Water Markets in Mula (Murcia), Spain
The goal of this project is to jointly explain why Mula and Lorca had markets to allocate water between the 13th and the 20th centuries, whereas all the other towns in the area allocated water with quotas. This observation is decomposed on three facts. i) Mula and Lorca moved to a market system; ii) Mula and Lorca used markets for over 700 years when all surrounding similar towns used quotas; iii) Mula and Lorca moved back to a system of quotas in the 1960s.
As argued below, the original institutional divergence (i) can be explained by a historical accident. Following Historia de la Región de Murcia by Miguel Rodriguez Llopis (1998), it is clear that a historical accident, the rebellion of two Muslim governors, created the institutional divergence. This lead us to the second question, why did this historical accident persisted for so long.
We answer this institutional persistence (ii) in Espín-Sánchez (2017). We show that a lack of incentives, due to the low savings accumulated by the farmers, created a solvency trap. This solvency trap created both the persistent of an inefficient institution to allocate water and lack of effort by the farmers, for over 700 years. We also show that farmers left the solvency trap in the 1960s (iii) due to an exogenous shock in the 1950s: increase in international demand for fruit increased their profits (temporarily) and their savings (permanently).
There is still one more thing needed to explain the return to quotas (iii). Even if farmers were no longer insolvent and could buy all the water property rights, they would only switch back to quotas if they thought that it was more efficient than the market. In Doval and Espín-Sánchez (2020), we show (theoretically) how a system of quotas could dominate the market system, if liquidity constraints are severe enough and utility is concave in water. Moreover, we show that when there is heterogeneity in both productivity and wealth (liquidity constraints) neither quotas or markets are efficient. We characterize the first-best mechanism and show that can only be achieved when liquidity constraints are sufficiently mild.
Once we have the theoretical predictions, the next step is to test in the data, whether quotas were more efficient than markets in Mula. Donna and Espín-Sánchez (2019) use this framework, where farmers are heterogeneous in two unobserved dimensions, productivity and wealth. Using structural estimation techniques, we conclude that quotas were more efficient than quotas. This result is a consequence of the presence of liquidity constraints, the relative homogeneity in farmers’ production function and the decreasing marginal returns of water in agriculture.
It may seem striking that such inefficient system persisted for so long without frictions. In Donna and Espín-Sánchez (2021) we study the system of self-governance of the farmers in Mula. The system was a self-organized commons, where conflicts were solve by a tribunal appointed by the farmers themselves. We show how punishments where low for first time offenders but much higher for recidivists. Moreover, punishments were higher when the defendant was wealthy and lower when the victim was wealthy. The evidence is again consistent with a model where farmers may face liquidity constraints. This system of leniency with harsher punishments for recidivists acted as a second-best form of insurance in the absence of quotas. Farmers could steal water without facing a severe punishment acted as a form of insurance. In order to prevent abuse in the system, recidivists were treated more harshly, which means that the insurance was only partial, because farmers that received two liquidity shocks would be punished harshly.
Finally, in Donna and Espín-Sánchez (2018) we show how the farmers were very sophisticated bidders. They did not collude to get lower prices of water, rather they compete fiercely and sometimes, even violently. The market-auction was efficient, and the seller did not try to reduce the supply of water to inflate prices, despite the market power that the seller, and probably due to the threat of rebellion. It is important to emphasize the sophistication and high level of competition in the market, in order to conclude that the reason quotas outperformed markets was not that markets were poorly run. Rather, quotas acted as an insurance mechanism during droughts.
History
During the reign of Ibn Hud (1228-1238), the Kingdom of Murcia enjoyed some prosperity and stability. After the weakening of the Almohads, Ibn Hud unified all the small muslim kingdoms (Taifas) in the peninsula. When Ibn Hud was murdered in 1238, the kingdom was dismembered into many Taifas. This same year Jaime I (King of Aragon) conquered Valencia and prepared to march south. Castile was also advancing to the south, expanding its territory at the expense of the now fragile Kingdom of Murcia, which territory was now a fraction of what it was under Ibn Hud. By 1242, Castile had conquered most of the kingdom. Ahmed, the son of Ibn Hud, traveled to Alcaraz (Toledo) to meet prince Alfonso (later known as King Alfonso X, the Wise ). They agreed that what remained of the Kingdom of Murcia would become a protectorate of Castile.
