Abstract
A new GDP series per capita for Sweden during 1560–1800 is presented, linked to slightly revised data for 1800–2000. Long-term stagnation up to the nineteenth century is revealed but with secular changes. Growth characterized much of the seventeenth century with modernization of state administration, industry and trade. In the next century, stagnation and even retrogression followed. Wars in the seventeenth century may have stimulated growth, but also exhausted resources. Despite stagnation, the structure of the economy shifted and created preconditions for the modern economic growth that took off in the nineteenth century.
1. Introduction
Recently, interest has grown in early modern economic growth, i.e., in the period before the nineteenth century. Primarily, European countries have been considered, but lately also extra-European countries. There are several reasons for this interest in studying the economic dynamics or lack of dynamics in the medieval and early modern period. One is to get a better understanding of the Industrial Revolution and its consequences in the industrialized economies. Another is a growing interest in the Great Divergence, i.e., the long-time divergence in productivity and living standards between Europe and Asia.
There are, however, different views on the European pattern of growth prior to the nineteenth century. In his estimates, Maddison (2001) depicts a slow but general European growth since the sixteenth century. This view has become much more differentiated over the last decade. Allen (2000) and van Zanden (2002) present growth only in the European centre around the Channel and a stagnating periphery, even with retrogression in the old Mediterranean centres of Italy and Spain.
In his seminal paper 2000, Allen outlines a broad pattern of European development 1300–1800 that essentially builds upon, firstly, urbanization rates and occupational structures and, secondly, on prices and wages combined in a model to estimate agricultural consumption and production. He identifies four major regions: the fast growing Britain; the moderately growing Netherlands and Belgium; the declining medieval centres of Italy and Spain; a large Continental group stretching from France over Germany into Central Europe from Poland down to Hungary characterized by low urbanization rates, high population growth and declining wages, indicating falling productivity in agriculture. In these Continental countries, there was some structural change with rural industrialization but this “should be thought of as a response to overpopulation and limited opportunities in agriculture rather than as a growth pole of the economy”, contends Allen while success around the Channel “depended upon a productive agriculture and a vigorous urban economy rather than on a large protoindustrial sector”.
The research on pre-industrial growth has stimulated the construction of GDP data and other series for a number of countries.1 In the present paper, new estimates of Swedish economic performance from the mid-sixteenth to the nineteenth century are presented together with slightly revised data for the following period. The construction of Historical National Accounts 1560–1800 combines the demand approach on agriculture with data of other sectors and of population. The construction indicates that Sweden, as a representative of Scandinavia, broadly fits to the Continental pattern with long-run stagnation on the GDP per capita level up to the early nineteenth century. Despite stagnation, however, there were structural changes that created a basis for the breakthrough of modern economic growth from the mid-nineteenth century.
2. Swedish GDP data
Sweden has a long tradition in the field of historical national accounts and belongs to the more advanced countries in this respect. Already in the interwar period a broad project was launched and a comprehensive data set for 1861–1930 was presented (Lindahl et al. 1937). A number of improvements and extensions of the series were then published. A new project was initiated in the 1980s with the aim of improving the methodology, extending the time period of the accounts and revising the data. Thus, Swedish Historical National Accounts back to 1800 now exist, built on detailed accounts of each sector, which were presented in nine volumes during the 1980s and 1990s. In 2007 a summarizing volume was published with GDP series based upon double deflation and chain indices with yearly links of a Paasche type as deflators.2 The accounts are to be further extended in a research project recently started on regional gross products back to the early nineteenth century.3
For the pre-1800 Sweden, data on the whole economy are scarce. A number of price and wage series are available for the dominant agricultural sector back to the sixteenth century and in some cases even further back. There are also harvest estimates. However, no data on the entire gross domestic product exist except for the benchmark of 1571 presented in Krantz (2004).4 When the benchmark of 1571 was related to the series from 1800, it appeared that GDP per capita was at about the same level in 1571 as in 1800, thus indicating long-term stagnation at the per capita level in Sweden prior to the nineteenth century. In the present paper, this conclusion will be put in a wider perspective through an estimate of annual GDP back to the mid-sixteenth century.
3. Estimates of HNA in Sweden back to 1560
A systematic construction of historical national accounts for Sweden during 1560–1800 is undertaken, using methods similar to those employed in recent European constructions of pre-modern GDP series. To a certain extent, the construction relies on circumstantial evidence such as series of wages and prices, but it is also based on historical records of metal production, foreign trade, domestic trade, and public services.
The new estimates will be linked to the GDP series of Krantz/Schön from 1800 with some revisions of the latter.5
3.1 Agriculture
Agriculture has been much studied in Swedish history. Most studies are of a qualitative kind, but some quantitative work has been done. Hence, there are some data for production, the most comprehensive being the tithes that exist for the whole period. They have been much discussed and criticized and no agreement on their quality has been reached.
An authoritative study on long-term changes is Hannerberg (1971), where a critical summary of the discussion on the tithes is given. Hannerberg's conclusion is that they mirror more or less correctly the allocation of harvests to various cereals. However, the long-term changes are not shown in a proper way.
This judgement has gained support in recent studies. Leijonhufvud (2001) has made a thorough scrutiny of the data and added them up to series for the entire country during 1559–1680. Her aggregated series indicates a constant level of production in a period when population roughly doubled. This indicates that the tithes series do not show long-time changes in a proper way.
Another recent work on harvests in the seventeenth and eighteenth centuries is Edvinsson (2009). He finds that tithes and harvest estimates register short-term fluctuations, but not long-term changes. He presents a series of annual per capita harvest 1665–1820 based on a calculation that combines harvest estimates and prices with an assumption of the long-term constant per capita level. Thus, one can conclude that tithes and harvest estimates are a relevant source for short-term fluctuations, but not for trends over longer periods.
A somewhat different perspective is provided by Olsson and Svensson (2010) with an estimate of agricultural production during 1700–1860, based on direct production data from farmers' accounts in the southernmost Swedish province, Scania. According to their results, there was stagnation on the per capita level until the 1780s; a first trend break occurred with the early enclosure movement in the 1780s followed by a strong expansion in the first half of the nineteenth century; a productivity increase per worker of 0.8 percent annually; and a per capita increase of 0.6 percent. Their estimates based upon a large sample of farm records points in the same direction as the Krantz and Schön (2007) estimate of a per capita growth of Swedish agriculture of 0.5 percent annually in the first half of the nineteenth century. One should emphasize, though, that Scania was the leading agricultural district and a pioneer in the enclosure movement.
