# Introduction

Except for studies conducted at the beginning of the 1850s by the Federal Department of the Interior, the first four national livestock counts, and individual counts by cantons, it can be said that the birth of Swiss agricultural statistics coincided with the formation of the Swiss Farmers Office (“Schweizerisches Bauernsekretariat” / “Secrétariat des paysans suisses”) in the year 1897. Under the dynamic guidance of Ernst Laur the Farmers Office immediately became very active in the area of statistics. The first attempt at determining a valuation of the agricultural sector already occurred in 1899; since 1922, the Farmers Office has published annual data subdivided by production area in its “Statistical Surveys and Estimates” (“Statistische Erhebungen und Schätzungen” / “Statistiques et évaluations”). While the early volumes of this journal were relatively small, those of the 1960s, 70s, and 80s contain a wealth of tables, only some of which could be integrated in this volume. While selecting data for inclusion, we generally favored official statistics, as long as they covered a period longer than three decades, and estimates performed by the Research Institute for Swiss Social and Economic History of the University of Zurich (“Forschungsstelle für schweizerische Sozial- und Wirtschaftsgeschichte der Universität Zürich”). Further, we attempted to avoid excessive overlaps with tables in Hans Brugger’s “Swiss Agricultural Statistical Handbook” (“Statistisches Handbuch der schweizerischen Landwirtschaft” / “Manuel statistique de l’agriculture suisse”) published in the second half of the 1960s. Consequently, we refrained, for example, from reporting on the development of agricultural insurance, the spread of animal diseases, or the yields of grain, potato, fruit, and wine harvests in the late 19th and early 20th centuries by canton. With the intent to offer an appropriate complement to Brugger’s long term series, we instead performed a partial processing of national livestock, cultivation, operation, and agricultural surveys at the canton level.

### The Estimates 1837–1945

The table section of this chapter starts with estimates arrived at within the framework of the National Fund project “Money supply and economic growth in Switzerland 1851–1913” (“Geldmenge und Wirtschaftswachstum in der Schweiz 1851–1913”): i. e. data for agricultural area, tree and livestock levels, physical production and its uses, and the valuation of the agricultural sector and its components.

While the agricultural production in Switzerland was estimated several times in the 19th century, paradoxically, true agricultural statistics did not appear until the first sector had already lost a large part of its earlier economic relevance. Certain production branches proved almost completely elusive to statistical valuation even at the start of this century, a fact that greatly increased the difficulty of calculating the creation of value added in the agricultural sector. We had no choice but to either extrapolate canton or local survey figures, or else to create an indirect estimation procedure. In the following sections we provide a rough outline of the methodology used to determine the contribution of the major agricultural “branches” to the value added in the first sector in the period of 1837–1945.

### Vegetable Production

We determined the value added of vegetable production by first multiplying areas by their yield per hectare and trees by yield per tree, respectively, and then subtracting feed and seed amounts from the result. The next step then consisted of multiplying the net yield by the prices. This task was relatively simple, as price developments are mostly well documented, with atypical curves in only a few exceptions. One aspect to be considered, however, was that part of the harvest yield was not usually realized until spring or early summer of the following year. Until approximately the turn of the century, we therefore used monthly price statistics in order to match prices to the harvest year.

There are few area statistics from the 19th century. For most grains, we were therefore forced to project national levels from value trends of individual cantons. Using, among others, the estimates from Werner Schlegel, we first attempted to determine the extent of the Swiss grain area reduction in the period of 1837–1911, then determined, with the help of canton statistics and data from contemporary literature for selected years, the proportion of individual grain types in the total national grain area.

The situation with production statistics for the 19th century is even worse than that for area statistics. Not a single canton had even reasonably reliable annual yield statistics before 1885. There were local recordings of variations in grain and potato harvests, but those figures, which are not even representative for the harvests in surrounding communities, can certainly not be used to project national harvest curves.

