development of groundwater markets in china: a glimpse ...€¦ · 1. introduction in recent years...

World Development Vol. 36, No. 4, pp. 706–726, 2008� 2007 Elsevier Ltd. All rights reserved

0305-750X/$ - see front matter

www.elsevier.com/locate/worlddevdoi:10.1016/j.worlddev.2007.04.012

Development of Groundwater Markets in China:

A Glimpse into Progress to Date

LIJUAN ZHANG, JINXIA WANG, JIKUN HUANGCenter for Chinese Agricultural Policy, Institute of Geographical Sciences and

Natural Resources Research, Chinese Academy of Sciences, Beijing, China

and

SCOTT ROZELLE *

Shorenstein Asia Pacific Research Center, Freeman SpogliInstitute for International Studies, Stanford University, USA

Summary. — The overall goal of the paper is to better understand the development of groundwatermarkets in northern China. Field survey shows that groundwater markets in northern China haveemerged and are developing rapidly. Developing in a number of ways that make them appear some-what similar to markets that are found in South Asia, groundwater markets in northern China alsodiffer by the impersonality and case bases. The privatization of tubewells is one of the most impor-tant driving factors encouraging the development of groundwater markets. Increasing water andland scarcity are also major determinants that induce the development of groundwater markets.� 2007 Elsevier Ltd. All rights reserved.

Key words — groundwater markets, determinants, equitable, northern China, South Asia

* The authors would like to thank Steve Beare, Amelia

Blanke, Anna Heaney, Qiuqiong Huang, Mark Giord-

ano, Tushaar Shah, and Aditi Mukherji for their insights

and helpful suggestions. We also would like to thank the

useful comments of four anonymous reviewers. We ac-

knowledge financial support from the Knowledge Inno-

vation Program of the Chinese Academy of Sciences

(KSCX2-YW-N-039), the National Natural Sciences

Foundation (70021001) in China, the International

Water Management Institute, the Food and Agriculture

Organization of the United Nations, and the Compre-

hensive Assessment of Water Management in Agricul-

ture. Scott Rozelle is a member of the Giannini

Foundation. Final revision accepted: April 12, 2007.

1. INTRODUCTION

In recent years groundwater has begun toplay an increasingly important role in irrigationin China, especially in northern China whereper capita water availability is less than one-twentieth of the world average (Liu & He,1996). Although surface water dominated Chi-na’s irrigation development in the 1950s and1960s, since the end of 1960s groundwatergradually has become the primary source ofirrigation water. According to official statistics,during 1965–2003, the number of tubewells in-creased from 0.2 million to 4.7 million (Minis-try of Water Resources & Nanjing WaterInstitute, 2004; Ministry of Water Resources,2003). Nearly all of the tubewells (95%) are innorthern China. Today, these tubewells provideabout 68% of the total irrigation water innorthern China (Wang, Huang, Blanke,Huang, & Rozelle, 2007).

The rise of groundwater in China not onlyfueled an expansion of sown area and rising

706

production (Huang, Rozelle, Wang, & Huang,2005), as the reliance on groundwater has in-creased, but also China’s groundwater econ-omy has become characterized by a growingwater crisis (Wang et al., 2007). Using datafrom a field survey conducted in six provincesin northern China, Wang, Huang, Huang,

DEVELOPMENT OF GROUNDWATER MARKETS IN CHINA 707

and Rozelle (2006) found that from 1995 to2004, the water table is falling sharply in morethan half of the regions in northern China. As aresult, there has been a concomitant rise in thecost of sinking a tubewell. In parts of the NorthChina Plain the shallow water table has beendropping at a rate of more than one meterper year (Ministry of Water Resources, 2002).The deep water table has fallen faster, decliningat a rate of more than two meters per year insome areas (Wang, Huang, & Rozelle, 2005).The cost of sinking and operating a tubewellhas at least doubled in many parts of northernChina (Huang et al., 2005).

During the 1990s with the falling of ground-water table, the ownership of tubewells also hasbegun to evolve. Before the rural reforms in the1980s, most tubewells were owned and oper-ated by the collective. For a variety of reasons,including the decline in the strength of the col-lective and the increased freedom of individualsto invest in their own farms, soon after the eco-nomic reforms began in the early 1980s theownership of China’s tubewell began to shiftsharply from collective to private (Shah, Giord-ano, & Wang, 2004; Wang et al., 2005). Thenumber of private tubewells increased from al-most nothing in the 1970s to nearly 40% by1990. The shift to private tubewell ownershipcontinued during the 1990s and beyond. Forexample, in 1995 collective ownership ac-counted for 58% of tubewells in the averagegroundwater using village in northern China(Wang et al., 2007). By 2004, private tubewellsrose to 70%.

While the rise of private tubewells has beenshown to lead to more efficient use of water,higher levels of irrigated area and more com-plex cropping systems (Wang et al., 2005,2006), it has also made access to irrigation wateran increasingly important issue. During the So-cialist era (1950–70s) when local leaders were incharge of allocating groundwater in almost allvillages, the equitable distribution of ground-water was not an issue. However, as tubewellshave been installed and begun to be operatedby private individuals, and as tubewells havebegun to be sunk to deeper levels (making thereal price of water rise), concern has arisen thatnot all farmers may have equal access togroundwater (Meinzen-Dick, 1996). It is possi-ble that the farmers that have access to the cap-ital are the ones that are more likely to sink andmanage tubewells. It is also possible that be-cause of this, these better endowed farmershave much better access to water than those

without tubewells. If so, it is possible that partof the gains from increased efficiency that ac-crues from the rise of private tubewell owner-ship is being offset by rising inequities in thedistribution of water and the associated gains.

The rise of private tubewells, however, doesnot have to lead to inequities if groundwatermarkets emerge and function well. While littlehas been written on groundwater markets inChina, outside of China groundwater marketshave long existed and recently have attractedthe attention of researchers. For example, mar-kets in groundwater have been found in manyparts of South Asia. In 1975 a World Bankstudy in Pakistan reported that nearly 30% oftubewell owners sold part of their pumpage toother farmers (Shah, 2000). In the early 1990sPant (1991) found that 86% of the householdsin eastern Uttar Pradesh purchased water forirrigation; in central and western Uttar Pradesh65% of farm households purchased water.More recently studies in Pakistan by Strosserand Meinzen-Dick (1994) and Meinzen-Dick(1996) have found groundwater markets perva-sive. While the analysis of many issues is com-plicated and the findings of many studies arecontroversial (meaning more study is neededon the management of groundwater markets),the South Asian experience has shown thatgroundwater markets in many places have pro-vided opportunities for the farmers withouttubewells to get access to water (Mukherji,2004; Shah, 1993; Sharma & Sharma, 2004;Strosser & Meinzen-Dick, 1994).

Despite the observations by field workersregarding the similarities between the rise ofgroundwater markets in China and those ofSouth Asia, almost no empirical studies havebeen done on the development of China’sgroundwater markets. In fact, searching of theliterature has found that there is almost no ref-erence in any work on groundwater markets ineither Chinese or English. Despite the absenceof research, policy makers and scholars havebegun to raise a series of questions. How preva-lent are groundwater markets in northernChina? More specifically, what is the propor-tion of the tubewells that participate in sellingwater? How much of their water do they sell?What are the characteristics of groundwatermarkets in northern China? Who are the buy-ers? Finally, why is that we observe water mar-kets in some villages but not in others? In otherwords, what are the determinants of groundwa-ter markets in northern China and what aretheir effects on the equity of access to water?

708 WORLD DEVELOPMENT

The overall goal of this paper is to answerthese questions to develop a better understand-ing of the development of groundwater marketsin northern China. To do so, the paper focuseson getting the data right and providing a profileof groundwater markets and their determinantsin northern China. To meet this goal, this paperhas four specific objectives. First, the paper de-scribes the evolution of groundwater markets innorthern China. Second, the paper explorestheir characteristics. In doing so, this papercompares China’s groundwater markets tothose that have emerged in South Asia. Third,the paper measures the determinants ofgroundwater markets in northern China andtries to understand why they have emerged insome villages but not in others. Finally, the pa-per seeks to understand what types of farmersare selling water and what types of farmersare buying water in an effort to understandthe equity implications of emerging ground-water markets. 1

To pursue objectives, the paper is dividedinto six sections. The first section introducesthe data. The second section uses the data tocompare the characteristics of groundwatermarkets in northern China with those in SouthAsia. The third and fourth sections descrip-tively analyze the determinants of the develop-ment of groundwater markets and present theresults of the multivariate analysis on the deter-minants of groundwater markets. The fifth sec-tion discusses whether or not groundwatermarkets are tending to help or hurt the poor.The final section concludes.

2. DATA

The analysis is based on two sets of data fromnorthern China. 2 The first survey, called theChina Water Institutions and Management sur-vey (or CWIM), includes two rounds of datacollection in 2001 and 2004. In the CWIM sur-vey, enumerators collected data and informa-tion from 338 households, 110 tubewellowners, and 68 canal managers in 80 villagesin three provinces (Hebei, Henan, and NingxiaProvinces). Since there is almost no irrigationby groundwater in the Ningxia Province sam-ple, in this paper, we only use the data from He-bei and Henan provinces. Located in the NorthChina Plain, the two provinces face seriouswater shortages and have the highest extent ofgroundwater irrigation (about 78% of irrigatedarea is from groundwater). Villages in Hebei

were chosen from counties near the coast, nearthe mountains, and in the central region be-tween the mountains and the coast. In Henan,villages were chosen from counties borderingthe Yellow River and from counties in irrigationdistricts varying distances from the YellowRiver. The most prominent feature of this sur-vey is that the study team was able to collectinformation on water allocation and water salesfrom tubewell owners and managers—boththose that sold water and those that did not.

