[*PG1413]WHY HAPPINESS?: A COMMENTARY ON GRIFFITH’S PROGRESSIVE TAXATION
AND HAPPINESS
Abstract: This Commentary examines three issues raised in Professor Thomas D. Griffith’s Article on the connection between progressive taxation and subjective well-being, focusing on the selection of happiness as the measure of the gains of redistribution, the ability to measure happiness or subjective well-being, and the implications of using happiness analysis in determining tax policy. After arguing that the progressive taxation debate would benefit from further exploration of why happiness is the appropriate measure of success, this Commentary raises concerns about relying on self-reporting of subjective well-being and how happiness studies should be interpreted and can be improved. Finally, this Commentary notes that studies of income and happiness may inform tax policy design by helping to determine the appropriate balance between taxes and expenditures, outlining a role for the government in informing taxpayers’ perceptions of happiness, and focusing additional research necessary for an effective progressive taxation policy.
Professor Thomas D. Griffith’s Article begins with the enticing proposition of linking taxation and happiness—words not often connected in the taxation literature. Taxation and happiness become intertwined in the analysis of progressive taxation and the arguments favoring redistribution of income. The Article accepts the premise that some level of redistribution is desirable and seeks to answer the remaining but important question of how much redistribution is optimal. A key component of this inquiry concerns the determination of the welfare gains from redistribution.1 Measuring gains from redistribution has always been a daunting task because the inquiry is empirically difficult and requires the incorporation of other fields of study. Nonetheless, if our goal is to develop progressive tax policy more [*PG1414]thoroughly, we must attempt to define what constitutes utility or welfare and determine how it should be measured.
Reflecting the core elements in Professor Griffith’s analysis, my comments fall into the following three categories: (1) establishing the relationship between the progressive taxation analysis and the selection of happiness as the measure of gains from redistribution, (2) examining the existing happiness research and literature (outside taxation), and (3) discussing the implications of using the happiness analysis for tax policy.
Professor Griffith does not contend that happiness is the exclusive measure for evaluating taxpayers’ gains from additional income, but the reliance on this single measure does raise questions about how we should envision utility or welfare. The beginning of the Article suggests loosely (but certainly quite plausibly) that measures of happiness from redistribution would be valuable in assessing tax policies.2 Other parts of the Article, however, seem to equate happiness affirmatively with the measurement of utility in the welfare analysis. Happiness is a likely component of utility, but its selection requires us to consider what the term does and does not capture. How, and in what ways, is happiness a good proxy for measuring utility? How does it compare to the concept of well-being, which the Article identifies but puts aside at the outset? It may be productive to revisit the choice of happiness as the focal measurement and distinguish it from other terms and criteria that could contribute to utility. In some cases, it may be that terms such as happiness or well-being are conceived broadly or are intended to be effectively synonymous. Regardless, though, additional clarification as to why happiness was selected and what other measures might also be considered, as well as how they could be coordinated, would move the progressive taxation debate toward concrete tax policy choices.
As part of the process of measuring happiness for use in tax policy, Professor Griffith turns to the growing empirical and psychological literature on happiness. Before drawing upon this fascinating body of work, Professor Griffith first considers potential reporting problems in the measurement of happiness.3 In addition to those outlined in the Article, three others may be worth exploring. First, recent research indicates that people tend to overestimate the intensity of both good and bad future events (“impact bias”).4 When this holds true, it becomes relevant whether we are measuring happiness before or after events occur. The assessment of happiness may be different at each point. Further, we must consider which point would be more relevant. The “before” state may affect the party’s behavior vis-à-vis the event, but the “after” state may be experienced for a much longer time period. Moreover, in cases of regularly repeated events (such as the withholding of income taxes from a paycheck every month), how does impact bias operate? This is a critical question because progressive tax reform would initially take the form of a future event whose impact individuals presumably would overstate (either positively or negatively). Once in place, however, the reform would constitute the existing state of the world. Asking about impact bias in this manner highlights the distinction between happiness with regard to an anticipated amount of income and happiness with regard to an anticipated tax regime. To the extent we are seeking to measure utility (whether through measures of happiness with income or some other criteria), we face the question of how to incorporate the effect of the tax regime itself (rather than simply the individual’s end of the day income) on self-reported happiness or satisfaction. Impact bias can exist in both cases, but it need not be identical.
