Water Conservation Potential from Irrigation Technology Transitions in the Pacific Northwest

Water Conservation Potential fromIrrigation Technology Transitionsin the Pacific Northwest

Glenn D. Schaible, C. S. Kim, and Norman K. Whittlesey

The effects of price changes on irrigation technology transitions and potential

agricultural water conservation in the Pacific Northwest are analyzed using Parks'

(1980) modified multinomial logit model. Results indicate that commodity price

effects are statistically significant, but they are relatively small with nonprogram crop

price effects greater than program crop price effects. Locational factors are also found

to affect technology transitions. In the absence of water policy changes, continued

irrigation technology adoption by year 2005 will result in average annual water

savings of approximately 404,000 acre-feet in the Pacific Northwest.

Key words: irrigation technology, water conservation, water demand, water policy.

Water resources development, inexpensive en-ergy, and economic policies as well as insti-tutional water resource arrangements have allcontributed significantly to agriculture's de-mand for water resources (Martin; Weather-ford and Ingram; Vaux; Just, Lichtenberg, andZilberman). Irrigated agriculture currently ac-counts for approximately 83% of water con-sumption in the 17 western states. However,growing demands for quality water resourcesby nonagricultural uses, including energy, mu-nicipal, commercial and industrial, recreation,fish and wildlife, and Indian and federal re-served rights, has heightened competition fora finite resource supply. Reallocating this scarcewater resource may be accomplished with re-fined water market structures, as well as throughresource policy-induced agricultural conser-vation (Howe; Vaux; Frederick; Bromley).

Removal of institutional barriers associatedwith Bureau of Reclamation water rights/usesis an important component of the reforms re-quired for the development of market-oriented

Glenn D. Schaible, an agricultural economist, and C. S. Kim, asenior agricultural economist, are with the Economic ResearchService, U.S. Department of Agriculture. Norman K. Whittleseyis a professor of agricultural economics, Washington State Uni-versity.

The views are the authors' and do not necessarily represent thoseof the authors' respective institutions.

water transfers (Wahl; Burness and Quirk).However, Bureau of Reclamation water deliv-ered to farms accounts for approximately 60%to 65% of surface water used for irrigation andfor only 25% to 30% of total water used forirrigation in the West.' In addition, the natureof the "politics of water," due to the commonproperty aspect of water supplies controlled bymany irrigation districts, will limit the effec-tiveness of some water markets (Rosen). Fur-thermore, upper-basin states, with water-de-pendent agriculture, are unlikely to provideunlimited support for significant interregionalwater market transfers. States have becomemore protective of their resources by broad-ening the "considerations" in reviewing ap-plications for changes in water rights (Mac-Donnell). These considerations involveaccounting for adverse effects to fish and wild-life, water quality, groundwater recharge, andthe regional economy. As a result, significantreallocation of western water resources to ei-ther higher valued uses, or uses justified on thebasis of equity and/or environmental policygoals, must now be resolved through policies

'Percents were derived using data from the Bureau of Recla-mation, 1988 Annual Report, and the Bureau of the Census, Farmand Ranch Irrigation Survey (1988).

Western Journal of Agricultural Economics, 16(2): 194-206Copyright 1991 Western Agricultural Economics Association

Water Conservation from Irrigation Technology 195

which promote greater agricultural conserva-tion.

Water-conserving irrigation technologies/water management practices are known to beable to significantly increase on-farm irrigationefficiencies by 10-30%, to significantly reducecrop water application requirements, and toreduce energy costs (Sweeten and Jordan; Jen-sen; Lyle and Bordovsky; Henggeler, Sweeten,and Keese; Homan, Skold, and Heermann;Wyatt). Therefore, as competition for waterresources continues to grow, irrigation tech-nology/water management substitutions forwater will be required to conserve greaterquantities for alternative demands. Caswell,Lichtenberg, and Zilberman confirm that en-vironmental considerations "may become amajor incentive for adoption of water-con-serving irrigation technologies..." (p. 889).

This study is concerned with estimating theseconservation/reallocation quantities. Theanalysis will estimate an irrigation technologytransition model in order to assess expectedconservation assuming past economic/insti-tutional environments (baseline) continue andthe degree to which technology adjustments,given changes in agricultural economic envi-ronments, will contribute to conservation/reallocation potential. Given that conserva-tion-oriented water policy changes will inducesubstitutions of irrigation technology/watermanagement for water beyond baseline esti-mates, results from the transition model thenestablish a basis from which to judge potentialconservation contributions of water policychanges.

The degree of resource substitution and thepotential for agricultural water conservation/reallocation is influenced principally by loca-tional (environmental) and economic vari-ables, such as commodity prices. Locationalfactors, such as climate, soil types, and topog-raphy, play a major role in defining the landquality and crop production character of a re-gion. Several studies have applied multino-mial or binomial logit, discrete choice modelsemphasizing farm-level locational character-istics in estimating descriptive, technology, orcropping pattern adoption models (Caswell andZilberman; Negri and Brooks; Lichtenberg).Given information on various farm charac-teristics a priori, such as climatic setting andland quality values, these logit models esti-mated the likelihood of adopting a particulartechnology on that farm. However, examina-

tion of the influence the agricultural economicenvironment has had on aggregate irrigationtechnology transitions has been sparse at best.

Just, Lichtenberg, and Zilberman extend theearlier works of Caswell and Zilberman, andLichtenberg, estimating an irrigation adoptionmodel to examine the effect farm programshave had on irrigation expansion and ground-water depletion. These authors suggest the needexists to focus more research on structural in-teractions between resource use and economicvariables. Finally, representative, farm-firmlevel activity models also have been used ex-tensively to examine micromanagement re-source adjustments to irrigated agriculturaleconomic environments (Hornbaker andMapp; High Plains Associates; Bernardo et al.;Ellis, Lacewell, and Reneau). However, thesestudies generally have used alternative exog-enous assumptions to constrain irrigation pro-duction technologies.

Previous studies have been site specific, haveused only firm-level, cross-sectional data, orare engineering studies based on experimentaldata. Also, data used in these studies generallyhave prevented a sufficient disaggregation oftechnology states. As a result, these studies havebeen micro oriented and have not examinedthe influence of prices on aggregate irrigationtechnology transitions. These studies have notaddressed irrigation technology transitionsfrom an aggregate, water conservation/reallo-cation policy perspective. In addition, previ-ous econometric studies modeled technologyadoption assuming only the traditional spec-ification error structure. This study examinesthe polychotomous character of the technologyadoption decision and estimates a modifiedmultinomial logit model, explicitly recogniz-ing the traditional specification error as wellas approximation error attributed to havingdata only on aggregate technology proportions,rather than on micro-level technology transi-tions (Parks 1980). Finally, emphasizing theinfluence of the agricultural economic envi-ronment on aggregate technology adjustmentsimposes recognition of the time-dependent na-ture of technology transitions and an autore-gressive error adjustment.

