I’ll present this paper in a week or so at the Water Forum, Wodonga. Comments more than usually welcome.
Drought, Uncertainty and Water Supply Planning
Visiting Fellow, Department of Economics, University of Melbourne, Parkville 3052.
Abstract. Water supply planning when there is a threat of uncertain drought is discussed in terms of anticipatory and adaptive policies as well as the links between these classes of policy. Sound anticipatory policies tend to be low cost and to have complementary effects in enhancing the productivity of adaptive policies as well as in promoting policy flexibility.
Droughts occur as a core part of Australia’s climate. The problem of addressing drought by means of pre-drought anticipatory and post-drought adaptation (including recovery) policies takes broadly the same form for both urban and agricultural sectors. The core forecasting issues for both sectors is for public and private sectors to cost-effectively minimize the impacts of uncertain drought events which have uncertain strength and duration. Anthropogenic climate change is likely to increase the frequency of droughts as well as their duration and intensity. This is a crucial, medium-term issue. But many core policy problems stem from high pre-existing Australian climatic uncertainty even without invoking climate change. Some climate change evidence suggests we may be moving towards a drier environment and perhaps an environment when a regime shift in the character of water supplies occurs: Internalising such thinking increases the benefits obtainable from sound policy but sound policy design involves “no regrets” options whose benefits increase as climate change occurs but which involve important economies even without it. Indeed it is clear that the evidence itself is mixed and regionally specific with some finding limited drying trends from 1950-2000 mainly in south eastern Australia but evidence of decreasing dryness in northern Australia. The recent report by Steffen (2015) confirms this picture – the major future impacts of drought will occur in south-western Australia and the country’s south east.
It is important to understand also that several Australian megadroughts – droughts lasting two decades or more – have been reported in the paleoclimatic record over the last millennium (IPCC, 2012, p. 170-171). The possibility for such events is not new.
Risk and uncertainty are intrinsic to the drought management question. It is not enough to simulate various possible future scenarios and to devise policies that are appropriate to each. When deciding on drought policies it is imperative to account for the fact that both anticipatory and adaptation options might get it wrong and hence on the need for caution in devising policy and for retaining policy flexibility. It is important also to understand that the effectiveness of adaptation options is impacted on by the effectiveness of anticipatory actions, an issue pursued in the economics of disaster relief literature (Clarke, 2015), and also discussed below.
Both anticipatory and adaptation policies must deal with risk and uncertainty. For anticipatory policies this risk and uncertainty arises from the likely incidence and duration of future droughts. For adaptation (including drought recovery) policies, risk and uncertainty arises from need to be forward looking and to anticipate the likely duration of a pre-existing drought as well as future droughts and water supply situations.
Both classes of policy can involve the determination of appropriate triggers to modify the way long-tailed events are incorporated into reliability assessments for standard large-scale, water infrastructure investments.
In political economy terms it is easier to gain support for accessing financial resources to address a drought once it has occurred rather than ex ante when it merely might plausibly occur. However there are substantial economies possible from initiating policy responses early. Clearly the productivity of ex post measures will be impacted on by policies taken ex ante.
Strongly risk-averse policy makers face restricted policy choices if they rely exclusively on ex post measures. Sound anticipatory policies deliver policy makers greater adaptation flexibility given that they would otherwise be forced to make choices under the pressures imposed by urgent post-drought political realities and policy myopia. A contribution of economic analysis can be to devise policies that relax the constraints that present themselves when a drought has arrived.
One class of anticipatory water supply policies are large, lumpy and irreversible infrastructure investments. With respect to such investments it is bad practice to deal with policy risk in a cost-benefit appraisal of options simply by replacing random climatic and policy payoffs by their expected values. Certainty equivalence procedures make no sense here. The irreversible character of certain anticipatory policies is crucial – desalination plants, water recycling schemes and dams (and indeed many public investments!) cannot be turned into amusement parks! Once such large-scale investments are undertaken they often (not always) have low-valued alternative uses outside their specific water management roles. Given risk, the possibility of future learning about commercial viability and risk-neutral policy preferences, creates a quasi-option value that provides arguments for delaying construction beyond the date that suggested by standard expected cost-benefit analysis. Indeed the requirement that expected benefits merely exceed expected costs needs to be strengthened so that quasi-option value is included as a component of costs. This is the “real options” viewpoint of, for example, Dixit and Pindyck (1994).
