The land price database (the PERVAL database) that we used was obtained from notaries and consists of all transactions of agricultural land that
84|LATRUFFE,PIET,DUPRAZ &LE MOUËL
occurred in the regions over the period studied. We considered only arable land and pasture which was non-built and already tenanted by a farmer or not. Because the smallest plots exchanged were sold at very high prices, reflecting the fact that future conversion to development use is anticipated for such plots, we restricted the database to plots with an area equal to or above ten hectares. During the period studied (1994–2011), 2,772 such transactions occurred in NUTS2 Brittany, 774 in NUTS2 Limousin and 739 in NUTS3 Meuse. Taking all regions together, plots sold of 10 hectares or above were, on average, 19.9 hectares in size and priced at €2,795 per hectare.
The occupations of both the seller and the buyer are some of the transaction characteristics which are available in the land sales database.
Two thirds of the plots are bought by farmers. In France, specific private bodies have the public mission of regulating the transactions in order to limit price speculation, avoid farm fragmentation and promote the settlement of young farmers. Each transaction is notified to these bodies, called the SAFER (Sociétés d’Aménagement Foncier et d’Etablissement Rural), which operate at the NUTS3 level. If the SAFER believes that a transaction is a threat to farm consolidation or settlement, or may be governed by price speculation, it can stop the transaction. It then tries to convince the seller and the buyer to change the transaction on an amicable basis and, if this is not possible, it pre-empts the plot and has five years to sell it back at a lower price or to another buyer. In the PERVAL database, the SAFER intervenes (by buying or re-selling a plot) in 16% of transactions.
The municipality in which the plot is located is also available in the PERVAL database, enabling each transaction to be related to agricultural subsidies and revenue as well as to other variables such as the municipality’s demographic characteristics and the zones it may come under.
4. Methodology
Because data regarding agricultural revenue and subsidies are not directly available from public statistics at the municipality level, they were estimated in a first stage of the analysis. The second stage of the analysis consists in regressing the transaction price on these proxies and the other variables mentioned above.
The dependent variable used for the second-stage estimation is the deflated price per hectare of agricultural land sold in plots with an area of ten hectares or more. The explanatory variables which were expected a
INFLUENCE OF SUBSIDIES AND REGULATIONS ON SALE PRICES OF FRENCH FARMLAND |85 priori to influence the land price are, first, the basic determinants of land price based on the present value model: on the one hand, revenue from agricultural use, which is separated into a market-based component (M) and a government-based component (G), and on the other hand, potential revenue from non-agricultural use.
An approximation of the agricultural revenue (M) and agricultural subsidies (G) at the municipality level was obtained through a first-stage regression. The revenue variable is the pre-tax profit from which we excluded subsidies to avoid double counting. Six types of subsidies could be considered, namely total agricultural subsidies and five different components: CAP first-pillar coupled direct payments to crops and herds;
CAP first-pillar land set-aside premiums; CAP first-pillar decoupled single farm payments (SFPs); CAP second-pillar less-favoured area (LFA) payments; and CAP second-pillar agri-environmental payments to extensive grazing livestock. The deflated revenue and subsidies were regressed on crop areas and herd numbers (observed at NUTS3 level) as a system of stacked equations using the seemingly unrelated regression (SUR) estimator. Then, the resulting estimated coefficients were used to generate projections at the municipality level from crop areas and herd numbers observed in the agricultural censuses. To account for the size of the municipality, the revenue and subsidies projections were divided by the municipalities’ UAA.
Potential revenue from non-agricultural use was not observed. For this reason, following the literature, we proxied it by two variables: the population density in the municipality where the plot is located, and a dummy indicating whether or not the municipality is part of an urban area.
In addition to these basic determinants suggested by the present value model, we controlled for the size of the plot sold, whether the buyer was a farmer, and the municipality’s area. We also included year dummies and NUTS3 region dummies. Finally, we considered regulations that may affect the price of agricultural land. The first regulation variable related to zoning based on the Nitrate Directive; the zoning dummy variable took the value 1 if the municipality was in the nitrate surplus zone, and the value 0 otherwise. The second regulation variable took the value 1 if the seller or buyer was a local SAFER, and 0 otherwise.
