Workshop on Agricultural Capital Stock and Related

Structural Statistics

13 November 2015, RomeFood and Agriculture Organization of the United Nations (FAO)

Background

Background/Mandate• New work on methodologically robust Agricultural Capital Stock statistics

funded by One of Five FAO Strategic Objectives (SO) for the 2014-2015 biennium :

To Enable inclusive and efficient agricultural and food systems (SO4)

• Output/Activity required from Statistics Division:

• To develop and implement methodologies for the measurement of capital stock and mechanization

• To develop, collect/compile, validate, and disseminate data on annual capital stock estimates

• Work and approach supported by FAO Legislated Bodies (Committee on Agriculture, African Commission on Agricultural Statistics, Asia-Pacific Commission on Agricultural Statistics, Latin America-Caribbean-FAO Working Group on Agricultural Statistics)

Variables in the databasea) Capital Stock related variables

Total economy Agriculture, Forestry & Fishery (AFF) Agriculture (Ag)

  Gross fixed capital formation ✔  ✔ ✔ Net capital stock ✘ ✔ ✔ Gross capital stock ✘ ✔ ✔ Consumption of fixed capital ✘ ✔ ✔

b) Macro-economic variables not in a) Value added¹ ✔³ ✔ ✔ Gross Output² ✘ ✔ ✔ Operating surplus, gross ✘ ✔ ✔ Operating surplus, net ✘ ✔ ✔ Compensation of employees² ✘ ✔ ✔ Employment² ✘ ✔ ✔

c) Other labour variables (for OECD countries only)Wages and salaries, N. of employees, Self-employed, Full-time equivalents - total engaged & employees, Hours worked - total engaged & employees

✘ ✔ ✔

¹ necessary for calculating the investment ratio and capital stock² supporting variables in the calculation capital stock³ it refers to the Gross Domestic Product (GDP)

Introduction

1. The components of the agriculture capital stock of fixed assets

Farm structures, including major improvements to other buildings and structures

Transport equipment

Machinery and equipment

Communications equipment, office machinery and computers

Produced intangible fixed assets, e.g. software, R&D

Major improvements to tangible non-produced assets, e.g. land

Livestock

Trees

2. The components of the agriculture capital stock of fixed assets

1947

1948

1949

1950

1951

1952

1953

1954

1955

1956

1957

1958

1959

1960

1961

1962

1963

1964

1965

1966

1967

1968

1969

1970

1971

1972

1973

1974

1975

1976

1977

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Share of equipment and structures in NCS and GFCF in the USA

Equipment-NCS Structures-NCS Equipment-Inv

3. Two FAO databases on Capital Stock with two different approaches

1. Current FAOSTAT database , with data up to 2007, based on aggregation of separate calculations of asset categories;

2. A new national-accounts based Capital Stock database.

Part IFAOSTAT Data on Capital Stock:

The Old Methodology

1. FAOSTAT – Net Capital Stock in percent of Gross Capital Stock

countries item 2000 2003 2005 2007

Denmark Land Development 98.0 98.0 98.0 98.0

Denmark Livestock (Fixed Assets) 100.0 100.0 100.0 100.0

Denmark Livestock (inventory) 100.0 100.0 100.0 100.0

Denmark Machinery & Equipment 87.5 87.5 87.5 87.5

Denmark Plantation Crops 95.5 95.5 95.5 95.5

Denmark Structures for Livestock 95.5 95.5 95.5 95.5

Denmark Capital Stock + (Total) 94.8 94.8 94.9 94.9

Ecuador Land Development 98.0 98.0 98.0 98.0

Ecuador Livestock (Fixed Assets) 100.0 100.0 100.0 100.0

Ecuador Livestock (inventory) 100.0 100.0 100.0 100.0

Ecuador Machinery & Equipment 87.5 87.5 87.5 87.5

Ecuador Plantation Crops 95.5 95.5 95.5 95.5

Ecuador Structures for Livestock 95.5 95.5 95.5 95.5

Ecuador Capital Stock + (Total) 97.8 97.9 97.9 97.9

1. FAOSTAT – Net Capital Stock in percent of Gross Capital Stock

countries item 2000 2003 2005 2007

France Land Development 98.0 98.0 98.0 98.0

France Livestock (Fixed Assets) 100.0 100.0 100.0 100.0

France Livestock (inventory) 100.0 100.0 100.0 100.0

France Machinery & Equipment 87.5 87.5 87.5 87.5

France Plantation Crops 95.5 95.5 95.5 95.5

France Structures for Livestock 95.5 95.5 95.5 95.5

France Capital Stock + (Total) 93.6 93.5 93.7 93.6

Niger Land Development 98.0 98.0 98.0 98.0

Niger Livestock (Fixed Assets) 100.0 100.0 100.0 100.0

Niger Livestock (inventory) 100.0 100.0 100.0 100.0

Niger Machinery & Equipment 87.5 87.5 87.5 87.5

Niger Plantation Crops 95.5 95.5 95.5 95.5

Niger Structures for Livestock 95.5 95.5 95.5 95.5

Niger Capital Stock + (Total) 98.9 98.9 99.0 99.0

2. Benchmarking FAOSTAT and OECD data on GCS (1)

Source: FAOSTAT     Sources: OECD and FAO SST

Gross capital stock, millions 2005 USD Gross capital stock, millions 2005 USD

        GCS Gross output Value added

Countries item 2005 DK/ EC   DK/ EC DK/ EC

 Denmark  Land Development 2,993 0.4       Denmark  Livestock (Fixed Assets) 3,693 0.6       Denmark  Livestock (inventory) 652 0.6       Denmark  Machinery & Equipment 4,296 9.4       Denmark  Plantation Crops 24 0.006       Denmark  Structures for Livestock 877 1.8      

Denmark Capital Stock + (Total) 12,535 0.6 63,065 1.8 0.9 Ecuador  Land Development 7,030         Ecuador  Livestock (Fixed Assets) 6,108         Ecuador  Livestock (inventory) 1,078         Ecuador  Machinery & Equipment 456         Ecuador  Plantation Crops 4,199         Ecuador  Structures for Livestock 486        

Ecuador Capital Stock + (Total) 19,358   4,820    

GCS for Ecuador is calculated from reported values of GFCF

2. Benchmarking FAOSTAT and OECD data on GCS (2)

Source: FAOSTAT     Sources: OECD and FAO SST

Gross capital stock, millions 2005 USD Gross capital stock, millions 2005 USD

        GCS Gross output Value added

Countries item 2005 FR/ EC FR/ EC FR/ EC

 France  Land Development 19,050 2.7       France  Livestock (Fixed Assets) 18,783 3.1       France  Livestock (inventory) 3,315 3.1       France  Machinery & Equipment 40,590 89.1       France  Plantation Crops 3,695 0.9       France  Structures for Livestock 10,724 22.1      

France Capital Stock + (Total) 96,156 5.0 227,347 17.3 13.0 Ecuador  Land Development 7,030         Ecuador  Livestock (Fixed Assets) 6,108         Ecuador  Livestock (inventory) 1,078         Ecuador  Machinery & Equipment 456         Ecuador  Plantation Crops 4,199         Ecuador  Structures for Livestock 486        

Ecuador Capital Stock + (Total) 19,358   4,820    

2. Benchmarking FAOSTAT and OECD data on GCS (3)

Gross Capital Stock 2005, millions 2005 USD Net Capital Stock, millions 2005 USD

 

FAOSTAT OECD (ISIC Rev.3:01)

FAOSTAT /OECD FAOSTAT OECD (ISIC Rev.3:01)

FAOSTAT/OECD

Australia 119,275 74,345 1.6 115,979 61,417 1.9Austria 15,412 59,601 0.3 14,017 46,344 0.3Belgium 6,912 17,349 0.4 6,623 8,056 0.8Canada 100,954 68,212 1.5 94,596 39,678 2.4Czech Republic 11,852 14,859 0.8 11,133 7,167 1.6Denmark 12,535 63,065 0.2 11,897 33,992 0.4Estonia 2,171 796 2.7 1,987 596 3.3Finland 12,265 30,403 0.4 11,101 15,219 0.7France 96,156 227,347 0.4 90,052 120,560 0.7Germany 84,125 290,611 0.3 76,597 146,516 0.5Hungary 11,370 31,888 0.4 10,766 18,203 0.6Iceland 985 991 1.0 898 754 1.2Ireland 20,470 19,766 1.0 19,573 11,104 1.8Israel 2,421 5,020 0.5 2,316 3,208 0.7Italy 85,127 426,349 0.2 78,328 217,319 0.4

FAOSTAT ≤ 50% of reported GCS FAOSTAT ≥ 150% of reported GCS

2. Benchmarking FAOSTAT and OECD data on GCS (4)

Gross Capital Stock 2005, millions 2005 USD Net Capital Stock, millions 2005 USD

 

FAOSTAT OECD (ISIC Rev.3:01)

FAOSTAT /OECD FAOSTAT OECD (ISIC Rev.3:01)