[*PG1416] Second, in terms of reporting accuracy, there may be a fundamental difference between asking about happiness (potentially more of a transitory concept) and satisfaction (potentially more of a big-picture concept). Generally, Professor Griffith uses the term “happiness,” but in some places, he seems to interchange it with “satisfaction,” without clarifying whether the terms are identical or, if not, how they are different. For example, “life satisfaction” seems to connote being in a good position or place, whereas “happiness” seems to connote more ephemeral feelings. Consider how many first year law students are “happy.” It might not be surprising if most students were “satisfied” with their situation in life while simultaneously declaring themselves “miserable.” To the extent these concepts are different, which should we seek to promote through the tax system?
The third reporting caveat derives from studies indicating that individuals perceive and tolerate pain differently depending on their perception of their ability to control the pain.5 For example, these studies indicate that when individuals have more direct control over the delivery of pain medication, they report less pain and anxiety and require less pain medication.6 These observations translate to the tax arena—if we find that taxpayers seem averse to progressive (or more progressive) taxation yet simultaneously purport to support redistribution in society (such as through charitable contributions) the apparent conflict may be due in part to the respondents’ perceptions of differences in control. One can cease voluntary charitable contributions (that promote redistribution) more readily than one can reduce one’s tax bill. This conclusion does not mean that taxes should be voluntary, but it can help explain potential gaps between reported views on redistribution versus progressive taxation.
Professor Griffith draws upon the following three categories of happiness studies to illuminate the redistribution question: cross-national comparisons, longitudinal studies, and individual assessments. Each category provides interesting insight into the relationship between income and happiness, although each also generates questions for how such insights ultimately should be interpreted.
The cross-national studies, which consider how countries rank in terms of happiness and income, treat each country effectively as an individual unit and then compare countries to see what conclusions can be drawn regarding income, happiness, and other possible factors. Although these studies are fascinating and useful, it would be valuable to consider how a country-based study differs from studies comparing actual individuals. A country is not an individual unit with a particular happiness and income level, but rather it is a measure drawn from all of the data points of happiness and income within that country (by median or mean). Thus, unique but critical features regarding the country may be disguised by the creation of the single data point for that country. Obvious possibilities include linguistic and cultural differences between and among countries that would influence the reporting within a country along a particular dimension and thereby be reflected in the final numbers for that country. Professor Griffith identifies a number of these concerns, but additional factors may also be influential. For example, in the European Union study covering the period from 1973 through 1998, residents of Denmark were found to be five times more likely to report that they were very satisfied with their lives than residents of France or Italy.7 Not only might this marked disparity reflect underlying cultural differences, it might also be due to the shape of the income distribution curve in the individual countries. That is, even if countries have relatively similar per capita income, the countries’ internal distributions of income may be quite different. To the extent that studies of individual assessments of happiness suggest that one’s happiness with a given level of income can be influenced by one’s relative income ranking in that society, the shape of a country’s income distribution curve may influence the happiness number emerging from that country.
[*PG1418] Another potential difficulty with relying on cross-national comparisons is the potential effect of “directional” or “trajectory” factors on the residents’ reported happiness. For example, in the World Values Survey discussed by Professor Griffith, Japan appears to be an interesting outlier.8 It has median income comparable to Denmark, Belgium, Iceland, and Norway, but reports significantly lower mean life satisfaction.9 Again, one possible explanation is a cultural reporting difference. Another independent possibility, however, is what might be termed a “directional” or “trajectory” explanation. Japan has experienced significant economic stagnation or decline for some period of time, and this reality may have shaped survey respondents’ views of their income situation. Having $X of income today may seem less attractive to someone who had $X+$Y of income several years ago. Moreover, if one’s view of the future is less optimistic, then a given amount of income today may generate less happiness than it would under a different economic forecast.10 This explanation could also account for the fact that residents of the former Soviet block seem more unhappy than their income would predict. If such residents perceive the stability and trajectory of their economy more negatively than residents of countries with comparable or even lower incomes, then residents of the former Soviet block may report lower happiness.
Finally, in an effort to eliminate potential factors influencing the income-happiness correlation, Professor Griffith cites studies arguing that there is almost no correlation between human rights and subjective well-being (after controlling for income). Depending on what the studies were measuring and comparing specifically, however, it is plausible that (1) where human rights violations are combined with low income, the respondents focus on subsistence needs (relating to arguments raised by Professor Griffith about the unique nature of money used for subsistence versus luxury goods), or (2) where human rights violations are significant, respondents fear reprisal from nondemocratic governments, and thus report differently in the studies.
