Introduction

Since the 1950 round of censuses, Statistics Division of FAO has published International Comparison Studies at the end of each round except for the 1980 round. For the1950 round the following subjects were covered in detail, including graphic presentation: Number and area of holdings, land tenure and land use; all classified by size of holding. For the 1960 and 1970 rounds detailed final reports were prepared covering all items proposed by the respective FAO Programmes. For the 1990 round a report covering selected items, considered important, was prepared (see the Report [2] and the Supplement [1]).

Some further studies on the 1990 round of censuses are included in this report. Subject of these studies can be seen from the list of seven tables shown in the contents. In selecting these studies an attempt was made to cover some of the issues not covered before. These are:

1. Interpolation of country data, for countries, which did not follow the proposed FAO classifications. This refers to the classification of holders by age groups (Table 6) and the classification of holdings by number of parcels (Table 7). Data on holders classified by age groups are not planned for collection in 2000 round of censuses, so in a way, this was a last opportunity to make this kind of analysis. The interpolation was done using the graph paper.
2. Observation of trends for selected items (Tables 2, 3 and 4). This refers to the central and structural characteristics of holdings, and data on land tenure and cattle. The trends observed in these studies refer to countries for which the 1990 census round data were available and, in addition, the 1980 and/or 1970 round data were also available.
3. Presentation of data not analysed before. This refers to the classification of holders by sex (Table 5). Similarly, central and structural characteristics (Tables 1 and 2), which were calculated only once in the past for the 1970 round of censuses [3], were calculated this time for three census rounds.

Census rounds include all agricultural censuses conducted in a ten years period. Thus, the 1990 round include censuses conducted in the period 1986 – 1995. A few exceptions were made in order to increase the coverage. The following censuses were, therefore, included in the 1990 round: Brazil 85 (Tables 1 and 2), Bangladesh 96 (Table 5), and Chile 97 and Venezuela 97 (Table 7).

There are several problems in data comparability. The comparability between countries depends upon different concepts, definition and practices applied. The comparability of the successive censuses in the same country is generally better as each country tends to keep the same concepts, but there are cases when there are such changes in data, which cannot be explained with the information available in the FAO Statistics Division.

One of the problems with the comparability of successive censuses is in the fact that many countries do not take regularly the censuses, and even those that take them regularly, do not provide always comparable data. Number of countries in tables showing successive data is, therefore, relatively small (Table 2: 48 countries; Table 3: 19 countries; Table 4: 40 countries).

The main source of data for present studies are tables from the ‘Supplement to the Report on the 1990 World Census of Agriculture - International comparison and primary results by country (1986 – 1995)’ [1], ‘1970 World Census of Agriculture, Analysis and International Comparison of Results’ [4] and unpublished International Comparison Tables: 1980 World Census of Agriculture, available in FAO Statistics Division [5]. At times, also individual country results were used.

The subjects selected for this studies are, so called ‘structural items’, which are not expected to change considerably in time. Nevertheless, in the 20-years period under observation some major change can be noted in many countries, concerning the number and area of holdings, and number of head of cattle in holdings raising cattle, in particular.

Central and Structural Characteristics

This refers to estimates of the following characteristics: (i) average size of holding, (ii) median size for number of holdings, (iii) median size for the area of holdings, and (iv) Gini’s (called also Lorenz) index of concentration. These estimates are shown in Table 1, for 74 countries for which 1990 census data were available. In order to study related trends, estimates are also shown in Table 2, for 48 countries for which 1980 and/or 1970 census data were available, in addition to 1990 data.

Brief explanation of these characteristics is given below. Readers interested in more details and exact definitions are suggested to see the FAO publication: ‘Agricultural holdings in the 1970 World Census of Agriculture: a statistical analysis’ [3].

Average size is the ratio between the total area of holdings and total number of holdings.

Median size for the number of holdings is the size dividing the number of holdings into two equal parts: 50 percent of the total number of holdings is below this size and another 50 percent is above.

Median size for the area of holdings is the size dividing the area of all holdings into two equal parts: 50 percent of total area of holdings is in holdings below this size and another 50 percent is above.

Gini’s (or Lorenz) index of concentration is a measure of concentration of agricultural areas in this application. It may vary from 0, when all holdings have the same area, to 1, when all agricultural land in a country is in one holding while the size of other holdings is zero.

The estimates of medians and the Gini’s concentration index were calculated using data on number and area of holdings classified by FAO size groups (see: Supplement to the Report on the 1990 World Census of Agriculture - International comparison and primary results by country (1986 – 1995); Tables 4.1 and 4.2 [1]).

The estimation of medians from the grouped data was made using the interpolation method based on log-normal properties of the distribution of the number and area of holdings by size. This method was used successfully for the interpolation of size groups for countries which provide different classification then that proposed by FAO. The method is not reliable, however, for the first size class (below 1 Ha) and the last size class (above specified value). For this reason estimates of medians were not possible in a few cases. For example, median for the number of holdings was not calculated for Nepal (92 census), as the first class, below 1 Ha, included 70 percent of holdings. Similarly, median for the area of holdings was not calculated for Paraguay (91 census) as the last class, above 1000 Ha, included 80 percent of all area of holdings. Concerning estimates for medians only first two significant digits are reliable; the estimates were rounded accordingly.

