along with the other Balkan countries, was a notable absentee from the economy of mass migration. Of course, some Bulgarians did emigrate. Enough indeed to provoke an anxious debate in Sofia.⁴⁶

Itinerant Bulgarian market gardeners were famous throughout southeastern and central Europe. Our figures show their contribution to rural incomes rising fourfold between 1892 and 1911. But in 1911 they still accounted for less than 4 percent of total farm income. The aggregate figures of population movement are no less stark. Viewed over the entire period between the 1870s and the 1930s, the combined overseas migration from Yugoslavia, Macedonia, and Bulgaria was no more than 135,063 people, compared to the 430,000 who left Greece over the same period.⁴⁷ Relative to population almost ten times as many Greeks emigrated. Over the course of the nineteenth and early twentieth centuries, more people emigrated overseas from the prosperous Benelux countries than from the poorest regions of the Balkans.

Such flows of population are clearly conditioned by a variety of factors including politics, information, cultural restrictions on mobility, and the initial availability of cash to buy tickets.⁴⁸

For all of these reasons it is not surprising that the Balkans were last in the emigration league. However, it is no less true that the pressure of poverty was extreme. And if we were to assume that per capita incomes in Bulgaria had actually fallen by as much as 20 percent in the generation prior to World War I, the low rate of emigration would surely be very puzzling indeed. If, however, extensive growth enabled output to be raised to match population, if the majority of the Bulgarian peasantry had the prospect of coming into the possession of land within their home country, then the lack of emigration is less surprising.

In short, our data suggest an internally consistent story of extensive growth as the most likely scenario for Bulgaria's economic development prior to 1914. Obviously, the stagnation of per capita incomes at a level precariously above the poverty line was not an attractive long-term scenario. Though its population was not becoming immiserated, Bulgaria was far behind the core countries of the liberal capitalism and it was falling further behind every year. Contrary to what Good and Ma suggest, Bulgaria was certainly not keeping pace with the rapidly converging countries of eastern and central Europe. In this respect it resembled Spain and Greece rather more than it did the revisionist view of Hungary or Italy. Furthermore, extensive growth could not continue indefinitely. However, our GNP estimates also lead us to reject Palairet's deeply pessimistic vision. In the generation prior to World War I, the Bulgarian economy was at least growing fast enough to hold the material standard of living in a precarious equilibrium.

THE IMPACT OF WAR

The Balkan wars of 1912 and 1913 and defeat in World War I dealt a devastating blow to the Bulgarian political system and to national self-confidence. In the wake of the wars, bourgeois spirits were further rattled by the triumphant rise to power of the radical peasant party, which took office in 1919, and the emergence of a significant communist party backed by the new Bolshevik regime in Russia.⁴⁹

Stamboliiski's peasant government was overthrown by military coup in 1923, heralding an uneasy return to parliamentary “normalcy.” But in truth Bulgaria's political system never recovered, sliding by the 1930s into monarchical authoritarianism.

What was the impact of this political turmoil on Bulgaria's economic development? It is clear that the economy suffered as a result of the war. Manpower and livestock were diverted from production, many never to return. Fields were left uncultivated. A severe inflation disrupted the fragile commercialization of agriculture. Fertile land was lost and the population was further swollen by an influx of 300,000 refugees from territories divided amongst Bulgaria's adversaries. But how severe was the impact of all this on aggregate economic output? Technically speaking, the answer to this question depends on the way in which we link Popov's national income estimate for 1911 to Chakalov's data, which begin in 1924.

As was already evident in Figure 1, according to Maddison, and Good and Ma, the impact of World War I on the Bulgarian economy was nothing short of devastating. In per capita terms Maddison claims a fall of no less than 40 percent between 1913 and 1924. Indeed, if we believe this version of events, virtually the entire interwar period is taken up with the recovery from World War I. The drama of the narrative is further enhanced by the fact that Maddison relies for the pre-1914 period on the highly optimistic growth rates projected by Good and Ma. The result, whether intentional or not, is a narrative of Bulgarian economic development in which World War I acts as an exogenous catastrophe, throwing a country, which had previously been developing at a rapid pace, profoundly off course.

But how solid is Maddison's bridge between 1913 and the early 1920s? For the period after 1924 Maddison is happy to rely on the growth path implied by Chakalov's figures, a choice with which we concur. The key issue is how to relate Chakalov's figures for 1924 to the common benchmark for 1913. If one follows the sources in Maddison's appendices, one is led back to the heavily revised 1957 edition of Colin Clark's Conditions of Economic Progress.⁵⁰

It is Clark who provides the warrant for the apocalyptic view of World War I, claiming a 23 percent drop in Bulgarian national income per capita between 1913 and 1926. To arrive at his even more gloomy figures for 1924, Maddison linked Chakalov's data for 1924–1939 to Clark's figure for 1926. But what were Clark's sources and how did he effect the translation of 1913 and 1926 leva into his standard international units? As his source for 1913 Clark gives an obscure article by Corrado Gini. For 1926 he cites an even more obscure publication by the Dresdner Bank.⁵¹ Clark makes no comment about how he adjusted for the effects of wartime inflation. By contrast, our methodology enables us first of all to match the nominal estimates of Popov for 1911 and Chakalov for 1924 and then to derive a constant price estimate for 1924 using exactly the same prices that were applied in the prewar period.

