tion with regards to implementing the proposed water allocation systems in England. 2495
In particular, identifying which farmers are more or less likely to engage with particular 2496
aspects of the proposals, such as water licence trading, may be used in deciding which 2497
catchments should be categorised as basic or enhanced. 2498
2499
Within the wider sphere of policy decision-making, behavioural typologies clearly 2500
offer greater value in supporting policy formulation compared to more traditional ty- 2501
pologies. However, limitations also exist with regards to the added complexity involved 2502
in developing a behavioural typology (i.e. measuring individual behavioural intentions), 2503
and the effects of the intention-behaviour gap. In this sense, behavioural farm typolo- 2504
gies may be more valuable for policy formulation (Guillem et al., 2012) rather than 2505
continual monitoring of policy interventions, and policy decision-makers should not 2506
rely solely on behavioural intentions but also on more traditional farm characteristics 2507
(Schmitzberger et al., 2005; Emtage et al., 2006, 2007; Gorton et al., 2008; Pike, 2008; 2508
Barnes et al., 2011; Poppenborg and Koellner, 2013). 2509
2510
7.3
Utility of the Agent-Based Model (ABM)
2511
This research has highlighted the importance of understanding farmers’ preferred be- 2512
havioural intentions with regards to the proposed water allocation systems in England. 2513
Furthermore, understanding the system level patterns of abstraction behaviour which 2514
could emerge from individual decision-making at the farm scale provides valuable infor- 2515
mation for policy decision-makers. However, although the ABM developed highlighted 2516
the potential effects of the proposed water allocation systems, several limitations of the 2517
model exist. 2518
2519
One of the main limitations of this model was its ability to only model in-season 2520
(short-term) water shortage and surplus preferred behavioural strategies, during a se- 2521
lected dry and wet year climate scenario, rather than model behaviours over a longer 2522
time period. With the latter approach, strategic (long-term) water shortage and sur- 2523
plus preferred behaviours could have been implemented, despite the added complexity 2524
of the model, to highlight the long-term effects of proposed policy changes. If the model 2525
had incorporated strategic behavioural strategies, then all farm types, during strategic 2526
(long-term)/water shortage scenarios, would have simply increased their application 2527
efficiency which would ultimately reduce abstractions at the system level during the 2528
growing season, regardless of policy or climate scenario. However, during strategic 2529
(long-term)/water surplus scenarios, more water would be abstracted during the grow- 2530
ing season as a result of those who preferred high water usage options (HWUO) growing 2531
the same crops but over a larger area. Furthermore, as the strategies for those who 2532
preferred low water usage options (LWUO), and those who had no preference, were to 2533
change nothing and sell surplus water for the duration of the growing season respec- 2534
tively, very little in regards to overall system level patterns of abstraction would have 2535
changed, as no trading occurred as buying was not a preferred behavioural strategy of 2536
any of the farmers. However, this highlights another limitation of the model. 2537
2538
The purpose of the model was to understand the potential system level patterns 2539
of abstraction behaviour which were likely to emerge based on farmers’ preferred be- 2540
havioural intentions. However, as none of the farmers’ preferred behavioural intentions 2541
involved buying more water, either strategically or in-season, no trading occurred in 2542
the model. Therefore, only the potential for trading at the system level was simulated 2543
based on surplus water available and shortage water indicated. Furthermore, although 2544
previous studies, such as that in southeastern Australia, have found that an increas- 2545
ing number of farmers participated in temporary water licence trading (i.e. weekly, 2546
monthly, or seasonal), very few participated in permanent water licence trading due to: 2547
differential tax treatment; policy uncertainty; institutional barriers such as administra- 2548
tive complexity and costs; and farmers’ perception that water rights are an inherent 2549
part of their property (Bjornlund, 2003). Nonetheless, the results of the simulations 2550
highlighted that even if trading was available, very few trades would likely occur, even 2551
if farmers growing seasons would be staggered, as very little overlap would exist with 2552
regards to when surplus water was available and shortage water was required. There- 2553
fore, despite these limitations, the results of the model would suggest that farmers, who 2554
do not already have storage available, are best investing in storage to make use of the 2555
surplus water available out-of-season. 2556
2557
Another limitation of the model concerns the lack of planning and learning to change 2558
adaptive strategies. A more sophisticated model would perhaps incorporate the abil- 2559
§7.3 Utility of the Agent-Based Model (ABM) 113
ity of farmers to better plan their irrigation water needs (IWN) during the year. In 2560
the current model, this is simply divided by the number of months their licence was 2561
available. However, in reality farmers know, approximately, from previous years when 2562
demand is likely to be greatest and when it is likely to be less. Furthermore, if farmers 2563
continually experienced particular water shortage or surplus scenarios they may very 2564
well change their behavioural strategies, which would certainly alter the system level 2565
patterns of abstraction behaviour. Therefore, a model which simulates abstraction be- 2566
haviour over consecutive years could incorporate planning and learning from previous 2567
seasons. Data regarding how farmers actually plan their abstractions each year could 2568
be obtained during focus groups used to measure salient beliefs. 2569
2570
Overall, the model presented, as with any, was a simplified representation of reality. 2571
Nonetheless, despite the limitations discussed, the scenario simulation results indicated 2572
that the proposed enhanced water allocation system was likely to achieve its intended 2573
aim of increasing efficiency whilst balancing the needs of licence users and the environ- 2574
ment, more than the current or proposed basic water allocation system. Furthermore, 2575
incorporating other licence users would likely open up greater opportunities for water 2576
licence trading. However, although the proposed enhanced water allocation system was 2577
designed to facilitate trading more easily than the proposed basic system, the method 2578
used to derive water shares under the enhanced system may result in fewer trades as 2579
licences adjust to reflect users needs, and therefore less surplus water would be available 2580
compared to the proposed basic system. 2581
2582
Previous studies have also used ABM to simulate different policy or climate sce- 2583
narios related to individual behavioural strategies and changes in land use or resources 2584
(Schl¨uter and Pahl-Wostl, 2007; Gal´an et al., 2009; Acosta et al., 2014). However, one of
2585
the main limitations that each of these studies concluded with regards to ABM concerns 2586
the fact that as the models are developed based on behavioural strategies of particular 2587
groups of individuals, often within a particular geographic location, then generalising 2588
the model for a wider population or different geographic location may be difficult. 2589
Therefore, although the simulation results presented in this study are designed based 2590
on farmers currently operating within the Great Ouse catchment, the framework of the 2591
model, after adapting it to another area, could be generalised to other catchments. 2592