The present estimate of agricultural production will follow a different methodology, namely the demand approach in Crafts (1980), Allen (2000), Prados et al. (2007), and Malanima (2011). It starts from an estimate of per capita consumption of agricultural products from the demand side in a model with positive income elasticity, negative price elasticity to agricultural products, and (a rather weak) positive cross elasticity to industrial products.
Wage and price data are probably the best indicators available of economic performance in Sweden prior to 1800. There are two consistent wage series for the analysis in the long term. One is for unskilled workers in Stockholm from the early sixteenth century up to the mid-nineteenth century constructed by Johan Söderberg (2009). The series is excellent in consistency and scope in time, but data are of course restricted to the only really urban region in an overwhelmingly agrarian economy. The other series is Lennart Jörberg's (1972) wages for agricultural day-workers from the 1730s up to the early twentieth century. These wage notations are available annually for all provinces in Sweden and thus provide an average for the whole economy.
With only Stockholm data prior to 1732, one may ask whether these wages are representative for the whole country. From figure 1, it is obvious that the two series are very similar in the long run between 1732 and 1850. There may have been specific periodical trends or fluctuations in building worker wages in the capital city, with, for instance, construction booms or slumps, but in the longer run it is reasonable that regional wage development goes in the same direction. There was probably an integration of the economy in the country at least in the central parts and thus it could be expected that the long-term wage changes were rather equal. We assume that these mechanisms of integration worked also in the pre-1732 period. It might be the case though that the strong expansion of the capital city in the seventeenth century periodically gives an upward bias in the wage estimate. On the other hand, the construction boom led to an inflow of workers that exerted a downward pressure on wages, particularly when recession set in.
Real wages of unskilled workers—urban during 1540–1850 (full line); rural during 1732–1850 (dotted line). Index 1570 = 100.
Real wages of unskilled workers—urban during 1540–1850 (full line); rural during 1732–1850 (dotted line). Index 1570 = 100.
In figure 1, the long-run wage series are deflated into real wages with the consumer price index constructed by Rodney Edvinsson and Johan Söderberg (2009). The real-wage series shows some distinctive long trends. From the 1540s to the turn of the century 1600, real wages were falling. From the low points in the early seventeenth century, the trend turned upwards until the 1690s, but fluctuations were very strong. From that point of time the level stagnated for a couple of decades with some deep falls during the wars in the early eighteenth century. From the 1730s and 1740s real wages clearly moved downwards quite persistently to the trough in the decade of the Napoleonic wars, to turn upwards again, but inconsistently, from the 1810s to the 1850s.
In the estimate of demand for agricultural products, price indices of agricultural and industrial goods are constructed from the major products of rye, barley, and butter and of bar iron and woolen cloth, respectively.6 Income is approximated by day wages (as in the real-wage series discussed above). Day wages may, however, deviate from income since the actual number of working days is unknown. For the short term, this is met by a rather low elasticity in demand from changes in real wages, which incorporates the circumstance that the effect on demand may be counteracted by variations in annual labour time, as noted by Malanima (2011).
There may, however, have been a more long-term shift in the number of productive working days over the year that would make day wages and income deviate. The onset of the “agrarian revolution” during the late eighteenth century, plausibly led to an increase in the number of productive working days per worker, particularly since the cultivation season was prolonged with a greater variety of crops and since investment activities increased [see Gadd (2011)].The intensification of agriculture was, however, mainly a nineteenth century phenomenon and had probably less effect in the eighteenth century. Consequently, we do not make any correction for this possibility in the estimate. Thus, while there might be a slight upward bias in the income estimate for the seventeenth century, there might also be a similar downwards bias for the eighteenth century.
All prices and wages are in real terms, i.e., deflated by the consumer price index. The elasticities used in the model are the same as in Malanima (2011): 0.4*d(wages) − 0.5*d(agroprices) + 0.1*d(indprices).7
The estimated annual per capita consumption during 1560–1800 is presented in figure 2. The long-term trends are quite clear. There is a severe decrease by roughly about a quarter in the estimated consumption from the mid-sixteenth century to the mid-seventeenth century, followed by stagnation but with heavy annual fluctuations up to about 1800. These trends are overall very much in line with the historiography of the Swedish economy. Heckscher (1935–1949) did early on characterize the drastic decline in consumption standards from the sixteenth century, while Myrdal and Morell (2011) recently conveyed a similar, very pessimistic view of the seventeenth century.
Estimated per capita consumption of agricultural products during 1540–1800. Index 1540 = 1. Sources: Wages from Söderberg (2009) and Jörberg (1972); the consumer price index from Edvinsson and Söderberg (2009). See text on the estimation model.
Estimated per capita consumption of agricultural products during 1540–1800. Index 1540 = 1. Sources: Wages from Söderberg (2009) and Jörberg (1972); the consumer price index from Edvinsson and Söderberg (2009). See text on the estimation model.
To go from consumption to production, data on foreign trade are needed. Annual figures on all important items in trade are available from 1732.8 Prior to 1732, Heckscher reports the composition of imports and exports at a number of benchmarks from the 1550s to the 1720s.9 With annual trade figures from 1732, and interpolated figures between trade benchmarks, the consumption estimates are turned into estimates of agricultural production during 1560–1800.
One can notice that trends in foreign trade are very similar to trends in the estimated consumption. A small positive trade balance in the mid-sixteenth century was turned into a negative balance in the following decades and at the turn of the century 1600, while a substantial trade surplus arose during the first decades of the seventeenth century. In the second half of the seventeenth century, and through the eighteenth century, the balance became negative again, with a culmination in imports around the turn of the century 1800. In the first decades of the nineteenth century, trends shifted once more with a new export surplus in the 1840s and 1850s.
3.2 Manufacturing industry and handicrafts
There are no complete production figures for the manufacturing industry, but information exists which can be combined in an estimation of a series for the whole industry. Only the most important industries are included in the estimate.
