New measures of the costs of unemployment: evidence from the subjective well-being of 3.3 million Americans.
A growing literature uses data on subjective well-being (SWB) to study macroeconomic determinants of life quality and relate them to policy discussions. Di Telia, MacCulloch, and Oswald (2001) use self-reported life satisfaction from the Euro-barometer surveys to estimate the unemployment-inflation tradeoff. Wolfers (2003) uses the same source of data to evaluate the cost of business cycle volatility. Di Telia, MacCulloch, and Oswald (2003) focus on European style welfare state policies. There is also an active literature on the social-norm effects of unemployment (Chadi 2014; Clark 2003; Clark, Knabe, and Ratzel 2010; Powdthavee 2007; Shields and Price 2005; Shields, Price, and Wooden 2009).
In this study, we focus on the indirect or spillover effects of unemployment on the SWB of U.S. residents, especially those who are still employed. Using two recent large surveys, we estimate the well-being costs of unemployment separately for different segments of the population, and decompose the total cost into monetary and nonmonetary costs of job losses, and the population-wide indirect effects. The indirect effects in the aggregate are found to be much larger than the direct effects. This suggests that more precise estimation and understanding of the indirect effects of unemployment are essential for any cost-benefit analysis of policies designed to mitigate the economic and social effects of unemployment.
The two recent surveys we use are the Gallup-Healthways Well-Being Index from 2008 to 2011 and the Centers for Disease Control and Prevention's Behavioral Risk Factor Surveillance System (BRFSS) from 2005, or in cases from the early 1990s to 2010. Both are large daily surveys, giving us a combined sample of more than 3 million U.S. respondents since 2005. The surveys include measures of SWB that cover both life evaluations and emotional reports. The surveys' fine-grained geographic identifiers allow us to relate variations in well-being to local labor-market conditions. These two surveys will add question variety and much sample size and richness to a literature in which U.S. studies were based mostly on the happiness question in the General Social Survey (GSS).
In addition to bringing in new survey data and finer-grain unemployment statistics, we experiment with a variety of identification strategies in order to provide more conclusive evidence and a better understanding on the spillover effects of unemployment. In the literature, Di Telia, MacCulloch, and Oswald (2001) and Wolfers (2003) find significantly population-wide negative effects using European and U.S. survey data. Clark (2003) and Mavridis (2010), focusing on the labor force, uncover no statistically significant effects from the British Household Panel Study surveys. In this study, we examine both the sample of employed workers and the wider population. More importantly, our analysis adopts a wide range of model specifications to make use of different sources of variations including those in official unemployment statistics, external industrial trends, unemployment by occupation, and workplace downsizing. These experiments not only help check robustness, but also shed light on the structure and dynamics of the spillover effects of unemployment. In particular, we find evidence that the anticipation of future increases in local unemployment has a negative impact on the population's well-being, and that job security is an important channel underneath the indirect effects.
The structure of the study is as follows. Section II reviews the literature. Section III describes the data and the estimation method. Section IV presents empirical findings. Section V concludes.
II. LITERATURE REVIEW
The literature on the macroeconomics of well-being can be traced back to the seminal paper by Easterlin (1974) showing that the rise of income in the United States since 1946 was not accompanied by an increase in its population's happiness. A more recent body of literature starts with Di Telia, MacCulloch, and Oswald (2001), which compares the costs of unemployment and inflation on happiness, using data from the EuroBarometer surveys. It is found that both unemployment and inflation reduce satisfaction but the coefficient on the unemployment rate is almost twice as large as the coefficient on the rate of inflation. Blanchflower (2007) also reports that the negative effect of unemployment is greater than that of inflation. Di Telia, MacCulloch, and Oswald (2003) expand the study to cover more macroeconomic factors, and report that both the level of and the changes in GDP have positive effects on life satisfaction. In addition, using the Euro-Barometer as the main data source, Wolfers (2003) extends the literature to include measures of economic volatility as explanatory variables, and finds that greater unemployment volatility lowers well-being.
There is an active literature on the social-norm effect of unemployment, whereby unemployed individuals may suffer less in areas where more people are unemployed. Clark 2003 reveals from British survey data that aggregate unemployment has a greater negative effect on employed workers than it does on unemployed workers, consistent with the social-norm hypothesis. Clark, Knabe, and Ratzel (2010) not only present consistent findings using the German Socio-Economic Panel, but also present evidence that the appropriate distinction may be between higher and lower levels of labor-market security, instead of between employment and unemployment. Powdthavee (2007) reports findings consistent with the social-norm effects using South African survey data. Shields and Price (2005) use data from the Health Survey for England and report that individuals who live in areas with high degrees of deprivation report lower levels of psychological well-being, with an inverse u-shape relationship. They also find evidence consistent with the social-norm hypothesis. Shields, Price, and Wooden (2009) find, from Australian survey data, a negative relationship between neighborhood deprivation and individual life satisfaction, although the unemployment rate does not stand out by itself. Chadi (2014) uses the German Socio-Economic Panel Study to study the interaction between individual and aggregate unemployment, and draws a conclusion that is very different from the literature, that being unemployed is more distressing in regions with higher unemployment rates.
The literature using U.S. data is more limited. Di Telia, MacCulloch, and Oswald (2001, 2003) and Wolfers (2003) use U.S. data from the GSS in addition to European data. The GSS has interviewed, on average, 1,500 respondents a year since 1972, and has a three-step happiness question "Taken all together, how would you say things are these days - would you say that you are very happy, pretty happy, or not too happy?" Di Telia, MacCulloch, and Oswald (2001) find that the average happiness in the United States is negatively correlated with year-to-year changes in inflation and in unemployment. Wolfers (2003) reports that state-level unemployment rate has a significantly negative effect.
Finally, we note a few examples that focus on the negative effects of individual unemployment. They are Winkelmann and Winkelmann (1998) and Kassenboehmer and Haisken-DeNew (2009). The focus of our study is the indirect effects of unemployment.
III. DATA AND THE ESTIMATION METHOD
A. Measures of Well-Being
Our first data source is the Centers for Disease Control and Prevention's BRFSS. The BRFSS is a state-based system of surveys collecting information on health risk behaviors, preventive health practices, and health care access. It collects information from more than 350,000 American adults (age 18 and over) a year in recent years. Starting from 2005, the BRFSS includes a question on life satisfaction: "In general, how satisfied are you with your life?" Respondents choose one of the following answers: very satisfied, satisfied, dissatisfied, or very dissatisfied. Oswald and Wu (2010) is a recent study that uses this measure of SWB. As of April 2012, the latest available year was 2010.
As highlighted in Kahneman and Deaton (2010), there are interesting differences between life evaluations (such as the life satisfaction described above) and reports of emotional experiences. To ensure that our study covers both aspects of well-being, we include an alternative measure from the BRFSS based on questions on mental health: "Now thinking about your mental health, which includes stress, depression, and problems with emotions, for how many days during the past 30 days was your mental health not good?" This question entered the BRFSS in 1993. We start the sample from 1994 in order to have a consistent set of income categories. About 4 million U.S. residents answered the question on mental health from 1994 to 2010.
Measured by the two indicators described above, the U.S. population is by and large happy. Overwhelmingly (93%), U.S. residents are satisfied or very satisfied with their lives; slightly more choose "satisfied" as opposed to the top category (49%-45%). Among the rest, 4.5% say that they are dissatisfied, only 1% choose "very dissatisfied." For the measure of mental health, most Americans (68%) say that they never have any days in the past 30 when mental health was not good.
The second survey that we use is the Gallup-Healthways Well-Being Index, a daily survey of U.S. residents that has interviewed about 1,000 adults every day since 2008. One of its primary measures of SWB is the Cantril Self-Anchoring Ladder (life ladder or ladder hereafter). It is the response to the following question: "Please imagine a ladder with steps numbered from zero at the bottom to ten at the top. Suppose we say that the top of the ladder represents the best possible life for you, and the bottom of the ladder represents the worst possible life for you. On which step of the ladder would you say you personally feel you stand at this time, assuming that the higher the step the better you feel about your life, and the lower the step the worse you feel about it? Which step comes closest to the way you feel?" The response thus has 11 levels from 0 to 10 in an ascending order, with higher values indicating better outcomes. From 2008 to 2011 (the latest available data as of April 2012), the Gallup-Healthways surveys provide a total of 1.4 million observations. Among them, 76% choose 6 or above (the middle rung is 5). The mode is 8 with a mass of 26%; 9 and 10 each accounts for about 9%. Among the rest, 14% choose 5 and 10% choose between 0 and 4.