The second set of data, called the NorthChina Water Resource Survey (or NCWRS),was also conducted in 2004. This survey of vil-lage leaders from 400 regionally representativevillages in Inner Mongolia, Hebei, Henan, Lia-oning, Shaanxi, and Shanxi provinces used anextended version of the community-level villageinstrument of the CWIM survey. Using a strat-ified random sampling strategy for the purposeof generating a sample representative of north-ern China, counties in each of regionally repre-sentative sample provinces were sorted into oneof four water scarcity categories: very scarce,somewhat scarce, normal, and mountainous/desert. 3 Two townships within each countyand four villages within each township alsowere randomly selected. Although the informa-tion on water sales in the NCWRS is somewhatless rich than the CWIM survey (given thatenumerators only surveyed village leaders andnot tubewell owners and households directly)by using the NCWRS survey the study teamwas able to generate the representative pointestimates of the prevalence of water marketsacross northern China.

The scopes of the two surveys were quitebroad. Each of the survey questionnaires in-cluded more than 10 sections. Among the sec-tions, there were those that focused on thevillage’s resource base (both the scarcity ofwater and the amount of cultivated land), theevolution of the ownership of tubewells, the vil-lage’s basic socio-economic conditions, andgovernment policies and regulations. Privatetubewells consist of two types of tubewells—individual and shareholding tubewells. If atubewell belongs to a single individual or a fam-ily, it is called an individual tubewell. In manycases, however, a tubewell is owned by a groupof individuals. Since in many of the groups, theindividual members are assigned shares thatindicate the investment stake that each memberowns in the tubewell entity, the groups are of-ten called shareholding groups and their tube-wells are called shareholding tubewells.


In addition, there was a section that focusedspecifically on groundwater markets. Ground-water markets are defined as localized, commu-nity-level arrangements through which ownersof tubewells sell pump irrigation services toother farmers of the village and neighboring vil-lages (i.e., they sell water to other farmers fromtheir wells for use on crops). This paper is onlygoing to examine ‘‘private’’ water markets orthe nature of groundwater markets that arebeing driven by individuals and groups of indi-viduals that sink tubewells. In adopting such adefinition, we are assuming that when villageleaders (the collective) provide water to villag-ers, it is being done under non-market condi-tions and is not a groundwater markettransaction.

For the section on groundwater markets, thestudy team designed many questions to elicitthe nature of groundwater markets and mea-sured their development. Enumerators askedvillage leaders if there were tubewell ownersthat sold water to farm households that didnot own tubewells. The survey also tabulatedthe number of tubewells in a village from whichwater was sold by its owner. Detailed informa-tion was elicited on the water sales activities ofeach village’s typical tubewell, including esti-mates of the total volume of water withdrawn,the volume water that was sold, and the infor-mation on how water-buying households paidfor the water that they bought from thewater-selling farmers. A series of questions wereasked of the village leaders about governmentregulations governing the price of water soldfrom tubewells. Finally, several questions fo-cused on asking water-selling households (ortubewell owners) if they sold water to farmersinside or outside of the village. 4

Both surveys were designed in ways that al-lowed the construction of data with an inter-temporal component. The surveys collecteddata on most of the variables for more thantwo years. For example, the CWIM survey cov-ered four periods: 1990, 1995, 2001, and 2004. 5

The NCWRS survey covered 1995 and 2004.The CWIM survey was conducted in tworounds. The first round was carried out in2001 which collected data and information on1990, 1995, and 2001. The second round ofthe survey was conducted at the end of 2004and collected data and information for 2004.In this sense the information from 2001 and2004 is the true panel data. The NCWRS sur-vey work was conducted in 2004 and collecteddata and information in 1995 and 2004.

3. GROUNDWATER MARKETS WITHCHINESE CHARACTERISTICS

This section measures the degree of the devel-opment of groundwater markets in terms ofboth breadth and depth as well as describestheir characteristics. According to the definitionof Shah (2000), ‘‘Breadth can be conceptuallydefined as the proportion of the farming com-munity that is participating in water trade, asbuyers or sellers or both; depth can be concep-tually defined as the quantitative significance ofwater transactions in the economies of individ-ual farm households of a community.’’ In therest of the paper the breadth of groundwatermarkets is measured by two indicators. Oneindicator is the share of villages that have anydegree of groundwater market activity. The sec-ond indicator is the share of tubewells fromwhich the tubewell owner is selling water towater-buying households. Depth is measuredby the ‘‘share of the volume of water’’ sold towater-buying households that is pumped fromtubewells that are selling water.

When using breadth indicators, groundwatermarkets have developed quickly in northernChina. According to the NCWRS survey, in1995 groundwater markets had emerged in only9% of the sample villages (Table 1, column 1,row 1). However, by 2004 there were ground-water markets in 44% of the villages (column2). During the same period the share of tube-wells from which owners sold water also in-creased. In 1995 water was sold from only 5%of tubewells; by 2004, however, this number in-creased to 18% (row 2). In addition, when usingindicators of the depth of groundwater mar-kets, the CWIM survey shows that by 2004,groundwater market activities were dominatingthe tubewell pumping activities of those farm-ers-cum-tubewell owners that were selling water(row 3). 6

Although there has been a lot more attentiongiven to the study of groundwater markets inSouth Asia (because groundwater markets havetraditionally been quite widespread), the datashow that China is catching up quickly. Forexample, a number of studies suggest thatgroundwater markets have become quite perva-sive in Pakistan (Meinzen-Dick, 1996; Strosser& Meinzen-Dick, 1994). These studies indicatethat 30–60% of tubewell owners in Pakistan sellwater. In India and other neighboring coun-tries, Shah (2000) showed that when tubewellowners sold water, they sold from 40% to70% of the volume of water they had pumped.

Table 1. Development of breadth and depth ofgroundwater markets in China, 1995 and 2004

1995 2004

BreadthShare of villages having

groundwater markets (%)9 44

Share of tubewells selling water (%) 5 18

DepthShare of water sold (%)

[conditional on tubewellowner selling water]

80 77

Data source: Data in row 1 and row 2 are from authors’survey of 68 randomly selected villages in four provinces(Hebei, Henan, Shanxi, and Shaanxi) of the NCWRS;Data in row 3 are from the authors’ survey of 50 ran-domly selected tubewells in two provinces (Hebei andHenan) of CWIM. We do not use data from all of thesample villages of the two surveys since the informationin the table is conditioned on villages that use ground-water to irrigate and that have private tubewells.


Hence, while these numbers may not be exactlycomparable, if assuming that the estimates arecorrect (i.e., water is being sold from 18% oftubewells in northern China; when water-sellinghouseholds sell water, they are selling 77% ofthe water from their tubewells), the develop-ment of groundwater markets in northernChina is approaching levels that are beingobserved elsewhere in the world.

(a) Characteristics of groundwater markets innorthern China

Although groundwater markets in northernChina have evolved more recently, there areat least three characteristics which appear tobe shared by groundwater markets in northernChina and South Asia. First, almost all ground-water markets in both places are informal.According to Shah (1993), a water market isinformal when transactions between water-sell-ing and water-buying households are donewithout legal sanction. In other words, farmersbuy and sell water without a contract and theiroral commitments cannot be adjudicated in acourt of law. According to the data, there werezero written contracts covering water salesamong agents in northern China; the same istrue according to the literature from India.

Second, groundwater markets in both north-ern China and South Asia are almost alwayslocalized. According to Shah (1993), the local-ized nature of water markets is almost univer-sal. In the survey data in China, water

transactions also are mostly limited to water-selling and water-buying households that liveand work in the same village. In fact, only 6%of water-selling tubewell owners (and a smallershare of the volume of water that they pump)sell water to farmers in other villages.

Third, groundwater markets in both north-ern China and South Asia are largely unregu-lated. In Shah (1993) the word unregulatedmeans the government exercises no direct influ-ence on the functioning of the market. Shah(1993) found little evidence of any interventionby any level of government in India. Based onthe NCWRS survey data, there were only for-mal regulations on the books about any aspectof groundwater markets in less than 25% of vil-lages (e.g., a price ceiling on the amount that awater-selling household can charge). Althoughsomewhat higher than the case of India, duringour field work and during interviews with tube-well owners, enumerators almost have neverencountered a case in which the tubewell ownerwas constrained by a government regulation;village leaders and tubewell operators almostnever were aware that there was any attemptby upper level officials to influence the func-tioning of water selling and buying. 7

While there are a number of similarities, itappears as if the different environments withinwhich groundwater markets have evolved innorthern China and South Asia have producedseveral differences in the nature of groundwatermarkets. First, although field work in many vil-lages has found evidence of impersonal mar-kets, groundwater markets in parts of SouthAsia appear not to be fully impersonalized.For example, Shah (1993) stated that transac-tions between water-buying and water-sellinghouseholds were impersonal in many cases. InIndia, this means that water-selling householdsin many villages do not distinguish among var-ious buyers in terms of price at which they sellwater and the quality of service provided. How-ever, other studies find that groundwater mar-kets can be personalized. For example, studiesin Pakistan (Jacoby, Murgai, & Rehman,2004), Bihar, India (Wood, 1995), and other re-gions in South Asia (Ballabh, Choudhary, Pan-dey, & Mishra, 2002; Pant, 2003) reported thatsellers charged some buyers one price and otherbuyers another price. In other words, afterobserving what ‘‘kind of a person’’ a potentialbuyer is, the price of water was set.