3.2.1 The metal industry
Export and production figures for iron for a couple of years are provided by Eli Heckscher (1935–1949) in his pioneering work on Swedish economic history. Additional figures were provided by Karl-Gustav Hildebrand (1957). Some other scholars have also published point estimates. However, available data were far from complete over time. No attempts were made to put the data together and elaborate long time series until recently when such series were presented in Olsson (2007). Estimates of the production of osmund, an old form of iron, and bar iron were constructed on the basis of available data. They are to a great extent based on export figures which are more frequent than output data. Besides, figures in Historisk statistik för Sverige (1972), from 1732 onwards were utilized. First, an annual export quantity series was estimated for the period 1368–1800. Then, with the help of point estimates of the total output for some years, export shares were arrived at and, thereby, output figures could be computed. Osmund iron was exported until the early seventeenth century and after that only bar iron.
For copper production, a quantity series from 1540 onwards was compiled by Lindroth (1955) in a comprehensive work on Swedish copper mining and manufacturing. Some lacunas in the series were filled by interpolation.
The iron and copper series were combined by a rough weighting procedure to form a volume series which may be assumed to represent the output of the Swedish metal industry. The weighting was performed on the basis of the price ratio between iron and copper. In the 1580s the copper price, according to Heckscher, was six times higher than the iron price. Of course this ratio varied up to 1800. However in the 1820s, according to Jörberg (1972), it was again around 6:1 and this ratio was used for the whole period.
3.2.2 The food industry
Data for this industry do not exist and, consequently, surrogate figures had to be found. Inputs to this industry emanate mainly from agriculture. Therefore, it was assumed that the output in the food industry followed that of agriculture.
3.2.3 The manufactories
A series that could contribute to shed light on industrial production before 1800 though it is too short to be included in the estimate pertains to the so-called manufactories. In Sweden in the seventeenth and eighteenth centuries these production units were founded with generous state support and were strictly regulated. They produced various goods, mostly textiles. The manufactory output in the eighteenth century was first estimated by Heckscher (1935–1949) and it was re-estimated by Krantz (1976). In table 1 the output of the manufactories is shown for some benchmark years.
The manufactories, volume of textile production 1745–1802, index 1779 = 100
| 1745 | 50 |
| 1759 | 91 |
| 1766 | 78 |
| 1779 | 100 |
| 1789 | 86 |
| 1795 | 92 |
| 1802 | 106 |
| 1745 | 50 |
| 1759 | 91 |
| 1766 | 78 |
| 1779 | 100 |
| 1789 | 86 |
| 1795 | 92 |
| 1802 | 106 |
Source: Krantz (1976).
The manufactories, volume of textile production 1745–1802, index 1779 = 100
| 1745 | 50 |
| 1759 | 91 |
| 1766 | 78 |
| 1779 | 100 |
| 1789 | 86 |
| 1795 | 92 |
| 1802 | 106 |
| 1745 | 50 |
| 1759 | 91 |
| 1766 | 78 |
| 1779 | 100 |
| 1789 | 86 |
| 1795 | 92 |
| 1802 | 106 |
Source: Krantz (1976).
After an initial spurt in the 1750s when the state subsidies were very marked, stagnation became characteristic of manufactory output. This supports the idea developed in the present paper that the eighteenth century saw stagnating or even slightly declining overall economic growth.
3.2.4 Total industry
Iron and copper production can be considered as representing the whole-metal industry, which, not least due to its exports, was a very important subindustry. So was also the food industry since it was relatively big. In 1800 their respective shares of the whole industry were 28.7 and 34.1 percent. These shares were also used in weighting them together. As a result a series considered to represent the whole industry was calculated. Since the manufactory production data show roughly the same performance as the whole industry, the assumption that the estimated series is a reasonable representation of the total is supported.
3.3 Building
For the nineteenth century, the building industry caused great problems of estimation due to the lack of data. The difficulties are even more pronounced for earlier periods and, consequently, an indirect technique was employed.
It is reasonable to suppose that manufacturing industry and handicrafts as well as agriculture via investments had a strong influence on building activities. Therefore, it is assumed that the series for industrial and agricultural output constitutes indicators of building. Furthermore, building of dwellings played an important role and this part of the building sector is supposed to be mirrored by population.10 Thus, the average of indices (1800 = 1) for agricultural and industrial production and population is supposed to show the performance of the building sector and this indicator was linked to the production value of building in 1800 in Pettersson (1987).
3.4 Transport
There are very few data on which to base an estimate on transport production for this period. Therefore, it was assumed that this sector's output could be represented by an average of index series (1800 = 1), constructed from 50 percent of the volume of material production (industry and agriculture) and 50 percent from the volume of domestic trade (as an indicator of transport intensity). The resulting series was linked to the value for 1800 according to Krantz and Schön (2007).
3.5 Private services
3.5.1 Domestic trade, etc
All commercial services are assumed to be included here, which means not only trade, but also banking, insurance, and hotels.
3.5.1.1 1624–1810
In this period domestic custom duties (“lilla tullen”) had to be paid in the Swedish realm. For all goods that were brought to market places or to towns to be sold, the sellers had to pay these duties (1/32 of the merchandize value). The control of the trade was strict, which resulted in records on all customs revenues. Most of the records have been saved in the archives and have been used to study various aspects of domestic trade, mostly local, but no systematic exploration of the whole data set was made until a few years ago. Then, the material was critically explored and data series for the whole country were presented by Andersson Palm (2008).
The weaknesses of the records were carefully scrutinized by Andersson Palm. One problem is that they comprise data from the entire country, which means that Finland is included. However, the Finnish share of the total was roughly the same during the whole period, around 10 percent. There are also a lot of other problems, but the data are of such a good quality that the series in constant prices can be used as an indicator of the volume of domestic trade.
Deflating domestic trade is a general problem when constructing historical national accounts and has been tackled in various ways. One is to use employment figures in volume calculations, which means that constant labour productivity or a specified productivity change is assumed. In the present estimation, this is impossible since employment figures are missing. Sometimes a consumer price index has been utilized as a deflator. However, Andersson Palm made a calculation of the volume. He chose to eliminate the changes in tariffs over time by applying the tariff of 1655 to the whole period. Thus, tariff changes were used as a deflator and his figures are accepted here. Data are missing for some years and, therefore, interpolations were made.
3.5.1.2 1560–1624
Prior to 1624 no data of the same quality as in the following period exist for domestic trade. Therefore, it was assumed that the trade followed the changes in the urban population. (Stads- och kommunhistoriska 2009).
3.5.2 Personal services
By personal services is meant all personal services specified in the historical national accounts after 1800 (Krantz and Schön 2007). However, due to the lack of data it is assumed that the changes in constant prices follow those of the urban population.
