Natural resources and economic growth: comparing nineteenth century Scandinavia and twentieth century Southeast Asia

ABSTRACT

This paper aims to bridge part of the gap that exists between the resource curse literature and economic historical research on natural resources by analysing four resource-abundant countries. The study proposes that at the sectoral level, the determinants of growth in resource-based industries were mostly similar in the late 19th and late 20^th centuries. However, we also argue that the relative contribution of natural resources to economic growth might have been declining during the late twentieth century. The evidence comes from an analysis of the forestry sector in Finland and Sweden between 1860 and 1910 and the palm oil industry in Indonesia and Malaysia between 1970 and 2016.

KEYWORDS:

• Natural resources
• forestry
• palm oil
• Finland
• Sweden
• east Asia

SUBJECT CLASSIFICATION CODES:

• n10
• n50
• q1
• q2
• o13

Acknowledgments

I am grateful for comments from Stig Tenold, Liam Brunt, Ola Honningdal Grytten, Jan Tore Klovland, Ingelin Orten, Karl Rolf Pedersen and Jari Eloranta. All remaining errors are solely mine.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 For recent literature surveys see Gilberthorpe and Papyrakis (2015), Venables (2016), van der Ploeg and Poelhekke (2017) and Badeeb, Lean, and Clark (2017). The present study defines natural resource similarly as Badeeb et al. (2017) as the natural assets that can be used for economic gains, which includes materials, minerals, forests, water and fertile land.

2 For instance, a real appreciation of the exchange rate (Corden & Neary, 1982) and reduced learning by doing (van Wijnbergen, 1984), decreased investment in human capital (Gylfason, 2001) and increased productivity (Bjørnland et al., 2019).

3 The paper title hinted at Scandinavia; however, Denmark and Norway are not included in this study. The omission of Denmark is due to its lack of a similar forestry sector, whilst Norway is omitted as the forestry sector was already in decline by the 1850s due to supply problems as many forests already had been exploited.

4 The reason for not exclusively taking the palm oil related trade figures is twofold. First, Indonesia and Malaysia were resource-abundant in many different natural resources meaning that taking all natural resources gives a better picture. Second, it is hard to know which manufacturing sectors were intensive in palm oil exclusively, as most resource-intensive manufacturing sectors were intensive in multiple natural resources.

5 One topic not considered in this paper is the difference between renewable and non-renewable resources in terms of resource depletion. For historical research in this vain see for instance Lindmark and Acar (2013).

6 Mehlum, Moene, and Torvik (2006), Torvik (2009), Mavrotas, Murshed, and Torres (2011), Sarmidi, Law, and Jafari (2014). Institutions, however, need not be exogenous, as resource rents in combination with corruption might reduce institutional quality over time. The term ‘Political Dutch Disease’ is often used to describe the negative effects of rent-seeking. Studies have explored how rent-seeking might increase corruption, political conflicts and income inequality. It can also isolate non-democratic regimes and decrease institutional quality (Hodler, 2006; Frankel, 2010; Sala-i-Martin & Subramanian, 2013; Bodea, Higashijima, & Singh, 2016). These mechanisms are, however, outside the scope of the present paper.

7 Iimi (2007), Ross (2007), Sachs (2007).

8 There many different ways in which the literature has defined both resource dependence and abundance. Resource dependence measures include primary exports over GDP, rents from natural resources over GDP, share of natural capital in national wealth. Resource abundance measures include total natural capital and mineral resource assets in USD per capita, total resource production per capita in USD per capita subsoil wealth. For examples of usages of these measures see Badeeb et al.(2017 tables 1, 2 and 3).

9 Obviously, these measures are not independent of each other, as the Spearman Rho correlation is 0.81, whilst the Pearson correlation coefficient is 0.55.

10 For studies using measures of resource dependence see for instance Sachs and Warner (1995) and Gylfason (2001). Studies that use resource abundance measures include Brunnschweiler and Bulte (2008) and Alexeev and Conrad (2009).

11 Wright (1990), David and Wright (1997) and Wright and Czelusta (2004).

12 Engen (2009), Holden (2013), Bjørnland et al. (2019), Ville et al. (2019).

13 Wright and Czelusta (2004), Ville et al. (2019).

14 Weber (2012), Gilje, Ready, and Roussanov (2016), Allcott & Keniston (2018) and Feyrer, Mansur, and Sacerdote (2017).

15 The third mechanism is also Wright and Czelusta’s (2004) main argument for the failure of many resource-abundant countries in the post-1950 period.

16 The Augmented Dickey-Fuller and the Phillip-Peron tests are standard tests for stationarity. The AIC is commonly used as a selection criteria for models. See for instance Gujarati (2015).

17 For some variables the different unit root tests gave conflicting results. However, no series was found to be I(2), regardless of the unit root test used.

18 $NATPROD$ is derived by multiplying the $NATSHARE$ with gross domestic product per capita and then taking the natural logarithm.

19 For the ARDL Bounds test to be valid, the variables can be either I(0) or I(1), but not I(2) (Pesaran, Yongcheol, & Smith, 2001).

20 The Pearson correlation coefficient between the Maddison data was 99.96 per cent for Finland (using Hjerppe, 1989 figures for GDP per capita), 97.46 per cent for Sweden (using Schön & Krantz, 2012 figures for GDP per capita), 99.76 per cent for Indonesia and 99.96 per cent for Malaysia (using the United Nations National Accounts, 2020).