The precise estimation of Gini’s index of concentration was difficult in cases when the small number of size groups was available. The main problem appeared in Africa and Asia in countries where the average size of holdings is small, sometimes below 1 Ha, while the FAO standard classification starts with the group below 1 Ha. In most of these cases more detailed classifications were applied from the original sources, whenever available. In case of India, Indonesia, Nepal and Philippines, which provided data for all three censuses detailed data were not found for all censuses. In these four cases classification starting with ‘below 1 Ha’ was applied. In this way the Gini’s index is a little overestimated in respect to other countries, but a more realistic observation of trends within countries is made possible.

An important factor, which has direct repercussion on estimation and interpretation of medians and Gini’s coefficient, is the criterion for classification by size applied by different countries. The FAO was recommending in all World Programmes classification of the number and total area of holdings by size of total area of holdings. This classification was not found suitable by some countries, which used classifications by agricultural land, cropland, cultivated land, arable land, etc. Furthermore, some of the countries not using total area as criterion for classification tabulated total area of holdings by size while others tabulated only agricultural or other area. Full information on country practices is given in Tables 4.1 and 4.2 in the Supplement [1], and may be consulted for proper understanding of values of medians and Gini’s coefficient. Differences between total, productive, agricultural, etc. area can be understood from the FAO Programme definitions of land use categories and land use data (see Frame 3.5.1 and Table 3.6 [1]).

The repercussions of factor mentioned above are not easy to explain and to understand. An attempt to explain repercussions is given below. In practice, not many serious comparability problems exist because of different classifications. Out of 74 countries shown in Table 1, 53 countries have classified the number and area of holdings by total area. In this number African countries were included as they have a special comparability problem in any case. Under the conditions of traditional land tenure in Africa, where ownership does not exist, generally total holding area is limited to the cultivated area. In most of other 22 countries using different classifications the problem is of limited importance as agricultural area and productive area are, generally, not much different from total area, the difference being often less then 10 percent. Criteria used in these 22 countries are shown in the table below. The only major problem exists in Finland as arable land used for classification in this country is only 20 percent of the total area of holdings. This is why the average size of holdings, based on total area, is not in line with the medians, based on arable land.

Countries using criteria different than total area for classification by size
(22 countries)
Countries Data classified by:
Area tabulated by size:

Reunion, Guadalupe, Martinique, French Guiana, Viet Nam, France, Germany, Greece, Ireland, Luxembourg, Netherlands, Norway, Portugal
Agricultural area
Agricultural area

Belgium, Denmark
Agricultural area
Total area

Austria
Productive area
Total area

Switzerland
Productive area
Productive area

Albania, Iran, Japan
Cultivated area
Cultivated area

Korea, Rep.of
Cultivated area
Agricultural area

Finland
Arable area
Total area

Estimates of medians and the Gini’s index of concentration for 74 countries, which provided data classified by size of holdings for the 1990 round of censuses are given in Table 1. All estimates were made for 58 countries, while in case of 16 countries data available were not sufficient for making some of the estimates. In two cases (Djibouti and Albania) data on area of holdings were not available. In other cases, either the lowest class included much more then 50 percent of all holdings, or the highest class included much more then 50 percent of all area of holdings. In three cases (St. Kitts & Nevis, Israel and French Polynesia) both factors happened.

The Gini’s index of concentration, in general, is lower in Africa and Asia and higher in South America then in other continents. It is often high in countries with high average size of holdings (e.g. USA, Argentina and Brazil). This index depends also upon the definition of holding. Thus in some countries, which included many holdings with little or no land, e.g., holdings with at least one beehive, or two head of pig, sheep or goats, … (Barbados, St. Vincent & Grenadines, …) the index is very high.

Data for 48 countries for which estimation of medians and the Gini’s index of concentration was possible for consecutive censuses, are given in Table 2. No major changes were observed in the Gini’s index in most of the countries. As expected, the increase of the Gini’s index of concentration was often accompanied by the increase of the median size for area and the decrease of the median size for number of holdings.

Some systematic changes in the index of concentration were observed in Asia, where this index decreased in India, Indonesia, Korea, Rep. of, and Nepal and increased in Japan, Pakistan, Philippines, Thailand and Turkey.

Major changes have been observed, however, in average area of holdings, which in the period of 20 years increased in developed countries and decreased in developing countries.

Land Tenure

Data on trends observed on number and area of holdings by tenure are available in Table 3, for19 countries. In this Table all holdings are classified as operated under one single form of tenure or under more than one form of tenure. There was a decrease in the number and area of holdings operated under more then one form of tenure in many countries, even those where the total number of holdings has increased considerably. These countries are Colombia, Peru, India and Pakistan. Exceptions were observed in Philippines, where the number of ‘mixed holdings’ and their area have increased. Number of holdings rented from others has decreased considerably in 9 out of 19 countries providing data. Decrease of over 50 percent was observed in American Samoa (96 percent), India (84 percent), Italy (61 percent), and Peru (57 percent).