To compare 1913 with 1924, we must also allow for territorial changes. As a result of the two consecutive Balkan wars in 1912 and 1913, and despite its involvement on the losing side in World War I, Bulgaria managed by 1919 to increase its overall national territory as a whole by roughly 6 percent. However, the initial gains in the First Balkan War were offset after World War I by the painful loss of the Dobrudja, the most fertile agricultural region. As a consequence, though Bulgaria's overall territory expanded between 1913 and 1919, it suffered a net loss of cultivated land of 6.81 percent.⁵²

Given the importance of agriculture to the overall estimate and the correlation between settlement and agricultural activity, we adjust the entire estimate of national income by this factor.

The result, as is evident in Figure 1 and Table 4, is a dramatic revision of the Clark-Maddison figures. In real terms we find very little difference between Bulgarian national income in 1911 and 1924, once we allow for territorial changes. The difference in per capita terms is entirely attributable to the continuing growth in population. The influx of over 300,000 refugees and the unabated demographic pressure were the factors driving the fall in per capita output.

If such a modest impact of World War I seems implausible we should remind ourselves that 60 percent of Bulgarian GNP was contributed by a largely self-sufficient rural sector, with the rest mainly being contributed by public services and small-scale trades. Agriculture did not depend heavily either on industrial inputs or the intensive application of animal manure. The enforced fallowing of land during the war may actually have increased fertility. According to the official figures the grain harvest was fully recovered by 1925. The harvest of grapes was significantly larger by the early 1920s than in the immediate prewar period. The production of tobacco, the new cash crop of Bulgaria agriculture, was ten times greater in 1922–1923 than in 1913. Furthermore, though the figures for the animal herd do show a modest decrease in horse power between 1911 and 1924, Bulgaria's holdings of pigs almost doubled. There can be little doubt therefore that the physical volume of production in Bulgarian agriculture rebounded quickly from the war. Much the same can be said of output in the industrial sector. The number of workers employed in protected industries was 40 percent greater in 1921 than in 1912. Coal output more than doubled between 1913 and the early 1920s.

LAND-SAVING INTENSIFICATION

Nevertheless, the war did mark a turning point in Bulgaria's economic history, but in a different sense from that implied by the Clark-Maddison figures. As compared to the extensive growth model of the prewar years, it is clear that in the course of the interwar years the Bulgarian economy shifted gear. This shift did not become visible in the aggregate GNP figures until the second half of the 1930s. It was obscured by the political turmoil of the 1920s and the 1930s, the impact of the Great Depression, and the subsequent integration of Bulgaria into the German sphere of influence. But the nature of this transition becomes starkly evident if we focus on the main drivers of the story: land, population, and agricultural output.

With World War I Bulgaria reached the limits of extensive growth. After 1914 there was no further extension of agricultural land to compare with that in the decades following 1878. The area under cultivation inched up of course, but not by enough to substantially affect aggregate output. By contrast, rapid growth in population continued for another 15 years, to reach its peak in 1926, the year in which the Bulgarian population registered a record annual crude growth rate of 2 percent. As a result, in the generation after World War I, land to labor ratios were progressively squeezed and farm sizes fell. If there was an era in which Bulgaria was genuinely in danger of tipping over the Malthusian cliff, with population growth outrunning the capacity of the economy to sustain it, it was the 1920s.

But this was a transitional situation. As economic conditions changed, so did demographic behavior. Beginning in the late 1920s, Bulgaria experienced one of the most rapid demographic transitions on record. By the mid-1930s the rate of population growth in Bulgaria had halved and continued to fall precipitately into World War II. As population pressure on the land intensified, agricultural practices changed as well. With land becoming increasingly scarce, Bulgarian farmers were forced to turn increasingly towards the land-saving strategies of intensification that had characterized the “green revolution” in Western Europe prior to 1914. The significance of this interwar shift is brought out particularly clearly in a long-run perspective offered by Figure 2, which shows an index of gross agricultural output per hectare of cultivated land and per head of rural population stretching from 1890 to 1950.

As Figure 2 clearly shows, the “take off” in output per unit of land began in interwar period. Taking a four-year average, between 1936 and 1939, returns per hectare were almost 32 percent higher than in 1911, which was by no means a bad year. In 1939 they were fully 40 percent higher and there is no reason to doubt that, but for the interruption of World War II, this trend would have continued, as it was to do in the 1950s.

