Adaptation of Swiss data for the World Input-Output Database Technical report

Adaptation of Swiss data for the World Input-Output Database Technical report Carsten Nathani Pino Hellmüller Tonio Schwehr Rüschlikon, October 28, 2016

IMPRINT Rütter Soceco Socioeconomic Research + Consulting Weingartenstrasse 5 8803 Rüschlikon Tel. 044 724 2770 Email: info@ruetter-soceco.ch Authors Carsten Nathani (project lead) Pino Hellmüller Tonio Schwehr This project was supported by the Swiss Secretariat for Economic Affairs (SECO). The content of this report does not necessarily reflect the opinion of the SECO. Disclaimer The Swiss data for the WIOD were prepared with due care within the given limited financial resources. The data are supplied as is and it is the responsibility of the users of the data to assess the usefulness of the data for their specific purposes. Rütter Soceco is not responsible for, and expressly disclaims all liability for, damages of any kind arising out of use, reference to, or reliance on the provided data. No guarantee is given that the information provided in the data set is correct, complete, and up-to-date.

CONTENT Content 1. Background and project goal 5 2. Adaptation of the supply and use table 2011 8 2.1 Supply table 8 2.2 Use table 8 2.2.2 Balancing of supply and use tables 10 3. Estimation of a supply and use table 2008 12 4. Generation of other input data 13 4.1 Data for the disaggregation of selected industries 13 4.1.1 Primary sector 13 4.1.2 Refineries 13 4.2 Employment data 13

BACKGROUND AND PROJECT GOAL 5 1. Background and project goal Background The World Input-Output Database (WIOD) provides time-series of world inputoutput tables for forty countries worldwide and a model for the rest-of-the-world, covering the period from 1995 to 2011. These tables have been constructed on the basis of officially published input-output tables in conjunction with national accounts and international trade statistics. In addition, the WIOD provides data on labour and capital inputs and pollution indicators at the industry level (http://www.wiod.org/new_site/home.htm, cf. also Timmer et al. 2015 for an introduction to the WIOD). The research consortium responsible for establishing the WIOD is currently preparing a new update. Time series of international input-output tables in current and previous-year prices as well as socio-economic and environmental accounts are generated for the time period 2005-2014. The new WIOD follows the concepts and definitions of the European System of National and Regional Accounts (ESA) 2010 and the industry classification NACE Rev. 2 resp. the product classification CPA 2008. For the first time, Switzerland will be included in the WIOD. The construction of the WIOD basically includes the following steps,! constructing time series of national supply and use tables in current and previous-year prices for each country,! constructing time series of international supply and use tables for each country in current and in previous year prices by disaggregating imports of a country by supplying country and industry,! constructing time series of integrated world input-output tables in current and in previous year prices,! constructing time series of socio-economic and environmental accounts that are compatible with the SUT. For the estimation of the Swiss supply and use tables in current prices, the WIOD team is planning to use the following published input data! the Supply table and the Use table at basic prices incl. net commodity taxes for 2011,! data from the national accounts on gross output and gross value added by industry for 2005-2014,! data from the OECD STAN database to further disaggregate certain industry data from the national accounts,! data from the national accounts on GDP by expenditure for 2005-2014, i.e. the contribution of final demand categories (and imports) to GDP. Thus, in order to estimate the initial SUTs for years other than 2011, structural data from the benchmark SUT 2011 are combined with the annual national accounts totals. The initial SUTs are then balanced with a SUTRAS approach. The use of Swiss data for the WIOD poses several difficulties:! The Swiss input-output tables (IOT) are classified by the Swiss Federal Statistical Office as experimental, since important basic statistics typically used to generate IOT are missing in Switzerland (e.g. product statistics, cost structure

