# Annex A. Brief description of the PEM-Norway model

The OECD Policy Evaluation Model (PEM) is an equilibrium displacement model that contains explicit product and factor markets (see (OECD, 2005[1]; OECD, 2015[2]) for further details on the general structure of the model). These markets, which inter alia include land, chemicals and fertiliser use, provide a direct connection between economic policy, farm activities and their environmental consequences, in particular as regards to water pollution and climate change.

PEM Norway distinguishes four outputs and 13 inputs. The outputs are wheat, coarse grains (barley and oats), milk and beef. Milk is processed further into fluid milk (e.g. drinking milk, yoghurt, cream) sold on the domestic market only and industrial milk (e.g. cheese, milk powder, butter) with is sold on both the domestic market and the international market.

No factor is assumed to be completely fixed in production, but land and other farm-owned factors are assumed to be relatively more fixed (have lower price elasticities of supply) than the purchased factors. There are three farm-owned factors: land, cows, and a residual “other farm owned factors”. The representation of the land market allows simulating payments based on area, payments based on non-current area (historical entitlements), and farm income. The set of purchased factors cover fertiliser, chemical use, interest, irrigation, feed, machinery and many others.

The PEM model for Norway follows the regionalisation used in the Norwegian Farm Accountancy Register and divides Norway into five regions. It is a stand-alone version of the model that takes world market prices as given and assumes that domestic market prices are fixed via negotiations with producer groups. Domestic consumer prices only adjust when needed to clear markets to avoid additional subsidised exports. The model is calibrated to the situation in Norway in 2017.

Norwegian agriculture has four major objectives: Food security, agriculture all over the country, value creation, and sustainability with lower greenhouse gas (GHG) emissions. For each objective, multiple indicators are produced using the model results.

Indicators for food security include self-sufficiency (on a calorie basis), farm land per 1 000 capita and cows per 1 000 capita. The energy content of food used in the calculations is 2 920 kcal per kg wheat, 703.5 kcal per kg milk and 1 697 kcal per kg beef. The numbers are taken from the agricultural sector model Jordmod (Mittenzwei, 2018[3]). The population in 2017 was 5.258 million inhabitants (SSB, 2020[4]).

The objective agriculture all over the country is reflected in 13 indicators. Seven measure the share of various variables in central regions: overall land use, land use to wheat, land use to grains, land use to milk, land use to beef, milk production, and beef production. Eastern lowlands and Jæren are considered central regions with the best natural and climatic conditions for agriculture. Land use at the regional level make up five additional indicators, while the last indicator measures overall land use compared to the baseline. This indicator is meant to cover whether land use in a region changes overall.

Value creation is measured via farm incomes and productivity. Fifteen productivity indicators are included, which differ with respect to the measurement of outputs and inputs. In general, productivity measures output in relation to input. The following formula has been used:

$\left(\frac{{Output}_{1}}{{Output}_{0}}-1\right)-\left(\frac{{Input}_{1}}{{Input}_{0}}-1\right)$

where superscripts 1 and 0 indicate the scenario and the baseline, respectively. Inputs and outputs are valued at base year prices both in the scenarios and in the baseline.

Total factor productivity measures productivity growth of all outputs and all inputs. Productivity indicators are presented for each of the four outputs (wheat, coarse grains, milk, and beef) for all inputs, purchased inputs and farm-owned inputs.

The fourth policy objective, sustainability with reduced greenhouse gas emissions, is reflected in indictors on GHG emissions, nutrient balances, and selected aspects of cultural landscapes. The parameters to calculate gaseous emission and nutrient balances are adapted from data for Switzerland (OECD, 2015[2]). These parameters are specific to plains, hilly, and mountainous areas. The parameters for plains regions are applied to the eastern lowlands and Jæren. The central lowlands and the southern valleys are associated with hilly areas, while the parameters for mountainous areas are used for northern Norway.

Nitrogen balances and phosphorus balances measure inputs and outputs of the two nutrients from all sources and are calculated on a regional basis. The data used to construct the environmental indicators come from the OECD AEI database along with additional calculations to disaggregate the environmental indicators to match each commodity covered by PEM. The N and P balance indicators were constructed following OECD and EUROSTAT guidelines (Eurostat, 2013[5]; OECD, 2013[6]).