In contrast, in the case of northern Chinagroundwater markets are almost fully imper-sonal. Based on our survey of 30 village leaders,


we found that within villages, only 7% of water-selling tubewell owners charge different pricesfor different types of buyers. In addition, inour survey of the tubewell owners, not one re-ported that they charged different prices for dif-ferent types of buyers.

Second, the patterns of payment withingroundwater markets in China and South Asiasometimes are different. In South Asia water-buying households often provide labor or offera share of their crop’s harvest in exchange forwater (Shah, 2000). In northern China, how-ever, water sold in groundwater markets is al-most always paid for on a cash basis. Suchdifferences may be primarily related to the dif-ferent land tenure arrangements that dominateboth countries. For example, in China the own-ership of cultivated land belongs to village col-lectives. Since the early 1980s, with theimplementation of household productionresponsibility in rural China, cultivated landwas allocated relatively evenly to each farmerwithin a village. As such, it implies that eachfarmer has land use right though they have noland ownership. However, in South Asia, landownership is private and many farmers haveno land to be used for their production. These

Table 2. Characteristics of groundwater mark

Village characteristicsPer capita income (yuan)

Groundwater marketsLevel of development (breadth/depth)

Share of private tubewells selling water (%)Share of water sold (%)

Policy interventions

Share of villages having subsidy (%)Share of villages receiving bank loans (%)Share of villages requiring well-drilling permits (%

Nature of groundwater markets

Informal (share of villages not having contract foLocalized (share of water-selling tubewells sellingto farmers in the villages, %)Unregulated (share of villages not having price ceImpersonal (share of water-selling tubewells ownesame prices for all types of buyers, %)Patterns of payment (share of water-selling tubewpay in cash for selling water %)

Data sources: The authors’ survey of 68 randomly selectedShaanxi) of the NCWRS and 13 randomly selected tubeweuse data from all of the sample villages of the two surveys sithat use groundwater to irrigate and that have private tuba Poor villages are those villages in which per capita incomevillages in which per capita income is more than 1625 yuan

landless farmers have to rent land from landowners or provide their labor to land owners.Therefore, in the same way that they pay theirrent with a share of the labor (i.e., through asharecropping contract), water-buying house-holds in South Asia also often provide laboror a share of their crop’s harvest in exchangefor water. In contrast, water-buying householdsin China do not exchange their labor for waterin such a way and in all cases in our sample(100%) pay cash for water.

Another important difference between Chinaand South Asia is the way in which electricity ispriced. This, too, may have a major impact onthe way groundwater markets work. For in-stance, in many Indian states, electricity ispriced on a flat rate basis and this does not al-ways reflect the scarcity value of water. InChina, however, electricity meters are in place,electricity prices are mostly at market rates andthe cost of pumping (and consequently the priceof water) reflects mostly the scarcity value ofwater (because it reflects the depth from whichthe water is pumped). Also, in India rural elec-trification is poor and, hence, many farmersdepend on diesel driven pumps and this maycreate a different configuration of groundwater

ets in poor and rich villages in China, 2004

Poor villagesa Rich villagesa

1008 2561

44 3774 91

5 47 7

) 66 70

r selling water, %) 100 100water 87 100

ilings, %) 80 74rs that charge 95 100

ells that 100 100

villages in four provinces (Hebei, Henan, Shanxi, andlls selling water in Hebei province of CWIM. We do notnce the information in the table is conditioned on villagesewells.

is less than 1625 yuan per capita; rich villages are thoseper capita.


markets than would appear when there are elec-tric pumps.

There may be one final difference betweenChina and India. Whereas, there are manyinstitutional aspects of groundwater marketsthat differ between villages that are relativelypoor and villages that are relatively rich inIndia, the same is not true in China. In fact,according to our data, most institutional fea-tures do not differ in a statistically significantway between villages that are relatively well-off in China and those that are relatively poor.We show this by dividing the sample into twogroups—according to the village’s per capita in-come. Our data show that income differs dra-matically between rich and poor villages;average income in the richer villages was250% higher than that in the poorer village(Table 2, row 1). Despite the large differencesin income, there is almost no difference (statis-tically) in the development of groundwatermarkets, the nature of policy interventions, orthe characteristics of groundwater markets be-tween these two types of villages. For example,in the rich villages 37% of tubewells sell water.This share is almost same as that in the poorvillages (in which 44% of tubewells sellwater—row 2). Therefore, there is evidence thatthe development of groundwater markets inboth rich and poor villages is more or less thesame. In addition, the share of villages havingbank loans or the share of village that are beingaffected by other policy interventions is alsosame in both poor and rich villages (rows 4–6). Finally, in both rich and poor villages,groundwater markets are unregulated, infor-mal, localized and impersonal; payment forwater (or the way in which transactions are set-tled) is also the same (rows 7–11).

4. GROUNDWATER MARKETS,TUBEWELL OWNERSHIP, AND

RESOURCE SCARCITY

In the small number of international papersthat have sought to understand the determi-nants of groundwater markets, a number offactors arise consistently and can be used as abasis for generating hypotheses that the papercan test using the survey data from northernChina. For example, Shah (1993) descriptivelyshowed that the availability of water resources,the scale of irrigation technology, and the ex-tent of land fragmentation were correlated withthe rise of groundwater markets. Strosser and

Meinzen-Dick (1994) set up a theoreticalframework that posits (among other factors)that the depth of the groundwater table andthe population density of a community areimportant factors affecting groundwater mar-kets. Shah (1993) and many others, such asMukherji (2004), also indicated the importanceof policy intervention in promoting or con-straining the development of groundwater mar-kets. Therefore, we would expect that if therewere regulations, it would constrain the expan-sion of the number of tubewells. In the case ofother policy intervention variables (e.g., theexistence of bank loan and grant programs thatare targeting those that invest in pumps andtubewells), when there are government grantand loan programs one would expect moretubewells and, as such, greater groundwatermarket activity. In other words, the observa-tions of the researchers working on SouthAsia’s groundwater markets suggest thatgroundwater markets are arising at least in partin response to the nature of the technologyneeded for sinking a well, the degree of resourcescarcity—both for land and water and policy. Ifthese observations and conjectures are pickingup more general relationships, then based onthe relationships a set of testable hypothesescan be generated: When the cost of sinking awell rises (either due to the falling groundwatertable or the relative competitiveness of largertubewells/pump sets) or when the attractivenessof sinking a well at a given cost declines (due tothe fact that a farmer may have an increasinglysmall parcel of land that is not able to utilizethe entire command area of a tubewell invest-ment, or due to a policy intervention by thegovernment), groundwater markets can be ex-pected to emerge.

According to descriptive statistics based onthe survey data from northern China, empiricalevidence for the hypotheses is present. Forexample, the data indicate that the develop-ment of groundwater markets maybe relatedwith water resource scarcity (Table 3). Whenthe water table falls in the NCWRS sample vil-lages over time (from 28 to 38 m—column 3),the share of tubewells from which water isbeing sold is higher (column 1). When dividingthe villages in the sample by the share of tube-wells from which water is sold into four groups,there is a positive correlation between theamount of groundwater market activity andthe level of the groundwater table (columns 1and 3, rows 3–6). 8 Likewise, when dividingthe tubewells in the sample by the share of

Table 3. Relationship between development (breadth) of groundwater markets and tubewell ownership/resourceendowment in China, 1995 and 2004

Share of tubewells selling water (%) Tubewell ownership Water scarcity Land scarcity

Share of privatetubewells (%)

Groundwatertable (m)

Per capita arableland (ha)

Grouped by yeara

1995 5 50 28 0.122004 18 81 38 0.10Grouped by share of tubewells selling waterb

0 0 68 28 0.110–30 12 46 45 0.1230–90 57 70 48 0.1190–100 100 100 48 0.09

Data source: The authors’ survey in 68 randomly selected villages in four provinces (Hebei, Henan, Shanxi, andShaanxi) of NCWRS. We do not use data from all of the sample villages of the NCWRS survey since the informationin the table is conditioned on villages that use groundwater to irrigate and that have private tubewells.a The number of observations used for each row in rows 1 and 2 is 68.b The number of observations used for each row in rows 3–6 is n = 100 (row 3); n = 10 (row 4); n = 8 (row 5); andn = 18 (row 6). Data are averages for two sample years.


water sold in three groups, there is also a posi-tive relationship between the amount ofgroundwater market activity and the level ofthe groundwater table (Table 4, columns 1and 4). One explanation of these trends is thatwhen the groundwater table is lower, the costof sinking a tubewell is higher, which couldkeep some farmers from investing in theirown tubewells even though they have a high de-mand for irrigation services. Alternatively(although mainly in a relative sense), it couldbe that the lower the groundwater table, thelarger is the size of the optimal tubewell/pumpset. In villages with larger tubewells/pump sets,other factors (including land size) held con-stant, there is less of a need for all farmers tohave their own tubewells.

Likewise, the data provide similar supportfor the hypotheses when looking at the relation-ship between groundwater activity and land

Table 4. Relationship between development (depth) of gendowment in

Share of water sold(%)

Tubewell owne

Share of individualtubewells (%)

Sharetu

Grouped by share of water sold a

0 0 190–90 48 4490–100 97 100

Data source: The authors’ survey of 50 randomly selected tWe do not use data from all of the sample villages of tconditioned on villages that use groundwater to irrigate ana The number of observations used for each row in rows 1

scarcity (Table 3, columns 1 and 4). During1995–2004 the average size of land per capitafor the sample villages fell from 0.12 to0.10 ha (rows 1 and 2). Coupled with the ob-served rise in the share of tubewells that areselling water, the descriptive data are consistentwith the idea that when farm size gets smaller,households have less of a need to invest in theirown tubewells. As a result, this could be onereason behind the rise in groundwater markets.The same trends appear when grouping obser-vations either by the share of tubewells fromwhich water is sold (rows 3–6) or by the shareof water sold (Table 4, columns 1 and 5).