For measures of emotional well-being, we use a set of questions in the Gallup-Healthways survey that ask about survey respondents' experiences the day before the interview. Examples include "Did you smile or laugh a lot yesterday?" and "Did you experience the following feelings during a lot of the day yesterday? How about worry?" We identify eight questions, primarily based on availability, that are evenly divided into positive and negative emotions. The four positive emotions are "smile or laugh a lot," "enjoyment," "happiness," and "learn or do something interesting." The four negative ones are "worry," "sadness," "stress," and "anger." From the answers to these questions, we derive a score of positive emotions and a score of negative emotions. Specifically, we count the number of "yes" answers to the first four questions to reach a positive score. The resulting score has five steps from 0 to 4. Between 2008 and 2011, 54% of the respondents report all four types of positive emotions and thus have the maximum score of four; 27% have a score of three; 14% have a score at two or one; only 4% report no positive experiences whatsoever. The negative score is constructed in the same manner based on the second set of emotions (worry, sadness, stress, and anger). About 50% of the respondents report zero negative experience; 20% have a score of one; 14% two; 9% three; leaving less than 5% reporting the maximum score of four negative emotions.
In addition to the two scores for emotions, we use the same set of emotional reports to construct an indicator for the dominance of negative emotions. It serves as a proxy for the U-index that was introduced in Kahneman and Krueger (2006), who raise concern about measuring life satisfaction with numerical scales, because "there is no guarantee that respondents use the scales comparably" (p. 18). Instead they proposed a U-index ("U" is for "unpleasant" or "undesirable") to measure the proportion of time an individual spends in an unpleasant state. The Gallup-Healthways survey does not allow a literal construction of the index, because it does not record minutes or hours associated with each mood or experience. Instead we construct a proxy by comparing the score of negative experiences to the score of positive ones. If the negative score is strictly greater than the positive one, we classify the respondent's day (before the interview) as an "unpleasant" one in the dichotomous manner advocated in Kahneman and Krueger (2006) and assign the value 1 to the index; otherwise the index is zero. In the survey, 11% of respondents have a pseudo u-index that is 1.
In total, our analysis makes use of six measures of SWB in two distinct categories. The life assessment category includes the 4-step life satisfaction and the 11 -step life ladder. The emotional experiences category includes the self-reported number of days when mental health is not good and the pseudo u-index that indicates the recent dominance of negative emotions over positive emotions. We also include the positive and negative components of the pseudo u-index as separate indices in the second category, as they may be differentially linked to unemployment and other factors influencing well-being.
B. Local and State-Level Statistics
Unemployment statistics at the county level come from the Local Area Unemployment Statistics program of the Bureau of Labor Statistics (BLS). There were 3,141 counties or equivalents in the 2000 Census; most are included in our data (more than 3,100 counties in Gallup-Healthways and more than 2,300 in BRFSS). There are other statistics that serve specific purposes, such as state-level unemployment rates, external industrial trends, and unemployment rates by occupation. We describe those data as they enter the analysis (Table S3 in the Supporting Information tabulates the sources).
C. The Relations among Unemployment Rates, Income, and Subjective Well-Being
This subsection presents simple data correlations before regression analysis in the next section. First we look at the cross-sectional relationship between average happiness and unemployment rates. Figure 1 shows the scatter plots between the two variables across states over the sample periods. When deriving average happiness, we exclude unemployed workers from the sample. The purpose is to focus on the indirect effect of unemployment on people who are not themselves unemployed. The relationships are negative in all cases, all with conventional statistical significance. The correlations are thus consistent with the hypothesized indirect effects of unemployment.
Next we look at variations over time by plotting the happiness trajectories according to local labor-market conditions. Specifically, we divide counties into quartiles according to changes in the unemployment rate from 2007 to 2010. Those in the top quartile are the hardest hit, those in the bottom quartile the least affected. We then compare the two groups (top and bottom quartiles) in the trajectories of their happiness measures. In addition, we exclude unemployed workers from the survey samples.
Figure 2 shows the plots. We note that the happiness measures are rather stable despite the severity of the recession. This is consistent with the findings reported in Deaton (2012). But national time series do not reflect the substantial differences at the local level. Here our hypothesis is that counties that were the hardest hit during the recession experienced larger declines in well-being relative to the least affected group. We also note that the average levels of the survey responses are affected by changes in survey design. Specific to the Gallup-Healthways surveys, Deaton (2012) reported that "Life evaluation questions are extremely sensitive to question order effects-asking political questions first reduces reported life evaluation by an amount that dwarfs the effects of even the worst of the crisis." The large jump in life ladder after the change in questionnaire also shows up in Figure 2 in the middle panel on the left-hand side. In our regression analysis, we use time dummies to remove the impact of conditioning effects caused by change in question order or in other aspects of the surveys.
Back to Figure 2 and the hypothesis that counties that were the hardest hit fared worse. The evidence is weak in BRFSS but stronger in the Gallup-Healthways surveys. In the BRFSS, the hardest-hit counties had lower life satisfaction even before the recession, starting from 2005 when the data became available. There is no obvious trend for the gap to narrow or widen thereafter. The BRFSS's other measure is the number of days with bad mental health. This measure began from the early 1990s. Here the evidence is also mixed. In the 1990s, the two groups were quite similar in this happiness measure. However, the two began to diverge around 2003. Since then there has been a statistically significant gap between the two groups in the expected direction; but the timing appears to be off.
There is stronger evidence in the Gallup-Healthways survey. For the Cantril ladder, the pseudo u-index and the index of positive emotions, there was little difference between the top and bottom groups in early 2008. A statistically significant gap, in the expected direction and consistent with the negative impact of unemployment, emerged in late 2008, at the onset of the financial crisis. The evidence is weaker for the index of negative emotions. According to this measure, the hardest-hit counties always fared worse than the better-off group even in early 2008. However, the difference widens somewhat during the economic crisis, consistent with the hypothesized impact of unemployment.
To summarize the observations from the correlations discussed earlier, both surveys, the BRFSS and the Gallup-Healthways, show strong cross-sectional correlations that are consistent with the negative impact of unemployment on the happiness of people who are not themselves unemployed. The evidence is less clear cut in terms of dynamic relationships. Consistent observations are made in the Gallup-Healthways data, but not in the BRFSS. Our later regression analysis will try to control for relevant factors at the individual and local levels.
Next we examine the relationship between SWB and income, which plays an important role in our analysis. We express the estimated effect of unemployment in terms of income equivalents, that is, the amount of monetary gains or losses that have the same effect on well-being as would a 1% higher unemployment rate. (1)
Figure 3 plots the income-happiness relationship. The top panels are for life satisfaction and mental health from the BRFSS, both plotted against the logarithm of household income. Life satisfaction increases steadily and linearly with log income over the entire range. The measure of mental health also rises with log income, but the relation is stronger at lower levels of income and weakens as income rises. This distinction between life evaluative measures and emotional well-being is similar to those reported by Kahneman and Deaton (2010) based on the Gallup-Healthways surveys. They found that while the life ladder has a positive and comparatively steady relation with log income, the relationship between emotional well-being and income flattens out after an annual household income of $75,000. In the BRFSS, there does not appear to be a satiation point even for the measure of mental health. Figure 3 also plots the four measures of well-being in the Gallup survey against log income. They are similar to those in Kahneman and Deaton (2010): life ladder has a positive relation with log income throughout, while emotional well-being increases little, if at all, at higher levels of income.