3.6 Public Services
3.6.1 Central government
3.6.1.1 1722–1800
The same subsectors as in Krantz (1987a, b) were distinguished. They are civil services, the court and the armed forces. Furthermore, the source, Åmark (1961), as well as the methods used by Krantz (1987a, b, p. 15f) for the period 1800–1809 were employed. In Åmark (1961, table 18) a special item is given for armament and ordering costs (“rustnings- och utredningsstater”) during the war periods 1757–1762 and 1789–1791. It is impossible to discern exactly how much of this item constituted wages. Therefore, it was assumed that this share was 50 percent.
In Krantz (1986) one item was added that was connected with the armed forces. There was an organization called indelningsverket which meant a special standing army supported in kind by the peasants and, thus, the costs were not visible in the state finances. Krantz calculated these costs for the nineteenth century up to the cancellation of the organization in the beginning of the 20th century (p 17ff). For the period up to 1800 a simplified procedure was chosen. The ratio between the total costs for the armed forces inclusive of the standing army and the costs given by Åmark (1961) for the period 1800–1809 was applied to the series for the period before 1800.
As remuneration constitutes value added, the series should be deflated with data pertaining to wages. However, to construct series for civil and military personnel was not possible since too much work was required. Therefore a wage series for workers in Stockholm provided by Söderberg (2009) was used. The Stockholm series is preferred since much, perhaps most, of the work in the public sector was performed in Stockholm.
3.6.1.2 1560–1722
No data collection corresponding to Åmark's exists for this period and it was considered as too laborious to go through the archival material concerning the state finances. Instead a simple estimation procedure was chosen. It was assumed that the public service production followed the population changes as estimated by Andersson Palm (2001). This means that the costs concerning indelningsverket are included for the whole period. This organization in a sustainable shape was established in the second half of the seventeenth century, but since there were more or less similar organizations also earlier, the estimation was the same for the whole period.
In the series for the period, after 1722 the values during wars, namely 1741–1743, 1757–1762, and 1788–1790 are considerably higher than for surrounding years. It is highly probable that the war periods before 1722 should also be distinguished by higher values than periods of peace. The war periods prior to 1722 were as follows: 1563–1570, 1590–1595, 1611–1613, 1628–1645, 1655–1666, 1674–1679, and 1700–1721. There were more phases that could be defined as war periods but the intensity of the warfare differed a lot and, therefore, a selection was made so that the more intensive wars were included (Sundberg, 1998). In the war periods in the eighteenth century public production was roughly double that of the eight surrounding years. Therefore, the assumption was made that the public production during the intensive war phases prior to 1722 should be doubled and, thus, a series for central government services was arrived at.
3.6.2 Local government
Sources for local government are to a great extent missing or extremely time consuming to deal with. Therefore a simplified procedure was opted for. The ratio between local government and central government production during 1800–1810 was assumed to be valid also for the entire period prior to 1800.
3.7 Services of dwellings
This sector comprises the return or output of the dwellings capital (Krantz 1991, p 151). For later periods, it was possible to make estimates on the basis of rents or imputed rents. This, however, requires knowledge about the stock of dwellings and its changes and this knowledge is lacking for earlier periods. Therefore, it was assumed that the sector's share of GDP for the period 1800–1810 was valid for the time prior to 1800.
4. Swedish GDP since 1560: long-term patterns
The new GDP per capita series, displayed in figure 3, gives new perspectives on the Swedish long-term economic development.
GDP per capita 1560–2000. Constant prices—SEK at the 1910–1912 price level.
GDP per capita 1560–2000. Constant prices—SEK at the 1910–1912 price level.
There are two main observations from the long-term pattern in GDP per capita. Firstly, there was hardly any growth prior to the mid-nineteenth century. Thus, the levels in the late sixteenth and early nineteenth centuries were about the same. Secondly, a secular pattern of variations in the early modern period is clearly visible. The last decades of the sixteenth century were marked by a downward tendency ending at the turn of the century and followed by a rise till the middle of the seventeenth century. Thereafter GDP per capita stagnated and from around 1720, i.e., after the long period of war, a downward tendency set in. After the slump during the Napoleonic wars, a period of new stability appeared but the level of the mid-seventeenth century was surpassed only during the 1870s. Thus the economic performance in a large part of the century can be seen as a recovery succeeded by modern economic growth and industrialization.
How reliable is this pattern of stagnation and secular fluctuations? This will be discussed from two angles: one is indicators of long-term economic and social development that are prevalent in European historical growth analyses, namely real wages and rates of urbanization, and the other is structural changes.
Real wages have been extensively used in the construction and are, of course, no real corroboration of the GDP estimate. Suffice it to say, that Swedish long-run changes of real wages have roughly the same pattern as long-run GDP per capita. There are only short term or periodic deviations as, for instance, in the first decades of the eighteenth century when war efforts raised public services while there was a slump in building activity affecting workers' wages.
The other indicator to be compared with the GDP per capita series is urbanization. In European estimates of early modern GDP, the rate of urbanization is often used as an indicator. Urban agglomerations were early on networks that provided overall scarce resources of knowledge and competence in industry and services, also to the rural economy. The degree of urbanization will therefore indicate levels and trends regarding the sophistication of the economy.
Sweden was for a long time a very rural country with few and small towns. It has been argued that the great dependence upon wood fuel for heating purposes in a cold climate like the Swedish and the widely distributed fuel with high transportation costs put a severe limit to the size of the Swedish towns (Thorburn 2000). Instead, the Swedish economic structure adapted to these circumstances with, for instance, a deliberate dispersion of the energy-intensive iron industry. Not until the second half of the nineteenth century was this natural barrier to urbanization broken with the advent of new means of transportation and the importation of coal.
Only a few sites with good transportation facilities and with strong positive agglomeration effects could early on climb over a threshold of 5,000 inhabitants. This level is common in European measures of urbanization and lends itself to comparisons. Clearly, Sweden was far below most European countries in terms of urbanization. By 1750 there was a European urbanization span from the leader Netherlands of 36 percent down to laggards such as Poland at 4 percent (Malanima 2011 table 5). Sweden was slightly above the Polish level at close to 5 percent of urban population.