The cities of Mula and Lorca rejected this agreement. In April 1244, Alfonso was in Murcia with his army ready to attack Mula (the closest of the rebel cities). After Mula was conquered, the army moved to Lorca, which surrendered by the end of June. The government of Mula and Lorca was given to the Order of Santiago and the Order of the Temple, respectively. The government of the city of Murcia was given, in part, to the descendants of Ibn Hud according to the terms of the Alcaraz Treaty. However, the Order of Santiago and the Order of the Temple had absolute authority over the cities of Mula and Lorca respectively, because the cities were conquered by force. In each city, the Orders separated the ownership of land and water, and created a corporation with tradable shares, whose owners got money from selling in public auctions the rights to use the water from the river.
Under these new institutions, the owners of the water property rights (Waterlords) were different persons than the land owners (farmers). The Waterlords in Mula established a well-functioning cartel (Heredamiento de Aguas) which lasted through the pre-modern era, despite the many political changes that occurred in Spain. The land owners were small proprietors, with family-size plots, who soon after created their own association, Sindicato de Regantes.
A Model of Persistence of Inefficient Institutions
In 1966, after over 700 years, the irrigation community in Mula (Spain) switched from auctions to quotas to allocate water from its river. This change happened in the absence of either political or technological change. Quotas were more efficient but required that farmers own water property rights. I develop a model in which poor farmers cannot credibly commit to purchase water rights. I show that empirical evidence on savings and prices is consistent with this interpretation. A temporary increase in output prices in the 1950s and better financial institutions allowed farmers to accumulate savings and solve the commitment problem.
![]() |
![]() |
Auction sheet from 1803 | Offering to buy all water rights for 4.2M |
- “Institutional Inertia: Persistent Inefficient Institutions in Spain,” The Journal of Economic History, 2017, Vol. 77, No 3, pp 692-723.
The Illiquidity of Water Markets
Water is an essential commodity and there is controversy over whether it should be allocated using markets. In 1966, the irrigation community in Mula (Spain) switched from a market institution, to a system of fixed quotas to allocate water from the town’s river. Due to the lack of output data, we estimate demand for water accounting for liquidity constraints (LC) and data during the auction period, and compute the welfare under each system. We get unbiased estimates for both farmers’ demand for water and their financial constraints. Ignoring the presence of LC biases the estimated demand and its elasticity downwards. Moreover, markets are efficient only in the absence of LC . We compute welfare under both institutions and find that the quota is more efficient than the market, due to concavity and idiosyncratic productivity shocks.
![]() |
![]() |
Agricultural Census (1954) | Real Estate Tax Registry (1954) |
- “The Illiquidity of Water Markets: Efficient Institutions for Water Allocation in Southeastern Spain” (Appendix), The Review of Economic Studies, forthcoming, with Javier Donna
Stability over time
We investigate the role of punishment progressivity and individual characteristics in the determination of crime. We model individuals’ response to judges’ optimal punishment in a dynamic setting. Agents have both persistent and transitory unobservable components. Judges observe a public signal that is correlated with the transitory component and update their beliefs about the persistent component based on past behavior. For the empirical analysis we examine a novel trial data set from a self-governed community of farmers in Spain in which: (i) judges usually impose low fines even though punishment is not costly; (ii) recidivists are punished harsher than first-time offenders for the same crime; (iii) judges vary the degree of punishments based on individual characteristics—such as when victims or accused have a Don honorific title indicating they are wealthy. Punishments are harsher when the accused is a Don and are milder when the victim is a Don. All these empirical facts can be explained by a model in which the community trades off deterrence and insurance.
- “Water theft as social insurance: south-eastern Spain, 1851-1948,” The Economic History Review, 2021, Vol. 74, No. 3, pp 721-753, with Javier Donna
Rational forward looking farmers
We use data on sequential water auctions to estimate demand when units are complements or substitutes. A sequential English auction model determines the estimating structural equations. When units are complements, one bidder wins all units by paying high price for the first unit, thus deterring others from bidding on subsequent units. When units are substitutes, different bidders win the units with positive probability, paying prices similar in magnitude. We recover individual demand consistent with this stark pattern of outcomes and confirm it is not collusive, but consistent with non-cooperative behavior. Demand estimates are biased if one ignores these features.
![]() |
![]() |
Substitutes | Complements |
- “Complements and Substitutes in Sequential Auctions: The Case of Water Auctions” (Appendix), with Javier Donna, The RAND Journal of Economics, 2018, Vol. 49, No 1, pp 87-127.
We are thankful to Fernanda Donna and Antonio Espín for their assistance. We are also thankful to Juan Gutiérrez for his help in finding the data and to Antonio and Pepe at the Municipal Archive in Mula.