6 BACKGROUND AND PROJECT GOAL statistics, statistics on intermediate inputs as well as trade and transport margins). Therefore various approaches have been developed to estimate the IOT with the existing data (Nathani et al. 2011, SFSO 2016a). Generally the data on sectoral output, value added and imports in the Swiss SUTs have a high level of confidence, data on the rest of the supply table and final demand (except for change of inventories) have a medium level of confidence and data on intermediate inputs have a low level of confidence.! For Switzerland, the SUT 2011 is currently the only table that follows ESA 2010 and NACE Rev. 2. The aggregation level of the Swiss SUT is higher than that of the typical Eurostat SUT.! Swiss national accounting differs from national accounting in other countries with respect to the aggregation of enterprise units into industries. Whereas ESA 2010 recommends the grouping of local kind-of-activity units (KAU) into industries, in Switzerland the industries Financial service activities (NACE 64, incl. part of NACE 66) and Insurances (NACE 65, incl. part of NACE 66) are largely equal to the respective subsectors of the institutional sector Financial corporations. The industry Public administration and defence; compulsory social security is equal to the respective institutional sector General government. The other industries are aggregations of enterprises instead of local KAU. This partly leads to large off-diagonal entries in the Swiss supply table.! Furthermore the industry totals of the SUT differ from the industry totals of the Swiss production account in the case of the industries public services (NACE 84) and education (NACE 85). Whereas in the production account the industry public services also includes public education, in the SUT public education has been integrated into the industry education.! The Swiss SUT 2011 is available only at basic prices incl. net commodity taxes. Tables at basic prices, that are needed for integration into the WIOD, are not published in Switzerland.! In the Swiss SUT supply and use of non-monetary gold (NMG) is included, whereas in the WIOD it is meant to be excluded. Imports and exports of NMG are highly relevant, since Switzerland is a major physical trading platform for gold in the world. Project goal The goal of this project was to improve the accuracy of the Swiss data in the WIOD by adapting the existing Swiss data to the needs of the WIOD team and by supplying additional Swiss data to the WIOD team. This includes the following work:! to adapt the supply and use table 2011 to a version in basic prices, that excludes supply and use of non-monetary gold,! to estimate an additional benchmark supply and use table in basic prices for 2008, that is compatible with the SUT 2011 and with the Swiss national accounts data,! to supply a time series of data on output and value added for the disaggregation of the primary sector and of manufacture of refined petroleum products from the chemical industry,! to estimate a time series of employment in full-time equivalents or in working hours by industry.

BACKGROUND AND PROJECT GOAL 7 These data sets are supplied to the WIOD team and serve as an additional input for the integration of the Swiss data into the WIOD database. Structure of this report This report briefly summarises the data sources and methods used for the generation of the above-mentioned data sets. Chapter 2 deals with the adaptation of the SUT 2011. Chapter 3 describes the work on the estimation of the SUT 2008. The estimation of the other data is included in chapter 4.

8 ADAPTATION OF THE SUPPLY AND USE TABLE 2011 2. Adaptation of the supply and use table 2011 As mentioned above, the published SUT 2011 is available in basic prices incl. net commodity taxes (NCT) and includes supply and use of non-monetary gold. In this chapter the adaptation of the SUT 2011 to basic prices and the exclusion of supply and use of non-monetary gold is described. In the official Swiss SUT 2011 the supply and use of non-monetary gold is recorded in the following way, according to information from Swiss Statistics. On the supply side, imports of non-monetary gold is based on data from the foreign trade statistics. Approximately 60% of the value is allocated to import of metals and 40% to import of financial services. On the use side net imports (imports - exports) are recorded as net acquisition of valuables. Regarding export of non-monetary gold, the allocation to specific product groups is not apparent. Instead, in the step of balancing the supply and use table, the export value of NMG was added to export totals and the allocation to specific product groups was left to the automatic balancing procedure. It is therefore not straightforward to extract the NMG on the export side. Furthermore this approach is likely to have distorted other transactions in the use table due to the large value of NMG exports (78.2 M CHF). 2.1 Supply table On the supply side the following adjustments were made:! In a first step to exclude the non-monetary gold, the above-mentioned import values for non-monetary gold were subtracted from the respective import values for metals and financial services.! Other adjustments to the supply table were made in the balancing process and are described below. 2.2 Use table Basic approach Since it was not straightforward to extract the non-monetary gold from exports, exports by product group had to be newly estimated. Furthermore other categories of final demand were also newly estimated as far as possible. This allows to avoid major distortions in final demand, stemming from the treatment of NMG in the original SUT 2011. Final demand values at purchasers prices were then transformed into basic prices. Regarding intermediate inputs we left the values unchanged. The adjusted supply table delivered the totals for product supply at basic prices. In comparison with the adjusted values in the use table imbalances remain that were first analysed, adjusted manually where possible and then finally balanced with the RAS procedure. In the following sections the working steps are explained in more detail.