Urbanization, furthermore, provides similar trends as estimated GDP per capita in Sweden, even if variation is within narrow margins, up to 1850 (figure 4). Both in urbanization and in GDP, there are low points about 1610 and 1810 with a peak of about 1690 and with rapid growth after 1850. Thus, there are two periods of urbanization in Sweden—in the seventeenth and the late nineteenth centuries. The first one coincides with the height of the Swedish Baltic Empire and the second one with modern industrialization, with stagnation in between.
Urban share of population in Sweden (post-1809 frontiers) during 1570–1890. Cities >5,000 inhabitants. Note: Benchmarks are 1570, 1610, 1650, 1690, 1730, 1770, 1800, 1810, 1820, 1850, 1870, and 1890, connected by interpolation. Source: Stads- och kommunhistoriska institutet (2009). Historisk statistik för Sverige (1969).
Urban share of population in Sweden (post-1809 frontiers) during 1570–1890. Cities >5,000 inhabitants. Note: Benchmarks are 1570, 1610, 1650, 1690, 1730, 1770, 1800, 1810, 1820, 1850, 1870, and 1890, connected by interpolation. Source: Stads- och kommunhistoriska institutet (2009). Historisk statistik för Sverige (1969).
While stagnation characterized long-term GDP, structural changes were considerable, as presented in table 2.
Sectoral composition of Swedish GDP 1565-1995, per cent, 11-year averages
| Agr | Ind | Build | Transp | Priv | Publ | |
|---|---|---|---|---|---|---|
| 1565 | 74.0 | 4.6 | 4.9 | 1.5 | 3.3 | 11.8 |
| 1600 | 71.6 | 5.5 | 5.6 | 1.5 | 3.5 | 12.3 |
| 1650 | 58.7 | 8.6 | 6.6 | 2.4 | 12.5 | 11.3 |
| 1700 | 55.0 | 7.4 | 5.8 | 2.5 | 15.9 | 13.4 |
| 1750 | 57.3 | 7.6 | 6.3 | 2.8 | 18.8 | 7.3 |
| 1800 | 53.5 | 7.1 | 5.7 | 3.3 | 19.9 | 10.5 |
| 1850 | 50.0 | 8.7 | 7.1 | 3.8 | 22.2 | 9.2 |
| 1900 | 32.6 | 24.2 | 6.5 | 7.5 | 23.3 | 5.9 |
| 1950 | 9.1 | 44.8 | 5.8 | 13.3 | 21.2 | 5.7 |
| 1995 | 2.3 | 48.3 | 3.2 | 17.8 | 23.8 | 4.5 |
| Agr | Ind | Build | Transp | Priv | Publ | |
|---|---|---|---|---|---|---|
| 1565 | 74.0 | 4.6 | 4.9 | 1.5 | 3.3 | 11.8 |
| 1600 | 71.6 | 5.5 | 5.6 | 1.5 | 3.5 | 12.3 |
| 1650 | 58.7 | 8.6 | 6.6 | 2.4 | 12.5 | 11.3 |
| 1700 | 55.0 | 7.4 | 5.8 | 2.5 | 15.9 | 13.4 |
| 1750 | 57.3 | 7.6 | 6.3 | 2.8 | 18.8 | 7.3 |
| 1800 | 53.5 | 7.1 | 5.7 | 3.3 | 19.9 | 10.5 |
| 1850 | 50.0 | 8.7 | 7.1 | 3.8 | 22.2 | 9.2 |
| 1900 | 32.6 | 24.2 | 6.5 | 7.5 | 23.3 | 5.9 |
| 1950 | 9.1 | 44.8 | 5.8 | 13.3 | 21.2 | 5.7 |
| 1995 | 2.3 | 48.3 | 3.2 | 17.8 | 23.8 | 4.5 |
Constant prices in 1910/12 price level. Services of dwellings excluded.
Source: Present estimate.
Sectoral composition of Swedish GDP 1565-1995, per cent, 11-year averages
| Agr | Ind | Build | Transp | Priv | Publ | |
|---|---|---|---|---|---|---|
| 1565 | 74.0 | 4.6 | 4.9 | 1.5 | 3.3 | 11.8 |
| 1600 | 71.6 | 5.5 | 5.6 | 1.5 | 3.5 | 12.3 |
| 1650 | 58.7 | 8.6 | 6.6 | 2.4 | 12.5 | 11.3 |
| 1700 | 55.0 | 7.4 | 5.8 | 2.5 | 15.9 | 13.4 |
| 1750 | 57.3 | 7.6 | 6.3 | 2.8 | 18.8 | 7.3 |
| 1800 | 53.5 | 7.1 | 5.7 | 3.3 | 19.9 | 10.5 |
| 1850 | 50.0 | 8.7 | 7.1 | 3.8 | 22.2 | 9.2 |
| 1900 | 32.6 | 24.2 | 6.5 | 7.5 | 23.3 | 5.9 |
| 1950 | 9.1 | 44.8 | 5.8 | 13.3 | 21.2 | 5.7 |
| 1995 | 2.3 | 48.3 | 3.2 | 17.8 | 23.8 | 4.5 |
| Agr | Ind | Build | Transp | Priv | Publ | |
|---|---|---|---|---|---|---|
| 1565 | 74.0 | 4.6 | 4.9 | 1.5 | 3.3 | 11.8 |
| 1600 | 71.6 | 5.5 | 5.6 | 1.5 | 3.5 | 12.3 |
| 1650 | 58.7 | 8.6 | 6.6 | 2.4 | 12.5 | 11.3 |
| 1700 | 55.0 | 7.4 | 5.8 | 2.5 | 15.9 | 13.4 |
| 1750 | 57.3 | 7.6 | 6.3 | 2.8 | 18.8 | 7.3 |
| 1800 | 53.5 | 7.1 | 5.7 | 3.3 | 19.9 | 10.5 |
| 1850 | 50.0 | 8.7 | 7.1 | 3.8 | 22.2 | 9.2 |
| 1900 | 32.6 | 24.2 | 6.5 | 7.5 | 23.3 | 5.9 |
| 1950 | 9.1 | 44.8 | 5.8 | 13.3 | 21.2 | 5.7 |
| 1995 | 2.3 | 48.3 | 3.2 | 17.8 | 23.8 | 4.5 |
Constant prices in 1910/12 price level. Services of dwellings excluded.
Source: Present estimate.
