Empirical Testing of Genuine Savings As an Indicator of Weak Sustainability: a Three-Country

Empirical testing of genuine savings as an indicator of weak sustainability: a three-country analysis of long-run trends.

Data Appendix

GDP and GDP deflator

Britain: Measuring Worth, http://www.measuringworth.com/ukgdp/ last accessed June 2013.

Germany: Pre-1975 data on German national product is available from Flora et al. (1983) and Hoffmann et al. (1965). GDP levels for later periods are taken from German Statistical Yearbooks (1999, 2008). Missing periods 1914-1924 and 1940-1949 were estimated using Ritschl and Spoerer’s (1997) GNP series. A GDP deflator was constructed using data from Hoffman et al (1965), Mitchell (2007) and the United Nations Statistical Division (2013).

US: GDP series, GDP deflator series were derived from Johnston and Williamson (2013).

Consumption

The present value of the change in consumption was calculated over three time horizons, 20, 30 and 50 years using country specific discount rate.

Britain: Feinstein (1972) and ONS. Deflated using a RPI from Measuring Worth.

Germany: Flora et al. (1983), German Statistical Office, downloadable under www.gesis.oreg/histat (Bundesamt, 2013), Ritschl (2005), Abelshauser (1998), and Harrison (1988).

US: Annual consumption data for 1869-2012 is from Rhode (2002) 1869-1900, Carter et al (2006) 1901-1962 and ERP (2012) 1963-2012. Consumption is deflated using US consumer price index from Johnston and Williamson (2013).

Population

Britain: From 1830 to 2010 calculated using the Measuring Worth UK population data minus the populations of Ireland from 1830-1920 and Northern Ireland from 1921 to 2010. From 1750 to 1830 Wrigley and Schofield (1989)’s annual population estimates of England and Wales are combined with those for Scotland, derived from Flinn (1977) and Census of Scotland from 1801 onwards.

Germany: Maddison database, http://www.ggdc.net/maddison/oriindex.htm

US: Derived from Johnston and Williamson (2013)

Genuine savings

NETPINV

Britain: 1765-1920 Feinstein & Pollard (1988), 1921-65 Feinstein (1972); 1966-2000 UK ONS publications.

Germany: Net investment from 1850-1959 is provided by Hoffmann et al. (1965). We estimated the gap during 1914-1924 using Kirner (1968) who reports investment in buildings, construction, and equipment by sector for the war and inter-war periods. The period 1939 to 1949 was estimated by using data on net capital stock provided by Krengel (1958). To estimate investment during 1960 to 1975 we used Flora et al.’s (1983) data on net capital formation. For the period 1976 to 2000 we use official World Bank (2010) and United Nations (2013) investment statistics to complete the series. Data on the change in overseas capital stock and advances is provided by Hoffmann et al. (1965). Gaps during war and inter-war periods were estimated using information on the balance of payments provided by the German central bank (DeutscheBundesbank, 1998, 2005). Remaining missing values were estimated using trade balances as a proxy for capital flows (DeutscheBundesbank, 1976; Flora et al., 1983; Hardach, 1973).

US: Net investment consists of produced capital and overseas investment. Gross fixed capital formation, inventories and net overseas investment from 1869-1909 were taken from Rhode (2002). Annual data for gross investment, inventories and net overseas investment for the years 1909-1929 was taken from the data appendix to Kuznets (1961). From 1929 to 1992 the data is from Carter et al (2006) and from 1992-2000 data was taken from the (ERP 2011). Capital consumption was from Kuznets (1961) for 1869-1929, from ERP(1963;1995;2011) for 1929-2000.

Green investment

We calculate the changes in natural capital by measuring the change in stocks multiplied by price minus average cost. Below we illustrate sources for renewable (forestry & land) and non-renewables (mining).

Renewables

Britain: Forestry stock estimates for the woodland area (hectares) and standing volume (cubic metres per hectare) are from British agricultural returns, Schlich (1904), Stamp and Beaver (1954), the 1923 and 1947 woodland censuses, Eurostat and the UK Forestry Commission. International trade prices per cubic metre are used to value the standing volume given Great Britain was a net importer of timber. A variety of prices estimates are combined, including UK import prices from 1847 to 1957 and US export prices from 1965 to 2000 (Hiley 1930; Bulfin 1974; Howard, 1974; MacGregor, 1950a, b, 1953, 1959). Average prices are used in the absence of the single long run series, for details see McLaughlin et al (2012). Felling costs are estimated as wage costs per m3. MacGregor (1946, p.30) shows labour costs were the ‘greatest direct influence on the cost of forestry operations’, but he reports daily rates, not the cost per m3. Labour costs per m3 are estimated from the employment of felling and forest workers and annual felling. Employment in forestry has been estimated for 1765-1840 assuming 5 workers per 100 acres (BPP, 1942-43), based on Heske’s (1938) claim that each 35 cubic feet of wood cut needed one day’s work. Census data provides employment of woodcutters from 1841 to 1921 as do Forestry Commission reports for later years. The felling data used to construct estimates of wage cost per m3 are from (Heskey 1938; MacGregor, 1959) and Forestry Statistics 2001.