In order to perform an estimate of agricultural sector value added despite this scarce statistical basis, we took a somewhat adventurous path: Knowing that high quality meteorological statistics, consisting of monthly values, exist since the late 1830s, and assuming that annual fluctuations in grain and potato harvests must have been influenced by temperature and precipitation levels at certain times of the year, we designed a linear regression model with multiple dependent variables that “explains” the harvest cycles using climactic variables. The coefficients of the individual equations were arrived at by regressing the grain and potato per-hectare yields for the canton of Berne against the monthly temperature and precipitation series of the Berne Meteorological Station. From the monthly values of Berne’s meteorological series, we built aggregate index variables whose composition we then varied until we arrived at an optimal approximation of the development of the corresponding harvest variable. In this way, we developed six different sets of variables, most consisting of four or five index series. In order to extend those indices backwards, we used the recordings of the meteorological stations in Berne, Basle, Geneva, and Zurich from earlier decades. By applying the coefficients calculated for the core period to earlier decades, we arrived at harvest fluctuation estimates in the canton of Berne for the years of 1837–1884.

The computed values of specific parameters of this regression model reveal that it imitates the actual development of per-hectare-yields in Berne with variable accuracy. Most successful was the retroactive prognosis of harvest fluctuation for winter grain (“Dinkel”) and for potatoes, and it was satisfactory for winter wheat as well. Winter rye and oat harvests could not be determined very accurately, but the “postdiction” was not a complete failure either. Barley estimates, however, were marginal if not inadequate.

We slightly corrected certain values of the estimate series in retrospect based on qualitative data from contemporary literature. Assuming that the development of the harvest curves for all of Switzerland closely correlated with the fluctuation of the canton of Berne per-hectare yields, we used the corrected estimate series to link together the national trend values which we had previously determined with the help of specialized literature. Once the development of perhectare yields for the major grains and for potatoes had been constructed, we only had to multiply the national yield series with the national area series in order to arrive at long term series of Swiss grain and potato production in the 19th and early 20th centuries.

By the second third of the 19th century, Switzerland had already comparatively well-developed canton viniculture statistics. Therefore, the processing of a large number of canton and local yield series into a national development indicator did not present any problem. Out of curiosity, and in order to be absolutely sure, we tested a linear regression model fed by weather and yield statistics for this “branch” of the first sector as well. Surprisingly, this model produced such convincing prognostical values, even over a period of over one hundred years (1814–1924), that the estimate series was included in the determination of national per-hectare yields.

In contrast, the regression model did not deliver good results for fruit. Since there are no canton, and only very few local, fruit yield series before 1885, we initially felt that an estimate of Swiss fruit production for the period 1837–1884 was impossible. The fact that, contrary to grain, there is a close correlation between price and production quantities for pomaceous – and to a lesser extent also stone – fruit, even in the early 20th century, gave us the idea to replace the atmospheric variables developed for the reconstruction of grain and potato crops with price series. The new model assumed a linear relationship between fluctuations in fruit yields and fruit prices in the canton of Berne in the period of 1885–1914. (The years 1915–1927 had to be excluded from the regression analysis because fruit prices jumped during World War I independent of fruit harvests). We were relatively satisfied with the results from the regression of harvest against price series. While the estimate series for cherries, prunes, and nuts could only be termed “poor”, the calculation for apples and pears, the most important fruits based on geographical distribution, yielded prognostic values every bit as good as those for wheat and wine.

Analogous to the methodology employed for determining grain and potato yields, we performed minor corrections on the estimate series by using reports on fruit harvests from qualitative sources.

### Livestock

For the period 1851–1913, Thomas Steiger estimated the most important components of livestock production. We used those estimates – covering cattle counts by breed, consumer and milk production, beef and veal production, milk, beef and veal prices, and the value added for milk, beef, and veal production – in virtually unchanged form. Since Steiger’s price series do not show absolute values, and since value added also is presented in indices rather than absolute numbers, there was a need to determine price levels for specific years and products. Additional assumptions and considerations were required in order to determine the value added for milk, beef, and veal production in the years 1837–1850 and 1914–1945.

Similar to what Steiger had done with his beef and veal production estimates, we estimated the value added for pork production by computing headcount variations, slaughter coefficients, reproduction rate, and export movements. For sheep, goat, horse, and bee farming, on the other hand, we were limited to rough production estimates.