Specifically, in this article we investigateeconometrically, using Parks' (1980) modifiedmultinomial logit model, the influence that ag-ricultural economic variables have on aggre-gate irrigation technology transitions to water-conserving technologies, while adjusting for

Schaible, Kim, and Whittlesey

Western Journal of Agricultural Economics

locational factors. The model is estimated forthe Pacific Northwest states of Idaho, Oregon,and Washington, a region with an increasingsense of scarcity due to growing nonagricul-tural water demands. We further examine eco-nomic implications, in terms of resource real-location potential, by forecasting technologytransitions and estimating agricultural waterconservation potential for alternative pricescenarios. These relationships are importantin identifying potential differences between re-gional dynamics of water resource adjustmentsand the conservation/reallocation potential offuture conservation-induced water policychanges. Finally, water policy implications aredrawn from the research results.

Polychotomous Discrete IrrigationTechnology Choice Model

Farm producers, assumed to be rational de-cision makers, make irrigation technologychoices consistent with utility maximizing be-havior. Irrigation technology choices, for aparticular plot (field), are discrete and mutu-ally exclusive. In other words, choosing onetechnology for a field excludes the simulta-neous use of other technologies on that field.Such discrete choice behavior is assumed tobe consistent with a well-defined, additive-separable, perceived utility function and con-sistent with random utility maximization(McFadden 1974, 1976, 1981; Pudney). Then,irrigators maximize their perceived utility byadopting the ith irrigation technology, where-in:

(1) U, = Max[Di(P, w, 4) + E(ei)T

> Dj(P, w, ) + e(ej)]

for i, j = 0, 1, ... , T; i - j, and where D andE are real valued functions, and Di(P, w, 4) =D[iri(P, w, 4)], Dj(P, w, 4') = D[rj(P, w, 4)], andrj(P, w, 4p) is the jth technology-specific per-ceived profit function, assuming competitiveinput/output markets and well-defined pro-duction technologies (Chambers and Just;Chambers and Foster). The vector sets for P,a vector of output prices; w, a vector of inputprices; and 4, a vector of location (land) andtechnology attribute/characteristics, representthe nonstochastic, observable values associ-ated with irrigation technology choices and ir-rigator perceptions of those values that definethe relative profitability of alternative tech-

nologies. The values ej represent the elementof the technology decision which reflects thevector of values for unobservable, unmeasur-able choice attribute/characteristics, plus theunobservable random values associated withirrigator perceptions of observed and unob-served choice factors.

Irrigators maximize their perceived utilityby maximizing perceived profits over tech-nology choices (Chambers and Foster; Caswelland Zilberman; Negri and Brooks), such that7ri(P, w, ') > irj(P, w, A) for i, j = 0, 1,..., Tand i = j. The random nature of irrigator per-ceptions of profits results in a stochastic utilityfunction (McFadden 1974) and, therefore, aprobabilistic irrigation technology choice.

Because e in equation (1) is stochastic, thefarm-producer decision is expressed as theprobability of selecting the ith irrigation tech-nology:

(2) Pi= Prob[Di(P, w, ) - Dj(P, w, ) > E(ej) - e(ei)]

for all i, j = 0, 1, ... , Tand i - j.To estimate the probabilities of alternative

technologies requires the specification of thefunctional form of D,(P, w, 4') and the distri-bution of e(ei). Specifying these modeling char-acteristics depends upon the nature of farm-producer technology decisions and consistencyof the assumption with respect to e(ej) - e(ei)and, therefore, e(ej) and e(ei), with utility-max-imizing behavior. Domencich and McFaddendemonstrate that if cj and Ei are independent,identically Weibull-distributed random vari-ables, then their difference is also Weibull dis-tributed and consistent with both the logisticfunctional form and utility maximization.

With respect to farm-producer decisions,farm producers have the option of choosingfrom among a multiple of on-farm irrigationtechnologies. Available technologies includesuch options as gravity application systemswhich deliver water across a field by the forceof gravity, either through a "field-flooding"technique or the use of more mechanical/man-agement techniques which could include si-phon-tube or gated-pipe systems, or surge-flowor cablegation systems. Some of these moremanagement-intensive systems could also in-clude the use of tailwater-reuse pits (the col-lection and reuse of irrigation water runoff).Sprinkler applications, which deliver water tothe field under pressure, include such tech-nologies as gun systems, hand or wheel movesystems, permanent systems, and high/low-

196 December 1991


pressure center-pivot systems. Drip/trickle ir-rigation, a low-pressure technology, distributeswater directly to the plant root zone throughwater emitters attached to small diameter tubesplaced above or below the field's surface. Thistechnology is most often applied to specialtycrop production (fruits and nuts, vegetables,etc.).

These technologies differ in their water-ap-plication efficiencies, i.e., the ratio of the crop'sconsumptive water requirement to the quan-tity of water applied to the field. Irrigationtechnologies also differ in their unit applica-tion costs as well as their effect on crop yields.While for agronomic reasons one technologymay be preferable to another, this does noteliminate the technology as a farm-produceroption. The water conservation and yield andunit cost effects, however, do result in onetechnology being relatively more profitablethan another. Therefore, the farm-producerdecision, together with a Weibull-distributedE, suggests the use of a polychotomous discretechoice model, i.e., a multinomial logit func-tional form (Caswell and Zilberman; Mc-Fadden 1974, 1981).

Following McFadden (1974), choices, or ir-rigation technology selection probabilities (Pi),are written in terms of the multinomial logitmodel as:

eDi (X)

(3) Pi =j ,

* o eDJ(j=o

i=0, 1, ... , J

where Dj(X) is an estimated utility function,Uj, for the farm producer with a probabilisticchoice set, Pi, where i = 0, 1, ... , J; and Xrepresents (simplified for later notational pur-poses) the aggregate set of relevant prices andlocation (land) and technology attribute/char-acteristic vectors, {P, w, i}, such that the farmproducer maximizes Uj = Dj(P, w, t) + cj, whereEj are independent random variations identi-cally distributed with the Weibull distribution.