Trips to the field and discussions with localofficials and water users, as well as the surveydata, also raise the prospect of one factor un-ique to China that may be behind the emer-gence of groundwater markets in the sample.According to the data, the private ownership

roundwater markets and tubewell ownership/resourceChina, 2001

rship Water scarcity Land scarcity

of shareholdingbewells (%)

Groundwatertable in 1995 (m)

Per capita arableland (ha)

81 13.6 0.12056 11.1 0.0910 17.6 0.089

ubewells in two provinces (Hebei and Henan) of CWIM.he CWIM survey since the information in the table isd that have private tubewells.–3 is n = 32 (row 1); n = 9 (row 2); and n = 9 (row 3).


of tubewells is correlated strongly with thedevelopment of groundwater markets. Ground-water market activity is higher as the share ofprivate wells has risen over time (from 50% to81%—Table 3, columns 1 and 2, rows 1 and2). Likewise, when observations are groupedaccording to the water sales activity of thetubewells, the share of private wells also risessharply (from 68% to 100%—rows 3–6). Specif-ically, when the share of individual tubewellsincreases, the share of water sold is higher(Table 4, columns 1 and 2). Interviews withvillage leaders and tubewell owners and manag-ers revealed that the main force driving thiscorrelation appears to be the incentives thatprivate tubewell owners face which encouragethem to produce earnings from their tubewellinvestments. Hence, in studies of China, anadditional testable hypothesis arises fromdescriptive statistics and observations in thefield: Groundwater market activities rise asthe share of private wells in a village increases.

5. DETERMINANTS OF GROUNDWATERMARKETS—MULTIVARIATE ANALYSIS

This section seeks to test more rigorouslythe determinants of groundwater markets.The paper is interested in identifying the factorsthat explain why some villages have groundwa-ter markets and others do not; this analysis isneeded because the findings will help us to bet-ter understand the forces that are creating theswell in groundwater market activity. This isimportant since it may help us predict as Chi-na’s villages confront the rising economic andenvironmental pressures of the nation’s devel-opment process (e.g., forces, such as the steadyshift towards private ownership of productiveassets, and increasing resource scarcity),whether or not institutions will emerge that willallow farm households to gain access to water,one of the most critical resources that they needfor production. To analyze the determinants ofgroundwater markets, the first part of this sec-tion introduces the methodology. In the nextpart the results will be reported.

(a) Methodology

Based on the descriptive analysis above andwork on groundwater markets in other coun-tries, the following econometric model is pro-posed to analyze the determinants of thebreadth of groundwater markets:

T jt ¼ aþ bOjt þ cW jt þ dLjt þ /Zjt þ ejt: ð1Þ

In Eqn. (1) Tjt represents the share of tubewellsselling water in village j in year t. The variableson the right hand side of Eqn. (1) are those thatexplain differences in the breadth of groundwa-ter markets (or the share of tubewells that sellwater) among villages and over time. The firstvariable, Ojt, represents the change of tubewellownership and is measured as the share of pri-vate tubewells in village j. The two variables,Wjt and Ljt, measure the resource endowmentsof the village (both its water and land re-sources) and are included to measure if increas-ing resource scarcity (or the cost of using theresource) helps induce the development ofgroundwater markets. Specifically, the waterresources variable (Wjt) is measured as the levelof the groundwater table. The degree of landscarcity (Ljt) is measured as cultivated landper capita.

In Eqn. (1) a set of control variables are in-cluded. The first set of control variables in-cludes three policy variables which areincluded to assess the effects of policy on thedevelopment of groundwater markets. The firstvariable, fiscal subsidies for tubewells, is a dum-my variable equal to one if there was a programof fiscal investment in the village that targetedtubewell construction (and zero otherwise).This government program, run by the local Bu-reau of Water Resources, is primarily targetedat individuals. A similar variable, bank loansfor tubewells, is included to control for whetheror not there was a loan program through Chi-na’s banks that gives preferential access tolow interest rate loans for investing in tube-wells. Unlike the fiscal subsidy program, mostbank loan programs target local villages andleaders; the loans are typically supposed to beused to be invested in collective wells. A finalvariable, well-drilling regulations, controls forthe presence of local regulations that would,ceteris paribus, slow down the construction oftubewells. 9 Although there is no explicit gov-ernment regulatory policy to encourage collec-tive tubewells at the expense of privatetubewells (or vice versa), without the supportof government, it likely that such regulations,if present (and enforced), would have a greatereffect on slowing down investment in privatetubewells. The hypotheses are that any govern-ment program that encourages (discourages)private tubewells relative to collective wells willencourage (discourage) the development ofgroundwater markets.


In explaining the development of the breadthof groundwater markets, the adoption of irriga-tion water conveyance technologies in the vil-lage is included as a way to control for thecost and efficiency of delivering water fromthe tubewell to the field. This variable is mea-sured as a dummy variable, equaling one ifthe village had adopted conveyance-inducingtechnology, such as surface (white dragons) orunderground pipe networks. It is thought thatif the conveyance of water is easier (and moreefficient), water markets will emerge more read-ily. Finally, several other factors are also con-trolled for. For example, village income percapita is included as a control for the village’ssocio-economic conditions. The symbols a, b,c, d, and / are parameters to be estimatedand ejt is the error term.

To analyze the determinants of developmentof the depth of groundwater markets, the follow-ing econometric model has been specified:

Mj ¼ aþ bOj þ cW j þ dLj þ /Zj þ ej; ð2Þwhere Mj represents the share of water sold fortubewell j. While the basic structure of Eqn. (2)is the same as Eqn. (1), because of the nature ofthe dependent variable (and differences in thesample—the breadth of water markets analysisuses village-level data; the depth of water mar-kets analysis uses tubewell-level data), the spec-ification is slightly modified. The first variable,Oj, represents the ownership of tubewell j; ifthe tubewell is owned by an individual (a singlefamily), it equals to 1; otherwise, the tubewell isowned by a group of individuals and equals 0.Since the demand by the individual farm house-hold for water from its own well is almost bydefinition less than the members of the share-holding group, a positive sign on the coefficientof the ownership variable would be expected(since there would be more of the excess capac-ity available for sale).

The relative scarcity of water and land alsomight be expected to affect the amount of watersold to other farmers. To control for water scar-city, the paper includes the variable Wj which ismeasured by the depth of the groundwater table.Since the cost of pumping (and, hence, the priceat which water can be sold to farmers) is directlyrelated to the depth of the well, a negative coef-ficient on the depth of the groundwater tablevariable would be expected. However, it is pos-sible that the sign is positive if as the groundwa-ter table falls (or is lower in level), it is moredifficult for farmers to sink their own wells andso groundwater markets emerge to meet the de-

mand. The analyst also needs to be concernedabout the endogeneity of such a variable sincethe development of groundwater markets mayinfluence the level of the groundwater table.Consequently, in the analysis the paper mea-sures Wj as the groundwater table of the villagein 1995, a time before the sample and a periodbefore the takeoff of groundwater markets.

In the same spirit, the paper includes a vari-able Lj to control for the degree of land scarcity(which is measured as cultivated land per capitain the village in which tubewell j is located). Thehypothesis here is that when land per capita islower, the benefits of investing in one’s ownwell fall and increase the demand for watermarkets.

In Eqn. (2), as in Eqn. (1), a set of three policyvariables and a set of control variables also areincluded. The first variable equals one if thetubewell owner (or the shareholding group) re-ceived a fiscally subsidized rebate after investingin the well (and was zero if it was fully self-fi-nanced). The second policy variable equals oneif the tubewell owner received a bank loan aspart of the investment financing package of thewell (and zero if not). Finally, a third policy var-iable equals one if the tubewell owner was issueda well-drilling permission certificate before thewell was drilled, and zero otherwise. It wouldbe expected that any policy that facilitates (dis-courages) the investment in tubewells would in-crease (decrease) the size and depth of the welland provide individuals with more (less) excesscapacity from which they are able to sell water.The definitions (and expected signs) of the othercontrol variables (village income per capita;dummy of adopting water delivery pipes) arethe same as those in Eqn. (1).

(b) Results

When estimating the determinants of thedevelopment of the breadth and depth ofgroundwater markets, a Tobit model is used. 10

This estimation strategy is needed since thedependent variables in both Eqns. (1) and (2)are in ‘‘share’’ form (i.e., between 0 and 1).There are also a number of villages (tubewells)in which the value of the dependent variable iszero. Using ordinary least squares approach(OLS) might generate a biased set of estimatedparameters.

In addition, we can also address several po-tential statistical problems that might arise inthe estimation of Eqns. (1) and (2). 11 Specifi-cally, two possible multicollinearity problems


come from using either the three policy inter-vention variables and/or the conveyance tech-nology variable along with the actual share ofprivate tubewells in the same regression (Eqn.(1)). There are two ways to assess whether ornot this is a problem. First, we use regressiondiagnostics to generate a condition numberfor the full model. Since the condition numberis only 39, we can conclude that there is no evi-dence of multicollinearity in a statistical sense.We can also assess the impact of ‘‘includingthe variables’’ by re-estimating Eqn. (1), butwithout including the policy intervention vari-ables (Table 5, column 2). As can be seen (com-paring the results of all of the other variables incolumns 1 and 2), the results vary little. Thesame is true when we re-estimate the equationwithout using conveyance technology (Table5, column 3) or without including conveyancetechnology and the policy intervention vari-ables (Table 5, column 4—that is, there arefew changes to the coefficients of interest). Weadopt the same strategy for assessing therobustness of the results of Eqn. (2). While wecan report that there is little change to the coef-ficients, due to space constraints, we do not in-clude the detailed findings in a paper.Therefore, we believe that any potential multi-collinearity problem in Eqns. (1) and (2) isnot serious.

