Cleaner Nudges? Policy Labels and Investment Decision-making.
|Report - Statistical table
Many governments have started to incorporate the lessons of behavioral economics in their policies. These behavioral interventions (or "nudges") are characterized by their non-pecuniary nature, including appeals to social norms, information provision, default options, and cash transfer labels (e.g., Schultz et al, 2007; Homonoff, 2013; Allcott, 2011; Allcott and Mullainathan, 2010; Ferraro and Price, 2013). The attractiveness of these interventions lies in the fact that they are simple and often inexpensive to implement, but at the same time produce considerable changes in behavior. Nudges have been found to be successful in a variety of settings, improving healthy and pro-social behaviors (e.g., Gine et al., 2010; Breman, 2011; Schultz et al., 2008), adoption of better technologies (Duflo et al, 2011), saving rates (e.g., Madrian and Shea, 2001; Thaler and Benartzi, 2004; Chetty et al., 2009; Chetty et al., 2014) and energy efficiency and use (e.g., Allcott and Rogers, 2014; Ayres et al., 2013; Jessoe and Rapson, 2014; Reiss and White, 2008; Brown et al, 2013).
Among these interventions, labeling manipulations have attracted the attention of social scientists and economists in recent years Newell and Siikamaki, 2014; Swartz et al., 2011; Mathios, 2000; Fischer, 2008; Kallbekken et al., 2013). Labels are often attached to cash transfers that, although unconditional to any specific use, are given a suggestive name with the intention to nudge recipients into socially desirable behaviors or to provide further information to consumers. Example of these types are the child tax credit in the U.S. or child benefits in the UK. The literature typically studies the effectiveness of the label in promoting such desired behaviors and stress that standard economics would predict unconditional labeled cash transfers to be equivalent to an unconditional unlabeled cash transfers. (1) However, recent evidence suggests that labeled transfers seem to be spent more than proportionally on the items suggested by the label (e.g., Beatty and Tuttle, 2014; Kooreman, 2000; Beatty et al., 2014; Blow et al., 2012).
This paper is different in that it asks whether labels may alter decisions on products related to the labeled good. The theory which underlies the disproportional impact of the label on the labeled good is mental accounting (Thaler, 1990). Mental accounting argues that households assign their income to categories of expenditure, thus when income is received that is labeled, it is disproportionately assigned into the labeled category. However it is unclear how individuals make "accounts" in their head and thus which goods the label impacts become an empirical question. Even in the case that it is clear that the account includes the goods in question, how related goods are impacted depend on the magnitude of counteracting income and substitution effects. The literature cited above classifies the labeling effect as encouraging behavior that policymakers would like to occur although labeling could be used for other purposes like agenda setting. Should the effect of the label "spillover" to decision on other goods, it becomes less clear that these altered behaviors are the decisions policymakers would like to encourage.
The UK Winter Fuel Payment (WFP) is an unconditional cash transfer into an individual's current/checking account designed to combat the excess elderly winter mortality and morbidity associated with cold indoor climates. It provides households, which have a member 60 years of age (or older) in the qualifying week of a given year, with a lump sum annual payment. A person who turns 60 in a given year, but after the qualifying week, does not receives the payment. The WFP is not means tested nor is it mandated that the payment be spent on fuel. Though the WFP transfers cash that could be utilized for any expenditure, the label of the transfer induces households to use a larger portion of it to pay their energy bills than a non-labeled transfer (Beatty et al., 2014). The rationale for this behavior is based on the framework of mental accounting. We interpret the theory that labeled income being assigned to a category as implying that households treat the WFP like a reduction in the price of energy. This affects directly the amount of money spent for fuel and can have indirect effects on substitute goods too. For instance, perceived lower price of energy might induce households to substitute away from more energy efficient technologies or renewables. While renewables are different than energy efficiency measures, and renewables are sometimes installed for different reasons (Andor et al., 2017), they have the potential to achieve the same goal set by the WFP, i.e., keeping elderly warm in winter, while also reducing the negative externalities from emissions. Our analysis shows that the WFP label has the unintended consequence of reducing the propensity to install renewable technologies. (2)
The identification strategy is based on the sharp eligibility criteria of the WFP which allows for an estimation of the casual impact of the WFP on the propensity to install renewable energy with a regression discontinuity design (RDD). (3) This analysis uses repeated cross-sections to compare thousand of households who are bom just before or after a qualifying week in each year from 2008 to 2011. The identifying assumption here is that these households bom few months apart are similar in their observed and unobserved characteristics and would have behaved similarly with respect to investing in renewable technology in the absence of the WFP. In other words, this assumption ensures that non-recipients represent a valid counterfactual to recipients near the discontinuity point (i.e., local randomization assumption).