The normalized multinomial logit model isexpressed as

e(di)Pi = J

1 + ~ e(d)j=1

for i= 1, ... , J, and

Po = 1 + z e(d),j=l

where the function di, the difference in farm-producer utility between choice sets (Pi) and(P0), is expressed as the following linear func-tion of the aggregate vector X plus the randomerror ei for choice set Pi:

(5)

(fori= 1,...,J).

Using equations (4) and (5), a multiregional,temporal logit equation for polychotomous ir-rigation technology decisions is expressed as:

K

(6) dimt , yimt ln(p,,nt /Pomt) = ik Xkm,k=l

+ E (Vimt + Uimt),

i = 1, ... , J (irrigation technology states);

m = 1, ... , M (cross sections);

t = 1, ... , T (years); and

where Yimt represents the normalized choice interms of the log of the odds of choice Pimt toPoit; X is the aggregate vector of relevant out-put and input prices, as well as location/tech-nology characteristics; fi is the vector of un-known parameters; and Vimt and uit are thespecification and approximation random-er-ror terms, respectively (Parks 1980).

Conventional logit analysis assumed that Ein equation (1) was equivalent to E(vimt). How-ever, this term is now appropriately recognizedas only the specification error component of E(Amemiya and Nold; Parks 1980). Becauseequation (6) uses observed proportions pi andnot actual selection probabilities Pi, the tech-nology logit equation involves the additionalapproximation error Uimt = ln(pimt/pomt) -In(Pimt/Pomt). The errors vimt and Uimt are as-sumed independent (Parks 1980), where forthe specification errors E(vimt) = 0, and

E(vimtvjmt) = E(vivj,):

= a for (mt) = r = y for all i and j= 0 for (mt) =r -#y for all i andj.

The approximation error structure is multi-variate normal with the mean vectorE(Uimt) = E[ln(pimt/pom) - ln(Pimt/Pom)] = 0,and the covariance for all i andj for each (mt)diagonal set, E(Uimt/ljmt) = (mt), where:

(7) Q(mt) = (l/n)mt


di = #ilXI + #i2X2 + - · · + OiKXK + Ei

(4)


/Pomt ... 1/Pomt

1/Pomt + 1/P2mt... 1/Pomt

1/pr ... 1./P + 1/P i

/Pomt ... 1/Pomt + 1/PJmtJ

Based on the use of Zellner and Lee's jointestimation procedure for discrete choice mod-els, the joint multinomial logit equations as-sociated with equation (6) can be expressed incompact notation as:

(8) Y = X: + E,

where Y is a (J x 1)MT vector, 0 is a (J xK) x 1 vector such that 3 = fl, 3., J),

and X is a block diagonal matrix such that:

and Xj(mt) represents the vector of MT obser-vations for the jth logit equation for the (1 xK) vector of explanatory variables indicatedin equation (6). The error structure of equation(8), c = (v + u), is expressed as E(mt) = 0 withthe covariance matrix for E in block diagonalform as:

(10) E(EE'

, +92 + z

= t + (I O ) = V

MT + 2

where Z is estimated by:MT

(11) , = S-[1/(MT)]. Qg,g=l

and Qg is estimated using equation (7), replac-ing Pi with p, (Parks 1980). Then, Aitken's es-timator, adjusting for heteroskedasticity andcross-equation correlation, for the modifiedmultinomial logit (MML) irrigation technol-ogy transition model is given by bMML =(X' V- X)-'X' V-y, and the Var(bMML) =(X' V-X)-1.

However, because cross-section, time-seriesdata were used for this study, the estimated

cross-choice covariance matrix, :, was alsocorrected for a first-order autoregressive errorstructure. This correction involved first apply-ing Parks' (1967) cross-section, time-series(CSTS) estimation procedure to the equationsindicated in equation (8) to acquire the trans-formed residuals e = y* - (X*)bcss. Second,the CSTS residuals are used to estimate theunadjusted MML residual covariance matrixin equation (11) as:

(12)S [ B e l}M T

S = I\\(MT) - 2 eig\\. -gL =1 -I

The estimator, bML, is both consistent andasymptotically more efficient than the stan-dard multinomial logit estimator (Parks 1980).

Application to the Pacific Northwest

Agriculture in the Pacific Northwest accountsfor more than 80% of total water withdrawals[U.S. Geological Survey (USGS)]. Irrigationduring initial development stages (early 1900s)emphasized the use of gravity systems. By theearly 1970s, technology transitions were evi-dent, with gravity and sprinkler systems ac-counting for 66% and 34% of regional irrigatedacres, respectively (Irrigation Journal).Throughout the 1970s and early 1980s, ad-ditional technology adjustments consisted pri-marily of center-pivot sprinkler technology. In1986 gravity systems accounted for only 42.7%,while sprinkler technology was used on 57.3%of regional irrigated acres (15.6% of which usedcenter-pivot sprinkler technology) (IrrigationJournal).2

This study examines these transitions to wa-ter-conserving technologies, estimating aggre-gate technology shares as a probabilistic func-tion of locational and time-dependenteconomic variables. Pooled data for the statesof Idaho, Oregon, and Washington, over theperiod 1974-86, are used to jointly estimatethe logit equation (8). Aggregate (grouped)technology shares are estimated using irrigated

2 Data for 1988, recently available from the Farm and RanchIrrigation Survey (1988) (Bureau of the Census), indicate that grav-ity systems accounted for approximately 36.1%, while sprinklersystems were used on 62.5% of regional irrigated acres. These datado not exist as an annual series and, therefore, could not be usedas part of the data base for the empirical model. However, theydo support the general nature of irrigation technology transitionsfrom gravity to sprinkler systems in the Pacific Northwest.

198 December 1991

·II

I


Table 1. Results for the Modified Multinomial Logit Model for the Log of Odds of IrrigationTechnology Transitions in the Pacific Northwest

Log of Deflated Output toEnergy Price Ratios for

Equation:a Constant WA* ID* Wheat Corn Alfalfa

LNRAT1Ob -4.6365 .4107 -. 1446 .1177 -1.2637 1.2499(.8719) (.0844) (.1031) (.2794) (.2245) (.1575)

LNRAT20c -10.8501 .9615 .7433 1.2965 -2.5546 1.8669(1.0132) (.1027) (.1066) (.2142) (.1983) (.1496)

Joint Equation Estimation R2= .9237

Covariance Matrix:

.0286 .03981 .0264 .0387-__ _L.0398 .1031J .0387 .0974 J

a Numbers in parentheses are estimated standard errors.b LNRAT10 represents the logit equation for the log of the ratio of conventional sprinkler systems (P,) to gravity systems (P0).cLNRAT20 represents the logit equation for the log of the ratio of center-pivot sprinkler systems (P2) to gravity systems (Po).* Locational variables for Washington (WA) and Idaho (ID). Oregon is the benchmark state.

acreage by technology for each state, publishedannually in the Irrigation Journal.