We note that there is uncertainty on whether the cross-sectional correlations between income and SWB truly reflect the effect of income on happiness. There are possibilities for both upward and downward biases. First, if SWB adapts to income changes over time, the effects of income would be stronger in the short run than in the long run. The cross-sectional correlations are more likely to pick up the weaker longer-term relationships, and thus to under-estimate the short-run impact. In the opposite direction, there may be omitted factors in the regressions that are common to both higher income and better happiness outcomes. This leads to over-estimation of the income effect. Despite the uncertainty, we use the income-equivalent representation for two reasons. First, the income equivalents provide a standardized representation of estimated effects across multiple measures of well-being that have different scales (0 or 1, 1 to 4, and 0 to 10). Without a standardized scale, we would not be able to evaluate whether the estimated effects are comparable across the six measures of well-being. Secondly, the positive relationship between income and SWB is one of the most robust findings in the happiness research and is relatively well-known (Kahneman and Deaton 2010). An income-equivalent presentation is thus a comparatively easy-to-understand choice of standardization. In addition, we hope that the income equivalents can provide some indication, crude and imperfect as it may be, of the economic importance of the spillover effects of unemployment.
Finally, we note that detailed summary statistics are in the Supporting Information.
D. Estimation Method
We employ a two-level regression approach for our analysis, using both individual and contextual information to predict individual well-being. The most important contextual variable is the county-level unemployment rate at the time of interview. The following equation describes the basic estimation, or Model-1.
[w.sub.(i,t),j] = [[alpha].sub.0] ln ([y.sub.(i,t)]) + ([X.sub.(i,t)])[[alpha].sub.1] + [[beta].sub.0][ur.sub.j,t] + [Z.sub.j,t][[beta].sub.1] + [D.sub.t][[beta].sub.2] + [u.sub.(i,t)].
The dependent variable [w.sub.(i,t),j] is the well being measure of worker i in county j who is interviewed at time t. In the subscript, we use a parenthesis to enclose i and t to highlight the fact that the surveys are not longitudinal. The time subscript t is in the unit of quarters.
The first variable on the right-hand side is the logarithm of household income, or ln(y([y.sub.(i,t)]).* (2) The vector [X.sub.(i,t)] has all other personal and demographic information including age categories, gender, marital status, educational attainment, race, and labor force status. The variable [ur.sub.j,t] is the unemployment rate in county j at time t. The vector [Z.sub.j,t] has other county-level information, including the log of average household income, the log of population density, the urbanization rate, the racial composition of each county's population, the percentage of owner-occupied housing (to measure the stability of population), and the longitude and latitude of the geographic centroid. It also includes dummy indicators for Alaska and Hawaii, so that the longitude and latitude variables reflect differences within the continental United States. Finally, we include a set of year-quarter dummies [D.sub.t] to capture time trends as well as possible framing effects resulting from changes in the survey questionnaires.
We estimate Model-1 with ordered Probit for all measures of well-being except for the number of days when mental health is not good, which is estimated linearly. All estimations use weights from the surveys and allow errors to cluster at the county level.
Besides the basic model, we employ a set of alternative specifications for robustness and experiments. We run a horse race between state-level employment rates and county-level statistics, and find the latter to have closer correlations with SWB. Other tests are listed below; we will provide more details as the analysis proceeds.
* Model-2 focuses on changes in unemployment rates instead of their levels.
* Model-3 uses fixed-effects models to remove unobserved local characteristics.
* Model-4 uses instrumental variables (IV) to remove unobserved local characteristics.
* Model-5 adds regional dummies to remove inter-regional correlations between happiness and unemployment rates.
* Model-6 uses occupational-specific unemployment rates to explore job security as a channel responsible for the indirect effects of unemployment.
We report estimates from the full sample and, separately, the sample of employed workers. All regressions on the full sample control for respondents' own unemployment status so that the local unemployment rate picks up the indirect effect. The BRFSS provides an indicator of unemployed persons in all years. The Gallup-Healthways poll, however, changed its questions on labor-market activity, starting from the second quarter of 2009, to match the unemployment definition in the Current Population Survey (CPS), creating a hurdle for identifying the unemployed in a consistent manner over time. We follow the CPS-based definition when the information is available. For the period before the second quarter of 2009, we define the unemployed as all those who are not working for pay, self-employed, full-time students, retired, home makers, or disabled. This approach likely undercounts unemployment, because some of the people who are studying or working at home will be classified as being unemployed under the CPS definition if they are actively looking for work. Indeed, the national unemployment rate in 2008 that we calculated in the Gallup-Healthways is only 3.8%, while it is 5.8% in the CPS. By contrast, the two sources generate almost identical estimates of the unemployment rate in 2010 (9.7% and 9.6%, respectively). For our interest in the indirect effect of unemployment, under-counting unemployment poses a challenge as it may inflate the estimated indirect effects. Fortunately, we can use our separate regressions that use only the sample of employed workers, thereby avoiding the need to identify unemployed individuals. The sub-sample regressions also highlight our interest in the indirect effects of unemployment.
IV. EMPIRICAL FINDINGS
A. The Indirect Effects of Unemployment
First the basic model, Table 1 presents estimates from the full sample and the sample of employed workers, in the top and bottom panels respectively. In all regressions, personal unemployment status is associated with lower well-being, while higher household income and higher levels of educational attainment are linked to higher well-being. Married couples are happier than the never-married singles, while the never-married are happier than the divorced, separated, or widowed. There is a robust U shape in age, with happiness falling as age rises before picking up again in later years. Men report lower life satisfaction and life ladder than females, while women tend to report higher scores of emotional experiences in both the positive and negative directions.
We use county-level unemployment (as a fraction of the labor force) to capture the indirect effects of unemployment. The coefficients on county-level unemployment are consistently negative across all of our measures of SWB. They are all statistically significant at the conventional levels (1% or 5%). Table 2 expresses the estimated indirect effects as monetary equivalents that are constructed as ratios of coefficients, with estimates from the basic model presented in the column labeled as Model-1. In our estimations, unemployment is in fractional terms, while household income is in logarithms. The ratio of the unemployment coefficient to that on log income is thus the monetary equivalent, in logs, for a fractional unit change in the unemployment variable. In turn, this means that the ratio of coefficients is the percentage income equivalent for a one percentage point change in the unemployment rate.
The first column of Table 2 shows the income equivalents from the basic model; they range from 2.7% to 4.2% in the full sample, and 2.7% to 6.3% in the sample of the employed. The point estimates from the employed sample tend to be higher than those from the full sample in five of the six measures of well-being, indicating that labor-market conditions may have weaker impacts on those outside the labor force, although in most cases the results are driven by a smaller estimated income effect among the working sample (i.e., a smaller denominator in the calculation of the equivalent income). There is no obvious pattern of differences between the evaluative measures (life satisfaction and life ladder) and the four measures of emotional reports. Take the life ladder and the pseudo u-index as an example: the estimates are similar and within two standard errors of each other. The average from the first four measures (leaving out the two components of the u-index) is 3.3% for the full sample and 3.8% for the sample of employed workers.
Our first robustness test is to repeat the regressions but with unemployment rates at the state level added as an extra variable, in addition to the county-level unemployment rate. This test serves two purposes. First, it presents a horse race between the two unemployment rates to see which one is more closely related to the measures of well-being. Secondly, it detects potential spillover effects from state-level unemployment beyond those at the local level. The results, presented in the Supporting Information as the first panels of Tables S7 and S8, suggest that county-level unemployment tends to have a tighter statistical relationship with SWB. In 9 out of the 12 regressions, the estimated coefficients on county-level unemployment are statistically significant, compared to 3 in the case of state-level unemployment. Importantly, however, we find that the state-level unemployment rates almost all have the same sign as the county-level unemployment rate. In most cases, other than the emotional reports in the Gallup-Healthways, they have roughly similar magnitude to those at the county level. Such findings are indicative of important spillover effects from statewide unemployment: it exerts a negative impact on well-being on a scale comparable to that of county-level unemployment.
The next model, labeled as Model-2 in Table 2, intends to shed light on the dynamic aspects of the indirect effects of unemployment. It does so by treating recent increases in unemployment rates separately from the levels. If the population has a tendency to adapt gradually to a higher level of unemployment, a recent increase in unemployment likely has a greater impact on well-being than the level per se. Specifically, we break time-t unemployment rate into a base component [ur.sub.j,t - 4] (where t - 4 is the same quarter last year), and a change component, [DELTA][ur.sub.j,t] = [ur.sub.j,t] - [ur.sub.j,t - 4]. The regressions then include both [DELTA][ur.sub.j,t] and [ur.sub.j,t - 4] on the right-hand side. Our interest is in the change component; the base serves as a control. Tables S7 and S8 present the estimated coefficients on both components, while the second column in Table 2 presents the income equivalents only for the change components [DELTA][ur.sub.j,t]. In all instances, recent increases in unemployment rates are negatively linked to well-being. The estimated effects are mostly statistically significant, except when life satisfaction in the BRFSS is the dependent variable, in which case it is the lagged unemployment rate that has strong statistical significance with the expected sign. In terms of magnitudes, the effects of recent changes tend to be greater than estimates based on the level of unemployment rates (the first column); but the confidence intervals overlap in all cases. The evidence for adaptation is thus indicative but not overwhelming.