The agricultural share fell decisively over the period, while the share of the secondary sector, i.e., industry and construction was overall rather stable. Instead, the share of private services rose, with the rise to a great extent concentrated to the seventeenth century, particularly its first part. Even if data for private services are scarce, some great changes seem plausible for this sector. The number of domestic servants—dominating private services around 1800—was much smaller in the beginning of the period than later and a low number signifies a less stratified agricultural society and a less diversified household sector. A similar feature of a less sophisticated economy in terms of commercialization and capitalization is also expressed in a much weaker position for private services in trade, finance and diverse personal services.
The conclusion is that there were clear structural changes between the late sixteenth century and the turn of the century 1800. Above all, private services rose in prominence, at the expense of agriculture. Thus, in a comparative sense the economy in the sixteenth century was more dependent upon resources from the primary sector and upon the further preparation of these resources within households with comparatively little specialization of labour. In the early nineteenth century, the economy had become much more sophisticated in terms of labour division and sector specialization.
The peninsula was very sparsely populated around 1570 and according to available estimates—admittedly uncertain—the number of inhabitants increased almost four times until 1800—a population increase much higher than the European average over these centuries. A constant real GDP per capita would mean that the average productivity per labourer remained constant over these centuries (assuming similar ratios of active labourers to the total population). It may seem paradoxical that GDP per capita remained almost constant between the sixteenth and the nineteenth centuries taking into account that those sectors that had a relatively high labour productivity in 1800 (such as manufacturing, transportation and private services) in all were smaller in the sixteenth century. This enigma may be explained by a relative shift in productivity over time. Population increase with recourse to marginal land and with diminishing returns to labour in agriculture would be balanced by a transfer of labour to sectors with constant returns to labour and on average returns at a higher level than in agriculture as in industry and services.
5. A reinterpretation of the seventeenth century
The GDP estimates present some new trends in relation to earlier historiography that may cast new light on the early modern Swedish economy. For the late sixteenth century, however, the new estimates are in accordance with the view early on proposed by Eli Heckscher (1935–1949) in his works on Swedish economic history, with a deterioration of the living standard up to the early seventeenth century.11
The great difference to earlier interpretations concerns the seventeenth century—by patriotic historians named the era of “Sweden as a Great Power”12. In economic history, the period has been characterized by continued deterioration and hardships for the population. In his recent survey of agriculture in early modern Sweden, Myrdal made a general characterization based on long-term agricultural change and population increase: “Sixteenth-century expansion, seventeenth-century stagnation” (Myrdal 2011, p. 102). This characterization seems not to be valid for the whole economy, however. In the second half of the century, the estimated GDP per capita even surpassed the levels from the mid-sixteenth century, despite the regression in agriculture. With industry and services included in the accounts, the picture is changed.
The basis for growth in the seventeenth century, according to the present estimate, was a reorganization of the economy, very much led by or supported by the state. Both administration and industry were modernized, with a technological transfer inter alia through skilled immigrants. Export of industrial goods was also of importance for economic growth especially in the first half of the century. Apart from bringing about the growth of industrial output, it enabled imports of various kinds and facilitated capital import. Furthermore, export prices rose more than import prices which, of course, was favourable for the Swedish economy (Hansson 2006, p. 164). As noted earlier, this was also a period of urbanization in Sweden.
Much of the new strength was directed to military ends. It might be the case that the difference in economic growth between the two centuries had to do with Sweden being a belligerent during most of the seventeenth century. This meant that not only human resources had to be mobilized, but industrial as well, for instance, for weapons, ammunition, and textile goods. Actually, the regime in this century has been termed “The Military State”.13 Among other things this is said to have implied heavy taxation and pauperization. A great number of soldiers were recruited but mostly from the lower societal strata. As a consequence, a shortage of farmhands emerged and women are said to have taken a more active part in agricultural work than before. Moreover, most of the soldiers were recruited from marginal land, which meant that the influences on the total production were comparatively small. Furthermore, the negative effects on the household economy were reduced since wars were fought outside the realm, with the absence of a part of the population diminising consumption pressure. It is also possible that productivity may have increased due to modifications of production methods (everyday rationalizations) especially in the first half of the century, which was most affected by conscriptions. Thus, the net effect could have been that total agricultural production was not negatively affected up to the late seventeenth century, particularly since the supply of iron at lower prices increased substantially, enabling the widespread use of better equipment in agriculture.
After the first decades of the next century, i.e., in what historians for some reason have named “The Era of Freedom”, there were only a few short and not very comprehensive wars. This could have contributed to stagnation or even a downward tendency of the per capita GDP. Attempts, especially in the second half of the century, to stimulate economic growth in a mercantilist manner did only to a certain degree counteract these tendencies (Magnusson 1996).
It might also be the case that the long war periods—where the tide turned from victories to defeats in the early eighteenth century—had exhausted the wells of public strength built up in the first half of the seventeenth century, and the focus of economic change shifted again to the rural areas.
There is also a contrast in opinions concerning the eighteenth century, which was characterized by Gadd (2011) as the time for “The agricultural revolution in Sweden” which lasted till 1870. According to the present estimate, stagnation plagued the eighteenth century, As population growth outran land reclamations and productivity growth in agriculture—despite the first stage of an agricultural revolution in pioneering regions—this meant a long-term stagnation or even decrease in the per capita level down to the trough during the last war period around the turn of the century 1800. The positive effects of the agricultural revolution upon income gained momentum only from the mid-nineteenth century.
Thus, the seventeenth century saw industrial and administrative modernization of the Swedish economy that produced an overall GDP growth in sharp contrast to estimated stagnation in agriculture and in food consumption. The eighteenth century was rather an epoch of ruralization and increased pressure upon the primary sector that only responded late with an agricultural revolution.
6. The nineteenth century growth in new perspective
Over the recent decades, a new interpretation of Swedish development in the first half of the nineteenth century has emerged that emphasizes the positive role of the agricultural transformation, structural changes and incipient growth in GDP per capita from the 1820s. The new long-term GDP series indicate, however, that the first half of the nineteenth century was merely a period of recovery from very low levels around the turn of the century, while the real breakthrough of modern economic growth occurred only in the late nineteenth century. This seems to some extent to support an old interpretation of Swedish development, namely that the first half of the nineteenth century was characterized by pauperization in an economy that had severe problems to support its population. According to this older view, the solution to this dismal situation was the combination of export growth and emigration in the second half of the century.
There were, however, a number of structural changes that intensified in the period from the 1820s, even if these changes were as yet only strong enough to lift the per capita GDP back to the seventeenth century level.