(i) Determinants of the breadth of groundwatermarkets

In estimating Eqn. (1) with the survey data,the econometric estimation performs well(Table 5, column 1). Most of the coefficientsof the control variables have the expectedsigns and a number of the coefficients are sta-tistically significant. For example, the coeffi-cient of well-drilling permit regulationvariable is positive and statistically significant(column 1, row 8). Even if we drop these pol-icy intervention variables (column 2) or if wedrop the conservation technology variable(column 3), econometric estimation still per-forms well and there are few differences amongthe estimation results.

More importantly, when examining the vari-ables of interest, research results show that thechange of tubewell ownership from collective tonon-collective induces the development ofgroundwater markets. The coefficient on theshare of non-collective tubewells variable is po-sitive and significant (Table 5, row 1, columns1–4). All other things held constant, when theshare of the non-collective tubewells in a village

increases, the share of tubewells selling waterincreases. If the share of non-collective tube-wells increases by 10%, the share of tubewellsselling water will increase by nearly 3% (column4). Although the paper cannot infer causality,this result shows the correlation between priv-atization and the rise of groundwater markets.One explanation of this is that in villages withmore privately owned tubewells, there is moreof an incentive to sell water. Another explana-tion of this relationship is that in villages witha higher percent of private tubewells, it maybe that there is less service provided by theoperators of the collective tubewells (for anynumber of reasons). Therefore, in these vil-lages, for farmers to gain access to water, itwould be necessary for farmers to access waterfrom sales from private tubewells.

Resource scarcity is also associated with theemergence of groundwater markets. Althoughit could have been that deeper water tablesmean higher water prices and less demands, infact, the coefficient on the depth to groundwa-ter table is positive and significant (Table 5,row 3, columns 1–4). When the groundwatertable drops by 1 m, the share of tubewells sell-ing water will increase by about 1% (column 4).Hence, the alternative interpretation is consis-tent with the findings: in areas in which thegroundwater table is deep, farmers’ demandfor water from groundwater markets is higher(relative to providing water from one’s ownwell). In simplest terms, if these results areindicative of underlying causal relations, thefindings are evidence of the hypothesis that invillages with scarce water resources, groundwa-ter markets develop more quickly.

Research results also show that land pressurehas increased the breadth of groundwater mar-kets. The coefficient on the per capita arableland variable is negative and statistically signif-icant (Table 5, row 5, columns 1–4). If per capi-ta cultivated land in the villages falls by 0.1 ha,the share of tubewells selling water will increaseby 10% (column 4). In other words, the resultsimply that with the decrease of per capita landresources, the share of tubewells selling waterhas increased. To understand the influence ofhousehold size on the breath of groundwatermarkets, the cultivated land per capita has alsobeen replaced by the cultivated land per house-hold (farm size) (Appendix Table, column 1,row 3). The results of the new analysis showthat the coefficient of this variable is also signif-icant in the regression analysis (as was the ori-ginal coefficient) of determinants of breadth of

Table 5. Regression analysis of the determinants of development (breadth and depth) of groundwater markets in China(using Tobit estimator)

Dependent variable: share oftubewells selling water

Share of watersold

(1) (2) (3) (4)

Tubewell ownershipShare of private tubewells 0.183 0.286 0.180 0.286

(3.86)*** (7.70)*** (3.83)*** (7.40)***

Dummy of individual tubewell 0.389(4.33)***

Water and land scarcityLog of groundwater table 0.003 0.006 0.003 0.006

(3.82)*** (5.06)*** (3.81)*** (4.96)***

Log of groundwater table in 1995 0.008(2.01)**

Log of per capita cultivated land �0.900 �1.036 �0.909 �1.036 �4.745(2.39)** (3.21)*** (2.40)** (3.10)*** (3.50)***

Policy interventionsDummy of fiscal subsidies for tubewell investment 0.051 0.041 �0.121

(0.46) (0.38) (1.58)Dummy of bank loans for tubewell investment 0.065 0.066 0.484

(0.59) (0.60) (3.02)***

Dummy of well-drilling permission regulation 0.116 0.117 0.045(3.09)*** (3.08)*** (0.46)

Other control variablesDummy of adopting water delivery pipes �0.025 0.008 �0.093

(0.64) (0.23) (0.94)Per capita net income of farmers �0.000 �0.000 �0.000 �0.000 0.000

(0.18) (0.88) (0.24) (0.85) (1.94)*

Constant �4.257 �3.853 �4.204 �3.918 �2.943(3.68)*** (4.74)*** (3.66)*** (4.76)*** (3.34)***

Observations 136 136 136 136 50Chi-square 35.19 96.41 35.30 94.29 46.37

Coefficients are marginal effects; absolute value of z or t statistics in parentheses.Data source: Data in the model ‘‘Share of tubewells selling water’’ come from the authors’ survey in 68 randomlyselected villages in four provinces (Hebei, Henan, Shanxi, and Shaanxi) in two years (1995 and 2004) of CNWRS.Data in the model ‘‘Share of water sold’’ come from the authors’ survey in 50 randomly selected tubewells in twoprovinces (Hebei and Henan) of CWIM. We do not use data from all of the sample villages of the two surveys sincethe information in the table is conditioned on villages that use groundwater to irrigate and that have privatetubewells.

* Significant at the 10% level.** Significant at the 5% level.

*** Significant at the 1% level.


groundwater markets (i.e., the equation thatused ‘‘share of tubewells selling water’’ as thedependent variable—Appendix Table, column1, row 3). These results imply that when theaverage holding of land in a village is small,there is more of a tendency for village’s tubewellowners to sell water. 12

(ii) Determinants of the depth of groundwatermarkets

The econometric estimation also performswell when estimating the depth of groundwater

markets (Table 5, column 5). The Chi-square is46, higher than that above explaining thebreadth of groundwater markets (column 1,row 13). Similarly, most of the coefficients ofthe control variables have the expected signsand a number of the coefficients are statisticallysignificant. For example, the coefficient on thevariable of farmer per capita net income issignificant (row 10). This means that farmers invillages with higher per capita income sell more.

In addition, similar to the regression resultson the determinants of the development of the

Table 6. Participation in groundwater markets by farmhouseholds in China, 2004

2004

Total households 173Number of household using groundwater 128Share of households having tubewellsthemselves to irrigate (%)

34

Share of households gettinggroundwater irrigation throughgroundwater markets (%)

20

Data source: The authors’ survey data from 46 randomlyselected villages in two provinces (Hebei and Henan) ofCWIM. We do not use data from all of the sample vil-lages of the CWIM survey since the information in thetable is conditioned on villages that use groundwater toirrigate and that have private tubewells.


breadth of groundwater markets, the develop-ment of the depth of groundwater markets isalso significantly associated with tubewell own-ership and water and land scarcity. For exam-ple, the coefficient on the dummy variable forindividual tubewell ownership is positive andsignificant (Table 5, row 2). This means thatcompared with shareholding tubewells, individ-ual tubewells sell a higher share of their water.In addition, the coefficient on the groundwatertable is also positive and significant (Table 5,row 4). Hence, in areas in which the groundwa-ter table is deep, farmers’ demand for waterfrom groundwater markets is higher. In sim-plest terms, if these results are also indicativeof underlying causal relations, the findings areevidence of the hypothesis that in villages withscarce water resources, groundwater marketsdevelop more quickly.

Research results also show that land pressurehas intensified the depth of groundwater mar-kets. The coefficient on the per capita arableland variables is significant (Table 5, row 5).Therefore, it appears that agricultural land isscarcer (making it less desirable for an individ-ual farmer to sink his/her own tubewell), theaverage tubewell operator sells a greater shareof water from his/her tubewell.

6. DO GROUNDWATER MARKETS HELPTHE POOR?

Based on the above analysis, the research re-sults show that groundwater markets in north-ern China have developed in terms of both theirbreadth and depth. Household data furtherindicate the importance of groundwater mar-kets for irrigation in northern China. Resultsshow that more than 70% of the sample house-holds depend on groundwater for irrigation(Table 6, rows 1 and 2). However, among allof the households using groundwater, only34% of them own tubewells. In contrast, 20%of households have to depend on groundwatermarkets to gain access to water for irrigation(rows 3 and 4). Although a fairly large numberof households still access groundwater fromcollective tubewells, as the strength of the col-lective diminishes, it is clear that the role ofgroundwater markets will become increasinglyimportant in the coming years.

As groundwater markets become increasinglyimportant, it is important to understand ifgroundwater markets are helping or hurtingthe poor and/or increasing or reducing inequal-

ity in rural China. If groundwater marketsemerge and function well, the rise of privatetubewells does not have to lead to inequities(although it may—the final answer is an empir-ical one). Elsewhere in the world, research hasshown that groundwater markets can be equityenhancing. For example, in Pakistan Meinzen-Dick (1996) demonstrated in their case studyarea that groundwater markets improved equi-ty of groundwater use by making water avail-able to small landowners or tenants andyounger households—those farmers who wereleast likely to own tubewells.

Our field surveys in northern China have pro-vided the similar evidence as that in Pakistanand other countries. According to the surveydata, groundwater markets have provided thegroundwater access opportunity to poor farm-ers and reduced potential income gaps byenhancing the access to groundwater. Specifi-cally, households in the sample that buy waterfrom groundwater markets are poorer thanwater-selling households. For example, the percapita income from cropping of water-buyinghouseholds is only 61% of that of water-sellinghouseholds; per capita total income of water-buying households is also lower than that ofwater-selling households (Table 7, rows 1 and2). Such results may imply that, although poorfarmers do not have enough money to sinktubewells, they can buy water from markets.If this is the case, groundwater markets almostcertainly are helping the poor.