A key assumption here is that decision-makers are risk neutral. This can be motivated by the Arrow-and Lind (1970) theorem applied to public investments: If there are many small investment projects being assessed and their probability distributions are independent then it makes sense to maximise expected payoffs since losses and gains will “average out”.
The assumption of “independence” is questionable when thinking about drought risks. El Niño phenomena create patterns of drought that cover large parts of Australia at the same time. For drought management policy-makers the underlying assumption of risk-neutrality required to implement both certainty equivalence and “real options” approaches, is inapplicable since they will place extra weight on very bad policy outcomes. Such risk-averse preferences imply policy makers will pay an insurance premium to avoid large social and economic stresses created by extreme droughts. Therefore, from the viewpoint of policy preferences, projects may be initiated earlier than would those selected using expected values in situations where expected benefits may even fall short of expected costs inclusive of a quasi-option value by a risk-premium. This is the “insurance viewpoint” that to many policy makers makes as much sense as views based on “risk-neutrality”.
The implications of the alternative “risk neutral – real options” and “risk-averse” insurance viewpoints drive the cost benefit case for large infrastructure projects policies in opposite directions. They create a numerical tug-of-war that determines which policy direction wins. Thus the case for capital-hungry desalination technologies is constrained by the scale of such large risky investments and by the prospect of learning more in the future about their viability and need. Given risk neutrality it suggests caution. However, risk-averse policy preferences are likely to favour early anticipatory responses and a willingness to proceed with projects that are not yet commercially viable in an expected return sense. Then projects should be implemented when expected benefits exceed expected costs (inclusive of quasi-option value) less the optimal insurance premium policy makers will pay.
How empirically important are the factors supporting the “real options” and “insurance” viewpoints? Revealed preference would suggest that insurance effects dominated in formulating actual policies decisions on desalination in Australia – the “go ahead” on such projects was implemented in all Australian capital cities while the “real options” approaches endorsed by groups such as the Productivity Commission (2011) suggested, if anything, the case for more policy restraint. Can the factors bearing on the theoretical indeterminacy here be narrowed down?
A key issue motivating the real options approach is the existence of learning so that a key issue is how much we can expect to learn about the determinants of drought severity and the length of their duration is crucial. Such knowledge drives our ability to forecast droughts. A constraint on our ability to learn about the key severe lengthy droughts that we are most interested in forecasting is that past experience provides few clues. These events are “long-tail” events that incur relatively infrequently so statistical evidence on their determinants is almost inevitably going to be poor. Pressures arise because such events are high cost so there are substantial pressures to act on poor information. The same too is true of attempting to forecast a regime switch, perhaps as a consequence of climate change, toward more frequent and more severe and lengthy droughts. There is no evidence that such a shift has yet occurred and only limited evidence that climate change is having such pervasive long-term effects perhaps through strengthened El Niño Southern Oscillations.
If waiting for a decade or more is unlikely to yield extra knowledge that strengthens our ability to forecast droughts and their duration then the “real option” viewpoint is redundant as is any case for waiting to improve our knowledge. The best policy prescription is to make decisions now on the basis of expected values and, if action is deemed inappropriate, to revisit the case for an active response over time.
The role of risk aversion in driving the demand for water supply insurance is also constrained by limitations in our current forecasting abilities and by scepticism over likely improvements in such abilities. We can certainly make contingent valuation estimates of how much citizens – and even politicians – would be willing-to-pay to insure against future water supply insecurity but, even then, unless we know the probability distributions of likely future droughts of different severities it is important to pose the right question that will pin down any precise measure of water supply security. Providing rainfall independent sources of water, such as desalination, does ensure adequate urban water supplies over time horizons that are long enough to allow eventual system expansion. It is important when considering such expansions to think about other risks that may impact on project viability such as higher future energy prices.