ADAPTATION OF THE SUPPLY AND USE TABLE 2011 9 Exports Since it was not straightforward to extract the non-monetary gold from exports, exports by product group were newly estimated. Exports of goods were calculated from the foreign trade statistics by converting the data from the HS classification used in the foreign trade statistics to the CPA classification used in the SUT with a correspondence table from Eurostat. Exports of services by service category were taken from the current account and allocated to product groups. Consumption expenditure of private households Consumption expenditure of private households in the national accounts is partly based on the results of the Swiss household budget survey and partly on other sources (cf. SFSO 2015 for details). In the estimation of household consumption we followed this approach as far as possible:! Consumption expenditures by product group in the consumption categories COICOP 01 to 05 and 07 to 10 were calculated on the basis of the household budget survey.! Health related expenditures (COICOP 06) were based on the statistics of costs and financing of health care (SFSO 2016).! Consumption expenditures in the categories COICOP 11 and 12 were taken from the published SUT 2011, since a new estimation was not feasible with the existing data. Government consumption expenditure Government consumption expenditures were derived from the government financial statistics, published by the Federal Finance Administration according to the methodology used in the estimation of earlier IO tables (Nathani et al. 2011). The government financial statistics are available by function and by type of expense. This allows to determine consumption expenditures by product group. Gross fixed capital formation Gross fixed capital formation by product group was calculated separately for construction and equipment. For construction, total expenditure for building construction and civil engineering according to the Swiss national accounts were used as a starting point. These were distributed to product groups according to cost structures used in the estimation of the Swiss IOT 2008. For equipment investment, a rough disaggregation by product groups is available in the Swiss national accounts that allowed an allocation to the NACE classification. Net acquisition of valuables According to Swiss Statistics inventory of national accounting methods, net acquisition of valuables (NAV) is calculated as net imports for certain products groups classified as valuables. Following this approach the total value for NAV could be allocated to SUT product groups with data from the foreign trade statistics.

10 ADAPTATION OF THE SUP PLY AND USE TABLE 2011 Other categories of final demand Data for the remaining categories of final demand, i.e. consumption expenditures for NPISH and changes in inventories, were adopted from the existing use table 2011. Conversion into basic prices Transactions that are recorded in purchasers prices (i.e. household consumption, gross fixed capital formation) resp. in fob prices (commodity exports) were transformed into basic prices by estimating trade and transport margins and reallocating these to the respective trade and transport industries. Net commodity taxes were subtracted from transaction values and reallocated to the respective row of the use table. Value added taxes were estimated with a matrix of transaction specific tax rates. Selected commodity taxes and subsidies could directly be assigned to specific relevant transactions. For other commodity taxes and subsidies, product group specific tax resp. subsidy rates were calculated from the supply table Intermediate inputs Intermediate inputs in basic prices were mainly adopted from the existing use table. 2.2.2 Balancing of supply and use tables The result of the above-mentioned steps were supply and use tables that were unbalanced with regard to total supply and use of product groups. The imbalances were analysed and specific entries in the supply and the use tables were adjusted to reduce the imbalances, when the reasons for imbalances were apparent or uncertainties were known to be particularly large. The adjustments refer to the following transactions: Supply table:! Shifts between production of public administration, education and health services by public administration,! shifts of manufacturing of chemical products between the chemical and the pharmaceutical industry. Use table:! Intermediate inputs of research and development services,! shifts between consumption expenditures of NPISH for social work services and other services,! export of hotel services,! export of administrative and support services. After these manual adjustments, the supply table as well as output and gross value added in the use table were fixed. The widely used RAS balancing procedure was used to equalise the remaining imbalances in the use table. Total product use as available from the supply table and total intermediate inputs resp. total final demand at purchasers prices as available from the Swiss national accounts were used as row and column totals. In order to account for the different levels of data uncertainty, the use table entries were partly fixed, i.e. excluded from the RAS

ADAPTATION OF THE SUPPLY AND USE TABLE 2011 11 balancing procedure. This mainly refers to final demand entries, of which between 80% and 100% (latter for net acquisition of valuables and exports) were fixed. Therefore the balancing more strongly affected the (more uncertain) intermediate inputs part of the use table. The final results of these working steps were a balanced supply and use table 2011 in basic prices, excluding non-monetary gold.