None of these questions regarding cross-national studies negates their role or use. Instead, the questions seek to highlight additional [*PG1419]areas of comparison or investigation that could further improve our understanding of the factors contributing to happiness and their relationship to an individual’s income. For example, to the extent that a deeper examination of cross-country comparisons supports the hypothesis that the income distribution curve for a country, as well as specific levels of income, are relevant to happiness, we may draw different conclusions about the amount and the structure of progressivity that we would seek to pursue in our own income tax system.
The longitudinal studies cited by Professor Griffith of per capita income and happiness in the United States over the period from 1972 through 1998 indicate that as per capita income increased (that is, the country experienced some significant economic growth), reported happiness did not similarly increase.11 Initially, this result (reflected also in the cited studies of France and Japan) seems inconsistent with the cross-country comparisons that demonstrate a connection between per capita income and happiness. As Professor Griffith argues, however, the cross-country comparisons reveal not only a basic connection between per capita income and mean happiness, but also demonstrate that among poorer nations, increasing per capita income has a more significant impact on happiness than it does among wealthier nations. Thus, Professor Griffith concludes that the longitudinal studies of countries such as the United States, France, and Japan can be understood as further demonstrating that additional income is not very significant for happiness in developed countries, but is substantially more significant for poorer nations. While that conclusion seems consistent with the cross-national studies and with some of the studies regarding individual assessments of happiness, there exists another possible, yet unexplored, influence on the happiness numbers for the developed countries. To the extent the longitudinal studies cover a substantial period of time, it is possible that “symbol drift”— changes in what is being studied—has occurred. Assuming demographic shifts in the respondents in each country (that is, if a large group of “baby boomers” had come of survey age), the nature of survey responses for the country may have changed accordingly.
The final set of happiness studies, which examines individual happiness within a nation, offers another window into the effect of income on happiness. These studies eliminate many of the variables— cultural, linguistic, political, or other macro factors—at the national level that can plague cross-national comparisons. At the same time, [*PG1420]these studies allow us to consider other factors that influence perceptions of income. For example, cited evidence that the average happiness of a reporting cohort remains fairly stable over lifetimes (even though income grows significantly) can be viewed, as Professor Griffith suggests, as consistent with the longitudinal data that higher income levels contribute little to happiness, at least among wealthier nations. The cohort data may point, however, to an alternative conclusion—one linked to the aging of the reporting group in cohort studies. It may be that one’s happiness with a given level of income operates partly as a function of age and expectations for that age level. Thus, as income increases over one’s life, happiness does not increase commensurately because the additional income in fact represents an expected and perhaps needed financial boost.12 Such an observation regarding cohort studies does not reject the contention that additional income does not markedly increase happiness, or that income effects on happiness have a rivalrous or comparative aspect. Rather, it contends that age and expectations may impact the effect of income on happiness. Further, it maintains that studies involving aging or shifting populations (regardless of whether the aging or shifting population was in fact an intended feature of the study) should account for such effects in drawing conclusions about the nature and strength of the income-happiness relationship.13 Moreover, as we consider a range of potential progressive taxation policies, the impact of income on happiness at different ages may guide the structure and design of tax policy.
Another feature of the income-happiness relationship, drawn from the studies of individuals, concerns the rivalrous nature of income. Studies indicate that individuals show more rivalry with respect to income than leisure. Part of the explanation for this observation, however, may derive not from some inherent difference in competitiveness over income as compared to leisure, but rather from the fact that individuals seek power as an end. Income levels that place individuals higher up in their national income chart can increase the power of those individuals, which has its own independent value. Al[*PG1421]though absolute income has a concrete role (for example, being adequate to relieve subsistence needs), relative income can offer power and influence in society (for example, increasing political access). Procuring more leisure than one’s friends may not translate so readily into more power. If we ascertain that income is rivalrous partly because of its power implications, we have an additional independent policy interest in curbing the burgeoning income gap in the country. Where accumulations of income and wealth confer power, a democratic society may seek to limit such power by limiting the income and wealth accumulations of the upper bracket taxpayers.14
The study referenced by Professor Griffith of a unique group of individuals experiencing increased income—lottery winners—seems in line with indications that greater income does not produce corresponding increases in happiness.15 It also appears consistent with the distinct psychological concept of impact bias noted above—that individuals overestimate the intensity of positive or negative feelings from an event. The case of lottery winners, however, raises the issue that how one obtains money—not just the level of income—influences the amount of happiness generated. If third parties show a greater willingness to press lottery winners for financial assistance, as opposed to pressing other high income individuals, then the additional income of lottery winners comes with burdens not typically associated with earning money through business or labor. An interesting line of investigation would examine whether lottery winners demonstrate less happiness than other individuals of equal income who acquired their money through labor or investment activity.16
Finally, studies of individuals based on the misery index and the impact of unemployment and inflation should be interpreted to call for different policies for different respondent populations. One of the studies Professor Griffith refers to indicates that unemployment re[*PG1422]duces welfare more than inflation.17 This observation, on its own, might suggest particular policies and allocations of resources; Professor Griffith specifically offers the proposal that government programs aimed at aiding the unemployed (or even reducing unemployment) may be desirable for increasing happiness. The conclusions drawn from this study, however, probably should be restated to reflect the profiles of different sets of respondents. It may be that inflation harms senior citizens, welfare recipients, and the chronically unemployed more than unemployment. In these cases, income support or measures attacking inflation may generate more benefits than those targeting unemployment.