The next two models both deal with possible concerns about unobserved local characteristics. The aforementioned findings are based on variations in unemployment rates and happiness at the county level. A concern is that there are unobserved local characteristics responsible for both unemployment and (un)happiness. We deal with this concern using both fixed-effects and instrumental-variables (IV) models. The fixed-effect model eliminates all cross-county variations, including any unobserved ones. The IV approach uses variations that are clearly driven by labor-market fluctuations.
The third column of Table 2, labeled as Model-3, has the income equivalents from the fixed-effects models. Tables S7 and S8 show the underlying estimates. (3) The income equivalents from the fixed-effects models are clearly weaker than those from the basic model. In the BRFSS, none of the estimates is significant. In the Gallup-Healthways survey, only four of the eight estimates are significant at conventional levels, with a fifth one having a 10% borderline significance. The size of the income equivalents are also smaller than those from previous models. Among the conventionally significant estimates, they range from 2.3% to 3.7%, just slightly over half of their counterparts from the basic model in column 1.
A fixed-effects model is not a costless way to handle unobserved factors. In a short time horizon, it has difficulty distinguishing stable local fixed effects from the well-being consequences of a persistent increase in unemployment that occurred before the sample period. Our samples indeed have short horizons: five of the six well-being measures have a sample period of either 4 or 6 years. Fixed-effects models are also vulnerable to what we call the anticipation effect, when future increases in unemployment are foreseen and such predictions reduce today's well-being (more discussion of this later).
An IV approach is another way to deal with the problem of omitted variables. We will instrument county-level unemployment rates with observable features in the labor market while leaving out the residuals including any unobserved components. Compared to the fixed-effects model, the IV approach allows us to keep using parts of the cross-county variation that are clearly driven by labor-market fluctuations. Specifically, we calculate the time series of likely employment losses for individual counties based on their shares of employment by industry and external statewide employment losses by industry. We then use the contemporaneous and lagged likely losses to instrument for local unemployment rates in standard two-stage regressions. This approach leaves out unobserved local characteristics except for those that are correlated with local compositions of industries or with statewide loss of employment by industries, the only two pieces of information used in the IV approach. (4)
We implement the IV approach using the industry classification at the level of 11 supersectors defined in the BLS. (5) The current likely loss and its 3-year lags, namely [LikelyLossRate.sub.j,s,t - k] for k = 0, 4, 8, 12, are then used as instruments in a two-stage least squares IV regressions. These likely employment losses are strong predictors of county-level unemployment rates: even without any co-variates, the likely losses explain 42% of the variation in the county-level unemployment rates since 2005.
We label the IV model as Model-4 in Table 2 for the income equivalents. Tables S10 and S11 have the underlying estimates, where 11 of the 12 indirect effects are significant at conventional levels. As income equivalents, they tend to be greater than those from the basic model, particularly so in the sample of employed workers. For example, local unemployment's impact on life satisfaction in the basic specification is 2.7% in the working samples. The corresponding IV estimate is 4.8%. In the case of life ladder, the estimated effect jumps from 4.4% to 8.2% with no overlap in the confidence intervals. The contrast is even greater between the IV estimates and the fixed-effects estimates in column Model-3. (6)
Why are the IV estimates greater than those estimates based on the actual unemployment rates? Two likely explanations are adaptation and anticipation. By adaptation, we mean that recent employment losses have greater impacts on well-being than the level, because people are able to adjust to higher unemployment; those who cannot adjust may choose to migrate. For those who remain on their jobs, the adaptation may in part reflect a lower threat of job losses compared to the period when unemployment is rising. Psychologists regard threat as a cause of "a drawn-out process of appraisal and reappraisal" leading to the feelings of helplessness and confusion (Lazarus and Folkman 1984, 92). A stabilization of unemployment, even if it is to a still-high level, can be beneficial for those who are still employed. This is a hypothesis that is supported in a previous robustness test, shown in Model-2, indicating that the negative SWB effects of recent rises in unemployment tend to be greater than those estimated based on contemporaneous levels of unemployment. Anticipation is about the threat of job losses in the immediate future. One of our later analyses shows that working in a company that is downsizing is more damaging to the emotional measures of SWB than is actually being unemployed. This is also consistent with psychologists' emphasis in the anticipatory aspect of threat, see chapter 2 of Lazarus (1966) for discussions and evidence. The instruments we use are predicted employment losses based on local industry shares and the wider area's employment trends. They very likely can predict a county's employment trends in the near future better than the current unemployment rate. In a panel of county-level unemployment rates, our instruments (namely [LikelyLossRate.sub.j,s,t - k] for k = 0, 4, 8, 12) predict the unemployment rises over the next year (namely, [ur.sub.j,s,t + 4] - [ur.sub.j,s,t]) with an r-squared of 23% in the 1990-2010 period (48% in the 2005-2010 period). The current employment rate's predictive power is much poorer with an r-squared that is 3% and 6% in those two periods, respectively. If the anticipation of future employment losses has a negative effect on well-being, we should expect the IV estimates to be greater than those from the basic model.
The anticipation effect can also explain why the estimates from the fixed-effects models are weak: if the anticipation effect is substantial, the fixed-effects estimates are misleading. Consider a sample with a short time horizon (a small number of years). Let's say that a county's unemployment rate rises substantially from the early years to the later years. The fixed-effect estimations will take an average of the two periods' unemployment rates and treat it as the county's long-term unemployment level. The earlier period is the "good" time because its unemployment rate is low; the latter period the "bad" time. But the anticipation effect concentrates more in the earlier period, when the residents can see that the bad times are coming based on industrial trends. The fear is in fact stronger in the "good" time. The anticipation effect thus weakens the relationship between current employment rates and well-being in the fixed-effect models. The larger is the anticipation effect, the weaker are the estimates from the fixed-effect models. Finally, we note that a strong anticipation effect is consistent with job security being an important channel behind the indirect effect of unemployment. When workers expect possible job losses in the future, the weakened sense of job security lowers well-being. After the employment losses actually take place, however, the still-employed workers may sense an improvement in their job security and thus experience a recovery in well-being. In the next subsection, we present analysis that focuses on the job security channel.
There is one more potential explanation for why county fixed-effect models return lower estimates: regional differences in happiness that are correlated with unemployment. The use of county-level dummies removes local effects and thus regional differences as well. What are the probable causes of regional differences? Some of those may have to do with differences in population density, urban shares, and race/ethnic composition. Our basic model already has those control variables, in addition to average income, longitude, latitude, and others. The model can thus handle systemic regional differences resulting from the observed factors. Problems arise, however, from unobserved or omitted factors. We can address this concern by adding regional dummy variables to the model, which will remove unemployment-happiness correlations between regions. Specifically, we use the Census Bureau's definition of regions: Northeast, Midwest, South, and West. There are substantial differences in unemployment rates among regions. In 2009, for example, the population-weighted unemployment rate was 8.4% in the Northeast, 9.7% in the Midwest, 8.9% in the South, and 10.1% in the West.
We report the test results in Model-5 of Table 2. Tables S7 and S8 show the underlying estimates. Adding regional dummies does indeed lower the estimates for the indirect effect of unemployment, indicating that the inter-region correlations between unemployment and happiness are steeper than those within regions. Almost all estimated income equivalents retain strong statistical significance, with only a few in the BRFSS dropping to the borderline significance of 10% confidence level. Quantitatively, the estimated income equivalents fall by an average of a quarter in the full sample, and a third in the sample of employed workers. The estimates now range from 1.2% to 4.4% with an average of 2.4% in the full sample and 2.8% in the sample of employed workers.