Greenhouse gas emissions in CO2-equivalents are produced using a conversion factor of 25 for methane and 298 for nitrous oxide following the AR4-report of the IPCC (GWP100). The GHG emission calculations were based on the national GHG inventory methods outlined in IPCC (2006), with the Tier 1 approach used to calculate N2O emissions from crops, and the Tier 2 approach used to calculate all other GHG emission sources.

Two environmental indicators shed light on aspects of cultural landscapes: the livestock density defined as the number of animals per unit of land devoted to milk and beef production, and grassland as a share of total land use. Both indicators are calculated at the regional level.

The data for the Norwegian version of the PEM model have been taken from many different sources. The OECD PSE-database, the OECD Aglink model, the Norwegian driftsgranskinger (i.e. counterpart to the EU Farm Data Network FADN) (Kristiansen, 2018[7]), the direct payment register of the Norwegian Agriculture Agency (2020) and a tool to calculate payments at the individual farm level (Mittenzwei, 2018[3]) have been most important.

The PEM model distinguishes five regions as presented in Chapter 1: eastern lowlands; Jæren; central lowlands; southern valleys; and northern Norway. The five regions are chosen to capture regional policy and geographic differences in order to provide a coherent analysis of the regional impact of policy reforms. Each region is relatively homogenous with respect to climatic and natural conditions for agriculture. Regions coincide with the regionalisation of the Norwegian driftsgranskinger and allow a straightforward calculation of factor shares for the major types of agricultural production. The regions also largely match the zones that exist for regionally differentiated payments in Norwegian agriculture.

Regional production volumes come from the price subsidy register of the Norwegian Agricultural Agency that collects data from dairies, slaughterhouses, and mills for the administration of regionalised output payments. Production volumes for processed products are taken from Norwegian Agriculture Agency that collects data on processed raw milk into different dairy products in connection with the administration of the milk price equalisation scheme. Domestic and international prices stem from the PSE database. Regional differences in output prices are insignificant and administrative prices are negotiated between the farmers’ organisations and the government.

Land use for wheat and coarse grains stems from the direct payments register. That register also contains data for fodder on arable land, surface-cultivated land and fenced pastures. It is assumed that 80% of that land is devoted to milk and beef. That factor is taken from the base year of the Norwegian agricultural sector model Jordmod (Mittenzwei, 2018[3]). It is further assumed that milk and beef occupy the same amount of grassland per animal unit. In sum, PEM covers about 80% of the utilised agricultural area in Norway.

The inputs are farm-owned capital, cows, land, concentrated feed, machinery and equipment, hired labour, chemicals, energy, fertiliser, insurance, irrigation, other purchased inputs and interests. The zero-profit condition in PEM facilitates that factor shares are sufficient to calibrate the model to the base year.

Factor shares are calibrated from the driftsgranskinger using the economic size unit of a farm as a selection criterion to identify a sufficiently large sample of representative farms for the four productions in the five regions. For grains and milk, farms with more than 99% and 80% of their total ESU from grains and milk, respectively, were selected. Beef production in Norway takes place either in combination with dairy cows or separate (i.e. suckler cows). Dairy farms with a share of more than 150% of their economic value from beef relative to milk and suckler cow farms with more 66% of their total economic value from beef are defined as beef farms. More detailed information on this data source can be found in (Mittenzwei, 2020[8]).

There is no distinction between factor shares for wheat and coarse grains. The shares for each production in each region are calculated based on the unweighted average of the inputs of the farms (Mittenzwei, 2020[8]). The costs for feed concentrates, chemicals, energy, fertiliser, hired labour, insurance, machinery and equipment, irrigation and interests are taken directly from the farm accounts. The cost for land is calculated as the sum of own land and rented land multiplied by the price of rented land. The cost for cows is half the value of cows in the balance and multiplied by the stipulated interest rate for debt. The cost of farm-owned input is calculated as a residual using the zero-profit condition. Farm-owned input is defined as the sum of market revenue and payments minus all other input costs.