The dependent variable, ln(pmt/p0mt), rep-resents the log of the odds of the relative sharesfor irrigation technology classes consisting ofgravity systems (POmt), conventional sprinklersystems (plmt) (including gun, boom, travelersystems, hand, mechanical, wheel move sys-tems, and towline and sideroll systems), andcenter-pivot sprinkler systems (p2mt)'

Explanatory variables include locationaldummy variables and three variables for com-modity prices for wheat, corn, and alfalfa haydivided by irrigation energy costs. The threeregional dummy variables for Idaho, Oregon,and Washington reflect nominal characteriza-tions of regional "locational" attributes in ag-gregate that result in differences in regionaltechnology shares. Crop price variables forwheat, corn, and alfalfa were chosen becausethey reflect regional program versus nonpro-gram irrigated crop diversity and a significantshare of both irrigated program crop and totalirrigated acreage (Bureau of the Census). Othercrop prices are either significantly correlatedwith wheat, corn, or alfalfa prices (other hayand alfalfa hay, for example) or acreage forthese crops is individually a relatively smallshare of total irrigated acreage.3 Electrical costs

3 Total regional irrigated corn, wheat, and alfalfa acres accountfor 36% of total regional irrigated acres. Wheat and corn acresaccount for 65.4% of total regional irrigated program crop acreage.Remaining regional irrigated crop production includes acres forbarley, other hay, sugar beets, Irish potatoes, vegetables, orchards,and an "other crop" category (8.7%, 6.3%, 2.6%, 8.4%, 3.9%, 5.2%,

(¢/kWh) paid by irrigators (Bonneville PowerAdministration) are used for irrigation energycosts. Crop price data, adjusted to 1984 dol-lars, are from the U.S. Department of Agri-culture (USDA). Indices of prices received forfeed grains, food grains, and hay products wereused to adjust output prices, and electrical costswere adjusted using an index of prices paid forfuels and energy (USDA).

Estimation Results

Table 1 presents the estimation results for thePacific Northwest. LNRA T10 indicates the es-timated parameters for the logit equation forthe log (LN) of the ratio (RAT) of conventionalsprinkler systems (P,) to gravity systems (Po),specifically, for ln(p 1mt/P0 mt). LNRAT2 0 indi-cates the estimated parameters for the equa-tion for the log of the ratio of center-pivotsprinkler systems (P2) to gravity systems (P0),specifically, for ln(p 2mt/p0om). The significanceof the price variables and the size of the jointequation estimation R-square value (.9237)indicate that irrigation technology transitionscan be explained with price variables.

The coefficients for the price variables maybe interpreted as the relative responsiveness

and 28.9%, respectively) (Bureau of the Census). The "other crop"category includes irrigated production for corn silage, dry-ediblebeans, and such small grains as oats and rye. Crop prices for pro-duction of barley, other hay, and the "other crop" category (43.9%of the remaining regional irrigated acreage) are assumed to bestrongly correlated with either corn or alfalfa prices.


Western Journal ofAgricultural Economics

of the odds of adoption of technology pi totechnology p, to changes in deflated relativeprices. Therefore, the log-log functional spec-ification results in the coefficients for a pricevariable being interpreted roughly as a relativeprice elasticity.

Price coefficients are statistically significant,except for wheat price in equation LNRAT10.Relatively small changes in real wheat pricesover the study period, 1.2 to 2.1 cents/bushelper year (USDA), due to a government sup-ported price, may be why wheat price is sta-tistically insignificant. However, most pricecoefficients are not significantly greater thanone. These coefficients may indicate that par-tial price effects on irrigation technology adop-tion are generally relatively small. In addition,price effects are greater for center-pivot sys-tems relative to gravity systems, than they arefor conventional sprinkler systems relative togravity systems. This is to be expected givengreater water-use efficiencies, reduced labor re-quirements, and the increased range of topog-raphy and soils suitability of center-pivot sys-tems, even though investment costs of suchsystems are often higher than conventionalsystems (Negri and Hanchar; Sweeten and Jor-dan).

The nonprogram crop price ratio (alfalfa/energy) has the greatest positive effect on tech-nology transitions for both conventional sprin-kler and center-pivot systems. This is probablydue to the fact irrigated alfalfa, accounting forthe most significant share (nearly 20%) of totalirrigated acres in the Pacific Northwest (Bu-reau of the Census), serves as a critical feed-stock input for the regional livestock sector.This result suggests that the transition to morewater-conserving technologies in the PacificNorthwest may be influenced more by the eco-nomic environment for nonprogram crops thanby that for program crops. The negative coef-ficients for the corn/energy price ratio, whileunexpected, probably reflect the fact that de-flated corn prices for each state declined overthe study period, on average from 4.2 to 8.7cents/bushel annually (USDA).

Results in table 1 also confirm the conclu-sions of the previous studies (Caswell and Zil-berman; Lichtenberg; Negri and Brooks), thatlocational factors play an important role indetermining irrigation technology adoption.The larger coefficient for Washington in equa-tion LNRAT20 indicates that locational fac-tors in Washington influence water-conserving

technology adoption to a greater extent thanin Oregon or Idaho. However, this probablyreflects the fact that a vast majority of irriga-tion in Washington is more localized within afairly homogeneous eastern region of the state,characterized by sandy soils and uneven ter-rain. Furthermore, much of this irrigation ex-ists as part of the Columbia Basin Project, us-ing a publicly financed distribution system toirrigate vast acreages far from the surface watersource. Much of this irrigated acreage devel-oped concurrent with center-pivot sprinklertechnology. On the other hand, irrigation inOregon and Idaho is either more geographi-cally dispersed with more privately financeddistribution systems or localized, but with alonger history of more intensive riparian de-velopment (USGS). This development is char-acterized by irrigation on heavier soils and onterrain more conducive to the use of on-farmgravity distribution sytems.

In addition, the greater size of the locationcoefficients for LNRAT20 than for LNRAT10suggests locational factors may play a moresignificant role in technology adoption the morewater conserving the technology. This merelyreflects the fact that newer, water-conservingtechnologies are less generic and depend moreheavily on management skills in concert withlocational factors in defining their adaptabili-ty. As a result, the greater "locational" em-phasis on the use of more management-inten-sive sprinkler technology in Washington, dueto soils, terrain, and development timing,would seem to preclude an ability to morereadily handle adjustments from high-pressureto low-pressure sprinkler technologies. Con-sequently, assuming no significant water policychanges, the less generic and more location-dependent character of newer technologiesmeans that these results may also suggest a lessinfluential role for prices in future regionaltechnology/water management adoption.

