The framework is applied to a stylized case based on the Waal River in The Netherlands. The river is leveed and surrounding communities experience flooding and related damages if design conditions are reached and the river inflow exceeds 14,000 m3/s. Climate change and socio-economic developments may increase the pressure on the available space and the potential flood damages through increased river inflow, and thus motivate action. Four alternative actions are considered in this case to illustrate the concept of transfer costs. The actions are based on the main options considered in the Netherlands (Fig. 1, Supplement section 4): (i) raising the dike with 0.5 m (low dike); or (ii) raising the dike further with 1 m (high dike); (iii) providing more room for the river by widening the riverbed at a small scale; or (iv) at large scale. These actions combine to create six adaptation pathways, comprised of four starting points and two possible transfer stations that are implemented depending on the rate of climate change. The fast climate change scenario portrays a rapid increase in river inflow, due to increased intense precipitation and higher temperatures that result in more rapid snow melt. This creates higher design conditions expressed in terms of river inflows. The slow climate change scenario portrays a later onset of climate change. Accordingly, the timing of an adaptation tipping point depends on the climate scenario. Besides climate change, socio-economic developments are considered resulting in economic growth of 2%, urbanization, and increased value of the area, which affect the avoided flood damage and transfer costs over time as explained below.
Two pathways are “linear” and represent no future flexibility. These are as follows: build a high dike (pathway 1) or widen the river to the maximum extent (room for the river) (pathway 6). Two other starting points allow for a staged adaptation to future conditions: build a low dike and then transfer to either a high dike (pathway 2) or to small room for the river (pathway 4), or start with small room for the river and then transfer to either a low dike (pathway 3) or large room for the river (pathway 5).
Each action is characterized by the amount of river inflow the action can accommodate without flood damage. This flow capacity determines the adaptation tipping point of an action or pathway as it is the threshold for achieving the objective; no flood damage at design conditions. An adaptation tipping point thus occurs if, in a particular scenario, the river flow exceeds the flow capacity of an action resulting in flooding; a new or additional action is thus needed. All pathways achieve objectives and thus result in the same river inflow capacity that can be accommodated without any flood damage.
The relative costs and benefits of the case are based on real world data, wherein raising a dike is less expensive than providing more room for the river, and raising a dike in two steps is more expensive than doing this at once (see Supplement section 4). The benefit of each action and pathway is assessed in terms of avoided flood damage compared to the current situation, assuming a linear increase of flood damage from current capacity of 14,000 to 20,000 m3/s when climate changes over time. This increase occurs over period of 80 or 100 years in, respectively, the fast and slow climate change scenarios (see time axes in Fig. 1), similar to expected changes of the design conditions for the Waal river in the Netherlands (Haasnoot et al. 2013).
The avoided damage is calculated by multiplying the extra (compared to current practice) river flow the pathway can accommodate with an assumed amount of avoided damage per discharge (0.75 avoided damage per m3/s extra river inflow). The additional flow the action can accommodate is corrected for the actual maximum discharge that can occur in the considered time period for a given scenario. As a result, the avoided damage is higher in a severe (faster) climate change scenario than in a mild climate change scenario (56 units/year avoided damage for the fast climate change scenario and 45 units/year for the slow climate change scenario). In the end, all pathways can accommodate 6000 m3/s extra discharge for a total of 20,000 m3/s. The moment at which this occurs depends on the scenarios. The “room for the river” actions have additional annual environmental benefits. These benefits are 0.5% of the avoided damage for small scale room for the river and 0.7% for large scale room for the river. The total benefits—monetized in a non-specified hypothetical monetary unit—are assumed to grow over time (at 2% per year) as the value of the protected area increases due to economic growth of 2% per year.
The transfer costs, considered in this case, represent the costs for removing and rebuilding urban area to implement the room for the river actions that require space. Over time, the area becomes urbanized in response to the flood management actions that protect the area from flooding if no actions are taken to reserve space. In addition, the value of existing urban area increases due to economic growth resulting in more costs to remove and rebuild infrastructure in another area. The transfer costs to shift from a low dike to small room for the river are considered to be larger than for a shift from small to large room for the river as the first transfer has more implications for the urban area than the second. To represent impacts of urbanization and economic developments, transfer costs are assumed to increase over time (see Supplement section 3.1).
The NPV of each pathway is calculated for two climate change scenarios with a 3% discount rate, consistent with the 2% economic growth per year. A sensitivity analysis has been done for the assumptions made in the case, including discount rate, economic growth, transfer costs, and the additional benefits of the room for the river actions (see Supplement section 5).
To illustrate the effect of the choice of evaluating alternative single-investments or alternative pathways for the ranking of the investment decisions, two time horizons, 40 and 80 years, are considered for the evaluation. A 40-year period does not encounter an adaptation tipping point and implements only one of the four initial options (low dike, high dike, small room for the river, large room for the river) and does not consider potential switching to other options, and thus also no transfer costs. An 80-year period encounters an adaptation tipping point for the low dike and small room for the river, allowing for the switch or transfer of management to large dike and/or large room for the river (Fig. 1).