The data also indicate that groundwatermarkets benefit weak farmers that are small, lesseducation, and older. Households that buywater have smaller holdings of cultivate landthan water-selling households. For example,the per capita land area of water-buying house-

Table 7. Characteristics of water-selling and water-buy-ing households in China, 2004

Water-sellinghouseholds

Water-buyinghouseholds

Per capita croppingincome (yuan)

1609 988

Per capita totalincome (yuan)

2891 2634

Per capita cultivatedland area (ha)

0.15 0.13

Education level ofhousehold head (year)

6.3 5.5

Age of householdhead (year)

47.6 50.0

The sample sizes of water-selling households and water-buying households are 36 and 25.Data source: Authors’ survey in 46 randomly selectedvillages in two provinces (Hebei and Henan) of CWIM.We do not use data from all of the sample villages of theCWIM survey since the information in the table isconditioned on villages that use groundwater to irrigateand have private tubewells.


holds is 0.13 ha while the land area of water-selling households is slightly larger, 0.15 ha (Ta-ble 7, row 3). Such a finding implies that house-holds with small holdings of land, who are notable to or choose to not sink a tubewell (andcannot utilize the entire command area of atubewell investment), are able to buy waterthrough groundwater markets. In addition, re-search finds that less educated and older farmersalso depend more on groundwater markets togain access to groundwater (rows 4 and 5). 13

Whether groundwater markets benefit thepoor or not may also be related with the struc-ture of the markets (i.e., whether they aremonopolistic or competitive). The poor shouldbe able to benefit more when markets are com-petitive than when water buyers face a singleseller. However, analysts do not always agreeon how to measure market structure. In otherresearches outside of China, researchersindicate that due to physical and topographicalconditions, groundwater markets may be frag-mented and could be monopolistic (Bagachi,1995; Campbell, 1995; Dhawan, 1988; Jacobyet al., 2004; Kahnert & Levine, 1994; Pant,1991; Shah, 1993). In another work, economistsattempt to measure the degree of competition.For example, to do so, following the work ofLerner (1970), Shah (1993) hypothesized thatthe ratio of water price to total variable costcan be used as a fairly good indicator of the levelfor monopoly profits (see Lerner (1970) for theproof of this proposition). In applying this to

the case of India, Shah and Ballabh (1997)found that the ratio was high, about 2.5–3.0 inMazaffarpur. Based on this, Shah and Ballabh(1997) concluded that groundwater marketswere not very competitive. Other researchersdo not agree with the approach of Lerner(1970) and Shah (1993). For example, both Fuj-ita and Hossain (1995) and Palmer-Jones (2001)also found fairly high ratios (water price to totalvariable cost) in their studies (averages 2.6).However, unlike Shah and Ballabh (1997), theyinterpreted the results to mean that, far frombeing monopolistic, water markets were com-petitive; the high ratio merely reflects the entre-preneur’s risk premium. In other words, basedon their observations, water markets were com-petitive and they are able to explain the highratio with an alternative explanation.

To examine the structure of groundwatermarkets in China, we follow the lead of otherresearchers from South Asia. First, followingthe approach of Shah (1993), we calculate theratio of water price to total variable cost.According to our data, the ratio is 2.2, rangingfrom 1.2 to 3.3. More than 70% of tubewellshave the ratios which were lower than 2.5. Thisaverage level is lower than the findings of Shahand Ballabh (1997) in India (range is from 2.5to 3.0) and Fujita and Hossain (1995), Pal-mer-Jones (2001). Hence, if the low ratio ofwater price to total variable cost does, in fact,measure competition, from this angle there isevidence that groundwater markets in Chinaare relatively competitive.

In addition, we also examine the data onwater prices and compare within village aver-ages of price variations with between villagevariations. Since groundwater markets arelocalized and most transactions are amongfarmers in the same villages. Therefore, if mar-kets are competitive, we would expect that mostprice variation should come from between vil-lages, not from within villages. In fact, we findthat the variation in the price of water is mainlydue to regional differences among villages, notfrom within villages. 14 For example, in one vil-lage the price of water from one tubewell ismore than 3.4 times that from one tubewell inanother village. However, within any of our vil-lages, the highest difference among the priceobservations is only 50%. In 75% of villages,water price differences among tubewells sellingwater within villages are much smaller. Theseresults are also consistent with the findings ofother researchers that believe water markets intheir study areas are competitive. For example,


after controlling for the influence of otherfactors, Kajisa and Sakurai (2003) found thatthe variation of water price mainly comes fromregional differences which lead them to theconclusion that groundwater markets are notmonopolistic. 15

Besides analyzing prices directly, we alsoexamine other types of data to provide evidencethat supports the finding of the non-monopolis-tic nature of groundwater markets. First, welook at profits from selling water. When usingour data, we are able to estimate both the fixedand variable costs that are associated withpumping and selling water. Accordingly, our re-sults demonstrate that (even when we do notconsider the value of family labor that is usedto pump and sell water) profits are generallysmall.

Second, we also look at the number of welloperators that are selling water and water deliv-ery conditions. Shah (2000) suggested thatwhen wells are sunk in a fairly dense manner,and when there are lined conveyance structuresin a village, there is less of a probability that asingle seller will have monopoly power and thatthe price of water will be relatively more com-petitive. Using this approach with our surveydata, we find that in almost all villages, thereare many tubewell operators selling water, notonly just one. On average, in each village, thereare 18 tubewells (and 13 private ones) and morethan 70% of the private tubewells are sellingwater. Furthermore, the adoption rate of sur-face pipe (or hoses) in groundwater irrigationregions of northern China (i.e., the use of effi-ciency-enhancing conveyance technologies) iscommon. Our survey found that more than70% of tubewell owners are using surface pipesto deliver water. The adoption of surface pipesgreatly increased the ability of farmers tochoose tubewells from which they want tobuy water. Therefore, based on these analyses,it seems that groundwater markets in northernChina almost certainly are not monopolistic.

7. CONCLUDING REMARKS

This paper has sought to understand thedevelopment of groundwater markets in north-ern China and examine the factors that deter-mine the development of groundwatermarkets. Using our two data sets, research re-sults provide strong evidence that groundwatermarkets in northern China have developed interms of both their breadth (the share of villages

in which there is groundwater market activity)and depth (the share of water which the averagewater-selling tubewell owner sells to others on amarket basis). Interestingly, although fewerpeople have worked on groundwater marketsin China, even compared to countries, such asIndia and Pakistan, which have better docu-mented groundwater markets, groundwatermarkets in northern China clearly have emergedand are almost equal in pervasiveness.

Our findings also demonstrate that whilethere are many similarities, there are also differ-ences. Groundwater markets in northern Chinahave many characteristics similar to those inSouth Asia; they are informal, localized, andmostly unregulated. At the same time, however,China’s markets appear to be less personalizedand transaction in China is done more on acash basis.

While the multivariate analysis is carried outmostly to understand descriptively the corre-lates of groundwater markets, there are a num-ber of robust findings that support thehypotheses of interest. The form of ownershipappears to be strongly correlated with the emer-gence of groundwater markets. Groundwatermarkets also appear in more villages and tube-well owners sell a higher share of the waterfrom their wells when the groundwater tableis deep (i.e., water is scarce) and land is scarce.All of these are suggestive that when the factorsthat affect supply and demand for groundwaterare in place, there is a tendency for markets toemerge. Policy has also played a role in China.

While much of the results are suggestive thatgroundwater markets are largely self-organiz-ing and unregulated, there also does appear tobe a role for the state. The findings show thatwhen the government makes it easier for indi-viduals and shareholding groups to get accessto capital and are not subject to local regula-tions, there is a greater level of groundwatermarket activity. Since our research results alsoshow that groundwater markets at the veryleast are not regressive and may, in some cases,be progressive, it may be that government-sponsored investment and banking programsthat allow individuals access to grants andloans to sink tubewell will further promotegroundwater markets with Chinese characteris-tics.

Finally, the research results indicate thatgroundwater markets in northern China dohelp the poor. Households that buy water fromgroundwater markets are poorer than water-selling households. Such a finding implies that


groundwater markets have provided greater ac-cess to groundwater to poor farmers and possi-bly help reduce income inequalities in ruralChina.

While it is beyond the scope of the paper tomeasure the impact of the emergence ofgroundwater markets on the groundwater tableit is possible that groundwater markets, to theextent that they encourage the greater use ofgroundwater, could accelerate the fall of thegroundwater table. If this is so, the question re-mains whether groundwater markets should beencouraged. What are the options? If ground-water markets were suppressed, given our re-sults, it could actually hurt the poor. So whatshould happen? To avoid hurting the environ-ment (if it was being hurt), instead of directlytrying to suppress groundwater markets, alter-native policies that would control the draw-down of the water table (e.g., water pricingpolicies) should be promoted. In addition, ef-forts to allow groundwater markets to emergecould help spread the benefits that come withgreater access to irrigation.

The analysis in this paper also has implica-tions that go beyond the water literature. First,the emergence of markets does not have to hurtthe poor. In fact, it is possible that they are pro-poor. According to our analysis, in the case ofChina’s groundwater markets, the poor havebenefited. Poor households have been involved

with both the supply and demand sides of themarket. This is somewhat different from whathas been the case in other parts of the worldwhere groundwater markets have emerged.

