In this section, we present an economic welfare analysis of avoidance and aversion behavior. The intent is to demonstrate that behavioral responses to disasters, even exaggerated ones, are amenable to such analyses, and hence can legitimately be included in BCA.
The workings of the three types of behavioral shifts presented in Section III as an underpinning for these phenomena could be analyzed using standard consumer theory (or production theory for the case of businesses) and expected utility theory. However, we skip this step and proceed to the more straightforward result of these behavioral shifts in terms of changes in supply and demand, since these are more operational aspects of the analysis in terms of actual measurement of behavioral linkages.Footnote 18
We illustrate examples of Mandatory Avoidance, Voluntary Avoidance, and Aversion Behavior in the figures below in terms of standard analysis of Welfare Maximization. Figures 2, 3, and 4 depict the change in consumer surplus and producer surplus, the combination of which represents the change in total economic welfare due to behavioral responses. In Fig. 4, this is dichotomized into cases of a shift in the demand curve and a movement along a truncated version of it.
Figure 1 is the standard market welfare analysis diagram. The market is in equilibrium at the intersection of Demand (D) and Supply (S), with the equilibrium quantity demanded (Q*) and equilibrium price (P*) determined simultaneously. This equilibrium is also an optimal allocation of resources since it maximizes the sum of consumer surplus (CS) and producer surplus (PS).
The case of Mandatory Closures, as in response to the COVID-19 pandemic, is very simple if the closures are complete. In that case, a truncated supply curve would be a vertical line at the zero axis, effectively indicating that no supply of the good or service is forthcoming. The welfare loss would be the entirety of the CS + PS. To illustrate the situation more generally, however, we refer the reader to Fig. 2, which depicts the case where some supply is forthcoming even under mandatory closures, which could arise, for example, in the financial service industry, where telework is an extensive option. In this case, the new supply curve is upward sloping (moving along the lower portion of S1) to the output level limit (Q2) but then becomes perfectly vertical (S2). (Of course, the marginal cost could also have increased, in which case the upward sloping portion of the supply curve would be higher than before until it goes vertical, but that does not significantly affect the analysis.)Footnote 19 The new equilibrium is the point at which S2 intersects the demand curve, D.Footnote 20 The shortage results in an increase in price from P* to P2 if prices are allowed to rise.Footnote 21 The welfare loss is that part of Consumer Surplus to the right of S2 plus that part of producer surplus to the right of S2. One interesting outcome is that the remaining consumer surplus (CS2) is now very small because of the price increase, and the new producer surplus (PS1 + PS2) is now proportionally larger than its consumer counterpart, as depicted by the crosshatched blue/green area to the left of S2 and bounded by P*and P2 (just the opposite of the original pre-closure equilibrium in our example). This also means, of course, that consumers have lost much more welfare than have producers, and that producers have captured some of the former consumer surplus. If the demand curve were very steep, however, this proportionality outcome would be reversed. However, it is unlikely that this market would be characterized by a very inelastic demand because that would be more consistent with the good or service being a necessity or “essential,” and hence less likely to be subject to mandatory closures in the first place.
The case of Voluntary Avoidance would be identical to that of Mandatory Closures as exhibited, if the avoidance were complete, which would involve the total elimination of both consumer surplus and producer surplus as in the case of the total truncation in supply under the initial Mandatory Closures case.Footnote 22 Again, however, we examine a more interesting case where avoidance is partial in Fig. 3. In this case, the supply curve remains fixed, but the demand is limited, such that it truncates at the level of output Q2 and is depicted as the vertical demand, D2, after that point, with the new equilibrium again being at price P2. This assumes that the high demanders do not change their demand, but it suddenly drops at Q2. (Alternatively, the entire demand function could shift downward so that only a small quantity is demanded, where the intersection might be a little above point C, and price would fall dramatically.)Footnote 23 Ironically, however, for the truncated case, the outcome is the same as in the Mandatory Closures case as depicted in Fig 2. Moreover, we can offer a similar insight regarding the elasticity of demand for cases other than strong necessities, which would represent a meaningful subset here. The difference is that the demand truncation is not likely to be as extreme for these goods; hence the producer and consumer surplus losses would not be as great. Moreover, there is the possibility that the more inelastic demand curve would result in the remaining consumer surplus exceeding that of the remaining producer surplus. Note that the demand change at output level Q2 in this case is at the same level as in the Mandatory Closure case. Q2 for the case of Voluntary Avoidance is likely to be at a higher level of output, however, because the closures in this case should allow for some responses that are not dominated by fear.
We split the case of Aversion Behavior into two parts. The first is depicted in Fig. 4A, which exhibits a downward shift in demand for the good (D2). For example, consumers are less likely to demand restaurant services at a given price in the aftermath of a dirty bomb attack or during and after a pandemic (in contrast to the avoidance cases, which are not price-sensitive). If the demand shift were parallel, then the reduction in consumer surplus and producer surplus would be proportional between the two groups), with the amounts determined by the new equilibrium intersection of demand and supply. However, we have drawn the new demand curve as relatively steeper to depict the hesitancy of consumers to partake of the good or service under dangerous conditions. In this case the remaining consumer surplus proportion is relatively larger than that of its counterpart producer surplus because of the price decrease.
