## The Irrigation Question

A persistent question about the Ricardian statistical estimates of how agriculture responds to changing climate has been whether they have adequately taken account of irrigation. One issue is whether benefits attributed to warmer climates are overstated because in fact these benefits reflect high values of land and output per hectare attributable instead to irrigation combined with the fact that there is a higher incidence of irrigation in warmer regions. Another issue is whether climate impact projections using these models address the availability of water for irrigation.

Evapotranspiration (the combined loss of moisture from soil through evaporation and plants through stomatal transpiration) increases with temperature. The need for irrigation rises as conditions become drier. It rises as a function of the difference between evapotranspiration and precipitation. Because global warming will increase both temperature and precipitation, the implications for soil moisture and the need for irrigation depend on the outcome of the race between rising temperature and rising precipitation.

An extremely simple test for the United States shows the incidence of irrigation is positively related to temperature and negatively related to

8. The ratios of net revenue per hectare to output per hectare used in chapter 5 range from a low of about 40 percent in the United States to a high of 78 percent in Africa. This ratio is applied to the percent change in land rental equivalent to obtain the corresponding percent change to be expected in output potential.

9. The MMSA Ricardian function is similar to equation (5.2) in chapter 5 (the MS function) but changes the coefficient on the logarithm of the ratio of carbon concentration to today's 350 ppm from 480 to 687 (Mendelsohn et al. 2000, 559).

precipitation. Using state data for agriculture and state capital data for temperature and precipitation, figure 3.1a shows the relationship across US states between the ratio of irrigated crop area to harvested crop area (percent) and annual average daily temperature (°C). Figure 3.1b shows the same incidence of irrigation as related to average annual precipitation (mm per year).10

Broadly the scatter diagrams show higher incidence of irrigation for higher temperatures and lower incidence of irrigation with higher precipitation. There are three states that are outliers to the temperature trendline: Nevada (irrigation incidence at 136 percent of harvested cropland), Utah (114 percent), and Wyoming (119 percent).11 All three have extremely high irrigation incidence but relatively low temperatures. The anomaly is explained by the low precipitation in all three, as they become the upper-left observations in figure 3.1b.12 Many of the states have low irrigation incidence, but some have extremely high incidence, as indicated by the difference between median (8.2 percent) and average irrigation (29.5 percent).

A simple statistical regression for these data shows the following, with t-statistics in parentheses:

Z = 24.1 + 3.73 T - 0.0455 P; adj. R2 = 0.21 (1.57) (3.22) (3.29)

Although the degree of explanation is moderate at only about 20 percent, the coefficients on temperature (T) and precipitation (P) are highly significant.

To anticipate the following climate analysis, for the United States the estimates in this study indicate that baseline global warming by the 2080s would cause the farmland-weighted averages for annual temperatures to rise by 5.4°C and the corresponding averages for precipitation to fall by 4.3 mm per year.13 If these changes are applied to the simple regression equation, the incidence of irrigation would need to rise by 20.3 percentage points as a consequence of climate change. The increase would be almost entirely from higher temperature; the slight decline in precipitation would have little effect, except in the sense that the failure of precipitation to rise would mean that the race between temperature and precipitation would be won hands down by temperature.

10. Temperature and precipitation are from NOAA (2007). Irrigated and harvested crop land are from USDA (2004).

11. Greater than 100 percent incidence implies that some irrigated land is used for pasture rather than crops and/or that some irrigated land is left fallow.

12. Note also that Carson City at 4,687 feet elevation and average annual temperature of 10.7°C may not be as representative of statewide conditions as is the case for most capitals. Thus, also in Nevada, Las Vegas at 2,028 feet has average annual temperature of 20.1°C.

13. Calculated from table 4.2 and appendix table E.1.

irrigated crop area as percent of harvested crop area

0 0