Adopting sustainable irrigation technologies in Italy: a study on the determinants of inter- and intra-farm diffusion

ABSTRACT

This paper analyses the drivers for adopting irrigation systems with water conservation and saving technologies (WCSTs) by Italian farmers. The agricultural sector in Italy, like in other Mediterranean countries, suffers from water scarcity and water endowment variability. Water resources play a decisive role in agricultural production and in implementing large-scale WCSTs capable of improving the resilience of the whole agricultural sector. This study uses a microeconomic panel data approach to estimate farmers’ decisions in adopting (inter-farm) sustainable irrigation technologies and assesses the intensity of (intra-farm) water-saving practices. Our analysis identifies the main determinants of adopting WCSTs for Italian farmers based on different socio-economic, physical, environmental and climatic variables.

KEYWORDS:

• Water conservation and saving technologies
• irrigation technologies
• technology adoption
• technology diffusion
• adaptation to climate change
• Italian farmers

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Reducing the number of water applications can decrease water stress, production input costs (e.g. water, fertilizers, pesticides), risks linked to crop yield losses as well as increase the overall value of irrigation by producing more value with fewer units of water.

2 The FADN datasets are collected randomly through the use of annual surveys of over more than 10,000 European farms. A representative sample is created for the Italian agricultural sector. The FADN datasets provide very precise and detailed information on farms’ georeferenced data: economic, productive, environmental, geographic, and social factors.

3 The climatic data was provided by the division of Impacts on Agriculture, Forests and Ecosystem Services (IAFES) of the Euro-Mediterranean Center for Climate Change with 0.5° × 0.5° grid cell spatial resolution (25 km²). Extracted from the ERA-Interim dataset of the European Centre for Medium-Range Weather Forecasts (ECMWF), this dataset includes seasonal climatic variables.

4 Calculated as the level of operating revenues of the total investments of a farm, this indicator should represent the capability of a farmer to measure the efficiency of an investment in obtaining high rates of return.

5 This variable suggests the farm’s access to credit, as well as the degree of farmer indebtedness with respect to internal and external financial resources. This indicator is considered a proxy for a farm’s financial strategy (Alcon, de Miguel, and Burton 2011).

6 Provinces are sub-regional administrative institutions of the Italian Republic (NUTS 2). They are second-level administrative divisions and fit between regions and municipalities. Italy has 107 provinces.

7 Soil texture – the combination of sand, silt, and clay – influences water availability in the soil’s layers and determines the rate at which water percolates through the soil. Moreover, it has an impact on crop water needs. It stands to reason that if the soil is mainly sandy, meaning the soil has reduced water retention, there should be an increase in the probability of adopting WCSTs since they are more efficient than traditional irrigation systems (i.e., flooding or furrow). Conversely, if soil water retention is high, as with clay soil, the opposite trend should be seen.

8 In other studies, weather variables are often introduced as yearly averages or variances of temperature and rainfall (see, among others, Asfaw et al. (2016); Huang et al. (2017); Knapp and Huang (2017)).

9 Reference evapotranspiration (known also as potential evapotranspiration) (ET₀) is the evaporative demand of the atmosphere independent of crop type, crop development and management practices. The value is independent of the water abundance in the area because it is only affected by climatic parameters. It is comparable in time and space with other ET₀ (Allen and FAO 1998) and it is measured in mm*day^–1. ET₀ is a measure of the evaporating power of the atmosphere from land surfaces in a specific area and time, independent of crop and soil characteristics. Its value represents the amount of water lost by evaporation and plant transpiration, and it is a proxy for the water demand of crops to compensate for natural water losses (Allen and FAO 1998). ET₀ is calculated using the Penman-Monteith method, which is based on a hypothetical grass reference crop, specific height, the soil’s shade tolerance, and water standards (Allen and FAO 1998). The standard ET₀ considers solar radiation (sunshine), air temperature, humidity and wind speed from a dataset of standard climatological records.

10 Following Woodill and Roberts (2018) and Knapp and Huang (2017), a five-year moving average is applied to each seasonal AI index (weather data are based on the years from 2007 to 2016 and the seasonal moving average indexes are computed by including data from the 4-years prior plus the current year). The seasonal AI includes AIJFM (January, February, March), AIAMJ (April, May, June), AIJAS (July, August, September) and AIOND (October, November, December). AIs are computed on the basis of each farm’s geographic ERA coordinates. The seasonal aridity indexes show that spring and summer are dry periods where the precipitation levels are well below the threshold of 0.65. In contrast, winter and autumn correspond to the humid period (Appendix Figure A1).

11 The AMEs of Probit models are analysed because they represent the change in the probability of WCST adoption at the regressors’ mean. Specifically, marginal, or partial, effects allow us to define the effects on the conditional mean of the dependent variable when a unitary change of a covariate occurs. In other words, the AMEs allow us to capture the change in the probability of adoption by, ceteris paribus, a unit change of a regressor (Wooldridge 2010).

12 We changed the timeframe for calculating the moving averages using three and four years and got similar results. We also tried to reduce the seasons to two main periods (cold = autumn + winter and warm = spring + summer), but the results remained similar.