We also consider the case where there is no shift in the demand curve for the good or service in question, but rather a limit in the quantity demanded. This could arise if there is some threshold level of contamination or exposure that can be tolerated. This results in a truncated demand curve similar to that depicted for the Voluntary Avoidance case (recall Fig. 3). We have drawn the limit at a higher level of output, however, than in the Voluntary Avoidance case, because this case is price-sensitive, thereby allowing for more latitude. In this case the new price would be at P2, though this does not convey the typical equilibrium properties of a demand and supply curve analysis; there is no reason the producers would lower the price to the intersection of D2 and S, if the marginal consumer is willing to pay P2. The outcome is similar to that in Fig. 3, in that the remaining producer surplus is larger than the remaining consumer surplus. However, we acknowledge that the aforementioned is only one of three possible interpretations. At the other extreme is the possibility that suppliers would bid down the price of the product because of an implied excess supply at P2 (though the fact that the supply is beyond the limit of demand makes this less plausible). In the extreme version of this case, the price would drop to the intersection of D2 and S, where the total reduced surplus is the same as in the first interpretation, but with consumers capturing the lion’s share. The third possibility is analogous to the outcome of oligopoly pricing. In effect, the demand curve drops from the point of its aversion limit to a horizontal zero level thereafter, thereby becoming a “kinked” demand curve overall, with the outcome being indeterminate somewhere along the vertical gap. But, again, total surplus is the same as in the other two cases; it’s just the distribution that differs.
Some general results are apparent from a comparison of these figures. First, behavioral responses are likely to result in greater losses of surpluses in the case of Mandatory Closures than in the other two cases, because of the more dramatic truncations in the former, ceteris paribus. In contrast to the other cases, Aversion Behavior is more likely to have two effects: a downward shift in the demand curve, and then a movement along it but only up to a limit and then an ambiguous equilibrium. Note also that we have simplified the analysis for sake of exposition, as the same event could trigger voluntary Avoidance behavior to some of those affected and Aversion behavior to others (see, e.g., Giesecke et al. 2012).
Note that since the direct behavioral reactions enumerated in the previous section can be interpreted in terms of demand and supply functions, they are also consistent with the basic conceptual approach of BCA. They are in fact amenable to estimation of standard economic welfare measures (metrics), in terms of approximations of consumer and producer surplus compensating variation and equivalent variation. This augurs well for their inclusion in BCA. In ECA, their inclusion is not dependent on welfare analysis; however, supply and demand shifts can simply be evaluated in terms of their impacts on GDP or income indicators that are deemed appropriate for ECA but not necessarily consistent with welfare measures.
Moreover, the general equilibrium (indirect, or supply-chain) effects of behavioral responses can also be measured in the standard manner as performed in non-disaster contexts, though there may be limits as to the extent that they can be legitimately included in BCA (U.S. EPA Science Advisory Board 2017; Farrow and Rose 2018). We know at the outset of this section that the legitimacy of this inclusion is often broader than intimated in standard benefit–cost analysis texts. For example, Boardman et al. (2018) advise against the inclusion of indirect effects because they do not account for the siphoning off of resources from existing activities. Only later in the exposition, and appearing as an afterthought, is it mentioned that this applies only to the case of full employment. We emphasize the asymmetry, however, of positive and negative economic impacts in the context of disasters. Most standard BCA texts focus on project evaluation or policy analysis, typically with expansionary ramifications, where the existence of full employment or the possibility that the project or policy will exceed the available labor (or capital) supply are serious concerns. Disasters are just the opposite, where the impacts of a negative effect of a decline in economic activity doesn’t usually run up against a labor supply ceiling, unless there are an extensive number of deaths/injuries or when Mandatory Avoidance is invoked. Thus, the standard caveat does not usually apply.
Direct estimation methods can be used to measure the reduction in demand and supply associated with behavioral responses to disasters. These methods include using actual data, adjusted for background conditions such as detrending for the on-going recession estimating the decline in airline travel in 2001 (e.g., Rose et al. 2009), or the use of survey methods to measure avoidance of mass gathering locations during COVID (e.g., Byrd and John 2021; Walmsley et al. 2021a), or experimental methods to measure the price inducements pertaining to aversion behavior (e.g., Giesecke et al. 2012).Footnote 24
Standard applied general equilibrium models can be used to measure the indirect effects of the behavioral responses, as in the aforementioned studies. The more difficult estimation pertains to resilience adjustments, typically in terms of spending on substitutes. This involves bringing in another market and having knowledge of substitution possibilities, but this is not a heavy lift using data transfer methods. The recent COVID pandemic relates to a variation on this theme—people do not necessarily substitute other goods for those not available during the mandatory business closures, but simply defer spending to a later date, i.e., pent-up demand (Walmsley et al. 2021a, 2021b).
In concluding this section, we note an important aspect of a broader analysis of avoidance and aversion behavior. These behavioral responses may lead to large welfare changes from non-market effects, such as lives saved, that should be evaluated, though separately (Rose 2021a, b).