12 ESTIMATION OF A SUPPLY AND USE TABLE 2008 3. Estimation of a supply and use table 2008 For the estimation of a second benchmark supply and use table for the reference year 2008 a more simple approach was used due to limited project resources. The basic idea was to combine the national accounts data for 2008 with structural information from the new SUT 2011. Solely imports, net acquisition of valuables and exports were freshly calculated from foreign trade data. Supply table The total output of industries, total net commodity taxes and total imports were available from the Swiss national accounts. Industry outputs and net commodity taxes were distributed to product groups with the respective structures from the supply table 2011. Imports by product group were estimated with data from the foreign trade statistics and the current account. Use table For the estimation of the use table the following data were available from the Swiss national accounts:! Output at basic prices and gross value added by industry,! total product supply by product group at basic prices from the supply table (equals total product use),! total consumption expenditures of private households for 12 consumption categories at purchasers prices,! total consumption expenditures of NPISH, government and social security funds at purchasers prices,! total gross fixed capital formation for construction and equipment at purchasers prices,! total net acquisition of valuables and inventory changes,! total exports, distinguishing commodity exports, travel services and export of other services. To estimate a preliminary use table, the above-mentioned values were combined with the respective structural information from the use table 2011. Exports and net acquisition of valuables were newly estimated using the same approach as for 2011 (cf. chapter 2). Sources for input data were the Swiss foreign trade statistics and the current account. The resulting unbalanced use table was then balanced with the RAS procedure, using the same approach as for the balancing of the use table 2011 (cf. chapter 2).

GENERATION OF OTHER INPUT DATA 13 4. Generation of other input data 4.1 Data for the disaggregation of selected industries 4.1.1 Primary sector In the published Swiss SUT, the primary sector is aggregated to one industry, whereas disaggregated data are used for the construction of the WIOD. As input data for the disaggregation we constructed a time series of output and gross value added of agriculture, forestry and fishery for the years 2005 to 2014, based on data from the economic accounts for the primary sector (SFSO 2016c). 4.1.2 Refineries In the published Swiss SUT the manufacture of refined petroleum products is aggregated with the chemical industry. In order to allow for disaggregation, we estimated a time series of output and gross value added of refineries for the years 2005 to 2014. Production of petroleum-based lubricants has a minor economic relevance in the Swiss refined petroleum product industry and was disregarded in the estimation due to missing data. Output of refineries was estimated by combining quantity with price data. Quantities of manufacturing of refined petroleum products were available from Swiss energy statistics (SFOE 2016). Production prices were derived from product specific import and export prices, calculated with data from the Swiss foreign trade statistics. Total intermediate inputs were calculated as the sum of expenditures for crude oil and other intermediate inputs, the latter estimated as a fixed value per ton of processed crude oil. Gross value added was calculated as output minus intermediate inputs. 4.2 Employment data tbd in the final version of this report

ABBREVIATIONS Abbreviations COICOP: CPA: IOT: NACE: NAV: NMG: NPISH: SFOE: SUT: WIOD: Classification of Individual Consumption by Purpose Classification of Products by Activity Input-output table Nomenclature statistique des activités économiques dans la Communauté européenne Net acquisition of valuables Non-monetary gold Non profit institutions serving households Swiss Federal Office for Energy Supply and use table World Input-Output Database

REFERENCES References Nathani, C., Schmid, Ch., van Nieuwkoop, R. (2011): Schätzung einer Input-Output-Tabelle der Schweiz 2008 (engl.: Estimation of a Swiss input-output table 2008). Final report for the Swiss Federal Statistical Office, Rütter Soceco, Rüschlikon. Download: https://www.bfs.admin.ch/bfs/de/home/statistiken/volkswirtschaft/input-output.assetdetail.348196.html. Swiss Federal Office of Energy (SFOE, 2016): Mineral oil balance. Tables for 2005 to 2014. Personal communication from the Swiss Federal Office of Energy. September 2016. Swiss Federal Statistical Office (SFSO, 2015): Inventory of methods used to estimate Switzerland s GDP and GNI. Neuchâtel. Download: https://www.bfs.admin.ch/bfs/de/home/statistiken/volkswirtschaft/volkswirtschaftlichegesamtrechnung/bruttoinlandprodukt.assetdetail.328585.html Swiss Federal Statistical Office (SFSO, 2016a): IOT 2011: Erläuterungen und Hinweise zur Nutzung der Schweizerischen Input-Output-Tabelle 2011 (engl.: IOT 2011: Explanations and notes for the use of the Swiss input-output table 2011). Neuchâtel. Download: https://www.bfs.admin.ch/bfs/de/home/statistiken/volkswirtschaft/input-output.assetdetail.350310.html Swiss Federal Statistical Office (SFSO, 2016b): Costs and financing of the health system 2014. Neuchâtel. Swiss Federal Statistical Office (SFSO, 2016c): Economic accounts for the primary sector. Neuchâtel. Download: https://www.bfs.admin.ch/bfs/en/home/statistics/agriculture-forestry/economic-satellite-accounts.html Timmer, M.P., Dietzenbacher, E., Los, B., Stehrer, R., de Vries, G.J. (2015): An Illustrated User Guide to the World Input Output Database: the Case of Global Automotive Production. Review of International Economics, 23, pp. 575-605.