A more drastic way to control for regional differences would be to add state dummy indicators, thereby eliminating all inter-state correlations in well-being and unemployment. We decide against this approach for two reasons. First, inter-state variations account for a large proportion of the differences in unemployment rates. In 2009, 56% of the variation in the county-level unemployment rates is among states. Second, excluding inter-state differences ignores the potential spillover effects from statewide unemployment on happiness. One of our earlier tests provides indicative evidence of the statewide spillover effect. It uses regressions that include both county-level unemployment rate and state-level unemployment rate on the right-hand side as predictors. The state-level unemployment rate has an additional contribution to the dependent variables on top of those from county unemployment. In most cases, the state-level effects are comparable in magnitude to the county-level effects. Therefore, removing the state-level spillover effect is likely to weaken the link between unemployment and happiness. However, we should note that even with the state-level dummies, we still find statistically significant estimates in the Gallup-Healthways that ranges from 1.5% to 4.6% in terms of income equivalent, with an average of 1.9% in the full sample and 3% in the sample of employed workers. In the BRFSS, however, all estimates lose statistical significance and are quantitatively small (see Tables S9 and S10). We note that there are more counties in the Gallup-Healthways surveys than in the BRFSS (3100 counties versus 2300) and consequently, more variation across counties within states. Another possible explanation is that happiness-unemployment correlations over time are lower in the BRFSS than in the Gallup-Healthways, as shown in Figure 2.
B. The Job-Security Channel behind the Indirect Effect
Why does local unemployment, or the anticipation of its increase, reduce SWB? One likely channel is job security: rises in local unemployment lead to fear of losing one's own job. (7) However, there are other possibilities. Higher unemployment may lead to deterioration of social conditions such as more crime. (8) Rapid reshuffling in the labor market can be disruptive in the short run. Although it is not easy to disentangle the mixture, our reasoning suggests an important role for the job-security channel. In this section, we present further relevant observations regarding the job-security channel.
We recognize that the estimated effect of local unemployment is a mixture of the effects of lower job security (fear of job losses because of a bad local economy) and of other local channels (say social disruption). The local unemployment rate alone cannot tell them apart. We need variations that are related to job security but are not local in nature. The unemployment rate by occupation meets these criteria. It is linked to job security. It may have some correlations with local unemployment as a result of industrial clustering. In a manufacturing recession, for example, residents in manufacturing regions face higher local unemployment rates and on average higher levels of occupation-specific unemployment rates. However, we can deal with these correlations by controlling for local labor-market conditions. If the orthogonal part of the occupational rate affects well-being, we attribute that particular impact to job security instead of local factors.
Next, we conduct the exercise using the Gallup-Healthways survey, which classifies workers into 11 occupation categories. (9) The quarterly time series of occupation-specific unemployment rates are from the Current Population Surveys (CPS). We cannot do the same for the BRFSS, which does not identify occupations. We run the regressions only on the sample of employed workers, as unemployment by occupation is less relevant for people outside the labor force.
The last column in Table 2 shows the income equivalents for the occupational unemployment rate. Table S8 shows the underlying estimates. In all regressions, local unemployment has negative and significant effects on employed workers' well-being. The occupational rate has similar effects on the life ladder and the pseudo u-index, as well as the positive component of the u-index. For the negative emotions, however, a higher occupational rate leads to a reduction instead of an increase. However, this beneficial impact is small and is completely overwhelmed by the reductions in positive emotions. As a result, a higher occupational unemployment rate significantly increases the pseudo u-index, or the chance
that the positive emotions become dominated. In terms of income equivalents, as shown in Table 2, the occupation-based estimates are broadly comparable to county-based estimates. For the life ladder, the comparison is 4.8% for the occupational rate and 4.4% for the county rate in column 1. For the pseudo u-index, it is 4.1% to 4.7%. In both cases, the confidence intervals overlap.
There is broad similarity between the occupation-based estimates and the county-based estimates. The job security channel is present in both estimates. If our assumption that the occupational rates affect well-being only through the job security channel is right, the quantitative similarity would then suggest that job security is the major driver behind local unemployment rate's indirect effect on well-being.
Our final test attempts to provide direct evidence on the well-being impact of job security. It uses a Gallup-Healthways survey question: "Based on what you know or have seen, would you say that, in general, your company or employer is hiring new people and expanding the size, not changing the size of its workforce, and letting people go and reducing the size." In regressions that are otherwise identical to Model-1, we find that workforce downsizing, comparable to "not changing the size of its workforce," has substantial negative effect on employed workers' well-being. Workforce expansion, on the other hand, is associated with greater happiness. Table S11 shows the estimates. The coefficients on the dummy indicators of downsizing and expanding are statistically significant in all cases. In terms of magnitude, the negative effects of downsizing are close to or exceed the impact from lowering the log of income by an entire unit. These extremely large estimated impacts are probably resulting from the immediacy of the danger of job losses. For the three measures of emotional reports, the impact of downsizing is greater than that of actual unemployment (in terms of income equivalents compared to their counterparts found in the baseline regressions). The fear of job losses, thus, appears to be more detrimental than the actual status of unemployment. This is not true for life evaluations, for which the actual unemployment status hurts more than downsizing.
Overall, the two tests discussed earlier demonstrate that job security is an important channel for the spillover effect of unemployment. This is also consistent with the emphasis and finding in Clark, Knabe, and Ratzel (2010) that regional unemployment has stronger negative effects for those with higher levels of labor-market security, and weaker ones for those who already fare poorly in the labor market.
C. Accounting for the Well-Being Impacts of Unemployment
The analysis described earlier focuses on indirect effects, but unemployment also has direct effects on the unemployed. This section decomposes the influence of unemployment on the SWB of a population into three parts. The direct monetary cost is the loss of well-being resulting from income losses from unemployment. The direct nonpecuniary cost is the further loss of well-being suffered by those who become unemployed. The indirect cost is the loss suffered by those who are not unemployed themselves. We conduct the exercise using estimates from the basic specification (Model-1), which generates estimates that are either smaller or similar to the IV model. Adding regional dummies tends to reduce the indirect effects by about one-third. Flowever, estimates from other specifications, based on recent changes, IV and occupational unemployment rates, tend to generate higher estimates. The baseline estimates, thus, present a reasonable median case. The estimates for the direct effects, on the other hand, are relatively invariant to model specifications when expressed in proportion to the effect of income on well-being.
To evaluate the direct costs on the unemployed themselves, we first estimate the reduction in income associated with unemployment by regressing the log of household income on personal unemployment status, together with all the covariates in Table 1 that include gender, age, education, race, and others. In the BRFSS data, personal unemployment status has a coefficient of -0.36 in the income equation, indicating a 43% income difference between the unemployed and others with similar characteristics. In the Gallup-Healthways survey, the estimated difference is 0.34 in logs or 40%. These estimate are probably higher-end estimates of the income losses from unemployment, as there could be unobserved factors responsible for a lower level of income as well as a higher chance of being unemployed. This potential bias strengthens our later argument that monetary loss is a small part of the costs of unemployment. How does the lower income affect well-being? The log income's per-unit effects on well-being are shown in Table 1. For life satisfaction, the effect is 0.2. A 0.36 reduction in log income has a negative impact of 0.36 x 0.2 = 0.072. The well-being equation in Table 1 also shows the coefficient on personal unemployment status. Those coefficients measure the nonpecuniary effect of personal unemployment, as the income variable is already controlled for. For life satisfaction, the coefficient of personal unemployment is -0.4. Hence, the ratio of nonpecuniary to pecuniary effects from personal unemployment is 0.4/0.072 = 5.6. (10)
There is an extra pecuniary effect on other household members of the newly unemployed. A job loss reduces household income, and hence negatively affects all members of the household according to our model specification. The average household size is 2.58 according to the 2010 Census Briefs. So there is a 1.58 times additional monetary effect on SWB through the within-household spillover effects.
To evaluate the indirect cost at the population level, we use the estimated coefficients on local unemployment in Table 1. In the life satisfaction equation, local unemployment has a coefficient of -0.59. Hence, the indirect cost of a one percentage point increase in unemployment is 0.59 x 1 %. As the U.S. labor force participation rate is about 65%, a 1% increase in the unemployment rate moves 0.65% of the population from the employment pool to the unemployment pool. The direct well-being loss resulting from monetary losses, as calculated before, is 0.072 individually and, thus, 0.072 x 0.65 at the aggregate. The direct nonpecuniary effect is 0.4 individually and 0.4 x 0.65 at the aggregate. The ratio of the indirect loss to the direct pecuniary and nonpecuniary loss is therefore 0.59/(0.072 x 0.65 +0.4 x 0.65)= 1.9. Alternatively, we can express the indirect effect as a multiple of the direct pecuniary effects on the unemployed. In this case, the indirect effect is 0.59/(0.072 X 0.65) = 12.6 as big.