Valid RDD estimation requires that other potential explanatory variables that affect the outcome are continuous around the treatment discontinuity point (see, e.g., Hahn et al., 2001 and Lee and Lemieux, 2010). We note that since WFP eligibility occurs around the same age as pension eligibility for females, there is the potential for a discontinuity in retirement for females to confound the effect of the WFP on propensity to install renewables. (4) Additionally, the decision-making process concerning renewable technologies may not rest with the older members of the household if they rent their home (instead of owing) and/or contain more than two adults (e.g., older person living with their own children). To ensure that the results obtained from the RDD estimation identify the effect of the WFP, and not one of the issues discussed above, we restrict the sample to homeowners, living in a household composed by maximum two members in which the oldest is a male (whose pension eligibility age is 65). (5) Outside of these identification concerns, RDD estimates are sensitive to the choice of bandwidth size (the window on either side of the cutoff) and functional form. For this reason, we present estimates from various combinations of different bandwidths (6, 8 and 10) and functional forms (linear, quadratic and cubic) in a parametric estimation and different bandwidths (6, 3, 2) in a non-parametric estimation.
Results consistently find a negative effect of the WFP on the probability to install renewables. Parametric models with optimal functional forms, as established by Akaike Information Criterion (AIC), show that the probability of installing renewables decreases by 1.2 percentage points for WFP recipients (from a baseline of 1.6 %. This drop corresponds to about 69% of households substituting away from renewable investments after receiving the payment (i.e., more than half of households who would have installed renewables are "discouraged" to do so just after receiving the payment). (6) Given the universality of the WFP this is a considerable distortion, however the impact seems to dissipate as the oldest person gains in age. Non-parametric models and parametric models with those of age 60 removed, known as the "doughnut hole" model, find a negative and statistically significant drop in renewable installation also. Additionally, placebo WFP eligibility ages other than 60 do not find statistically significant changes in the propensity to install renewables. Other falsification tests show that the WFP has no effect on the probability to invest in one's home through other large items, such as remodeling their kitchen, and that extra income that is not labeled does not effect the probability of installing renewables. Finally, we show that adding loft insulation (the most common energy efficiency improvement) is not affected by the WFP and it is argued that this is due to the plethora of programs by local governments and utilities to encourage loft insulation installation.
This analysis is the first to estimate the indirect effects of a labeled cash transfer payment by looking at the potential distortionary effects on related goods. These indirect effects are especially important in energy issues, in which negative externalities are pervasive and policies that seems at first effective, may ultimately lead to socially inefficient outcomes. We find that households are nudged away from sources which generate relatively cleaner energy. In other words, a distortionary effect on the renewable market is not a necessary evil to achieve the public health objective of reducing cold-related mortality and morbidity. The label can be changed such that it promotes harmony between health and energy policy. The impact of the WFP label on renewable energy investment is particularly concerning given current UK energy policy. The UK Committee on Climate Change (CCC) has outlined ambitious goals for improved household energy efficiency and uptake of renewable energy technologies. The scenario envisioned by the CCC for the UK to meet their carbon budgets requires substantial savings from the building sector. As a result of...
To continue readingRequest your trial
COPYRIGHT GALE, Cengage Learning. All rights reserved.