### Uses

For the 19th and early 20th centuries, we performed rough estimates of the percentage of seed and feed for grain and potato (including shrinkage and losses) based on data gathered from specialized literature. Systematic surveys on agricultural production uses did not start until after World War I: for milk and fruit in the beginning of the 1920s; for potatoes ten years later; and for grains not until the year 1941. Our estimates, as well as the Swiss Farmers Office’s numbers, about the uses of grain and potato crops, should be interpreted with caution, as even today there are no exact figures for seed usage per hectare or for the percentage of “shrinkage and losses”.

We managed to estimate butter production back to 1886, and cheese production to 1851 by using, on the one hand, data published on national butter and cheese production quantities in the livestock counts of 1866, 1886, and 1911, and, on the other hand, canton statistics on the portion of milk processed in cheese factories and the weight of the cheese produced from that milk. The chapter “Industry and Trades” reports on quantitative production, gross production values and value added of these and other food products.

### Forestry

Beginning in the 19th century, information on gross proceeds per hectare of forest was included in the governmental reports of a significant number of cantons. We used those series to determine canton indices which we then projected to the national level. This resulted in an estimate series depicting the progression of the gross forestry production value for the period of 1841–1918. Next, we determined absolute values for marker years from reports of the Federal Department of the Interior on the status of forestry activities at the middle of the century; from a canton area, production, and price statistic developed for the Swiss National Exposition of 1883; and from official tallies from later decades. In a third step, we connected the absolute values with the progression indicator arrived at from the extrapolation of the canton indices, and made some additional assumptions in order to be able to determine lumber production as a percentage of overall wood production.

### Hunting and Fishing

We could only estimate the annual proceeds of these two first sector production branches in a very rough manner. We therefore have to admit that the valuation series of the first sector shown in the table portion of this chapter represents, to a certain degree, a projection.

### “Final Raw Proceeds”/Value added

Since 1923, the “Statistical Surveys and Estimates” carry a so-called final raw proceeds. The Swiss Farmers Office uses this term to describe the total value of agricultural production; the latter being defined as the “monetary value of all agricultural products leaving farming operations either through the sale to non-farmers or through the use in farming households themselves, calculated according to prices achieved or achievable through sales at farms”. (Agricultural Yearbook 1969 and 1984: “The final raw proceed of agricultural production by canton”). Since the Farmers Office did not include “those products used by the original producer, or others, for ongoing agricultural production, such as seed, rough feed, feed grain, feed potatoes, feeding milk, litter, eggs for hatching, utility animals, etc.” into the final raw proceeds, we assume that it can be equated to the overall value added of the agricultural sector.

Within the framework of his “Commission for the Preparation of Future Economic Contracts” (“Enquête zur Vorbereitung der künftigen Handelsverträge”), Ernst Laur tried to estimate the final raw proceeds, or value added, of agricultural production and its main components for the years 1885 and 1895. When comparing those numbers with our own estimates, we found substantial deviations in certain cases. This can be explained by Laur’s use of different sources for the determination of absolute price levels and intermediate consumption portions (feed and seed amounts), and also partly by his use of different assumptions about the relationship between individual agricultural production parameters (e. g. slaughter coefficients).

To simplify, we defined overall forestry value added as the sum of the gross proceeds from wood production and ancillary forestry uses. This means that our estimates consider the value of possible intermediate consumption to be negligible.

### Results of Bookkeeping Operations 1901–1991

The Farmers Office began an annual Statistics of Bookkeeping Operations already at the turn of the century. Until 1977, this abundant body of data was printed in the Agricultural Yearbook; for the period of 1978–1991 we needed to consult with the “Statistical Surveys and Estimates” and the annual reports of the Federal Research Institute for Agricultural Economy and Operations in Tänikon prepared by Karl Hostettler. Apart from the main results of the bookkeeping operations, we found especially interesting the reporting of active capital gross profit by size of operation and agricultural uses, and by zones and production structure. In order not to make the table section of this chapter too voluminous, we provide this profit index in the form of averages, rather than as long-term time series.

SOURCE: «Agriculture and Forestry» in Ritzmann/Siegenthaler, Historical Statistics of Switzerland, Zürich: Chronos, 1996, 519-526