Table 3 fists the decomposition for all the measures of well-being. Its last two columns show the ratio of indirect effects to direct effects, and the ratio of total nonpecuniary effects to pecuniary effects. They show that the nonmonetary cost of unemployment is about six to nine times as large as the monetary costs, and that the indirect effects of unemployment at the population level are substantially greater than the direct loss suffered by the unemployed themselves. Individually, the indirect effects are small, but they affect a much broader population.
We can summarize the accounting exercise using the averages from the first four measures of well-being: if the direct monetary loss of the unemployed themselves is 1, the additional SWB loss of the unemployed is about 5, while at the population level the indirect or spillover effect is about 16 including the impact of monetary loss to other household members. All together, the total well-being costs of unemployment are about 20 times as large as those directly because of the lower incomes of the unemployed.
This paper estimates the impact of aggregate unemployment on SWB, using two recent large-scale American surveys, the Gallup-Healthways Well-Being Index from 2008 to 2011 and the Centers for Disease Control and Prevention's BRFSS in recent years and since the early 1990s in some cases. We contribute to a literature that is comparatively thin in U.S.-based evidence with larger samples, finer-grained contextual information, and alternative model specifications.
We find robust evidence that unemployment has significant indirect effects on the population, particularly so for those who are still employed. The basic specification correlates individual well-being with unemployment statistics at the county level. However, the evidence holds up well in a variety of models, including those that use alternative sources of variation based on external industrial trends and occupational unemployment rates. We find weaker estimates of the indirect effect from the fixed-effects model and higher estimates from instrumental variables based on external industrial trends that are strong predictors of unemployment losses in the near future. We interpret the latter result as indicating that the anticipation of future employment losses has negative effects on SWB. We also explore the channel behind the indirect effect, and find evidence indicative of the importance of job security, including substantial negative effects of workplace downsizing on the well-being of still-employed workers. Overall we find stronger, more clear-cut evidence from the Gallup-Healthways surveys than from the BRFSS. The former provides consistent evidence both cross-sectionally and over time. The BRFSS shows strong cross-sectional relationships, but much weaker relationships over time.
In the aggregate, the spillover effects are substantially greater than the direct well-being costs for the unemployed themselves. For those who are still employed, a one percentage point increase in local unemployment has an effect on SWB close to that of a 4% fall in household income. The income equivalents are calculated based on the estimated cross-sectional relationship between household income and SWB, which potentially suffer biases of uncertain directions. The estimated income equivalents serve only as a crude indication that the spillover effects of unemployment are economically meaningful.
BLS: Bureau of Labor Statistics BRFSS: Behavioral Risk Factor Surveillance System GSS: General Social Survey SWB: Subjective Well-Being
Blanchflower, D. G. "Is Unemployment More Costly Than Inflation?" NBER Working Papers 13505, National Bureau of Economic Research Inc., 2007.
Chadi, A. "Regional Unemployment and Norm-Induced Effects on Life Satisfaction." Empirical Economics, 46(3), 2014, 1111-41.
Clark, A. E. "Unemployment as a Social Norm: Psychological Evidence from Panel Data." Journal of Labor Economics, 21(2), 2003, 289-322.
Clark, A., A. Knabe, and S. Ratzel. "Boon or Bane? Others' Unemployment, Well-Being and Job Insecurity." Labour Economics, 17(1), 2010, 52-61.
Deaton, A. "The Financial Crisis and the Well-Being of Americans." Oxford Economic Papers, 64(1), 2012, 1-26.
Di Telia, R., R. J. MacCulloch, and A. J. Oswald. "Preferences over Inflation and Unemployment: Evidence from Surveys of Happiness." American Economic Review, 91(1), 2001, 335-41.
--. "The Macroeconomics of Happiness." Review of Economics and Statistics, 85(4), 2003, 809-27.
Easterlin, R. A. "Does Economic Growth Improve the Human Lot?," in Nations and Households in Economic Growth: Essays in Honor of Moses Abramovitz, Vol. 4, edited by P. A. David and M. W. Reder. New York: Academic Press, Inc., 1974.
Ferrer-i-Carbonell, A., and P. Frijters. "How Important Is Methodology for the Estimates of the Determinants of Happiness?" Economic Journal, 114(497), 2004, 641-59.
Helliwell, J. F., and H. Huang. "How's the Job? Well-Being and Social Capital in the Workplace." Industrial and Labor Relations Review, 63(2, article 2), 2009, 205-27.
Kahneman, D., and A. Deaton. "High Income Improves Evaluation of Life but Not Emotional Well-Being." PNAS, 107(38), 2010, 16489-93.
Kahneman, D., and A. B. Krueger. "Developments in the Measurement of Subjective Well-Being." Journal of Economic Perspectives, 20(1), 2006, 3-24.
Kassenboehmer, S. C., and J. P. Haisken-DeNew. "You're Fired! The Causal Negative Effect of Entry Unemployment on Life Satisfaction." Economic Journal, 119(536), 2009, 448-62.
Knabe, A., and S. Ratzel. "Quantifying the Psychological Costs of Unemployment: The Role of Permanent Income." Applied Economics, 43(21), 2011, 2751-63.
Lazarus, R. S. Psychological Stress and the Coping Process. New York: McGraw-Hill, 1966.
Lazarus, R. S., and S. Folkman Stress, Appraisal, and Coping. New York: Springer Pub. Co, 1984.
Luechinger, S., S. Meier, and A. Stutzer. "Why Does Unemployment Hurt the Employed? Evidence from the Life Satisfaction Gap between the Public and the Private Sector." Journal of Human Resources, 45(4), 2010, 998-1045.
Mavridis, D. "Can Subjective Well-Being Predict Unemployment Length?" Policy Research Working Paper Series No. 5293, The World Bank, 2010.
Oswald, A., and S. Wu. "Objective Confirmation of Subjective Measures of Human Well-Being: Evidence from the USA." Science, 327, 2010, 576-78.
Powdthavee, N. "Are There Geographical Variations in the Psychological Cost of Unemployment in South Africa?" Social Indicators Research, 80(3), 2007, 629-52.
Raphael, S., and R. Winter-Ember. "Identifying the Effect of Unemployment on Crime." Journal of Law and Economics, 44(1), 2001, 259-83.
Shields, M. A., and S. W. Price. "Exploring the Economic and Social Determinants of Psychological Well-Being and Perceived Social Support in England." Journal of the Royal Statistical Society Series A, 168(3), 2005, 513-37.
Shields, M., S. W. Price, and M. Wooden. "Life Satisfaction and the Economic and Social Characteristics of Neighbourhoods." Journal of Population Economics, 22(2), 2009,421-43.
Winkelmann, L., and R. Winkelmann. "Why Are the Unemployed So Unhappy? Evidence from Panel Data." Economica, 65(257), 1998, 1-15.
Wolfers, J. "Is Business Cycle Volatility Costly? Evidence from Surveys of Subjective Well-Being." International Finance, 6(1), 2003, 1-26.
Additional Supporting Information may be found in the online version of this article:
FIGURE S1. Measures of well-being from the BRFSS; 2005-2010 for life satisfaction; 1994-2010 for mental health; unweighted histograms.
FIGURE S2. Measures of well-being from the Gallup-Healthways Well-Being Index, 2008-2011; unweighted histograms.
TABLE S1. Data sources.
TABLE S2. Distribution of household income in the 2005-2010 BRFSS.
TABLE S3. Distribution of annual household income in the 1994-2010 BRFSS.
TABLE S4. Distribution of household income in the Gallup-Healthways survey 2008-2011.
TABLE SS. Summary statistics for other variables in the 1994-2010 BRFSS.
TABLE S6. Summary statistics for other variables in the 2008-2011 Gallup-Healthways Poll.
TABLE S7. Alternative model specifications-full samples.