So why might this be the case? One thought isthat markets work well and are competitive andexpand the opportunities for access to re-sources for the poor (and rich) when they arecomposed of agents that all have access to min-imum amount of resources—both land andcapital—and the market environment is rela-tively unregulated. In the case of China, allhouseholds in each village have land and thegovernment has instituted programs that haveoffered loans and grants to those that want tosink a well. In addition, given the initial invest-ment by the government in water (in the pre-and early reform era), the incomes of mostfarmers were already high enough to allowsome farmers to gain access to enough capitalfor investment (and so had access to sufficientliquidity) that they were able to afford to buywater when it was provided in a competitivemarket environment. Therefore, when ground-water markets emerge in such an environment,buyers and sellers can both benefit, and overallaccess to water can raise production and thewelfare of all participants. Such a case, how-ever, may not occur in places in which re-sources—such as land and capital—are lessequitably distributed.

NOTES

1. Due to limitations on the nature of the data, and thelength of this paper, we are not examining the impacts ofthe rise of groundwater markets. While it is an extremelyimportant topic to measure the effects of groundwatermarkets on cropping patterns, crop productivity and thewater table, it is really beyond the scope of this paper. Inaddition, in the past there had been another work doneon groundwater markets and a paper by Mukherji(2004) provided an excellent review of the literature thatdoes not need replicating here. According to Mukherji,there were a number of papers that have looked at thegroundwater markets from an institutional point of view(Dubash, 2000, 2002; Kajisa, 1999; Palmer-Jones, 1994).In this paper, however, we take a neo-classical econo-mists approach as having a number of other economists(e.g., Shah, 1985, 1989, 1991, 1993).

2. In our studies northern China is defined to includethe following regions: northern China (huabei), North-east China (dongbei), and Northwest China (xibei).

3. Since water resource endowments in mountainous ordesert regions are complex and we do not have detailedsecondary data to sort these areas into the other threecategories, we decided to classify mountainous anddesert regions into an independent category. During thecourse of sample selection, it should be noted that we didnot select any desert sample counties. By design, wechose only a limited number of counties that areclassified as mountainous (10%).

4. There is a potential asymmetry when trying to matchwater-selling and water-buying households with house-holds that own tubewells and those that do not owntubewells (henceforth, non-tubewell owners). While anon-tubewell owner can only be a water-buying house-hold and cannot be a water-selling household, a tubewellowner can be both a water-selling household and awater-buying household (e.g., in the case when he/she iscultivating a plot that is in a location that cannot besupplied by his/her own tubewell).


5. In China, nearly all villages have trained accoun-tants that maintain written records about many aspectsof the village life, including information on demograph-ics, the economic structure of the village, income, land,etc. In addition (and importantly), most villages in thesample kept detailed, community-level records aboutvillage water issues. We also held detailed interviewswith multiple stakeholders in the village—leaders, tube-well owners, and farmers—who had been village resi-dents for the entire sample period (during 1995–2004).Given the importance of water, villagers had littletrouble remembering their ‘‘well histories.’’ In otherwords, we were able to rely on both leader/wellowner/farmer recall and accountant records to enumerate thenumber of wells and pumps in past years.

6. We also note that the share of water sold reduced from80% in 1995 to 77% in 2004 (Table 1, row 3). This is eitherpossibly due to the increase in the number of wells sellingwater or the reduction in the demand of water with theincrease in number of tubewells in the region. Accordingto our data, either of these two interpretations is allplausible. During 1995–2004, both the number of wellsselling water and the total number of wells have increased.In the 68 sample villages that are in the NCWRS, thenumber of wells selling water increased from 75 in 1995 to342 in 2004; at the same time, the total number of wellsalso increased from 1472 to 1967.

7. In the villages that claimed that there was a formalregulated ceiling on the price of water, typically, theregulation was said to have been set by the price bureau.When such an announcement was made, in many casesthe regulation was mainly targeted at all water users inthe county (including—often especially—at water uses inindustry and municipal water districts). Because of thisin some villages even though leaders said that there wasa regulation ‘‘on the books,’’ it did not always mean thatthe rule was ‘‘implemented’’ in practice in the village.

8. We want to emphasize that in this part of the sectionof the paper, we are only examining correlations withour descriptive data. We are not suggesting causality.The most that we can say with descriptive statistics isthat we are showing data trends that are consistent withthe hypotheses. We include a multivariate analysis in thenext section.

9. In our study areas well-drilling regulations are therules and directives produced by local officials that seekto influence the behavior of farmers before they drill newwells. When these are enforced, a farmer is supposed toapply for a permit for drilling a new well from the localgovernment. The permit often says where and at whatdepth the well is to be drilled. Once it is granted, thefarmer is legally allowed to sink a new well. Although

there are such regulations in many, if not most, regionsof China, these regulations are rarely implementedeffectively. During our survey, we asked village leadersif farmers in their villages officially are supposed to applyfor such permits before sinking new tubewells in theirvillages. If the answer was ‘‘Yes,’’ the variable well-

drilling permit is coded as ‘‘1,’’ otherwise it is coded as‘‘0.’’

10. The Tobit model was developed by Tobin (1958) totreat data sets when the dependent variable was char-acterized as being censored. The Tobit model, in theform of y = f(X) = Max(0, Xb + U), is a hybrid of theProbit model and the linear regression model (McDon-ald & Moffitt, 1980). The use of Tobit model has twoadvantages. First, it overcomes the information losswhich would occur if a Probit model was used withcensored data that was translated into binary data.Second, it overcomes the violation of the OLS assump-tions that would occur if an OLS estimator was usedwith censored data. Recent social science research hasshown a growth in applying Tobit model (Gunderson,1974; Helms & Jacobs, 2002; Langbein, 1986).

11. Besides multicollinearity problems, we might alsobe concerned about the potential endogeneity problemsin Eqns. (1) and (2). It is possible that the estimatedparameters of our variables of interest are biased sincethe share of tubewells selling water may be determinedsimultaneously with the share of private tubewells. Itcould also be that the estimated coefficient is affected byunobserved heterogeneity. In such a situation, theestimate of the coefficient on the share of privatetubewells variable could be biased. To control for anyof this endogeneity, we use a fixed effects model(Appendix Table, column 2). The results of the fixedeffects model show that most coefficients (especially thecoefficients of interest) do not change much (comparedwith the findings in column 4, Table 5; and column 2,Appendix Table). We take the same strategy for assess-ing the robustness of the results of Eqn. (2). While wecan report that there is little change to the coefficients,due to space constraints, we do not include the findingsin a paper. Therefore, we believe that any potentialendogeneity problem in Eqns. (1) and (2) is not serious.

12. What are the implications of these results? They donot mean necessarily that only small households arebuying water. In China, one of the effects of the natureof the nation’s initial de-collectivization movement wasthat there was not much difference in the size of the landthat was allocated to farmers within the village. There-fore, what we are seeing are inter-village differences. As aresult, this means that the villages that have mostly smallhouseholds have more sales; as opposed to villages withmostly large households. This feature of China makes it


different from other countries—especially those in SouthAsia.

13. We also conducted statistical tests (t-tests) tounderstand whether or not the differences betweenwater-selling and water-buying households are signifi-cant statistically. Results show that none of them aresignificant. The t statistics for per capita land area is1.18, education level of the household head is 0.52, andthe age of the household head is 1.17. However, thisresult does not in itself prove causality. If thesecharacteristics influence the decision of buying or sellingwater, we need to do further multivariate analysis.

14. The shares of total expenditures that are accountedfor by water are also relatively homogeneous within

villages. Whether a household purchased water from atubewell operator or whether he/she supplied water fromhis/her own well, our results show that within the samevillage there are only narrow differences.

15. According to our survey in the villages that havecollective tubewells and private tubewells selling water,we found that water price of collective tubewell is almostsame to private groundwater markets. On average, thedifference between the price of water being sold fromcollective tubewells and that being sold from privatetubewells is less than 15%. In addition, we have also notfound a significant level of difference between theprice of water being sold by private tubewell ownersand that being sold by the owners of shareholdingtubewells.

REFERENCES

Ballabh, V., Choudhary, K., Pandey, S., & Mishra, S.(2002). Groundwater development and agriculturalproduction: A comparative study of Eastern UttarPradesh, Bihar and West Bengal. Paper submitted toIWMI-Tata Water Policy Programme, Anand, Guj-arat.

Bagachi, K. S. (1995). Irrigation in India: History andpotentials of social management. New Delhi: Upa-labdhi Trust for Development Initiatives.

Campbell, D. (1995). Design and operation of small-holder irrigation in South Asia, World Bank Tech-nical Paper No. 256. Washington DC: The WorldBank.

Dhawan, B. D. (1988). Irrigation in India’s agriculturaldevelopment: Productivity, stability, equity. NewDelhi: Sage.

Dubash, N. K. (2000). Ecologically and socially embed-ded exchange: Gujarat model of water markets.Economic and Political Weekly, 35(16), 1376–1385.

Dubash, N. K. (2002). Tubewell capitalism, groundwaterdevelopment and agrarian change in Gujarat. NewDelhi: Oxford University Press.

Fujita, K., & Hossain, F. (1995). Role of the ground-water market in agricultural development andincome distribution: A case study in a NorthwestBangladesh village. The Developing Economies, 33(4),442–463.

Gunderson, M. (1974). Training subsidies and disad-vantaged workers: regression with a limited depen-dent variable. The Canadian Journal of Economics,7(4), 611–624.

Helms, R., & Jacobs, D. (2002). The political context ofsentencing: An analysis of community and individualdeterminants. Social Forces, 81(2), 577–604.

Huang, Q., Rozelle, S., Wang, J., & Huang, J. (2005).Irrigation, agricultural performance and povertyreduction in China. Food Policy, 31(1), 30–52.

