Belgium

The described methodology was used until 2002. Then Belgium reduced the pig surveys from four to two and furthermore administrative data for the production forecasts were used. The characteristics of the database that contains administrative data (Sanitel) are also explained in this study.

Belgium carries out four pig surveys a year, in April, May, August and November. The April and August surveys are concentrated on pigs while the surveys in May and November cover various aspects of agricultural activity.

The surveys in April, August and November are samples while the one in May is a full survey. With the surveys in April and August 23 % of the total pig population are covered. The survey in November covers 44 %. In November the average sampling error for the total pig population is 0.6% (confidence interval 68 %). The sampling errors for April and August are not known.

Within the April and August surveys one selects 60 communes. All enterprises which are located in these communes are part of the survey. A rotation system is installed; one commune can only be selected once every two years. To determine the sample size the communities are selected in both regions in accordance with the regional subdivision. The sample is then distributed among the four half-years. The regional subdivision will initially be made in proportion to the pig population recorded (agricultural census in May). This breaks down into 4.04% for the Walloon region and 95.96% for the Flemish region. Since the concentration of breeding farms is lower in the Walloon region, the proportion of communities to be recorded must be higher than the proportion of pigs.

The sampling selection for the November survey is based on the data of the previous year's agricultural census in May. For the construction of the sample basis, types of holding are excluded from the sampling frame which do not have arable crops or livestock and horticulturists (farming types 2, 3 and 6) which did not record crops or livestock in the November survey.

A stratified random sample is taken. The following criteria are applied: language region, economic size and techno-economic orientation of the enterprise. The sample design covers 24 strata.

The sample holdings were assigned to the different strata using the Neyman method. This method optimises the estimate of the overall value of a specific variable. It is based on the spread of these variables in the various strata. Since several variables must be estimated for this survey, an optimum result cannot be achieved for a specific variable, but it can be achieved for general criteria such as the standard gross margin.

In April and August, about 2 500 holdings (4% of the total number of holdings) are surveyed. In November, the survey is carried out at 15 000 holdings (25%).

In November the survey takes the form of a postal enquiry. In April and August the samples are carried out by officials of the local administrations. The survey in May is done in the same way. The respondents are committed by law to respond and therefore the response rate is 100 %. The surveys are carried out by the Institut National de Statistique (National Statistics Office).

In the medium term (up to 2005), Belgium plans to partially replace statistical surveys on the pig population by the use of administrative data. Sanitel and Animo are available as a data source.

In the future (beginning in 2002) Belgium will carry out only two surveys a year, at six-month intervals in the months of May/June and November/December. Belgium will use administrative information from the Sanitel system to compile its gross indigenous production forecasts.

The basic conditions determining the use of the Sanitel administrative database in compiling gross indigenous production forecasts are:

A cooperation group made up of officials from the National Statistical Institute (Institut National de Statistique) at the Belgian Ministry of Economic Affairs, from the Centre for Agricultural Economics (Centre d'Economie Agricole) at the Ministry of Small Enterprises, Traders and Agriculture, or from corresponding bodies set up under the regionalisation programme, and from the European Commission shall supervise the use of the Sanitel administrative database in compiling gross indigenous production forecasts.

In particular this group shall ensure that the procedure for updating the Sanitel register will continue to guarantee adequate coverage and representativeness compared with the results of the pig livestock surveys. This group shall always undertake a detailed examination each time there is a significant change in the Sanitel administrative database.

Before 31 December 2004, Belgium shall transmit to the European Commission a report on the experience acquired through use of the Sanitel administrative database in compiling gross indigenous production forecasts.

SANITEL is a computerised system intended to manage automatically the permanent inventory of animals by taking account of the entries into, and departures from, each herd. It also provides information on:

health status (eg. contagious diseases, brucellosis);

the status concerning residues (eg. detection of antibiotics or hormones on slaughter);

the status concerning contaminants.

Note: the difference between the latter two points lies in the responsibility of the keeper. In the case of residues, the substances are in the animal following an action by the keeper. This is not the case with contaminants.

This third status has had to be added following the different crises which the farming world has suffered.

This system shows the situation as regards the inventory of herds of the different species (currently cattle and pigs) and traces the movements of animals. The objective of SANITEL is above all the health and epidemiological monitoring of the animals concerned.

SANITEL is a national database but one which is decentralised in regional posts (called federations) corresponding to the following provinces:

Lier (Province of Antwerp);

Bertem, including Brussels (Province of Flemish Brabant);

Braine (Province of Walloon Brabant);

Torhout (Province of Western Flanders);

Drongen (Province of Eastern Flanders);

Mons (Province of Hainaut);

Loncin (Province of Liège);

Rocherat (Eastern Cantons);

Alken (Province of Limburg);

Marloie (Province of Luxembourg);

Ciney (Province of Namur).

However, the management and uniformity of the system are handled in a centralised way by the Central Association for Animal Health (ACSA) under the supervision of the public health administration.

SANITEL can constitute a tool with a view to a reduction in the number of surveys. However, this database is not designed either for statistical processing or for counting on a specific date. In fact, pig monitoring is not individual but on a herd basis.

Every keeper of pigs is required to complete a health certificate showing the capacity of his holding. Subsequently, every three or four months approximately, he has a visit from an approved veterinarian so that he can declare the type and number of animals actually present. The exact number of pigs is only known when the visit reports are made or where there is movement! The veterinarian is required to send his reports to the Federation every week (or every day in a crisis period).

Since all the veterinarians do not visit all holdings at the same time, a method allowing the number of pigs by category to be determined precisely at a given time has been developed.