The “agricultural revolution” from the late eighteenth century up to the mid-nineteenth century is by now well documented (e.g., Gadd 2000). In the light of this documentation, the agricultural growth rate of some 0.3–0.4 percent above the population growth in the first half of the nineteenth century is reasonable, resulting in the change from the import to an export surplus of grain together with stable food consumption per capita. The transformation entailed increased commercialization of agricultural production, the introduction of new crops, more intensive cultivation and rotation, enclosures and reclamation of land. These changes meant reduced seasonal variations in labour demand in agriculture. Even if the marginal productivity of labour was reduced during peak seasons, the number of productive working days per annum was increased. Involuntary unemployment or employment with other low productive ends could thus be reduced. Therefore, the agricultural revolution may have produced an “industrious revolution”—further induced by a higher preference for income due to an enriched supply of consumption goods. The fall in real day wages recorded in the 1830s and 1840s may very well have been consistent with an average stable real income for wage earners, while real income undoubtedly grew from the 1850s.
Another indication of growth was the consumption of industrial goods, which expanded strongly, with industrial consumption goods supplied both from domestic industries and from imports. Textiles made up a large part of this consumption but over the decades the range of goods widened.
Substitution effects were presumably particularly strong in textiles, where, for instance, the import of cotton at rapidly falling prices induced stagnation and regional decline in the traditional cultivation of flax and in the domestic production of linen. In Schön (1979, 1980) the total production of textiles during 1826–1870 was calculated, including production within households for their own use. The substitution effects of factory cloth for domestic household production was, however, very limited before 1850. Instead, factory cloth and domestic household production expanded in parallel while growth slowed down in proto-industrial forms of production such as putting-out (see table 3). There was also a shift over time in the quality composition of textile growth, from luxurious to more plain or coarse qualities, indicating a shift also in income distribution and involvement of a larger part of the population in consumption growth. In the 1850s, this structural change became even more pronounced.
Annual percentage growth in the supply of cloth from different types of production 1826–1859
| Factory production | Household for sale | Household for own use | Imports | Total supply | |
|---|---|---|---|---|---|
| 1826/28–1837/39 | 4.6 | 4.2 | 2.1 | 9.4 | 3.7 |
| 1837/39–1847/49 | 4.2 | 3.3 | 4.1 | 3.1 | 3.9 |
| 1847/49–1857/59 | 3.4 | 0.8 | 4.5 | 7.2 | 4.1 |
| Factory production | Household for sale | Household for own use | Imports | Total supply | |
|---|---|---|---|---|---|
| 1826/28–1837/39 | 4.6 | 4.2 | 2.1 | 9.4 | 3.7 |
| 1837/39–1847/49 | 4.2 | 3.3 | 4.1 | 3.1 | 3.9 |
| 1847/49–1857/59 | 3.4 | 0.8 | 4.5 | 7.2 | 4.1 |
Volumes in constant prices, price base 1838–1840.
Source: Schön (1979).
Annual percentage growth in the supply of cloth from different types of production 1826–1859
| Factory production | Household for sale | Household for own use | Imports | Total supply | |
|---|---|---|---|---|---|
| 1826/28–1837/39 | 4.6 | 4.2 | 2.1 | 9.4 | 3.7 |
| 1837/39–1847/49 | 4.2 | 3.3 | 4.1 | 3.1 | 3.9 |
| 1847/49–1857/59 | 3.4 | 0.8 | 4.5 | 7.2 | 4.1 |
| Factory production | Household for sale | Household for own use | Imports | Total supply | |
|---|---|---|---|---|---|
| 1826/28–1837/39 | 4.6 | 4.2 | 2.1 | 9.4 | 3.7 |
| 1837/39–1847/49 | 4.2 | 3.3 | 4.1 | 3.1 | 3.9 |
| 1847/49–1857/59 | 3.4 | 0.8 | 4.5 | 7.2 | 4.1 |
Volumes in constant prices, price base 1838–1840.
Source: Schön (1979).
The agricultural and early industrial expansion clearly had its roots in the growth of the domestic market but a new impetus came in the 1850s with the growth of foreign demand for goods that suited Swedish conditions very well. Rising prices of oats and timber stimulated Swedish agriculture and forestry and spread commercialization both regionally and socially. These trends came to the fore from the mid-nineteenth century, but the commercialization of agriculture and the growth of domestic markets in the preceding period had prepared the ground that made reactions to new and favourable price movements both swift and broad throughout the country.
7. Sweden in a European comparison
When historical GDP levels in different countries are to be compared, principles of estimation and deflation14 ought to be scrutinized. One such principle of great importance is what is included in the estimate. Basically, it is the value added of the main activities that everyone performs to earn a livelihood. In all societies, however, a number of activities are directed to own consumption (or the household's consumption). These are in principle not included in national accounts. Historically, demarcations are less clear. Should, for instance, all weaving or sewing in the households be estimated and added to the accounts? If so, why not baking, brewing, churning, tidying, baby care, etc. for your own household?
The Swedish historical national accounts have followed the principle of estimated total primary production (agriculture with subsidiaries) and estimates of the recorded or registered (from censuses or other official records) activities in secondary and tertiary sectors (including a number of supplementary estimates where these records have been incomplete). With a low level of specialization in the economy, however, such a principle will as a rule underestimate the level of final consumption. It will, in particular, underestimate contribution from female work that for a long historical period to a large extent was directed to consumption within the household. To counteract such a tendency, value added of unpaid domestic work from 1800 onwards was estimated in Krantz (1987b).15 The rationale was that such an estimate would include all types of domestic work, not singling out one or two activities in particular.
As was shown both in Krantz (1987c) and Schön (2000, 2012), the inclusion of unpaid domestic work has a strong effect upon the level of GDP in the nineteenth and early twentieth century. It is also important when analysing structural or social change. It has, however, a rather small effect upon annual growth rates in the long run—only in the order of 0.2 percent annually.
Since the level of economic specialization in Sweden prior to the twentieth century was rather low, the exclusion of unpaid domestic work may have a comparatively large effect upon the Swedish level. This is particularly so if the accounts of other countries have included some (but certainly not all) of the unpaid activities. For this reason, the Swedish GDP per capita is estimated here both exclusive and inclusive of unpaid domestic work (see table 4 and figure 5).