As Figure 2 also shows, however, this positive trend in per hectare productivity did not translate into an immediate increase in rural standards of living. Per capita output levels struggled to return to their prewar levels for most of the 1920s, an alarming finding given the low levels of income already prevailing prior to 1913. Nor do these figures, which do not distinguish between self-consumed output and marketed output, adequately reflect the savage squeeze on money incomes suffered by the Bulgarian countryside as a result of the worldwide collapse in commodity prices after 1929. There was some rebound in output per capita in the course of the 1930s. However, even then the pre-1913 level was not significantly breached prior to World War II. It was not until the rural population actually began to decline in absolute terms in the early 1950s that per capita output finally took off.

The changes in Bulgarian agriculture in the interwar period did not go unnoticed by contemporaries. The British journalist Hubert H. Tiltman, who traveled widely in eastern Europe, described in admiring terms how Bulgaria was transforming itself from a nation of peasants into a nation of high-value gardeners.⁵³

This view was shared by the émigré agronomist Slavcho D. Zagoroff, a former Bulgarian Finance Minister and ambassador to Berlin, who in the 1950s served as the editor of the Stanford Food Institute's survey of Danubian agriculture. In Zagoroff's words Bulgaria in the interwar period had undergone “a transition to a higher stage of farming,” though he stopped short of calling it a “transformation.”⁵⁴ There can certainly be no doubt that the attitude of the Bulgarian political elite towards agriculture was dramatically changed from the early 1920s. Though Stamboliiski and his peasant party were brutally removed from power, the priority of industry in Bulgarian economic policy was overturned once and for all. From the 1920s onwards, the Bulgarian state and its politicians concentrated their attention firmly on agriculture.⁵⁵

But one should surely not exaggerate the influence of policy. The most straightforward and compelling explanation for this pattern of lop-sided agricultural advance was simply that Bulgarian agriculture was conforming to the brutal logic of factor proportions. Given the increasing pressure of rural population, it would have been surprising if agricultural productivity growth had not been biased towards land- rather than labor-saving. It was land that was scarce not labor, and peasant families adjusted accordingly.

Capital accumulation was certainly part of this story of land-saving intensification. The import of agricultural machinery, interrupted by World War I, was resumed in the 1920s. Most significantly, as we have already noted, there was a sustained investment in new ploughs. Whilst providing rewards from deeper tilling, these also demanded a significant increase in the quantity of labor per hectare. This flow of modernizing equipment was slashed during the depression, as a result of the collapse in the money earnings of agriculture. However, imports revived rapidly in the 1930s and they were now accompanied by an unprecedented import of artificial fertilizers as well (see Figure 3).

The common source of both crucial inputs to Bulgarian agriculture was Germany. Though the clearing system instituted in Bulgarian-German trade in the 1930s, was in many ways unsatisfactory to Bulgaria, Hitler's Germany did at least offer a large and secure market for Bulgarian agricultural exports, which was not available elsewhere. And this in turn facilitated a marked shift in the composition of Bulgarian agricultural output.

As before 1914, Bulgarian farmers in the interwar period showed remarkable flexibility in responding to market signals. In the 1920s Bulgarian farmers had shifted rapidly into tobacco, a crop particularly suited to the soil and climate of some of the poorest areas of the country. In the 1930s, following the collapse of global tobacco prices, a new phase of diversification brought a shift into fruit and vegetables. This latter move, of course, was fully in line with the trend towards land-saving, labor-intensive farming. And as in Italy, where farmers had pursued this line since the late nineteenth century, it produced rising yields per hectare amidst substantial rural poverty (see Table 5).⁵⁶

Given its limited impact on per capita incomes, the progress in output per hectare in Bulgarian agriculture in the 1920s and 1930s was hardly a recipe for general economic development. It was, at best, a rational response to the increasing pressure of population on a limited stock of land. What was crucial therefore was that Bulgaria was, by the interwar period, capable of combining a new pattern of more labor-intensive agriculture with a sustained investment in human capital.

By the 1930s Bulgarian society began to reap the benefits of two generations of investment in mass literacy. The results were spectacular. For the early years following independence, Bulgarian education statistics show that literacy did not exceed 10 percent. The situation was particularly severe for women, of whom in the 1890s only 6 percent knew how to read. By 1900 general literacy had risen to 18.4 percent. By 1934, 55 percent of the population were literate. The 1946 census recorded almost total literacy for both sexes (91.7 for male and 91.8 for female).⁵⁷

Within the space of only two generations Bulgaria had achieved a quantum leap in human capital. Both as a factor in its own right and as a general index of the penetration of external forces into the previously closed world of village life, this remarkable literacy drive suggests a radical shift in Bulgaria's prospects for accelerated economic growth.