Moreover, the pig data are encoded by optical reading at the time visit reports are made. If, unfortunately, the veterinarian introduces an erroneous datum into his encoding sheet, the system will not detect it and it will be difficult to correct it later. With a view to detecting and correcting these errors, a second procedure has been developed.

breeding boars;

weaned piglets;

rearing pigs;

fattening pigs;

sows.

Sows are defined as being covered sows and other sows which are not covered. This category does not include young sows which have not been covered which are included in the category of rearing pigs.

Surveys on the number of cattle

Belgium carries out two cattle surveys a year, in May and November. With the surveys multiple objectives are established.

The May surveys is a full survey; the one of November is a sample survey. The average sampling error for the total bovine population is 0.3% (confidence interval 68 %).

A stratified random sample is taken. The following criteria are applied: language region, economic size and techno-economic orientation of the enterprise. The sample design covers 24 strata.

The survey is carried out at 15 000 agricultural holdings, i.e. 25%. It records 33% of the total bovine population.

In November the survey takes the form of a postal enquiry. In May the samples are carried out by officials of the local administrations.. The respondents are committed by law to respond and therefore the response rate is 100 %. The surveys are carried out by the Institut National de Statistique (National Statistics Office).

In the medium term (up to 2005), Belgium has no plans to use administrative data as a substitute for statistical surveys on the bovine population.

Surveys on the number of sheep and goats

Belgium carries out one survey of the sheep and goat population a year - in May. The surveys are carried out jointly in May each year (integrated surveys).

In May the samples are carried out by officials of the local administrations. The respondents are committed by law to respond and therefore the response rate is 100 %. The surveys are carried out by the Institut National de Statistique (National Statistics Office).

In the medium term (up to 2005), Belgium has no plans to use administrative data as a partial substitute for statistical surveys on the sheep and goat population.

Statistics on slaughterings

Belgium compiles monthly statistics on the number and slaughter weight of animals slaughtered in abattoirs for human consumption. These cover pigs (total), calves, heifers, cows, bulls, steers, sheep (total), lambs and goats (total).

All abattoirs report their slaughterings to the Institut National de Statistique every month. The results are available two months after the reference month.

Data on external trade in live animals

	Number	Slaughter Weight
Pigs, total	X	X
Calves	X	X
Heifers	X	X
Cows	X	X
Bulls	X	X
Steers	X	X
Sheep, total	X	X
Lambs	X	X
Goats, total	X	X

Since 1 January 1999 data on external trade no longer refer to the Belgo-Luxembourg Economic Union (BLEU), but to Belgium.

The National Bank of Belgium (BNB) has collected data on external trade since 1 January 1995. For live animals information is provided on number, weight and value (€).

Before the creation of the internal market and the abolition of tax and customs formalities at the internal borders of the European Union on 1 January 1993, data on external trade were collected on the basis of customs documents. For extra-EU trade this continues to be the case. For intra-EU trade a new survey method called Intrastat has been introduced.

An important feature of this new method is that enterprises send data directly to the BNB. Up to the end of 1997, Belgian enterprises were not obliged to report their figures before the month in which the annual dispatches or arrivals exceeded BEF 4 200 000 (104 115 €). In 1998 the threshold was raised to BEF 10 000 000 (247 894 €) (only trade flows exceeding this threshold must be reported).

The Intrastat declaration must arrive at the BNB no later than the 20th working day of the month after the reference month. The data are usually available after two and a half to three months.

To calculate estimated production volumes, the data on external trade in live animals are converted into carcass equivalents using special conversion coefficients.

Production forecasts (gross indigenous production)

Gross indigenous production is calculated according to the official EU definition, i.e. total slaughterings plus the export of live animals, minus the import of live animals.

To make its forecasts, the BEE used a demographic model based on the results of the surveys and censuses carried out regularly during the year. A breakdown of the various types of animal by weight (pigs) or age (cattle) and gender was used, and reproduction potential and productivity were also taken into account.

The model was validated by taking into account movements in appropriations (chronological series), previous observations and market conditions (prospects).

These initial estimates were presented to experts for their opinion and confirmation.

Beginning in 2002, census data will be replaced by administrative data, those from SANITEL. Another model for calculating the GDP forecast which is simpler and more rapid to use has been devised. This links the number of sows to the GDP in such a way as to establish GDP forecasts directly and uniquely on the basis of this population. The main question which arises is the quantification of this link, in other words the determination of the coefficient "ct" in the following equation

SANt-7 = number of sows in production in month t-7, a number calculated on the basis of SANITEL data.

The 7-month deviation corresponds to the normal period between farrowing and slaughter (in other words: 70 days-weaning- + 140 days -fattening-). The model has been tested with other periods (11 months, in other words by adding the gestation period; and 6 months), but it is with t-7 that the strongest ratio is obtained.

Poultry statistics

The overall poultry population of the individual farmers is known from the annual agricultural and horticultural surveys in May and December and the surveys on the structure of the agricultural holdings.

Prodcom industrial statistics provide businesses with less than ten employees or with an annual turnover of BEF 100 000 (2 479 €) monthly data on their production in terms of monetary units (Belgian Francs) and weight units (CPA code 15.12). In 1997 these results covered approximately 93% of the sector's turnover.

Poultrymeat consumption is monitored via household budget surveys.

In accordance with a European Union Regulation, the department of the Ministry for Agriculture which is responsible for rearing provides the Commission with data on hatching eggs and day-old chicks. The "Centre d'Economie Agricole" evaluates these data to produce meat balances (using technical coefficients).