TABLE S8. Alternative model specifications-samples of employed workers.
TABLE S9. Estimates from the full samples-linear regressions with state dummy indicators.
TABLE S10. Estimates from the samples of employed workers-linear regressions with state dummy indicators.
TABLE S11. Workplace expansion and downsizing--samples of employed workers.
(1.) Both BRFSS and Gallup-Healthways collect their income information as household income in categories. We turn the information into continuous values by estimating a monetary value for each category, assuming that the overall income distribution is lognormal. We do so for each individual year to allow the midpoint to grow over time. We then smooth the year-specific estimates using 3-year moving averages centering on the current year, before turning them into constant 2010 dollars using the Consumer Price Index.
(2.) We turn categorical income information into continuous values under the assumption that the income follows a lognormal distribution. We then assign the estimated mid-point value to each of the income categories. The midpoint estimate is likely to be less accurate for open-ended brackets, so we add to the regressions a dummy indicator for the top income category. We did not include a dummy indicator for the lowest income category, because the respondents in the bottom category are either few in number (in BRFSS) or removed before regressions (in Gallup-Healthways; more later on this). The top bracket presents a greater concern because it has a much larger concentration of survey respondents. The BRFSS's top bracket starts from S75,000 in annual terms and includes about a quarter of the respondents in recent years. The Gallup-Healthways survey's top bracket starts from 5120,000 in annual terms and includes about 10% of the respondents. Following Kahneman and Deaton (2010), we excluded the respondents in the Gallup-Healthways survey whose reported monthly incomes were lower than 5500, as such values were unlikely to be serious estimates of household income. The lowest BRFSS income bracket is 510,000 a year or below; it includes about 4% of the sample in recent years. We keep those observations in our analysis. Both surveys have nontrivial portions of respondents with missing income information (about 10%-20%). We included a dummy indicator for missing income in all regressions.
(3.) We switch to linear models in order to take advantage of STATA's built-in command to handle large dummy-variable sets (about 3000 counties in our case). The choice is also motivated by the incidental parameters problem that renders fixed effects probit models inconsistent. The choice of linear vs probit models makes little qualitative difference in SWB regressions, as documented in Ferrer-i-Carbonell and Frijters (2004). However, it does mean that the estimates in Tables S7 and S8 for the fixed effects models are not strictly comparable to those from the probit models. The income equivalents in Table 2, on the other hand, are comparable as they are expressed as income equivalents (or ratios of coefficients). Ratios of coefficients are robust to the choice of probit or linear models, because switching from one to the other tends to affect estimated coefficients proportionally (see Helliwell and Huang 2009).
(4.) Formally, the likely rate of employment losses from the same quarter last year (i.e., from t - 4 to t) in county j of state s is [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. The loss rate is expressed as a fraction. The denominator on the right-hand side is the total employment in the county j at t - 4, expressed as the number of employed workers N summed across industry x. The numerator is the likely employment losses, summed across industries. The likely losses are in turn the product of two factors: one is the proportional employment losses by state and industry (excluding the influence of county j); the other the number of workers by industry in county j in the base period. We note that the subscription - j in the numerator means that we exclude county j when calculating the state-wide industrial trend in ([N.sub.x,-j,s,t]/[N.sub.x,s,-j,t-4] - 1) to ensure that the state-wide industrial trend is strictly external to county j itself.
(5.) They are construction, education and health services, financial activities, information, leisure and hospitality, manufacturing, natural resources and mining, other services, professional and business services, trade, transportation, utilities, and the unclassified.
(6.) There is a difference in sample period between the IV model and other models; but that sample period is not responsible for the observed pattern of difference in estimates. The IV model (column Model-4) does not include 2011 data, because as of April 2012, the necessary quarterly census of employment and wages full-year data were not yet available. This is not a concern for the BRFSS data, since its sample period ended with 2010. For the Gallup-Healthways data, Model-1 and Model-3 are based on the 2008-2011 sample; the IV model uses the 2008-2010 sample. We can remove the difference in sample periods by running Model-1 and Model-3 using only the 2008-2010 data. From such regressions, the estimated income equivalents among employed workers are 4.3% in the case of life ladder, 4.5% for the u-index, 5.8% for the positive emotions and 3.4% for negative emotions. In the fixed-effects model, they are 1.6%, 2.2%, 3.8%, and 0.07%, respectively. Without exception, they are either smaller or similar to those from the 4-year estimates. As a result, they are all substantially smaller than the IV estimates, the pattern of difference that we focus on.
(7.) In a recent study, Luechinger, Meier, and Stutzer (2010) try to disentangle the effects by comparing the wellbeing of public-sector workers, who have greater job security, to that of private-sector workers. They find that the former are less affected by high levels of unemployment. Such findings are certainly consistent with the job-security channel. See also Clark, Knabe, and Ratzel (2010).
(8.) The evidence linking crime to unemployment is mixed. At least one study (Raphael and Winter-Ember 2001), using defense contracts and state-specific exposure to oil shocks to instrument for unemployment rates, found significantly positive effects of unemployment on property crime. On the other hand, the U.S. crime rates continued their downward trend during the Great Recession. The 2010 FBI Uniformed Crime Report indicates that property crimes in the United States fell from 9.8 million in 2008 to just above 9 million in 2010. Even though national figures cannot reveal local heterogeneities, it does raise questions about the importance of business cycles in crime trends.
(9.) The following is the list: 1. Professional worker--lawyer, doctor, scientist, engineer, nurse, accountant, computer programmer, architect, investment banker, stock brokerage, marketing, musician, artist; 2. Manager, executive, or official--in a business, government agency, or other organization; 3. Business owner; 4. Clerical or office worker in business, government agency, or other type of organization; 5. Sales worker 6. Service worker--policeman/ woman, fireman, waiter or waitress, maid, nurse's aide, attendant, etc; 7. Manufacturing or production worker; 8. Construction or mining worker; 9. Transportation worker;
(10.) Installation or repair worker; 11. Farming, fishing, or forestry worker.
(10.) Our approach does not distinguish between temporary and permanent effects of income changes from unemployment. Knabe and Ratzel 2011 suggest that not making such distinction leads to overestimating the nonpecuniary costs of unemployment by about one-third. But because the nonpecuniary costs in our data are on average several times as big as the monetary costs, adjusting the estimates downward by one third would not change the picture substantially.
JOHN F. HELLIWELL and HAIFANG HUANG *
* The research underlying this study is part of the "Social Interactions, Identity and Well-Being" program of the Canadian Institute for Advanced Research, and we gratefully acknowledge the intellectual and financial support thereby available to us. We are also grateful to the Gallup Organization for access to data from the Gallup-Healthways daily poll, and for helpful suggestions from George Akerlof, Andrew Oswald, and Rainer Winkelmann.