Germany: Germany had an established reputation as one of the most advanced nations involved in forestry management and inspired British and U.S. developments in silviculture (e.g. see Schlich (1904), Zon (1910), Hiley (1930), B.P.P. (1942-43), Heske (1938)). Information on German forestry stock were taken from Zon (1910), Zon et al. (1923), Hoffmann et al. (1965), and Endres (1922).

US: Forestry: Changes in forestry stock were obtained by estimating the area of forestry and the standing volume of timber (m3). The area of forests were obtained from Carter et al (2006, series CF101-118 and Cf135-144). Estimates of standing volume were obtained from (Zon (1910), Zon &Sparhawk (1923), Clawson (1979), Oswalt et al. (2007), USDA (1997), Smith & Darr (2002), Smith, et al. (1997), USDA (1997), Carter et al (2006). The earliest estimate of standing volume was 94.59 cubic metres per hectare in 1920. It was assumed that this was constant from 1850-1920. The forest area was multiplied by this estimate. From 1920 to 2000 the area of forestry was multiplied by the standing stock of timber (m3) per hectare. The change in the standing volume of timber was valued at market prices minus average costs. For the period 1869-1904 forestry prices were derived from Warren & Pearson (1932) and stumpage prices from 1905-2000.(Carter et al 2006). Employment estimates and annual lumbering were derived from the Carter et al (2006) and Lebedys (2004), the wages used to calculate wage cost per m3 were unskilled wages derived from Officer (2012) and David & Solar (1977).

For the US we have also calculated the economic value of gains/losses in farmland because of the increasing expansion of US agricultural land over the ninteenth century compared with Britain and Germany which were older established agricultural economies. We use estimates of rents (i.e. profits) per acre discounted over time. In any year the value of appreciation or depreciation in the stock of farmland is given by the physical change in area (we cannot distinguish different types of crop land) valued using the present value of rents over a 30 year forward period. Land value data was obtained from Carter et al (2006). Data from Lindert (1988) indicated that the rental value of land was 15 per cent of the land value, a rental value of us farmland was estimated assuming that this rental value was a constant ratio of the value of agricultural land. Rents were as far as 2030 were forecasted using an ARIMA (5,1,1). In the USHS there are two separate series for farmland DA5 & DA17. DA5 is an annual series from 1900-1999 and DA17 is a quinquennial series. From 1900-1949 DA5 is a linear interpolation using DA17 but from 1950 onwards there is a change in the series which leads to a distortion in the number of land farm. This distortion is not reflected in DA17, thus this series has been chosen

Non-renewables

Britain: Estimates of coal extraction are from Pollard (1980), Flinn (1984), Church (1986), Mitchell (1984, 1988) and from UK Mineral Statistics and UK Mineral Yearbook. Pithead prices per tonne are from Church (1986), Mitchell (1984, 1988), Clark & Jacks 2007, Supple 1987, Ashworth 1986, NCB reports, UK Mineral Statistics and UK Mineral Yearbook. Wage estimates were taken from (Flinn (1984), Church (1986), Mitchell (1984, 1988), Ashworth (1984) and National Coal Board reports. The 19th century data are for hewers and were reported as daily wages in Flinn (1984)[58] and shift rates in both Mitchell (1984) and Church (1986). Labour force numbers were taken from the Annual Returns of Mines from 1874 onwards Mitchell (1988), from census returns (Mitchell (1988) & Taylor (1961)), and estimated assuming productivity of 250 tons per worker year pre-1874.

Extraction data for iron ore are from the official series beginning in 1854 and earlier estimates from Hyde (1977) and Riden (1977). Mine head prices from 1854 onwards are also reported in the Mineral Statistics. The integrated organisational structure of British iron industry makes it difficult to ascertain iron ore prices pre-1854. We assume the price of iron ore was a ratio of the price of pig iron, adopting 10% of the pig iron price, which was the average ratio 1857-1914. British iron production dwindles in importance by 1900, and US prices are used for the period 1915-2000 from Kelly et al. (2010), to value the small quantities of British production. Daily wage rates across all the mining industries were similar (Bur (1984)), though wage costs per ton differed. From the 1907 census of production output per man year (OMY) for iron ore miners was 611 tons (BPP (1910)) versus an OMY of 321 tons for coal miners ((BPP 1909)). Labour productivity in iron ore mining was around twice that of coal and therefore their labour costs per ton would have been about half. We use this relativity to estimate wage costs per ton for iron ore mining. Data on tin, copper, lead and zinc extraction came from Mitchell (1988) and UK mineral statistics and mineral yearbook. There are no separate employment data for these mining operations and wage costs per ton, given the similar extraction technology, are assumed equal to those for iron ore mining.