Before we switch to discuss adaptation policies that assist in dealing with drought once it occurs it is important to understand that such adaptation policies can (or should) be driven by prior anticipatory policies. Rational water pricing and policies that encourage demand management and water conservation are important in both urban and rural sectors. Drought preparation policies are specifically important in rural settings. These are substitute policies for ex post drought relief policies.
A specifically important class of anticipatory actions lies in providing information and developing management skills.
One issue is forecasting rainfall and drought outcomes. This is a low cost research option that has complementary effects in improving the productivity of other anticipatory policies.
Anticipatory policies also include information gathering and preliminary investments that would be essential prerequisites to larger scale policy actions were a severe drought situation to develop. Melbourne’s desalination plant at Wonthaggi took as long to plan and to assign contracts as it took to build. Many of the requirements for such a project can be undertaken prior to doing it – making engineering and economic choices among available options, securing land purchases, obtaining environmental approvals and so on. This increases option flexibility ex post and overall can reduce policy costs at the expensive of relatively low upfront costs.
Finally, it should also be recognized that private anticipatory decisions can forestall the potentially disastrous impacts of drought. Specifically if farmers anticipate the prospects of drought by constructing dams, limiting stocking levels and by selecting crops that, if threatened by drought, bring about only transitional losses, then again policy flexibility for policy makers ex post improves. The Productivity Commission (2011) (largely following the arguments of Freebairn (1983)) emphasised that drought assistance policy should primarily be directed towards improving the ability of farmers to themselves anticipate and improve their adaptation capabilities in relation to drought. These capabilities will depend on business and other skills that can be fostered on an ongoing basis.
The advent of drought creates a need for adaptation responses. The key forecasting issue is to take actions appropriate to the scale and expected duration of the drought. Not much is known about the latter although improved knowledge of the evolving scale of El Niño events and the autoregressive character of drought events provide some guidance. If policy-makers have no basis for forecasting the duration of a drought then it is likely that the appropriate policy will involve a desired sequencing of policies as well as their mix.
A key issue in assessing adaptation responses is the flexibility with which a range of options can be pursued. It can be predicted that almost any measure irrespective of its costs will be advanced by politicians – and perhaps civil servants – if a crisis situation is envisaged. Politicians face short term possible costs that might stem from drought events and face undue incentives to take inefficient actions whose costs get masked in overall tax/expenditure programs or distributed as interest costs on debt that are distributed over decades.
Drought relief policies have conventionally been used to assist farmers facing drought largely as a consequence of political pressures imposed by the farm lobby and by the short-term incentives politicians face. Financial assistance via drought assistance and low interest loans provide no incentives for farmers to take anticipatory actions and encourage farmers to remain in business and maintain such things as stocking densities even when climatic circumstances are adverse. The intention of drought assistance policies in Australia to encourage self-reliance are not being advanced by policies which improve the capabilities of adapting to drought – indeed incentives for self-reliance are thwarted by such policies . In 2007-08, during the Millennium drought, 23 per cent of farmers received public assistance totally over $1b (Productivity Commission, 2009). This assistance did not encourage self-reliance and encouraged environmentally unsound farm management practices.
Given self-reliance objectives the Productivity Commission opposed all such assistance on the grounds that it was not incentive compatible. If assistance was to be awarded it should at least be paid conditional on farmer’s undertaking their own private anticipatory policies to deal with drought via a “Mutual Responsibility Contract” which required of farmers an appropriate degree of anticipatory response. The “Exceptional Circumstances” (EC) criterion governing the award of drought assistance never makes sense since, apart from the fact that it was more generously interpreted than was planned, it did nothing to improve the self-reliance of farmers in preparing for, adapting to and recovering from drought. Under EC those who make preparations were unlikely to qualify for assistance so the incentives were to overgraze and overspend in good times and to then receive assistance when times were bad.
Other sound anticipatory policies are “no regrets” options that make sense without the prospect of drought but whose effectiveness increases substantially if drought occurs.