Per capita GDP in seven European countries 1600–1800
| England | Netherlands | Spain | Italy | France | Germany | Sweden excl. unpaid domestic work | Sweden incl. unpaid domestic work | |
|---|---|---|---|---|---|---|---|---|
| 1565 | 1062 | 1498 | ||||||
| 1600 | 1350 | 1860 | 1440 | 1350 | 1300 | 1150 | 847 | 1195 |
| 1700 | 1890 | 2160 | 1430 | 1440 | 1440 | 1210 | 1251 | 1765 |
| 1750 | 2150 | 2260 | 1310 | 1560 | 1490 | 1250 | 959 | 1353 |
| 1800 | 2010 | 2050 | 1290 | 1430 | 1410 | 1260 | 929 | 1311 |
| England | Netherlands | Spain | Italy | France | Germany | Sweden excl. unpaid domestic work | Sweden incl. unpaid domestic work | |
|---|---|---|---|---|---|---|---|---|
| 1565 | 1062 | 1498 | ||||||
| 1600 | 1350 | 1860 | 1440 | 1350 | 1300 | 1150 | 847 | 1195 |
| 1700 | 1890 | 2160 | 1430 | 1440 | 1440 | 1210 | 1251 | 1765 |
| 1750 | 2150 | 2260 | 1310 | 1560 | 1490 | 1250 | 959 | 1353 |
| 1800 | 2010 | 2050 | 1290 | 1430 | 1410 | 1260 | 929 | 1311 |
Swedish GDP exclusive and inclusive of unpaid domestic work. (1990 international GK dollars PPP).
Source: Malanima (2011), table 5, and the present estimate.
Note: Since annual fluctuations were strong in Swedish data, 11-year averages in GDP are used.
Per capita GDP in seven European countries 1600–1800
| England | Netherlands | Spain | Italy | France | Germany | Sweden excl. unpaid domestic work | Sweden incl. unpaid domestic work | |
|---|---|---|---|---|---|---|---|---|
| 1565 | 1062 | 1498 | ||||||
| 1600 | 1350 | 1860 | 1440 | 1350 | 1300 | 1150 | 847 | 1195 |
| 1700 | 1890 | 2160 | 1430 | 1440 | 1440 | 1210 | 1251 | 1765 |
| 1750 | 2150 | 2260 | 1310 | 1560 | 1490 | 1250 | 959 | 1353 |
| 1800 | 2010 | 2050 | 1290 | 1430 | 1410 | 1260 | 929 | 1311 |
| England | Netherlands | Spain | Italy | France | Germany | Sweden excl. unpaid domestic work | Sweden incl. unpaid domestic work | |
|---|---|---|---|---|---|---|---|---|
| 1565 | 1062 | 1498 | ||||||
| 1600 | 1350 | 1860 | 1440 | 1350 | 1300 | 1150 | 847 | 1195 |
| 1700 | 1890 | 2160 | 1430 | 1440 | 1440 | 1210 | 1251 | 1765 |
| 1750 | 2150 | 2260 | 1310 | 1560 | 1490 | 1250 | 959 | 1353 |
| 1800 | 2010 | 2050 | 1290 | 1430 | 1410 | 1260 | 929 | 1311 |
Swedish GDP exclusive and inclusive of unpaid domestic work. (1990 international GK dollars PPP).
Source: Malanima (2011), table 5, and the present estimate.
Note: Since annual fluctuations were strong in Swedish data, 11-year averages in GDP are used.
GDP per capita exclusive (full line) and inclusive (dotted line) of unpaid household work during 1560–2000. Constant prices—SEK at the 1910–1912 price level. Note: Assumption of a fixed ratio between the series per capita during 1560–1800.
GDP per capita exclusive (full line) and inclusive (dotted line) of unpaid household work during 1560–2000. Constant prices—SEK at the 1910–1912 price level. Note: Assumption of a fixed ratio between the series per capita during 1560–1800.
Clearly, Sweden was behind Western Europe in GDP per capita during the whole early modern period, very much so if unpaid domestic work is excluded. With the unpaid work included Sweden still lagged behind Western and Southern Europe for most of the period—the only exception is the very high level that Sweden attained at the end of “The Great Power Period” around the turn of the century 1700. The position very close to Germany around 1800—with unpaid work included—suggests, however, that this inclusion raises the Swedish level comparatively too much. One can conclude that a systematic inclusion of unpaid work also in other countries would raise their levels too, but probably not as much as for Sweden.
8. Conclusion
The characteristics of the growth process in the nineteenth century and onwards in Sweden have been much studied and different opinions have evolved especially on the roots of the process. In contrast to this, very little research has been done on the question of growth or stagnation in the Swedish economy prior to 1800. Thus, a long-time perspective that could deepen the understanding of the start of modern economic growth in Sweden has until recently been missing. The guesstimate by Maddison implied that there was a slight growth from the sixteenth century but the quantitative foundation of this is weak. Moreover, it was contradicted by the GDP estimate for 1571 which indicated a long-term stagnation towards 1800. In this article, a new GDP estimate provides a view on Swedish pre-modern economic change that might be put into a broader European pattern, based upon newly constructed historical national accounts.
For Sweden the data indicate that long-term stagnation prevailed in the per capita income up to the early nineteenth century in contrast to the development in the region around the Channel. Swedish economic performance, however, has not been even over the centuries. Rather, a certain economic expansion together with urbanization and administrative modernization in the Swedish “Great Power Period” took place during the seventeenth century followed by stagnation or even retrogression per capita during the eighteenth. Thus, trends in urbanization and industrial production indicate that the state and its war economy played an important role to stimulate modernization in the seventeenth century but also to waste resources in the eighteenth century.
The structure of the economy shifted, notwithstanding the long-term stagnation at per capita GDP level. Strong population growth during the two centuries contributed to increased specialization, manifested inter alia by the growth of domestic services and manufactories, as one precondition for the agricultural transformation that swept the country in the first half of the nineteenth century. Together with the subsequent export expansion, long-run stagnation gave way for modern economic growth.
Acknowledgements
We thank the participants at the workshop QUANTIFYING LONG RUN ECONOMIC DEVELOPMENT, The University of Warwick in Venice, Palazzo Pesaro Papafava 22–24 March 2011, and the members of the seminar in Economic History, Lund University. We gratefully acknowledge Mats Olsson for his advice as well as two anonymous referees. Financial support from the Swedish Research Council (dnr 421-2008-2023) is also gratefully acknowledged.