Finally, the predictive efficiency of the es-timated logit equations was also tested usingTheil's U-coefficient. The relatively smallU-coefficient values (table 2) indicate that theestimated equations performed reasonably welland are reasonably valid relationships. The highR 2 for the joint equation estimation, statisti-cally significant coefficients, and small U-co-efficients all seem to suggest, at least for thePacific Northwest, that the MML irrigationtechnology model can be useful as a predictorof the effects of exogenous economic changes.

200 December 1991


Table 2. Theil's U-Coefficients

For Predicted Proportions

State/Region p0 pi P2

Idaho .0416 .0520 .1304Oregon .0520 .0596 .1245Washington .1067 .0594 .1137

Pacific Northwest .0628 .0580 .1191

Note: Values equal to zero indicate the simulated results are per-fect. Values equal to one indicate no relationship (Chan).

Simulation Analysis

In 1988 nearly 12.3 million acre-feet of waterwere applied to produce a variety of crops inthe Pacific Northwest. Table 3 indicates actualaggregate water-use coefficients of alternativeirrigation systems. These coefficients reflectcurrent average water-use efficiencies givenreal-world application difficulties, constraints,and mismanagement. While these coefficientsare similar for Idaho and Oregon (for overallsystems), Idaho uses the greatest quantity ofwater (6.1 million acre-feet) within the region.Average per-acre application for all irrigationsystems was 1.9 acre-feet in Idaho and Oregon,compared to 2.3 acre-feet for Washington (Bu-reau ofthe Census). In all three states, per-acreapplication rates were lower for sprinkler ir-rigation systems than for gravity flow systems.Continued adoption of sprinkler irrigation sys-tems (especially center-pivot systems) will re-sult in agricultural water conservation throughincreased water-use efficiency.

Simulation analysis is conducted to provideinformation on agricultural water conserva-tion assuming continued technology adoption,

Table 3. Water Application Rates in the Pa-cific Northwest

Irrigation Technology

AllAll Gravity Sprinkler

State/Region Systems Systems Systems

................. Acre-Feet/Acre ------------------Idaho 1.9 2.2 1.7Oregon 1.9 2.1 1.6Washington 2.3 2.6 2.1Pacific Northwest 2.0 2.3 1.8

Source: Bureau of the Census, Farm and Ranch Irrigation Survey(1988).

Table 4. Annual Average Real (Deflated)Price Changes for the Period 1974-86

Corn Wheat Alfalfa EnergyState ($/bu.) ($/bu.) ($/ton) (¢/kWh)

Idaho -. 042 .017 3.03 .002Oregon -. 087 .021 4.20 .002Washington -.066 .012 2.10 .001

Note: Computed from data in Agricultural Prices: Annual Sum-maries 1973-1988, USDA.

however, in the absence of water policychanges. The analysis indicates the degree ofchange to be expected, from the 12.3 millionacre-feet of regional agricultural water use, un-der varying price assumption scenarios. Theestimated coefficients from table 1 [for equa-tion (6)] along with equation (4) are used tosimulate the effects on technology proportionsfor alternative price scenarios for the period1986-2005. Then, the technology proportionsfor each price scenario are further evaluatedby comparing their water conservation poten-tial.

Four annual price scenarios were used tosimulate technology proportions. Annual priceratios for the estimated logit equations (table1) for the period 1987-2005 were computedusing historical average real (deflated) pricechanges over the period 1974-86 (table 4). ForScenario I, wheat price rises annually by thehistorical average annual price change, whileprices for corn, alfalfa, and energy are heldconstant at 1986 levels. Scenario II is similarto Scenario I but with a 30% increase in wheat'shistorical average annual price change. Sce-nario III, which closely proxies a baseline sce-nario, involves annual price increases forwheat, alfalfa, and energy by their historicalaverage annual price change, while corn priceis held at the 1986 level. And Scenario IV issimilar to Scenario III but with the historicalaverage annual price change increased by 30%.

Results of the technology share simulationsare presented in table 5. These results indicatethat rather modest technology transitionswould continue with program crop (wheat)price increases (Scenario I), with declines ingravity and conventional sprinkler systems andslight increases in center-pivot sprinkler sys-tems for all three states. Center-pivot sprinklersystems would increase from 1.3-1.5% from1986-2005, with average annual real price in-creases for wheat ranging from 1.2 to 2.1 cents



per bushel. With a 30% increase in the averageannual price change for wheat (Scenario II),center-pivot sprinkler systems would increasefrom 1.7-1.9%. These results seem to suggestthat price increases for program crops in thePacific Northwest have minimal effects on ir-rigation technology transitions. 4

The combined effect of price increases forwheat, alfalfa, and energy, however, are moresignificant. If annual prices for wheat, alfalfa,and energy increase by their historical averageannual price change (Scenario III), the declinein gravity-irrigated acreage would range from8% in Washington to 15.9% in Idaho from1986-2005. Increases in sprinkler-irrigatedacreage during this time would range from 5.1%and 2.9% for conventional and center-pivotsystems, respectively, in Washington to 10.1%and 5.8% for conventional and center-pivotsystems, respectively, for Idaho. Increasing theaverage annual price changes by 30% (ScenarioIV) results in a slight increase in these tech-nology transitions. Sprinkler irrigated acreagewould increase by an additional 1.71%, 2.27%,and 3.22% for Washington, Oregon, and Ida-ho, respectively. These relatively minor in-creases probably reflect the effects of institu-tional barriers to resource adjustmentscommon to western irrigated agriculture. Inother words, they reflect the conservation dis-incentives inherent with beneficial use criteria,i.e., "use it, or lose it," and the lack of insti-tutional arrangements for conserved waterrights. Rather than risk losing water rights,these institutions (or the lack of them) promotestability with irrigators' initial investments, i.e.,technology in place tends to remain in use.However, results for Scenarios III and IV doseem to suggest that nonprogram crop pricechanges have a more significant effect on tech-nology transitions in the Pacific Northwest thando program crop price changes.5 This is dueto the larger share of the region's cropping pat-tern accounted for by nonprogram crops, whichis invariably influenced by their regional com-parative economic advantage.