Adaptation tipping points and transfer costs occur for pathway 4, which starts with a low dike and then shifts to creating small scale room for the river, and for pathway 5, which starts with small scale room for the river and then shifts to large scale room for the river to avoid flooding caused by increasing river inflow in response to climate change. A transfer later in time is more costly than the one early in time, due to economic growth and increased urbanization. In pathway 4, the raised dike has to be relocated to implement the small scale room for the river, and houses must be relocated. The cost of house relocation is likely to increase over time with economic growth and with urbanization (if urban development is not prevented in this area), especially as the area becomes safer as a result of to the raised levee. The latter is known as the levee-effect; people and assets tend to accumulate in protected areas, in turn, requiring higher protection, in a feedback loop (De Moel et al. 2011). Pathway 5 has transfer costs to enable the implementation of additional room for the river measures later on.
Over a 40-year period, wherein only the initial action of each of the pathways is implemented, the low dike pathway has the highest NPV (Fig. 2). In contrast, the more expensive options, such as a high dike and large scale room for the river, rank lowest; they are an overinvestment within this short-time horizon. With the short-time horizon, the ranking is the same in both climate scenarios.
Over a longer time period (> 40 years), an adaptation tipping point is reached for the low dike and the small scale room for the river as too much flooding damage occurs. The functional lifetime of these initial actions is thus exceeded and some transfer costs will be incurred, thereby changing the NPV ranking of the pathways. Over an 80-year period, pathway 3 ranks highest, i.e., the best initial investment is small scale room for the river and a transfer to a low dike at the adaptation tipping point, while in the 40-year period, wherein no transfer occurs, the low dike ranks first. This result illustrates how a policy action preferred in the short-term may become suboptimal over a longer time horizon that requires transfer costs associated with adaptation to new conditions. In other words, it is economically more efficient to add a dike to the room for the river under uncertain climate change, rather than starting with a low dike which prevents an initial, long-term economic investment in room for the river. Note that pathways 3 and 4 have the same end situation in terms of actions implemented, but differ in NPV due to a different sequence of the actions that influence the transfer costs, and timing of the amount of investments. Without transfer costs, they rank similarly, despite that the higher short-term costs for its starting action of pathway 3 (see Supplement section 5).
The situation is similar to the Netherlands, where there is a long history of raising dikes to prevent the increasingly populated delta from flooding, generating significant path-dependency. Despite the intention to provide more room for the river, from a cost-benefit perspective, this action is generally not viable due to lower costs of incremental dike heightening and the high transfer costs of relocating activities to make room for the river. These additional costs (that often grow over time) are part of the room for the river investment cost, making it economically unviable to invest in room for the river compared to raising the dike. To avoid further lock-in, transfer costs of room for the river could be mitigated by preparatory actions, such as zoning to prevent urban development in areas that might later serve as flood zones. Also, if the additional benefits of enhanced quality and quantity of ecosystem services and recreation provided are included, then the room for the river option becomes more attractive. Whether and how transfer costs and additional benefits are considered and at what timeframe is used are often a political choice but can highly determine the outcomes of an evaluation of options.
Our analysis also shows that the ranking of the policy actions and pathways can be sensitive to the projected magnitude of climate change. For example, under a slower rate of climate change, pathway 4 performs less well than pathways 5. However, if climate change occurs more rapidly, both pathways are equally appropriate long-term strategies. Differences in the ranking of pathways between the slow and fast climate change scenario will also occur if the chosen time horizon is such that the slow climate change scenario does not have an adaptation tipping point and related transfer costs, while the fast climate change scenario does (for example 2045 in the case presented here). As discussed earlier, since time horizons for evaluation are often fixed and not related to the operational life time of investments, this can have consequences on the choice of actions.
In the case presented here, pathways that include multiple decisions and have the flexibility to adapt over time score best. Still, an action that limits losses in all scenarios whatever the cost (large room for the river in the short-term) could be preferred if there is a window of opportunity (for example political or economic) to invest in the short-term and a risk of lack of funding or support in the future. In New Zealand, for example, stakeholders preferred the implementation of a more robust action today even if it might be an overinvestment because of lack of trust that further actions will be implemented in the future (Lawrence and Haasnoot 2017; Lawrence et al. 2018). This uncertainty about the future ability to implement action could be treated as an uncertainty in the analysis.
In addition to the impact of the time horizon, transfer cost, and climate change, we further explored the sensitivity of the outcomes for assumptions used for additional benefits, economic growth, and discount rate (see Supplement section 5). In practice, additional benefits such as ecological values are difficult to quantify and doing so may be a political choice, but including them can alter the ranking of the pathways significantly. Without considering the ecological values for the room for the river actions, implementing a small dike becomes the preferred short-term option. Different values for economic growth and discount rate will alter the NPVs but not the ranking, if they are both changed consistently following the Ramsey rule (Barro and Sala-i-Martin 2003) as recommended by the World Bank (World Bank OPSPQ 2016). Using a higher discount rate and not changing the economic growth accordingly (for example to represent higher pure preference for the present) may change the ranking, as it lowers the contribution of future transfer costs, resulting in less difference between pathways 3 and 4, and also lower future benefits resulting in pathway 2 that will rank highest. Increasing the economic growth does not alter the ranking, but increases the difference between the pathways with and without these costs, as in this case, the transfer costs are affected by the economic growth.
Whether the approach makes a difference for ranking the pathways thus depends on the relative amount of transfer costs compared to costs of the actions themselves, when they may occur and the extent they are related to socio-economic conditions. From the Optima dam and the Maeslant Barrier, we have seen that these costs can be considerably and should thus be considered and estimated.