Helliwell: Senior Fellow, Canadian Institute for Advanced Research and Professor Emeritus of Economics, Vancouver School of Economics, University of British Columbia, Vancouver, BC V6T 1Z1, Canada. Phone 604-228-9534, Fax 604-822-5915, E-mail: firstname.lastname@example.org
Huang: Assistant Professor of Economics, Department of Economics, University of Alberta, Edmonton, Alberta T6G 2H4, Canada. Phone 780-248-1323, Fax 780-4923300, E-mail: email@example.com
TABLE 1 Estimates from the Full Samples (the Top Panel) and from the Samples of Employed Workers (the Lower Panel Showing Only Estimates of Interest) Life Days of Bad Satisfaction Mental Health Full Samples Log of household income 0.2 -.90 (0.008) *** (0.03) *** LFS: Unemployed -.40 2.38 (0.009) *** (0.06) *** Unemployment (fraction) in county -.59 2.45 (0.13) *** (0.78) *** Male -.05 -.96 (0.004) *** (0.02) *** Age 18 to 29 0.12 0.2 (0.007) *** (0.03) *** Age 50 to 64 0.06 -.72 (0.005) *** (0.03) *** Age 65 or above 0.23 -2.19 (0.007) *** (0.04) *** Edu: High school or below -.04 0.02 (0.006) *** (0.02) Edu: University degree 0.15 -.63 (0.004) *** (0.02) *** Married/with partner 0.3 -.27 (0.007) *** (0.03) *** Divorced/separated/widowed -.04 0.7 (0.007) *** (0.04) *** Log(average income in county) -.03 -.06 (0.02) * (0.09) Log(pop./sq. mile in county) -.02 0.09 (0.003) *** (0.02) *** Other variables: see footnotes Observations 1,939,405 3,310,113 [R.sup.2] 0.08 F statistic 501.37 422.2 Samples of Employed Workers Log of household income 0.22 -.74 (0.01) *** (0.04) *** Unemployment (fraction) in county -.60 2.57 (0.19) *** (0.9) *** Other variables: see footnotes Observations 863,833 1,604,097 [R.sup.2] 0.03 F statistic 295.4 204.1 Life Pseudo Ladder u-Index Full Samples Log of household income 0.24 -.27 (0.004) *** (0.004) *** LFS: Unemployed -.34 0.3 (0.007) *** (0.009) *** Unemployment (fraction) in county -1.01 0.91 (0.1) *** (0.11) *** Male -.17 -.06 (0.002) *** (0.004) *** Age 18 to 29 0.19 -.26 (0.005) *** (0.008) *** Age 50 to 64 0.02 -.04 (0.003) *** (0.005) *** Age 65 or above 0.35 -.57 (0.004) *** (0.009) *** Edu: High school or below -.01 0.08 (0.004) *** (0.005) *** Edu: University degree 0.16 -.10 (0.003) *** (0.006) *** Married/with partner 0.09 -.06 (0.004) *** (0.006) *** Divorced/separated/widowed -.11 0.14 (0.005) *** (0.007) *** Log(average income in county) -.01 -.007 (0.01) (0.02) Log(pop./sq. mile in county) -.02 0.03 (0.003) *** (0.004) *** Other variables: see footnotes Observations 1,283,025 1,267,079 [R.sup.2] F statistic 714.5 341.36 Samples of Employed Workers Log of household income 0.25 -.20 (0.005) *** (0.006) *** Unemployment (fraction) in county -1.07 0.91 (0.12) *** (0.15) *** Other variables: see footnotes Observations 640,286 634,859 [R.sup.2] F statistic 414.18 113.01 Positive Negative Emotions Emotions Full Samples Log of household income 0.17 -.22 (0.003) *** (0.003) *** LFS: Unemployed -.13 0.29 (0.007) *** (0.007) *** Unemployment (fraction) in county -.61 0.61 (0.08) *** (0.08) *** Male -.07 -.12 (0.003) *** (0.003) *** Age 18 to 29 0.24 -.07 (0.006) *** (0.005) *** Age 50 to 64 0.01 -.15 (0.004) *** (0.004) *** Age 65 or above 0.21 -.68 (0.005) *** (0.006) *** Edu: High school or below -.15 -.02 (0.004) *** (0.004) *** Edu: University degree 0.1 -.02 (0.004) *** (0.003) *** Married/with partner 0.07 -.002 (0.004) *** (0.004) Divorced/separated/widowed -.07 0.13 (0.005) *** (0.005) *** Log(average income in county) 0.02 0.05 (0.01) (0.01) *** Log(pop./sq. mile in county) -0.02 0.02 (0.003) *** (0.003) *** Other variables: see footnotes Observations 1,270,822 1,284,254 [R.sup.2] F statistic 344.83 612.69 Samples of Employed Workers Log of household income 0.1 -.16 (0.004) *** (0.004) *** Unemployment (fraction) in county -.66 0.58 (0.11) *** (0.1) *** Other variables: see footnotes Observations 636,078 640,257 [R.sup.2] F statistic 117.06 217.1 Notes: (1) Standard errors in parentheses. (2) Other variables: Not all estimates are shown in the table. All regressions, including those in the lower panel on the samples of employed workers, have on their right-hand side a set of year-quarter dummies, a set of race-ethnicity dummies, the indicator of top income bracket, the indicator for missing income information, county-level average household income, population density, share of urban population, of owner-occupied housing, of black residents, of Hispanic residents, and of other minorities, the longitude and latitude of county centers, and indicators for Alaska and Hawaii. (3) The second column uses survey linear regression; others use survey ordered probit. All use survey weights and cluster errors by county. *, **, and *** indicate statistical significance at 10%, 5%, and 1% levels, respectively. TABLE 2 The Estimated Income Equivalents, as Percents, of the Indirect Effects of Unemployment, with Standard Errors in Parentheses Measure of SWB Population Model-1 Model-2 Model-3 Life satisfaction All 3 1.2 -1.7 (.6) *** (1.1) (1.2) Employed 2.7 1.5 -1.8 (.8) *** (1.4) (1.4) Days of bad mental All 2.7 3.9 -.4 health (.9) *** (1.4) *** (1) Employed 3.5 4.1 .4 (1.3) *** (2.1) *** (1.4) Life ladder All 4.2 5.9 2.3 (.4) *** (.7) *** (.6) *** Employed 4.4 5.9 2.4 (.5) *** (.9) *** (.9) ** Pseudo u-index All 3.4 4.4 1.1 (.4) *** (.8) *** (.8) Employed 4.7 6.7 3.3 (.8) *** (1.7) *** (1.6) *** Positive emotions All 3.7 3.5 1.6 (.5) *** (.9) *** (.9) * Employed 6.3 6.8 3.7 (1) *** (1.9) *** (1.8) ** Negative emotions All 2.8 4.2 .3 (.4) *** (.7) *** (.7) Employed 3.6 5.9 .8 (.6) *** (1.3) *** (1.2) Measure of SWB Model-4 Model-5 Model-6 Life satisfaction 4.6 1.6 (1.2) *** (.6) ** 4.8 1.2 (1.7) *** (.75) * Days of bad mental 4.6 2.5 health (2) ** (D * 4.4 2.8 (2.8) (1.5) * Life ladder 7.9 3.2 (.9) *** (.4) *** 8.2 3.2 4.8 (1) *** (.5) *** (.3) *** Pseudo u-index 3.4 2.3 (.9) *** (.4) *** 6.6 2.9 4.1 (1.6) *** (.7) *** (.6) *** Positive emotions 2.2 2.9 (1.1) ** (.4) *** 5.6 4.4 11.2 (2.1) *** (.9) *** (.9) *** Negative emotions 3 2 (.8) *** (.3) *** 4.8 2.3 -1 (1.3) *** (.6) *** (.4) ** Notes: The table shows the estimated coefficients on the unemployment-rate variables of interest (in fractions) expressed in proportion to the estimated coefficients on logged household income. Standard errors are calculated using the Delta method. Model-1: estimates from the basic specification with county- level unemployment rates; Model-2: estimated effects of recent increases in county-level unemployment rates; Model-3: estimates from the fixed-effects model. Model-4: estimates from the IV model; Model-5: adding regional dummies to the regressions; Model-6: estimates based on occupation-specific unemployment rates; In Models -1,-3, -4, and -5, the unemployment variable of interest is the level of local unemployment rates (actual or instrumented). In Model-2, the unemployment variable is the change in local unemployment rate from same quarter last year. In Model-6, the unemployment variable is the occupation-specific unemployment rate. *, **, ***: Significance at 10%, 5%, and 1%, respectively. TABLE 3 Comparing the Direct and Indirect Costs of Unemployment--with Direct Cost to the Unemployed Workers Because of Monetary Losses Normalized to 1 Direct Costs Indirect Costs Monetary: Nonmonetary: Monetary Nonmonetary Same Population Loss Loss Household Wide Life satisfaction 1 5.6 1.58 12.6 Days of bad mental 1 7.3 1.58 11.6 health Life ladder 1 4.2 1.58 19 Pseudo u-index 1 3.3 1.58 15.2 Positive emotions 1 2.2 1.58 16.2 Negative emotions 1 3.9 1.58 12.5 Ratios Indirect to Nonmonetarv Direct to Monetary Life satisfaction 2.1 7.1 Days of bad mental 1.6 7.3 health Life ladder 4.0 9.0 Pseudo u-index 3.9 7.2 Positive emotions 5.6 7.1 Negative emotions 2.9 6.4
|Printer friendly Cite/link Email Feedback|
|Author:||Helliwell, John F.; Huang, Haifang|
|Date:||Oct 1, 2014|
|Previous Article:||Entry and exit, product variety, and the business cycle.|
|Next Article:||Switching regression estimates of the intergenerational persistence of consumption.|