Jacoby, H., Murgai, R., & Rehman, S. (2004). Monop-oly power and distribution in fragmented markets:The case of groundwater. Review of EconomicStudies, 71, 783–808.

Kahnert, F., & Levine, G. (1994). Groundwater irrigationand the rural poor: Options for development in theGangeic basin. A World Bank Symposium. Washing-tong, DC: The World Bank.

Kajisa, K. (1999). Contract theory and its application togroundwater markets in india. PhD thesis, MichiganState University.

Kajisa, K., & Sakurai, T. (2003). Determinants ofgroundwater price under bilateral bargaining withmultiple modes of contracts: A case of MadhyaPradesh, India. Japanese Journal of Rural Economics,5, 1–11.

Langbein, L. (1986). Money and access: Some empiricalevidence. The Journal of Politics, 48(4), 1052–1062.

Lerner, A. P. (1970). Principles of welfare economics.New York: Augustus M. Kelley.

Lin, J. Y. (1992). Rural reforms and agricultural growthin China. The American Economic Review, 82, 34–51.

Liu, C., & He, X. (1996). Water issues in the 21thcentury. Beijing, China: Science Publishing House.

McDonald, J., & Moffitt, R. (1980). The uses of Tobitanalysis. Review of Economics & Statistics, 62(2),318–321.

Meinzen-Dick, R. (1996). Groundwater markets inPakistan: Participation and productivity, ResearchReports 105, International Food Policy ResearchInstitute, Washington, DC.

Ministry of Water Resources and Nanjing Water Insti-tute (2004). Groundwater exploitation and utilizationin the early 21st century. Beijing: China WaterResources and Hydropower Publishing House.

Ministry of Water Resources (2002). China WaterResources Bulletin.

Ministry of Water Resources (2003). China WaterResources Bulletin.

Mukherji, A. (2004). Groundwater markets in Ganga–Meghna–Brahmaputra basin: Theory and evidence.Economic and Political Weekly, 31, 3514–3520.

Palmer-Jones, R. (1994). Groundwater markets in SouthAsia: A discussion of theory and evidence. In M.Moench (Ed.), Selling water: Conceptual and policy


debates over groundwater markets in India. Ahmeda-bad: VIKSAT-Pacific Institute-Natural HeritageInstitute.

Palmer-Jones, R. (2001). Irrigation service markets inBangladesh: Private provision of local public goodsand community regulation. <http://www.sas-net.lu.se/palmer_jones.pdf>.

Pant, N. (1991). Ground water issues in eastern India. InR. Meinzen-Dick, & M. Svendsen (Eds.), Futuredirections for Indian irrigation: Research and policyissues. Washington, DC: International Food PolicyResearch Institute.

Pant, N. ( 2003). Key trends in groundwater irrigation inthe eastern and western regions of Uttar Pradesh.Paper submitted to IWMI-Tata Water Policy Pro-gramme, Anand, Gujarat.

Shah, T. (1985). Transforming groundwater marketsinto powerful instrument of small farmer develop-ment. ODI Irrigation Management Network PaperNo 11d. London: Overseas Development Institute.

Shah, T. (1989). Groundwater grids in the villages ofGujarat: Evolution, structure, working and impacts.WAWANA, January, 14–29.

Shah, T. (1991). Water markets and irrigation develop-ment in India. Indian Journal of Agricultural Eco-nomics, 46(3), 335–348.

Shah, T. (1993). Groundwater markets and irrigationdevelopment: Political economy and practical policy.Bombay: Oxford University Press.

Shah, T., & Ballabh, V. (1997). Water markets in NorthBihar: Sxi village studies in Muzaffarpur district.Economic and Political Weekly, 32(52), A183–A190.

Shah, T. (2000). Groundwater markets and agriculturaldevelopment: A South Asian overview. In GWP,Pakistan water partnership, Proceedings of regionalgroundwater management seminar, October 9–11,2000, Islamabad (pp. 255–278).

APPEND

Regression analysis of the determinants of deve

Tubewell ownershipShare of private tubewells

Water and land scarcityLog of groundwater table

Log of cultivated land per household

Log of per capita cultivated land

Shah, T., Giordano, M., & Wang, J. (2004). Irrigationinstitutions in a dynamic economy: What is Chinadoing different than India? Economic and PoliticalWeekly, 31(July), 3452–3461.

Sharma, P., & Sharma, R. (2004). Groundwater marketsacross climatic zones: A comparative study of aridand semi-arid zones of Rajasthan. India Journal ofAgricultural Economics, 59(1), 138–150.

Strosser, P., & Meinzen-Dick, R. (1994). Groundwatermarkets in Pakistan: An analysis of selected issues. InM. Moench (Ed.), Selling water: Conceptual andpolicy debates over groundwater markets in India.Ahmedabad: VIKSAT–Pacific Institute–NaturalHeritage Institute.

Tobin, J. (1958). Estimation of relationships for limiteddependent variables. Econometrica, 26, 24–26.

Wang, J., Huang, J., Blanke, A., Huang, Q., & Rozelle,S. (2007). The development, challenges and manage-ment of groundwater in rural China. In M. Giord-ano, & K. G. Villholth (Eds.), The agriculturalgroundwater revolution: opportunities and threats todevelopment (pp. 37–62). Trowbridge, UK: CromwellPress.

Wang, J., Huang, J., Huang, Q., & Rozelle, S. (2006).Privatization of tubewells in north China: Determi-nants and impacts on irrigated area, productivity andthe water table. Hydrogeology Journal, 14, 275–285.

Wang, J., Huang, J., & Rozelle, S. (2005). Evolution oftubewell ownership and production in the northChina plain. Australian Journal of Agricultural andResource Economics, 49(2), 177–196.

Wood, G. D. (1995). Private provision after publicneglect: Opting out with pumpsets in north Bihar.Presented at the International conference on politicaleconomy of water in South Asia: Rural and urbanaction and interaction, held at Madras Institute ofDevelopment Studies, January 5–8.

IX A

lopment breadth of groundwater markets

Share of tubewells selling water

Tobit Fixed effect Tobit

0.275 0.654 0.189(3.86)*** (7.95)*** (4.09)***

0.006 0.008 0.003(3.49)*** (2.06)** (3.52)***

�0.377(2.44)**

�0.827 �0.927(0.75) (2.37)**

http://www.sasnet.lu.se/palmer_jones.pdf

http://www.sasnet.lu.se/palmer_jones.pdf

Appendix A—continued

Share of tubewells selling water

Tobit Fixed effect Tobit

Policy interventionsDummy of fiscal subsidies for tubewell investment 0.033

(0.32)Dummy of bank loans for tubewell investment 0.063

(0.57)Dummy of well-drilling permission regulation 0.116

(3.04)***

Other control variablesDummy of adopting water delivery pipes �0.009

(0.24)Per capita net income of farmers �0.000 �0.000 �0.000

(0.20) (0.27) (0.10)Well density �1.255

(0.72)Share of sown area for grain crops 0.152

(1.47)Constant �7.302 �0.739 �4.691

(2.96)*** (1.59) (3.85)***

Observations 136 136 136Chi-square 22.78 40.28R2 0.73

Coefficients are marginal effect, absolute value of z statistics in parentheses.Data source: Data in the model ‘‘Share of tubewells selling water’’ come from authors’ survey in 68 randomly selectedvillages in four provinces (Hebei, Henan, Shanxi, and Shaanxi) in two years (1995 and 2004) of NCWRS. We do notuse data from all of the sample villages of the NCWRC survey since the information in the table is conditioned on avillage that uses groundwater to irrigate and has private tubewells.

* Significant at 10%.** Significant at the 5% level.

*** Significant at the 1% level.


APPENDIX B. A CASE STUDY

To illustrate how the broader institutionalframework affect the emergence and develop-ment of groundwater markets in northern Chi-na, during the survey in Hebei provinceenumerators visited one village in one of thesample counties and conducted additionalinterviews. In this village groundwater is theonly source of water for irrigation. In 1983,there were eight collective tubewells. These col-lective tubewells were able to provide irrigationto about 30% of cultivated land. Things chan-ged, however, after 1983. In the early 1980s,reformers in the province de-collectivized agri-culture and land was contracted to each house-hold in the village. The land was distributedequally to all of the farmers (about 4.5 mu or

0.3 ha per household). These rural reforms pro-vided greater incentives for farmers to exert ef-fort, which led to increased agriculturalproductivity and higher incomes (Lin, 1992).However, in such villages access to water wasstill one of the main constraints to even greaterproductivity increases. More importantly, thewater table in the village was falling duringthe early and mid-1980s. By 1985, the level ofwater had dropped so far that a number ofthe collective tubewells could not be operated.To maintain the original level of irrigation ser-vices, some farmers began to dig their owntubewells. Farmers at this time would have pre-ferred the village to re-sink the well, but in thisvillage the leadership had access to few fundsand so there was no alternative but to sink


the wells themselves. Generally, each tubewellin the village was able to irrigate about 50 mu(or 3.3 ha). This means that each tubewellcould supply irrigation to about 10 households.Since some farmers either lacked the financialresources or were too busy to invest in theirown tubewells, they began to buy water fromtheir neighbors who had tubewells. Under suchan environment, groundwater markets have

emerged gradually. Later in the early 1990s,government loan programs helped farmers tofinance tubewell construction. By 2004 thenumber of wells had grown to 27, and 89% ofthem were private. Due to the development ofprivate tubewells and groundwater markets,by 2004 more than 60% of land in the villagecould be irrigated, a level much higher thanduring the early 1980s.

Available online at www.sciencedirect.com

development of groundwater markets in china: a glimpse ...€¦ · 1. introduction in recent years...

Documents