FAOSTAT/OECD

Japan 293,689 477,191 0.6 266,445 309,337 0.9Luxembourg 496 2,071 0.2 445 1,027 0.4Netherlands 12,057 84,645 0.1 11,591 48,290 0.2Norway 9,157 18,079 0.5 8,345 14,612 0.6Poland 74,080 83,994 0.9 67,304 23,597 2.9Portugal 14,785 14,691 1.0 13,976 6,976 2.0Republic of Korea 16,332 92,978 0.2 15,318 61,129 0.3Slovakia 6,513 8,285 0.8 6,249 3,127 2.0Slovenia 2,861 3,729 0.8 2,588 1,450 1.8Spain 83,135 117,695 0.7 78,239 86,362 0.9Sweden 14,662 30,878 0.5 13,656 18,808 0.7Switzerland 8,535 33,961 0.3 7,945 17,208 0.5United Kingdom 49,011 77,269 0.6 45,891 37,062 1.2United States of America

617,633 727,166 0.8 577,252 397,855 1.5

FAOSTAT ≤ 50% of reported GCS FAOSTAT ≥ 150% of reported GCS

LAND DEVELOPMENT

“Land development = ∑ {(arable land) x (unit price) + (irrigated land) x (unit price)} Plantation crops = ∑ (land under permanent crop) x (unit price)”   AGRICULTURE MACHINERY “Machinery and equipment = ∑ {(number of machinery for i) x (unit price of machinery for i) +

(economically active population in agriculture) x US $35)} Where i stands for tractor, harvester & threshers and milking machine. US $ 35 has been taken

from 1995 series after adjusting for price rises.”

3. FAOSTAT asset calculation

Part IIThe New FAO Database on

Agricultural Capital Stock and Related Structural Statistics (SST)

1. National-accounts based Capital Stock databaseOutputs Produced

• Global database on Agricultural Capital Stock and related structural statistics,

• Covering 223 countries and territories for 1970-2013, with supporting metadata/documentation

• Industrial Coverage: Total Economy; Agriculture, Forestry & Fishing (AFF), ISIC Rev.3:A+B; Agriculture (Ag) subsector, ISIC Rev.3:01

• Variables

• Capital Stock related variables: GFCF, Net Capital Stock (NCS), Gross Capital Stock (GCS), Consumption of Fixed Capital (CFC)

• Other SNA variables: Value-Added¹, Gross Output², Gross Operating Surplus, Net Operating Surplus, Compensation of Employees², Employment²

• Other Employment Variables (for OECD countries only): Wages and salaries, Number of employees, Number of self-employed, Full-time equivalents - total engaged, Full-time equivalents – employees, Hours worked - total engaged, Hours worked – employees

• Other indicators: Investment Ratio (IR), and Agriculture Orientation Index (AOI)

¹ necessary for calculating the investment ratio and capital stock² supporting variables in the calculation capital stock

2. Key Findings (1)• Agriculture, forestry 

and fishery GFCF in 2013 estimated at $363 billion globally (2005 USD), up 5% from 2012 and 50% from 2000.

• Asia showed highest level of agricultural investment in 2013, with GFCF of $131 billion (2005 USD) in 2013, surpassing Europe which led until 2008. 

• Note: Analysis begins in 2000 due to missing or unreliable data for many countries prior to 20002000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

0

20,000

40,000

60,000

80,000

100,000

120,000

140,000

Gross Fixed Capital Formation in Agriculture, forestry and fishery by region, constant 2005 USD, 2000-2013

AfricaAsia & PacificEuropeLatin America & CaribbeanNorthern AmericaOther Developed

2005

USD

(bill

ions

)

*Other Developed includes Australia, Japan and New Zealand

2. Key Findings (2)

• In developed countries, the GFCF Agriculture Orientation Index (AOI) usually exceeds 1, indicating a higher level of investment in AFF relative to its share of the economy.  

• Developing countries have a GFCF Agriculture Orientation Index (AOI) less than 1, indicating a lower share of agricultural investment relative to its economic share.  

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 20130.0

0.5

1.0

1.5

2.0

2.5

Agricultural Orientation Index (AOI) in Agriculture, forestry and fishery by region, constant 2005 USD, 2000-2013

Africa Asia & PacificEurope Latin America & CaribbeanNorthern America Other Developed

*Other Developed includes Australia, Japan and New Zealand

Introduction to the new methodology

3. Methodological Approach

1. Data Source: selection varies by country and variable• UNSD: National Accounts Estimates (UNSD : NAE), Official Country Data (UNSD : OCD)• OECD: Structural Analysis database (OECD-STAN), National Accounts Database (OECD-NA)• World Input Output Database (WIOD)