“Scarcity pricing” of water means charging more for water as it becomes scarcer. (It is interesting to note that water resource economics has developed its own jargon to cover common sense ideas in economics). As an efficiency-directed tool this is the best way of addressing supply demand imbalances in general but specifically for growing water shortages that stem from drought. The pricing can either be directed to clearing the water market or of efficiently allocating a stock of water with low renewal prospects over time. In the latter case the marginal cost of water from whatever backstop technologies are available – for example, dams, a desalination plants or water from recycling or aquifers – provide a terminal price that existing water prices should converge to at a rate reflecting the social discount rate.
Such scarcity pricing rather than pricing scarce resources cheaply facilitates postponement of expensive irreversible infrastructure investments thereby saving present value and increasing the prospect that such investments will become unnecessary because the drought breaks.
In rural settings freely tradeable sustainable water allocation permits along with steeply rising prices during periods of drought effectively mimic scarcity pricing initiatives.
Scarcity pricing is routinely rejected in urban settings because of the “special” character of water as an “essential good”. This ambiguous idea partly reflects the legitimate distributional concern that high water prices impact adversely on low-income households. Multi-part tariffs can address such distributional concerns but these pose efficiency issues because then a uniform water price is not being be charged to all.
The alternative to scarcity pricing is water rationing and water restriction policies. These have been widely used in Australia and to meet short-term and medium-term shortage issues. Economists routinely dismiss restriction policies as having high deadweight losses in the order of hundreds of millions of dollars but, as policies, sharing-the-burden schemes have a high measure of community acceptability and effectiveness.
The deadweight losses imposed by such schemes need to be assessed against the savings they create in terms of reducing the quasi-option value losses imposed by pre-mature infrastructure development.
The ideal is to devise a comprehensive package of drought policies that recognise links between anticipatory and adaptive policies. Adaptation decisions taken when the pressure of drought bears on decision making by those with short-term policy horizons can lead to hasty and costly policy responses that disadvantage the community.
Compared to infrastructure costs there are relatively low costs of R&D into the determinants of droughts and their character. Such R&D efforts can also be directed towards researching drought policy – though workshops such as the present one – and for improving understanding of candidate technological responses. Such efforts should be ongoing.
The productivity and effectiveness of drought relief adaptation responses, taken once a drought has occurred, will be improved if farmers are incentivized to themselves prepare for drought by making relief contingent on such preparation.
The economics of substantial infrastructure projects (desalination, dams, water recycling, use of aquifers, pipeline construction) can be substantially improved if they can be delayed using anticipatory policy. This gains option value. This can be achieved without sacrificing social insurance objectives if scarcity pricing of urban water supplies and water trading among rural users is set in place before a drought crisis. Triggers can be relaxed with effective demand management since the availability of extant supplies will be stretched out further if water becomes more expensive. Measures such as water supply restrictions and rationing remain viable medium term options if market-based controls are ruled out on ideological or distributional grounds. The efficiency costs of such policies needs to be assessed against their high social acceptability and their ability to deliver gains in terms of option value.
- Arrow and R. Lind, “Uncertainty and the Evaluation of Public Investment Decisions”, American Economic Review, 60, 1970, 364-378.
- Clarke, “Economics of Disaster Preparedness and Relief”, mimeographed, October 2015.
- Clarke, “Planning Urban Water Investments with an Uncertain Climate”, Economic Papers: A Journal of Applied Economics and Policy, 32, 4, December 2013, 426-439.
- Clarke, “Real Option and Insurance Approaches to Evaluating Infrastructure Projects under Risk and Uncertainty: A Checklist of Issues”, Australian Economic Review, 47, 1, 2014, 147-156.
- Freebairn, “Drought Assistance Policies”, Australian Journal of Agricultural Economics, 27, 3, 1983, 185-199.
Intergovernmental Panel on Climate Change (IPCC), Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation, Cambridge University Press, Cambridge 2012.
- Dixit & R. Pindyck, Investment Under Uncertainty, Princeton University Press, Princeton, 1994.
Productivity Commission, Government Drought Support, Final Inquiry Report, Report No. 46, Melbourne, 2009.
Productivity Commission, Australia’s Urban Water Sector, Final Inquiry Report, Report No. 55, Canberra, 2011.
- Steffen, Thirsty Country: Climate Change and Drought in Australia, Climate Council of Australia, 2015.