21 Lindmark and Acar (2013) did construct rent data for Sweden 1850-200, but we do not have similar data for Finland.

22 ISIC A, B and C comprise agriculture, hunting, forestry, fishing and mining.

23 For similar control variables see Alexeev and Conrad (2009) and Bjorvatn and Farzanegan (2013). One control variable not included in the study was an institutional indicator, which would be natural given the literature on the resource curse. However, few indicators on institutions cover both periods, meaning that a meaningful comparison cannot be made. The few indicators that do cover both periods, for instance the Polity IV indicator for democracy, have little variation over time, meaning that little information can be derived using time-series techniques.

24 Söderlund (1953), Björklund (1984).

25 The key difference between Finland and Sweden were the institutional reforms in the first half of the 19th century, and was most likely a key factor in determining the differences in resource-led growth. While Swedish institutions increasingly became more market-oriented, Finland retained many mercantile institutions up until the mid-19th century. One example is the development of the financial system, which came earlier in Sweden, and played an important role in financing the forestry and related industries following 1870. For the emergence of the Swedish banking sector, see for instance Sandberg (1978) and Adams, Andersson, and Lindmark (2005); and for the emergence of the Finnish banking sector, see for instance Palo (2004). The Finnish Diet did not meet in the first half of the 19th century, and first started to assemble from the 1860s and onwards (Ojala & Karonen, 2006, p. 103). The lack of an active political body meant that reforms were not enacted before.

26 The lack of property rights was especially prevalent in Eastern, Western and Northern Finland (Michelsen, 1995; Palo, Uusivuori, & Lehto, 1999). By 1900, the demarcation of property rights was mostly completed (Palo et al., 1999).

27 Swedish restrictions on forestry had existed since the 18^th century. These restrictions were in place to protect the mining industry. However, the invention of the Bessemer process in the early 19th century led to the abolition of restrictions on the use of forests, as access to forests was less important for the mining industry. The Bessemer process allowed for the production of steel without the use of wood, meaning the forests were no longer of strategic importance to the Swedish mining industry.

28 The Russian emperor allowed the Finns semi-autonomous rule, and the Swedish laws simply remained in effect during a transition period into the Russian empire that lasted until the mid-19th century. As Finland had few mines, these restrictions were largely redundant.

29 In addition, a number of vocational training schools were set up in the country, and the number grew from 35 at the end of the 19th century to 66 in 1908–09 Ahlström (1992, p. 7).

30 The first chemical pulp mills were established in the 1870s, and by 1913 Sweden was the largest pulp exporter in the world (Blomstrom & Kokko, 2007).

31 Järvinen, Ojala, Melander, and Lamberg (2012).

32 Järvinen et al. (2012).

33 Skippari, Ojala, and Lamber (2005), Lamberg and Ojala (2006) and Järvinen et al. (2012).

34 Palo et al. (1999), Palo (2005) and Nykänen (2018)

35 Sachs and Warner (1995) and Gylfason (2001).

36 Brunnschweiler and Bulte (2008) and Alexeev and Conrad (2009).

37 The palm oil tree was imported from its indigenous West Africa, and the first commercial production was in 1911 for Indonesia and from 1917 in Malaysia (Moll, 1987).

38 Government-owned plantations did exist. The FELDA programme granted land to landless people.

39 Bruno (2017, p. 37) indicates that the yearly yield of crude palm oil per hectare increased from 1.43 metric tonnes in 1950 to 3.05 metric tonnes in 1970.

40 Rasiah (2006), on the basis of interviews with 50 palm oil companies, concludes that the role of PORIM was critical in training and encouraged new-product development.

41 In 1968, non-Malaysians controlled 78.2 per cent of all plantation land (Bruno, 2017, p. 10). Foreign- owned plantations disagreed with the government policies of promoting Malaysian refineries, preferring instead to have the oil processed in Europe. By 1983, non-Malaysians owned only 8.6 per cent of all planted plantation land. The state purchased most of the foreign-owned palm oil plantations to increase the equity share of indigenous ownership (Pletcher, 1991, pp.630–631). Among the most well known examples are the takeovers of Sime Darby in 1976 and Guthrie in 1981 (Martin, 2005, p.255). Although the takeovers were not achieved through expropriation, they were still aggressive.

42 In 1998, PORLA merged with PORIM to form the Malaysian Palm Oil Board.

43 Malaysia established the state-owned oil company, Petronas, in 1974. Since its establishment, Petronas has expanded into upstream activities as well as some downstream projects. From the 1990s onwards, the company became an important international oil company, with overseas operations in 35 countries. The company also invested in shipping and became the world’s largest single owner-operator of LNG vessels (von der Mehden & Troner, 2007). Petronas’ growing market share and ability to enter new markets gave the company a reputation of being a relatively efficient oil company. Indonesia established its state-owned company, Pertamina, in 1968. However, Pertamina experienced little technological learning from foreign companies, resulting in high exploration costs for oil. Following the transition to democracy in 1998, Pertamina lost its monopoly power in upstream activities in the domestic Indonesian market. Since the company lacked a competitive environment prior to 1998, it was unable to compete with foreign competitors coming into the Indonesian market, among them Petronas. The low efficiency and, thus, lack of competiveness has resulted in decreasing market shares and political influence (Hertzmark, 2007). Pertamina is an example of how excessive protection and a lack of incentives for technological upgrading can hurt a company long-term when the economic environment changes.

44 In fact, Isham et al. (2005, p. 164) wrongly characterize Malaysia has having diffuse, rather than point resources.

45 The average share of primary products as a share of world merchandise exports for the 1870–1910 period was 62 per cent, only dropping by 3 per cent from 1860 to 1910 (Federico & Tena-Junguito, 2016). The corresponding primary product share for the 1970–2016 period was an average 31 per cent, dropping by more than 10 per cent from 1970 to 2016 (WDI, 2020).

46 Hill (2000); Felker, Jomo, and Rasiah (2013).