4 Irrigated program crop acreage accounts for only 25.3% of totalirrigated acres in the Pacific Northwest (Bureau of the Census).Irrigated wheat acreage accounts for the largest portion (57%) ofthis irrigated program crop acreage and nearly 15% oftotal irrigatedacres. Only irrigated alfalfa acreage exceeds irrigated wheat acreagerelative to total irrigated acreage.

5 Irrigated alfalfa acreage accounts for nearly 20% of total irri-gated acres in the Pacific Northwest and represents the largest singlenonprogram crop, accounting for 26% of nonprogram crop acreage(Bureau of the Census).

Finally, these results provide some evidenceof the potential price effects of commodity sup-port programs on regional irrigation technol-ogy transitions. At least for the Pacific North-west, increases in commodity program supportprices (specifically for wheat) would haveminimal effects on any "land-augmenting"irrigation technology adoption. While this ev-idence differs from previous research (Lich-tenberg; Just, Lichtenberg, and Zilberman), itis not surprising. Both studies emphasize rel-ative profitability of irrigated program cropsas critical determinants of land-augmentingtechnology adoption. However, program cropprices affect profitability for a much smallerportion of irrigated crop production in the Pa-cific Northwest than in western Nebraska. Ir-rigated program crop acreage accounts for only25.3% of total irrigated acreage in the PacificNorthwest, and approximately 73% of waterfor irrigation comes from surface sources (Bu-reau of the Census). In addition, purchasedwater costs for off-farm water sources for thePacific Northwest are relatively low, averagingless than $22 per acre.6

Agricultural water use associated with tech-nology transitions for the Pacific Northwest isestimated by applying 1988-level irrigated acres(Bureau of the Census) and water-use rates (ta-ble 3) to the projected technology shares fromthe model simulation results (table 5). Thequantity of annual agricultural water conser-vation by 2005 due to continued irrigationtechnology adoption is the difference in wateruse for 1986 and 2005. Annual water conser-vation estimates by 2005 for each state (andthe Pacific Northwest) and MML model sce-nario are presented in table 6.

Annual water conservation by the year 2005for the Pacific Northwest will amount to ap-proximately 404,000 acre-feet, assuming thathistorical average annual price changes con-tinue (Scenario III). This level of water con-servation amounts to 3.3% of 1988 water use.If the average annual price change increasedby 30% (Scenario IV), agricultural water con-servation would amount to 3.9% of 1988 ag-ricultural water use. This level of water con-servation is relatively modest. The additional

6 Purchased water costs for Bureau of Reclamation water in thePacific Northwest averaged $13.31 per acre in 1986 (McGuckin,Moore, and Negri), while the Farm and Ranch Irrigation Survey(1988) (Bureau of the Census) indicates that purchased water costsfrom off-farm sources averaged $22 per acre for the Pacific North-west.

202 December 1991


,w0

0

CUQ

ri"

0

C16

cIl

16

404

riA0

'CU

CU

P.-

.P"

CdF-4

cna30

40

a

H-

0

i

0

d)

QV)3

o

C

01-)

0

00

0-0

0CZaau4)0

:1

0 o0 e

* Il

000 o00 O oent ml v

r- e -0- O,Oa 00 C\

*'t . Ce0

''t 0 no o 00

00oo

c"l u" N

oo 'I oloo t o

Il 00 C(l

00 N 00

. .( o

oooo

N4 o C'- . eo

0

0-4

en trn -00 o o

1t - 0T

o

oN00o0

T ON\ -T 00 ON^ T 00

00oo

0

4O

00 N CN00 O t00 00 00

N in 00

ON 00 N

00 00 00

rN o\ ONo 00 o

*- '. It

0

Cl 'I SOt- In mn

- - I- I

00 (ON O't (O nloo o ,'

e ("o -

N 'N 0- .l e'

(en en 0000 I cO

t en 00I<t ON en' oC Clen (N cN

q cn 0C

0n eN 0000 t- -

In tD D

*'I D Cl(en en Cl(n (en e'n

0\ _ 00- 0 00ON - CN- C" Clq. . .

- (N -(Cr tn 00tn .) >n

00\o

0

kn WN 'os40 C OC00 ^n 00cq C -

oN 0 00- so (

" N N. . .

00

C1

I

n In tn

00 'IT 0)cl cl Cl

0000

0ION"

O 0 000NN 00- .l .

00

I

0

Cl ("'. M- > NN - -

kn 00 004O Cl 000000 ooCl Cl Cl

ON00

IN

t - - ON'-00 t

00 oo Nt. 00 I'

1 N * nn

Co.~

so e( 0oo00 oo00 00Cl Cl Cl

04O\O

4)0 4)0 ac 40D 8Im oD o CD

Ct S ¢ ? | I ISO «D »D V S SW D D \ O .S p 'O ) n oo0oo O U oo ° ooN o U0oo , 0oo o, oU 00

ONN 0\ 0CON I ONONO 0C N UONONO N0

o-i0C


C)

0

~00-\ a

000

a)

d C4 )aB J

CdC(;A

o ON

a 0

0a 4).

S d

,,ja,04)

4) a 4

'c d

CF d4)

~a0cd4) a

>Cd~ a1

,a4) cdE

cd0aa0

III'da .a

(d d

CO) d c

0a

z

,'0 a' 6C444

a

C~ g00 .gd 0

COaa o aoaa. dC.)i 0 ~~000 0ao.4; II

0 Q mozck c

094c

0

o

f, o


Table 6. Annual Water Conservation in Year2005 from Irrigation Technology Adoption inthe Pacific Northwest

Using MML SimulationResults from:

Sce- Sce- Sce- Sce-nario nario nario nario

I II III IV

..............------- (1,000 acre-feet) --------------------Idaho 15.9 20.0 252.9 304.1Oregon 5.8 7.6 89.4 106.1Washington 4.3 5.0 61.3 74.6

Pacific Northwest 26.0 32.6 403.6 484.8

water conservation expected from increasingonly wheat prices will also be relatively small.Finally, results in table 6 indicate that the ma-jority of agricultural water conservation(62.7%) will occur in Idaho (Scenario III). Thisis to be expected, given that in 1988 50.8% ofregional irrigated production occurred in Ida-ho (3.2 million of 6.3 million acres) (Bureauof the Census). In addition, a larger share ofirrigated production in Idaho (41%) is basedon gravity technology, while only 31% of ir-rigated production in Oregon and Washingtonis based on gravity technology. Furthermore,2.6 million acres of Idaho's irrigated crop acre-age produces lower valued crops, relative to1.3 and 1.1 million acres for Oregon andWashington, respectively. Therefore, profit-ability is regionally more critical in Idaho.