Cattle

Data on number of holdings reporting cattle and number of head of cattle are shown in Table 4 for the 1990, 1980 and 1970 rounds of censuses. Such data were available for 40 countries.

A process of specialisation of cattle raising holdings can be observed, although the total number of cattle did not change much in most of the countries. Thus, the percentage of holdings raising cattle has decreased in 27 out 40 countries. In some developed countries like Canada and many European countries this decrease was considerable. Consequently, in most of these countries the average number of cattle in cattle raising holdings has increased. This increase was more then double, from 1970 to 1990 round of censuses, in 12 countries: Reunion, Israel, Belgium, Denmark, France, Germany, Greece, Italy, Netherlands, Norway, American Samoa and Australia.

The average number of head of cattle in cattle raising holdings, in 40 reporting countries, was ranging from 2.3 in Reunion to 143.3 in Australia in 1970 round of censuses. In 1990 round of censuses it was ranging from 2.2 in Philippines to 308.8 in Australia.

Holders by Sex and Age

The FAO definition of holder is:

The holder is a civil or a juridical person who exercises management control over the agricultural holding operation and takes major decisions regarding resource use. The holder has technical and economic responsibility for the holding and may undertake all responsibilities directly, or delegate responsibilities related to day-to-day work management to a hired manager.

Data on classification of holders by sex were available for 41 countries in the 1990 round of censuses and are shown in Table 5. In four countries less then 5 percent of female holders were reported (Guinea - 2.0 percent, Western Samoa - 2.4 percent, Bangladesh - 3.5 percent and Morocco - 4.4 percent). In seven countries more than 30 percent of female holders were reported (Namibia - 57.1 percent, French Guiana - 40.1 percent, Cape Verde – 36.2 percent, Botswana – 36,2 percent, Grenada – 33.8 percent, Sao Tome & Principe – 33.6 percent and Bahamas – 31.0 percent. Apparently, females are more frequently holders in countries where males tend to go abroad for work.

Data on holders classified by age groups were available for 40 countries in the 1990 round of censuses and are shown in Table 6. Data available for 18 counties refer to different age groups then those recommended by FAO and interpolation was required. The interpolation was made using the graph paper. Interpolated data are shown in italics in Table 6. The average age of holders, at the time of census enumeration, was also estimated and is shown in Table 6. In 23 out of 40 countries providing the information on age of holder, the average age of holders is between 47 and 54 years of age. In 12 countries the average age of holders was below 47 years (42 years: French Guiana; 44 years: Ethiopia; 45 years: Congo, Dem. Rep. of, Reunion, Nepal and Northern Mariana Is. and 46 years: Tanzania, Uganda, Zambia, St. Vincent & Grenadines, Fiji and Western Samoa). In five countries the average age was above 54 years (55 years: Barbados: 56 years: Martinique and Puerto Rico; 57 years Virgin Islands (US) and 64 years: Guinea). It may be noted that French Guiana, the country with lowest average age of holder: 42 years, has a high percentage of female holders: 40.1 percent.

Considering definition of holder, given above, biases might have been introduced by enumerators in some countries, favouring males and old people when deciding who is the holder in an agricultural holding.

Fragmentation

The agricultural holdings are fragmented into one or more parcels. The FAO definition of parcel is:

A holding parcel is any piece of land entirely surrounded by other land, water, road, forest, etc. not forming part of this holding. A parcel may consist of one or more fields adjacent to each other.

Data on number of holdings classified by number of parcels were available for 25 countries in the 1990 round of censuses and are shown in Table 7. Data for 3 counties refer to different groups then those recommended by FAO and interpolation was required. The interpolation was made using the graph paper. Interpolated data are shown in italics in Table 7. The total number of parcels was not available for 7 countries and was estimated. The estimated totals are also shown in italics.

The average number of parcels per holding vary from 1.1 parcels per holding in Namibia and Barbados, and 1.2 parcels in Venezuela, to over 5 parcels in Turkey (5.4), Portugal (5.7), Spain (8.1) and Niue (8.7).

Bibliography

1. Supplement to the Report on the 1990 World Census of Agriculture - International comparison and primary results by country (1986 – 1995); Statistical Development Series 9a, Rome, 2002.
2. Report on the 1990 World Census of Agriculture - International comparison and primary results by country (1986 – 1995); Statistical Development Series 9, Rome, 1997.

Summary:
Report on the 1990 World Census of Agriculture

Agricultural Census Database:
http://www.fao.org/es/ess/census/

3. Agricultural holdings in the 1970 World Census of Agriculture: a statistical analysis, FAO Economic and Social Development Paper, No. 47, Rome, 1984.
4. 1970 World Census of Agriculture, Analysis and International Comparison of Results, Rome, 1981.
5. Unpublished International Comparison Tables: 1980 World Census of Agriculture, available in FAO Statistics Division.