Oil and gas extraction are from Energy Trends 2002. Historic oil prices per barrel are from the BP Statistical Review of World Energy and converted to price per tonne, taking a barrel to equal to 0.136 tonnes. Dollar prices are converted to pounds with the historic exchange rates from Officer (2013). The marginal costs of oil and gas extraction are assumed to be zero.

Germany: Fischer (1989) and Fischer and Fehrenbach (1995) provide detailed data on German mining activities (hard coal, brown coal, crude oil, iron ore, copper ore, zinc ore, lead ore, silver ore, tine ore and nickel ore). These data also include the number of employees in mining, covering the period until the 1970s. Information on quantities and market prices by commodity on an annual basis are available. Additional information were collected from Mitchell (2007). Data provided by Fischer (1989) and Fischer and Fehrenbach (1995) are also available by German state, which allows subtracting contemporary contributions of the mining sector of Alsace-Lorraine between 1871 and 1918. Moreover, the statistical offices of the German Empire and the Federal Republic of Germany provide information on the 1914 to 1923 as well as the post-1962 periods, respectively (Bundesamt, 2013; Germany. Statistisches Reichsamt., 1925).

To assess the costs of depletion the number of employees in mining and their average wage were used. Data on the labour force in mining is provided by Fischer (1989), Fischer and Fehrenbach (1995), and the German Statistical Office (2013). Wages of mining workers are reported by Hoffmann et al. (1965), Kuczynski (1947), Mitchell (2007), and official contemporary statistics (Germany. Statistisches Reichsamt., 1925).

US: From 1880-2000 mining (fuel, metals and minerals) data is from US historical statistics (Carter et al 2006). Fuel is comprised of coal Bituminous, Coal Subbituminous, Coal Lignite, Coal Pennsylvania Anthracite, Crude Petroleum, Natural Gasoline and Cycle Products, and Liquefied Petroleum Gases, Natural Gas Marketed, Uranium Concentrate. Metals are comprised of Iron Ore, Copper, Zinc, Manganese Ore, Chromite, Tungsten Concentrates, Molybdenum Ores and Concentrates, Vanadium Ores and Concentrates, Nickel, Bauxite, Aluminum Primary, Magnesium Primary, Gold, and Silver. Minerals are comprised of Crude Gypsum Mined, Lime, Sand and Gravel, Stone, Sulfur Production from Frasch Mines, Pyrites Production, Salt, Potash sold by producers, and Phosphate Rock.

From 1869-1880 mining output was estimated using data from US historical statistics and (Herfindahl 1966) and (Gallman 1960), and valued at international prices. Commodities used were iron ore, copper, lead, zinc, gold, silver, coal and crude petroleum.

Mining wage cost per tonne was calculated from data on mining wages in coal and the annual relative productivity between coal mining and mining. Coal output, employment and wage data from 1869-2000 were obtained from Carter et al (2006). Mining employment and output were also obtained from Carter et al (2006). Over the period 1855-2000 the mean relative productivity difference between coal and all forms of mining was 1.06.

Education

Britain: Carpentier (2013).

Germany: Hoffmann et al. (1965) and Diebolt (1997, 2000).

US: Carter et al (2006) from 1869-1996. Supplementary data was obtained from the National Centre for Education Statistics.

Total Factor Productivity

For each country we computed a time series of US total factor productivity over time 1860-2013, and use its trend growth rate to construct a measure of the value of technological progress. We used calculated the present value of future changes in TFP over 20 horizons using country specific discount rates.

Britain: TFP: real gross capital stock from Feinstein (1972) and Feinstein & Pollard (1988) and O’Mahony (2009). Labour hours from Crafts (1985), Voth (2000), Wrigley (1989), Flinn (1977), Feinstein (1972), and O’Mahony (2009). Factor shares, used to measure the output elasticities assuming wages equate to marginal product of labour, are from Crafts (1985), Matthews et al (1982) and ONS (2006). The factor shares are: 1760-1860, α= 0.50; 1856-1920, α=0.58; 1920-1951, α=0.70; 1951-1973, α=0.73; 1973-2000, α=0.64. Annual TFP has been calculated except for 1760 to 1860 where an annual series has been interpolated from decadal data. The trend TFP is a Kalman filter of the TFP growth rate. TFP trend post 2007 is forecast using an ARIMA (3,1,3) forecast.

Germany: Information on capital stock for the period 1850 through 2000 is provided by Metz (2005). Missing values during and after WW2 have been estimated on the basis of Krengel (1958). Data on labour hours worked and real GDP is taken from Greasley and Madsen (2006). Factor shares used were from Greasley and Madsen (2006), capital share is 0.60 and labour 0.40. A Kalman filter of the TFP growth rate was estimated and this was forecast using an ARIMA (2,1).