Returning to the ultimate questions driving Professor Griffith’s analysis—how much progressivity should we have in our tax system and how it should be designed?—the studies of income and happiness point in a number of valuable directions. First, in theory, we can have net redistribution without progressive taxation by working solely through the expenditure side.18 Certainly many factors affect the decision of how best to combine the tax side and the spending side, but does the happiness literature have implications for that balancing question? For example, does a non-progressive tax system generate less hostility for redistributive public spending? How might a taxpayer’s conceptions of (or reactions to) a progressive system constitute a “negative” on the happiness dimension? Recalling the psychology literature related to impact bias, if taxpayers believe that having progressive rates (that is, potentially higher rates) will be very unpleasant, how should this effect be factored in, even when the actual pain is less than anticipated due to impact bias and the declining marginal utility of money? Are we measuring happiness before or after the dollars are reallocated? Is this contrast muted by the annual nature of the progressive taxation “event”?
Second, is there a viable role for government to affirmatively shape and inform taxpayers’ calculations of happiness? Perhaps we should envision the government’s role in progressivity policy as more than decision making about where and how dollars should be col[*PG1423]lected and allocated. Maybe the role should include providing explanations of the policy based on knowledge about happiness, income, and cognitive biases. For example, if the reason that European countries evince more popular support for redistribution draws from their greater homogeneity, then perhaps a more explicit policy conversation about the psychological influence of homogeneity and heterogeneity can educate and shape popular American views on the subject. Similarly, a variety of studies indicate that increased levels of income, past some basic point, do not provide corresponding increases in happiness; in fact, higher taxes with money spent on certain public goods generate more happiness. Can this information be disseminated in such a way as to begin to shape individuals’ perceptions of what will, in fact, make them happy?19
Third, an important part of the redistribution debate concerns the ultimate calculus—how the gains from redistribution compare to the efficiency costs of higher taxes (such as a decline in work effort). Although focused on the former (calculating the gains), Professor Griffith’s Article does make the observation in the context of the cross-national and longitudinal studies that economic growth does not bring increased happiness. Essentially, the theory maintains that if economic growth is less valuable than we may have thought (because it does not really increase happiness), then redistributive policies that increase happiness through the reallocation of income, but with the side effect of decreasing economic growth, may not be such a serious concern. This position must be evaluated in both the short term and the long term. The net short-term result is plausibly the image described above—we achieve gains created from reallocating income at minimal cost to happiness.20 In the long term, however, the picture may be notably different. Even if economic growth does not have a linear relationship with increased happiness, failure to achieve certain levels of national economic growth could result in a country being overtaken (literally or figuratively) by other countries and left without the ability to recover. In this long-term picture, taxpayers would be quite unhappy when their deteriorating national situation began to provide them with fewer or starker choices, even if they did not fully appreciate economic growth when they had it.
The happiness literature overall seems to support the traditional marginal utility of income arguments espoused for progressive taxation. The analysis of this literature, however, also indicates that we may arrive at more nuanced conclusions about the impact of income on happiness and that these insights should be taken into account in moving from broad support for progressive taxation to advocacy of particular levels of progressivity and related regime design decisions. This initial investigation into the happiness literature yields a number of valuable observations for tax policy design, including the potential importance of the income distribution curve in a country and the potential variation among groups of taxpayers with regard to how income affects happiness. As psychological research into happiness develops, tax law can both draw upon that research for its own policy needs and help shape the research by more sharply defining the information needed to craft effective progressive tax policy.