We performed regressions on a sample consisting of all three regions together, and on the samples of each region separately. In addition, for all four samples, we performed one regression including the total subsidy
86|LATRUFFE,PIET,DUPRAZ &LE MOUËL
variable and one regression including the five different types of subsidies instead.
5. Findings
First, we found that agricultural revenue generally has no significant influence on land price, contrary to what can be expected from the present value model. One reason may be that the original variable (i.e. pre-tax profit) used to construct our proxy variable, which is the only one that was available from the statistics, may not be the best representation of income generated by farming activities on land because it is too low in the accounting balance sheet. The gross margin would be a better candidate but was not available in the original database. Another reason may be that the revenue variable was proxied at the municipality level and not at the level of the plot itself.
Second, we found evidence that agricultural subsidies actually capitalised at least to some extent in the price of land in the regions studied over 1994-2011. However, the magnitude of such a capitalisation depends on several factors. One varying factor is the region; for the sample including the three regions together, we found a positive but relatively small capitalisation effect of the total subsidies per hectare. However, this effect is differentiated according to the region in question. In NUTS2 Brittany, the positive effect is significant only for plots located in the nitrate surplus area and is greater than in both other regions. As for these two other regions, the effect is greater in NUTS2 Limousin than in NUTS3 Meuse. Another varying factor is the type of subsidy. When considering all regions together, we found that only land set-aside premiums significantly capitalise into the price of land, whether the plot is located in a surplus zone or not. The capitalisation effect is high, suggesting a scarcity effect due to the requirement to withdraw land from production. In addition, SFP has a significant positive capitalisation impact only for plots located in a surplus zone.
Third, we found a significant influence of regulations on land price.
Regarding land transaction regulations, plots purchased or re-sold by SAFER were found to be significantly more expensive. This finding is counterintuitive, as SAFER are expected to contribute to alleviate speculation on land prices. One reason may be that SAFER do not always use their pre-emptive right with a view to keeping land price low; they may also pre-empt land that is up for sale to change the buyer, to limit farm fragmentation, or to support the settlement of young farmers. Another
INFLUENCE OF SUBSIDIES AND REGULATIONS ON SALE PRICES OF FRENCH FARMLAND |87 reason may be that SAFER’s intervention with a view to keeping the price low may occur for specific land, which is more expensive than the average agricultural land. The interactions with the SAFER variable and the subsidy variables were not significant. By contrast, we found some significant effect of subsidy variables interacting with the zoning regulation variable: in NUTS2 Brittany, where the nitrate surplus zoning is implemented, the capitalisation of subsidies is significant for plots located inside the zone but not for plots located outside the zone, revealing a restriction on land mobility in the surplus areas. This suggests that public intervention in the form of nitrate zoning regulations may affect land mobility in favour of a specific use of land and may increase the degree of capitalisation of subsidies in agricultural land price, possibly an unintended consequence as it goes against the government objective of supporting farmers’ income.
References
Cavailhès, J. and S. Degoud (1995), “L’évaluation du prix des terres en France: une application de la réforme de la PAC”, Cahiers d’Economie et Sociologie Rurales, 36:49–77.
Goodwin, B. and F. Ortalo-Magné (1992), “The capitalization of wheat subsidies into agricultural land values”, Canadian Journal of Agricultural Economics, 40(1): 37–54.
Latruffe, L. and C. Le Mouël (2009), “Capitalisation of government support in agricultural land prices: what do we know?”, Journal of Economic Surveys, 23:659–691.
Latruffe, L., Y. Desjeux, H. Guyomard, C. Le Mouël and L. Piet (2008), Study on the Functioning of Land Markets in the EU Member States under the Influence of Measures Applied under the Common Agricultural Policy – Report for France, Brussels: Centre for European Policy Studies.
Latruffe, L., L. Piet, P. Dupraz and C. Le Mouël (2013), “The Influence of Agricultural Support on Sale Prices of French Farmland: A comparison of different subsidies, accounting for the role of environmental and land regulations”, Factor Markets Working Paper No. 51, Centre for European Policy Studies, Brussels.
Patton, M., P. Kostov, S. McErlean and J. Moss (2008), “Assessing the influence of direct payments on the rental value of agricultural land”, Food Policy, 33:397–
405.
88 |