There are many different budgetary payments in Norway. PEM covers the most important of these, as well as milk quota. Certain legal and regulatory constraints are also built into the model structure. Most importantly, the Soil Act requires all arable land, surface-cultivated agricultural land, and fenced pastures to be kept in food production. The aim of the Act is to produce food, maintain the soil’s production capacity and to keep up the agricultural landscape. Less than 1% of the utilised agricultural area is denoted “out of production” in the direct payment register (Norwegian Agriculture Agency, 2020[9]).

The standard procedure in PEM is to take payment information directly from the PSE-database. This is not adequate in the case of Norway. In addition to the regionalised nature of payments, there is also a farm structural component in the payment system. This means that payment rates are higher for the first animals than for later animals. In other words, per unit payment rates are negatively correlated with farm size. The rationale is to incentivise farmers not to fully exploit economies of scale. The payment rates in the Norwegian version of PEM are therefore based on a detailed calculation of the most important payments into six payment groups for all active farms in Norway (Mittenzwei, 2018[3]).

The payments within each of the six Norwegian payment groups are linked to single types of support in the PSE database (Table A A.4.). Output payments coincide with output payments in the PSE database. Income support to dairy farms is a scheme where only the first five dairy cows and the first 40 suckler cows of a farm receive support. This payment is categorised as a payment based on non-current animal number with production required, because virtually all dairy farms in Norway have more than five dairy cows. Acreage payments are split between payments based on current area and payments based on non-current area. The latter cultural landscape payment is provided with a uniform payment rate for all crops in all regions. Animal payments and welfare payments belong to the category of payments based on current animal numbers where production is required. Finally, “other payments” contain all payments that cannot be linked directly to the most prominent land uses or animal numbers. Investment support, organic payments, income tax deduction, and fuel concession are examples of those payments.

The payment amounts in Table A A.4. show the payments for Norwegian agriculture. As the PEM model includes only wheat, coarse grains, milk and beef, payment amounts need to be adjusted to account for that selection. The PEM model covers about 78% of the total payment amount or NOK 10 807 million. Income support for dairy farmers is slightly higher than reported in the PSE database; the reason may lie in additional payment regulations that are not covered in the calculations for the individual farms.

The four productions included in PEM account for about half of all output payments. They allocate also almost all acreage payments and two-thirds of the animal payments. Half of the welfare payments can be traced to milk and beef, while nearly all other payments are related to the four productions.

The regional profile of the payment system is clearly visible with lowest payment rates in Jæren and highest payment rates in northern Norway for most payment categories. Regional differences in payment rates are smaller for crop products (Mittenzwei, 2020[8]).

## References

[5] Eurostat (2013), “Methodology and Handbook Eurostat/OECD. Nutrient Budgets, EU-27, Norway, Switzerland”, No. 112, Eurostat, Luxembourg.

[7] Kristiansen, B. (2018), “Driftsgranskingar i jord- og skogbruk. Rekneskapsresultat 2017”, NIBIO Bok 5(10), https://nibio.brage.unit.no/nibio-xmlui/handle/11250/2580084 (accessed on 4 August 2020).

[8] Mittenzwei, K. (2020), PEM model for Norway. Background report for the Agriculture Review.

[3] Mittenzwei, K. (2018), Økonomisk modellering av klimatiltak i jordbruket: Dokumentasjon og anvendelser i CAPRI og Jordmod. Versjon 1.0 av 30.04.2018, https://nibio.brage.unit.no/nibio-xmlui/handle/11250/2496992 (accessed on 4 August 2020).

[9] Norwegian Agriculture Agency (2020), Om produksjonstilskudd - Landbruksdirektoratet, https://www.landbruksdirektoratet.no/no/produksjon-og-marked/produksjonstilskudd/om-produksjonstilskudd (accessed on 4 August 2020).

[2] OECD (2015), OECD Review of Agricultural Policies: Switzerland 2015, OECD Review of Agricultural Policies, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264168039-en.

[6] OECD (2013), OECD Compendium of Agri-environmental Indicators, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264186217-en.

[1] OECD (2005), The Six-Commodity PEM Model: Preliminary Results, OECD, Paris, https://one.oecd.org/#/document/AGR/CA/APM(2005)30/en?_k=pgr4y0 (accessed on 8 March 2018).

[4] SSB (2020), Population, by sex and one-year age groups, https://www.ssb.no/en/statbank/table/11727 (accessed on 4 August 2020).