Research Summary and PolicySuggestions

Results from this study indicate that com-modity prices affect irrigation technology ad-justments. However, while the price parame-ters for the Pacific Northwest are statisticallysignificant, most are relatively small, with pric-es for nonprogram crops having a greater in-fluence on irrigation technology transitionsthan prices for program crops. Locational fac-tors, such as climate, topography, soils, anddevelopment timing, also affected PacificNorthwest irrigation technology adjustments.Nonetheless, economic variables will be im-portant when evaluating potential agriculturalwater conservation from future irrigation tech-nology transitions.

The results indicate that irrigation technol-ogy transitions in the Pacific Northwest havebeen and will continue to be (in the absenceof water policy changes) relatively slow. As-suming that real prices for wheat, alfalfa, andenergy increase annually by their historical av-erage annual real price change, gravity irrigat-ed acreage declines by as much as 15.9% inIdaho to 8% in Washington by year 2005 (Sce-nario III). Conventional and center-pivotsprinkler systems will increase by 10.1% and5.8%, respectively, in Idaho, and by 5.1% and2.9%, respectively, in Washington. These shiftswill result in annual water savings by year 2005of nearly 404,000 acre-feet for the PacificNorthwest. Increasing the average annual realprice changes by 30%, however, results in onlya slight increase in technology shifts, rangingfrom 1.7% to 3.2% across states. These tech-nology shifts increase water savings from 3.3%to 3.9% of 1988 agricultural water use. Finally,results indicate that irrigation technology tran-sitions due to changes in real prices for wheatare relatively insignificant.

However, considering that the estimatedtechnology transitions and the implied con-servation are indicative of past irrigation ef-ficiencies, future conservation may be greaterbecause future irrigation technologies will bemore efficient. Recent studies examining theeconomics and risk aspects of water-conserv-ing technologies and water management strat-egies, including improved furrow irrigationsystems, low-pressure center pivots, low-en-ergy precision application (LEPA) systems, andirrigation scheduling, indicate potentially sig-nificant savings in water use and increased on-farm returns (Lee, Ellis, and Lacewell; Ber-nardo et al.; Homan, Skold, and Heermann;Harris and Mapp; Hornbaker and Mapp).Adoption of these technologies has variedthroughout the West, with adoption being par-ticularly slow in the Pacific Northwest.

In 1988, less than 19% of irrigated acreagein the Pacific Northwest involved the use ofsuch water-conserving technologies as LEPA,surge-flow, or cablegation systems, etc. (Bu-reau of the Census). Even fewer acres involvedthe use of such water management techniquesas soil moisture sensing or commercial irri-gation scheduling services. Therefore, the fu-ture adoption of newer water-conserving tech-nologies/management practices means that thewater conservation estimates in this article areprobably conservative. Future application of

204 December 1991


this study's research approach in regions wherethese technologies have become more domi-nant should reveal their conservation poten-tial.

Finally, the historically slow pace of irriga-tion technology adoption in the Pacific North-west should not be considered all that unusual.The availability of major surface water sources(the Columbia and Snake River basins) andthe institutionally protected status of agricul-tural water use, both legally (legal preferencefor agricultural uses) and economically (seefootnote 6), has resulted in the perception bymany irrigators of unconstrained water re-source supplies. The lack of adequate marketforces to transmit information to the agricul-tural sector on the increasing scarcity value ofwestern water resources has resulted in marketindicators (relative prices) playing a reducedrole in the past in promoting water-use effi-ciency/conservation in agriculture.

The results of this study indicate that loca-tional and economic parameters do influenceirrigator technology transitions to water-con-serving technologies. However, within an eco-nomic environment of relatively low real cropprice increases and perceived unconstrainedwater supplies, transition to water-conservingtechnologies is relatively slow. This means, thatin the absence of policy-induced changes, rel-atively small conserved quantities can be ex-pected to be available in the future to meetincreasing nonagricultural demands. There-fore, due to the stability of irrigation technol-ogy in the Pacific Northwest, conservation in-centive-oriented water policies, either subsidiesor institutional changes (rights to conservedwater, for example), will be needed to promoteadoption of water-conserving technologies toacquire more significant gains in agriculturalwater conservation for reallocation.

[Received November 1990; final revisionreceived July 1991.]

References

Amemiya, T., and F. Nold. "A Modified Logit Model."Rev. Econ. and Statist. 57(1975):255-57.

Bernardo, D. J., N. K. Whittlesey, K. E. Saxton, and D.L. Bassett. "An Irrigation Model for Management ofLimited Water Supplies." West. J. Agr. Econ. 12(1987):164-73.

Bonneville Power Administration (BPA). Correspon-dence received from J. R. Wilkins, agricultural econ-omist. The data were assembled by D. Hollister, RateAnalysis Section, BPA, Portland OR (August 1989).

Bromley, D. W. "Land and Water Problems: An Insti-tutional Perspective." Amer. J. Agr. Econ. 64(1982):834-44.

Bureau of Reclamation. 1988 Annual Report. U.S. De-partment of the Interior, Washington DC, 1989.

Bureau of the Census. Farm and Ranch Irrigation Survey(1988). Special Report Series AC87-RS- 1, Vol. 3, U.S.Department of Commerce, 1990.

Burness, H. S., and J. P. Quirk. "Water Laws, WaterTransfers, and Economic Efficiency." J. Law andEcon.23(1980): 111-34.

Caswell, M., and D. Zilberman. "The Choices of Irriga-tion Technologies in California." Amer. J. Agr. Econ.67(1985):224-34.

Caswell, M., E. Lichtenberg, and D. Zilberman. "TheEffects of Pricing Policies on Water Conservation andDrainage." Amer. J. Agr. Econ. 72(1990):883-90.

Chambers, R. G., and W. E. Foster. "Participation in theFarmer-Owned Reserve Program: A Discrete ChoiceModel." Amer. J. Agr. Econ. 65(1983):120-24.

Chambers, R. G., and R. E. Just. "Estimating Multiout-put Technologies." Amer. J. Agr. Econ. 71(1989):980-95.

Chan, M. W. L. "A Markovian Approach to the Studyof the Canadian Cattle Industry." Rev. Econ. and Sta-tist. 63(1981):107-16.

Domencich, T. A., and D. McFadden. Urban Travel De-mand: A Behavioral Analysis. Amsterdam: North-Holland Publishing Co., 1975.

Ellis, J. R., R. D. Lacewell, and D. R. Reneau. "EstimatedEconomic Impact from Adoption of Water-RelatedAgricultural Technology." West. J. Agr. Econ.10(1985):307-21.