2. Data Bridging• Bridging across data series within a single data sources • Bridging across ISIC revisions 3. and 4.• Bridging across data sourcesNote: Bridging was essential to create a long time series for capital stock related variables

3. FAO Estimation/imputation• Imputation/estimation missing observations in an existing series• Estimations of capital stock related variables, and variables required for these estimations

(Value-Added, Gross Output)Note: No estimation or imputation was conducted for other variables

4. Number of countries with data in the UN-NAE and UN-OCD

INDUSTRIES COVERED ISIC Rev 3

 AGRICULTURE, FORESTRY, FISHERY A+B

 AGRICULTURE HUNTING AND RELATED SERVICIES A01

  No. of countries with data    Total economy A+B A01

I. NATIONAL ACCOUNTS ESTIMATES (UNSD) GDP (current and fixed prices, LCU and $)   >200 GFCF (current and fixed prices, LCU and $)   >200 GFCF deflator   >200 GDP/capita (current prices, LCU and $)   >200 Value added ISIC Rev3: A+B (current and fixed prices, LCU and $) >200

Value added deflator   >200

II. OFFICIAL COUNTRY DATA (UNSD)   OUTPUT   148 98  Less Intermediate consumption VALUE ADDED, GROSS   157 105 COMPENSATION OF EMPLOYEES 126 81OPERATING SURPLUS, GROSS     134 78MIXED INCOME, GROSS   25 18  Less CONSUMPTION OF FIXED CAPITAL 98 56OPERATING SURPLUS, NET     104 55MIXED INCOME, NET   15 11GROSS FIXED CAPITAL FORMATION   87 54CLOSING STOCK OF FIXED ASSETS   23 22EMPLOYMENT   70 50

Data for additional countries from STAN, OECD-NA and WIOD

5. Data Selection: How data sources were used

• For OECD countries , by order of preference:

1. OECD: Structural Analysis database (OECD-STAN)2. OECD - National Accounts Database (OECD-NA)3. UNSD : Official Country Data4. World Input Output Database (WIOD)

• For all other countries, by order of preference:

1. UNSD : Official Country Data2. non-OECD countries covered in OECD-STAN and OECD-NA3. World Input Output Database (WIOD)

6. Bridging (1)

1. WHAT IS THE BRIDGING ?

It is a process that connects two or more :•fully specified data stores •for a limited time or on an ongoing basis

N.B.: Before the bridging is undertaken, we “rebase” the constant values using the 2005 as a base year. Moreover, only values in LCU are bridged.

2. WHY WE USE THE BRIDGING ?

In order to have long data series, as data in different series / ISIC Revisions / Sources are not comparable. 

6. Bridging (2)

3. WHEN WE USE THE BRIDGING ?

i. Series-bridging This involves UNSD - Official Country Data, as data are reported in different series.

Different series numbers are used to store different time-series versions of national accounts statistics, e.g. different SNA national accounts methodology, different currencies, fiscal years, or by different sources.

• NB: Priority is given to the most recent series

ii. ISIC Revisions-bridgingThe present dataset reports data in ISIC Rev.3, as most of the countries still report information in ISIC Rev.3.

The bridging of various revisions of ISIC involves the UNSD - Official Country Data & the OECD – STAN database. If no data for a certain year is available in ISIC Rev.3, but it is available in ISIC Rev.4, then the latter will be bridged into ISIC Rev.3.

6. Bridging (3)

iii. The sources-bridgingIn order to expand the time series, different sources are bridged.

How do we use the sources ?

1. OECD (STAN and National Accounts) : it is used as the primary source for OECD countries (the dbs include also some non-OECD countries, e.g. China, Brazil)

NB: data from OECD-NA are in ISIC Rev.4 only - converted into ISIC Rev.3.

2. UNSD (Official Country Data) : it is usually used for non-OECD countries.

3. WIOD : it is used when no other source is available.

6. Bridging (4)

4. HOW THE BRIDGING IS UNDERTAKEN ?

The bridging is possible only when there is overlap between at least two data for the same year.Once the two series to be bridged have been identified, we calculate their conversion factor (as a ratio of the two series):

• We have used the last available year or the last 3-years average, depending on the trend of the series: if the trend is more or less stable, then we assume that the last available year is sufficient, otherwise if the values varies significantly over the years, then an average has been taken.

The bridging is not recommended when the two series differ significantly, e.g. when the conversion factor is greater than 1.5 or is smaller than 0.5.