Frederick, K. D. "Overview." In Scarce Water and In-stitutional Change, ed., K. D. Frederick. WashingtonDC: Resources for the Future, 1986.

Harris, T. R., and H. P. Mapp. "A Stochastic DominanceComparison of Water-Conserving Irrigation Strate-gies." Amer. J. Agr. Econ. 69(1986):298-305.

Henggeler, J. C., J. M. Sweeten, and C. W. Keese. SurgeFlow Irrigation. L-2220, Texas Agricultural Exten-sion Service, College Station TX, 1986.

High Plains Associates. Six-State High Plains OgallalaAquifer Regional Resources Study. A Report to theU.S. Department of Commerce and the High PlainsStudy Council, Austin TX, March 1982.

Homan, D. M., M. D. Skold, and D. F. Heermann. "Mon-itoring and Control of Center Pivot Systems with Mi-crocomputers." Water Resour. Bull. 23(1987):903-09.

Hornbaker, R. H., and H. P. Mapp. "A Dynamic Analysisof Water Savings from Advanced Irrigation Tech-nology." West. J. Agr. Econ. 13(1988):307-15.

Howe, C. W. "Economic, Legal, and Hydrologic Dimen-sions of Potential Interstate Water Markets." Amer.J. Agr. Econ. 67(1985):1226-30.



Irrigation Journal. Annual January issues, 1974-87. VanNuys CA: Gold Trade Publications, Inc.

Jensen, M. E. "Improving Irrigation Systems." In WaterScarcity: Impacts on Western Agriculture, eds., E. A.Engelbert with A. F. Scheuring, pp. 218-36. BerkeleyCA: University of California Press, 1984.

Just, R. E., E. Lichtenberg, and D. Zilberman. "The Ef-fects of the Feed Grain and Wheat Programs on Ir-rigation and Groundwater Depletion in Nebraska."Paper presented at the Commercial Agricultural andResource Policy (CARP) Symposium in Baltimore MD,4-5 May 1989.

Lee, J. G., J. R. Ellis, and R. D. Lacewell. "Valuation ofImproved Irrigation Efficiency from an ExhaustibleGroundwater Source." Water Resour. Bull. 21(1985):441-47.

Lichtenberg, E. "Land Quality, Irrigation Development,and Cropping Patterns in the Northern High Plains."Amer. J. Agr. Econ. 71(1989):187-94.

Lyle, W. M., and J. P. Bordovsky. "LEPA Irrigation Sys-tem Evaluation." Transactions Amer. Soc. Agr. Eng.26(1983):776-81.

MacDonnell, L. J. The Water Transfer Process as a Man-agement Option for Meeting Changing Water De-mands. Vol. 1. Natural Resources Law Center, Uni-versity of Colorado School of Law, April 1990.

Martin, W. E. "Evolving Water Institutions in an Ur-banizing West: Discussion." Amer. J. Agr. Econ.68(1986): 1152-54.

McFadden, D. "Conditional Logit Analysis of Qualita-tive Choice Behavior." In Frontiers in Econometrics,ed., P. Zarembka, pp. 105-42. New York: AcademicPress, 1974.

"Econometric Models of Probabilistic Choice."In Structural Analysis of Discrete Data with Econo-metric Applications, eds., C. F. Manski and D.McFadden. Cambridge: MIT Press, 1981.

. "Quantal Choice Analysis: A Survey." Ann. Econ.and Social Measure. 5(1976):363-90.

McGuckin, T., M. Moore, and D. Negri. Informationsummarized for the Pacific Northwest from a surveyof western irrigation water districts conducted in 1989by T. McGuckin, New Mexico State University, M.Moore and D. Negri, Resources Technology Division,Economic Research Service, U.S. Department of Ag-riculture.

Negri, D. H., and D. H. Brooks. "Determinants of Irri-gation Technology Choice." West. J. Agr. Econ.15(1990):213-23.

Negri, D., and J. J. Hanchar. "Water Conservation

Through Irrigation Technology." Agriculture Infor-mation Bulletin No. 576, Economic Research Service,U.S. Department of Agriculture, November 1989.

Parks, R. W. "Efficient Estimation of a System of Re-gression Equations When Disturbances are Both Se-rially and Contemporaneously Correlated." J. Amer.Statist. Assoc. 62(1967):500-09.

."On the Estimation of Multinomial Logit Modelsfrom Relative Frequency Data." J. Econometrics13(1980):293-303.

Pudney, S. Modelling Individual Choice: The Economet-rics of Corners, Kinks and Holes. New York: BasilBlackwell Ltd., 1989.

Rosen, M. D. "Property Rights and Public Choice inWater Districts: An Application to Water Markets."Ph.D. Dissertation, Department of Agricultural Eco-nomics, University of California, Davis, 1990.

Sweeten, J. M., and W. R. Jordan. Irrigation Water Man-agementfor the Texas High Plains: A Research Sum-mary. Tech. Rep. No. 139, Texas Water ResourcesInstitute, Texas A&M University, College Station TX,1987.

Theil, H. "On the Estimation of Relationships InvolvingQualitative Variables." Amer. J. Socio. 76(1970):103-54.

U.S. Department of Agriculture. Agricultural Prices: An-nual Summaries 1973-1988. Statistical ReportingService, Crop Reporting Board.

U.S. Geological Survey. "State Summaries of Water Sup-ply and Use." National Water Summary 1987. Wa-ter-Supply Paper 2350, 1990.

Vaux, H. J., Jr. "Economic Factors Shaping Western Wa-ter Allocation." Amer. J. Agr. Econ. 68(1986): 1135-42.

Wahl, R. W. "Promoting Increased Efficiency of FederalWater Use Through Voluntary Water Transfers." Disc.Pap. Ser., FAP87-02, Resources for the Future, Wash-ington DC, September 1987.

Weatherford, G. D., and H. M. Ingram. "Legal-Institu-tional Limitations on Water Use." In Water Scarcity:Impacts on Western Agriculture, eds., E. A. Engelbertwith A. F. Scheuring. Berkeley CA: University of Cal-ifornia Press, 1984.

Wyatt, W. A. The Cross Section. High Plains Under-ground Water Conservation District No. 1, LubbockTX, 1990-91.

Zellner, A., and T. H. Lee. "Joint Estimation of Rela-tionships Involving Discrete Random Variables."Econometrica 33(1965):382-94.

206 December 1991

Documents

Water Conservation Potential from Irrigation Technology Transitions in the Pacific Northwest