6. Bridging: examplesSeries-bridgingGross Fixed Capital Formation, ISIC Rev.3:A+B

• No overlap• Different currency Bridging not possible

The bridging is possible

7. FAO Estimation/imputationImputation GFCF

a. If reported data are available only for a limited number of years : the missing years' data are imputed based on data on the most recent, or an average of the most recent, available investment ratio (GFCF/VA).

b. If there are no data at all :• imputation for GFCF in Ag are based on (1) value-added ratio  for Ag over AFF to GFCF in AFF;• imputation for GFCF in Ag & GFCF in AFF are based on (2) regression equations (linear and logarithmic) in which the endogenous variable is the investment ratio and the exogenous variable GDP/capita.

It would have been preferable to use VA/EMPL as explanatory variable but many employment series are missing or are of dubious quality.

The parameters of the equation are estimated based on the data from the countries from which data are available. The R2 value of the regressions vary between 0.9 (logarithmic regression for middle and high income countries) and 0.74 (linear regression for low income countries).   

When reported country data on CFC are available these are used for estimating GFCF, when the latter are missing, as the averages of CFC and GFCF over a medium term period are normally of the same magnitude.

7. FAO Estimation/imputationCalculation of Capital Stock

Data on Net Capital Stock (NCS), Gross Capital Stock (GCS) and Consumption of Fixed Capital (CFC) is available only for a limited number of countries (mainly OECD countries). 

For all other countries data have been calculated by FAO using the Double Declining Balance Method.

In applying this method, assumption were made of depreciation rates which range from 0.03 to 0.08 depending on the economic level of the countries.

There is large degree of arbitrary judgements in applying these deprecation rates.  Hence, data should be interpreted with great care as they only might give an order of magnitude and not information about the exact level.

7. FAO Estimation/imputationThe estimation

b) Macro-economic variables not in a)Agriculture, Forestry & Fishery (AFF) Agriculture (Ag)

Value added Missing data are replaced by bridged data of Value Added, UNSD-NAE.

If there are no Official Country Data at all then the UNSD-NAE have been used.

Missing data are replaced by bridged data of Value Added, UNSD-NAE from those years when data are available.

If there are no Official Country Data at all, then the whole data series are imputed, using UNSD-NAE for Agriculture, forestry, fishery (ISIC Rev.3:A+B) as the basis and then multiply by the value-added ratio of Agriculture (ISIC Rev.3:01) over Agriculture, forestry, fishery (ISIC Rev.3:A+B) for neighbouring countries with the same level of economic development and structure.

Output Missing data in country series of Gross Output, Official Country Data, are imputed using the ratio of Gross Output/Value Added of adjacent years.

In case the data series are empty then Gross Output is estimated using the ratio of Gross Output/Value Added for neighbouring countries with the same level of economic development.

Missing data points in country series of Gross Output, Official Country Data, are imputed using the ratio of Gross Output/Value Added of adjacent years.

In case the data series are empty then Gross Output is estimated using the ratio of Gross Output/Value Added for neighbouring countries, with the same level of economic development and structure, applied to available data on Value added, National Accounts Estimates for Agriculture, forestry, fishery (ISIC Rev.3: A+B)

8. ConclusionsChallenges & possible improvements/development

The fact that a large amount of data on GFCF is imputed or estimated is not the only weakness of the database.

Another weakness is that we have no information on how large share of GFCF is machinery and how large share is structures.

The present database on Gross Fixed Capital Formation is a test database. The methodology and input data will continuously be improved - in particular when FAO data on machinery investment become available.

Part III Calculating capital stock and initial stock

Calculation method to be used for the FAO capital stock measures

Double-declining balance method (proposed by the OECD Capital Stock Manual, the SNA 2008 and BEA)

Where WtE and WtB are the end-year and beginning-of-the year net capital stocks, It  is gross fixed capital formation, δ (It/2+WtB) is consumption of fixed capital,δ is the depreciation rateδ=R/TA  where TA is the average service life of an asset, and R is a parameter around 2. 

WtE = WtB + It – δ (It/2+WtB) = It (1 - δ/2) + WtB (1 – δ)

Initial stocks

A starting stock for some period t₀ has to be computed

• Using capital survey information or

• Estimate for the long-run growth rate of volume investment when geometric age-efficiency or age-price profiles apply net stock at the beginning of the benchmark year t₀ can approximated by

Wt0 = It0 / (δ+ θ)

Θ = the long-run growth rate of volume GDP, or in our case, value added in agriculture

Some methodological issues for discussion

Some methodological issues for discussion

Depreciation rates for different groups of countries

Assumed growth rates for the initial capital stock

Deflators

Imputations when there are missing data and when there are no data at all

Stability of bridging data series (Rev 4 to Rev 3 and between different sources)

How to get reliable data on GFCF in agriculture machinery (the instability of 

COMTRADE and PRODCOM data)